text stringlengths 14 5.77M | meta dict | __index_level_0__ int64 0 9.97k ⌀ |
|---|---|---|
February 10, 2020 - 5:00 am EST 2 years ago
The Pin Sheet – February 10, 2020
Welcome to The Pin Sheet, a quick guide for your day in golf that pairs beautifully with our Clubhouse Newsletter.
Between this daily aggregation and the Newsletter, you'll find everything you need to know that's going on in the world of golf to be the most well-informed player in your foursome.
Without further ado, now on the tee…
Today's Clubhouse Newsletter
LPGA Cancels Asian Swing Due To Coronavirus
Due to the continued health concerns and recent advisories surrounding the coronavirus in some Asian countries, the LPGA and its partners have made the decision to cancel the 2020 Honda LPGA Thailand and the 2020 HSBC Women's World Championship in Singapore.
RELEASE ⬇️
— LPGA (@LPGA) February 10, 2020
The LPGA Tour announced on Sunday that due to ongoing health concerns in Asia as a result of coronavirus, they have canceled the Honda LPGA Thailand (Feb. 20-23) and the HSBC Women's World Championship in Singapore (Feb. 27-March 1).
These two events join the Blue Bay LPGA, which was scheduled to be played on China's Hainan Island March 3-5, as canceled events due to the outbreak.
"It is always a difficult decision to cancel events and the LPGA greatly appreciates the understanding and all the efforts made by our title sponsors (Honda and HSBC) as well as IMG to host incredible events for our players," the LPGA Tour's statement read. "The health and safety of our players, fans and everyone working on the event is always our highest priority. While we are disappointed that these tournaments will not take place this season, we look forward to returning to Asia soon."
These three tournament cancelations mean the LPGA will not return to Asia until the Olympics in August, and there will be an unexpected break following this week's Women's Australian Open. The Tour schedule will pick back up on March 19 at The Founders Cup in Arizona.
Rory No. 1, Brooks No. 2, Spieth Qualifies For Mexico, Phil Doesn't (Yet)
The new world rankings. As expected, Rory is now at the top and Jordan Spieth is back inside the top 50. And we have a big week ahead at Riviera. @McIlroyRory @JordanSpieth pic.twitter.com/o0ekrJYbtG
— Jay Coffin ⚰️ (@JayCoffinGC) February 10, 2020
It was a busy weekend for big names in the Official World Golf Rankings. As expected, Rory McIlroy supplanted Brooks Koepka as the game's top-ranked player, while Jordan Spieth's exceptional final round at the AT&T Pebble Beach Pro-Am did enough to get him back inside the top-50 in the world, squeezing in at No. 49.
Phil Mickelson required a T2 to slide back inside the top-50 in the world two weeks ahead of the first World Golf Championship of the year, but a solo third-place finish kept him on the outside looking in at No. 55.
Jason Day's 4th-place finish was enough to secure his ticket to Mexico, moving him up eight spots into 38th in the OWGR
Taylor Holds Off Star-Studded Chasers To Win At Pebble
The first Canadian to win the @ATTProAM. ??
Congratulations, @NTaylorGolf59! ? pic.twitter.com/YEGpkyms2A
— PGA TOUR (@PGATOUR) February 10, 2020
Nick Taylor won on the PGA Tour for the second time, holding off final-round playing partner Phil Mickelson and a litany of others who made ill-fated runs at the 31-year-old Canadian over a windswept Pebble Beach.
Taylor entered the final round with a one-shot lead over Mickelson, but a 4-under par front nine ballooned his lead to five heading for the incoming nine. Taylor stumbled a bit making back-to-back bogeys on Nos. 11 and 12 and then doubled the par-5 14th hole, but he was far from alone in having trouble navigating the 40-mile-per-hour gusts.
His lead never shrunk to less than two, and a chip-in birdie on the par-4 15th was all the cushion he needed.
"That was amazing," Taylor said. "I believed I could do it because I've done it before. But to do it in that fashion, playing with Phil, gives me a lot of confidence going forward."
"It's disappointing certainly to have not won, but I got outplayed," Mickelson said. "I mean, Nick played better than I did. He holed a couple of great shots. That eagle on 6, the putts he made on 4, 5 and 7 … he just really played some great golf."
Lee Makes It Three For The Family At The Vic Open
2014/2018: @minjeegolf ?
2020: @Minwoo27Lee ?
Brother and sister's names are now both on the #VicOpen trophy. pic.twitter.com/K79ms3s2p3
— The European Tour (@EuropeanTour) February 9, 2020
It's hard to imagine that a former U.S. Junior Amateur champion would be struggling to keep up with his big sister, but that's the dynamic that Min Woo Lee has found himself in. On Sunday, the 21-year-old Australian joined his sister, Minjee Lee (she won the U.S. Junior Amateur as well), by putting his name on the Vic Open trophy thanks to a two-shot win over Ryan Fox.
Minjee, the ninth-ranked player in the world, was fresh off of a 6th-place finish in the women's event, which was held on the same course as the European Tour's event, and watched on with her phone out taking pictures like a proud big sister. Minjee has won the Vic Open twice, first in 2014 and again in 2018.
"My sister and I winning the same tournament, it's pretty special," Min Woo told EuropeanTour.com. "I've got bragging rights now so it's even better."
"I was super, super proud of him," Minjee said. "It was really cool to just even be here with him and even watch him the last two holes. Just to see him play, I haven't really seen him play that much, so to be able to have a win here is really cool."
Join hundreds of thousands of golfers who started their day with the SwingU Clubhouse. Subscribe to the best newsletter in golf.
By subscribing, you accept the Terms of Service and Privacy Policy. You may unsubscribe at any time. | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 3,078 |
Intrusive Thoughts: What Causes Them?
Intrusive thoughts are something that almost everyone experiences. The main difference is the way that people deal with them. Stacey Colino's article on health.usnews.com explains intrusive thoughts as doubts, images, ideas, or impulses that are involuntarily drawn up. She continues to explain that these thoughts are usually based off of things that are important to us, and often involve stressful situations.
Click here to read the full article by Stacey Colino at health.usnews.com. | {
"redpajama_set_name": "RedPajamaC4"
} | 6,920 |
Q: Generate serial number in mysql
How can I generate serial number like in image in MySQL?
A: For all i know, out of the box mysql allows autoincrementing columns by 1.
maybe you can divide the number into two columns , use autoincrementation on the decimal part and update the other part manually
A: You can take a reference of the following MYSQL query and generate serial numbers
select @prev_value := TotalLoanAmt,
@row_num := IF(@prev_value=TotalLoanAmt, @row_num + 1,1) AS RowNumber,
TotalLoanAmt
FROM Loan, (select @row_num := 0) x, (select @prev_value := '') y
order by TotalLoanAmt
;
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 6,888 |
Q: Future.get cannot block the thread in forkjoinpool/workstealingpool I set the size of workstealingpool as 1. It seems that the future.get() doesn't block the thread.
@Test
public void runAsyncThenApplyExample() {
ExecutorService executor = Executors.newWorkStealingPool(1);
CompletableFuture cf = CompletableFuture.supplyAsync(
() -> {
//assertTrue(Thread.currentThread().isDaemon());
System.out.println("func: " + threadName());
Callable<Long> callable = () ->stub();
Future<Long> future = executor.submit(callable);
try {
future.get(); <<<<< **I think this should block the thread**
} catch (InterruptedException | ExecutionException e) {
e.printStackTrace();
}
return 1;
}, executor).thenAccept(s -> {
System.out.println("accept: " + threadName());
});
//assertFalse(cf.isDone());
System.out.println("main: " + threadName());
sleep(10);
assertTrue(cf.isDone());
}
private Long stub() {
System.out.println("stub: " + threadName());
return 1L;
}
private String threadName() {
return Thread.currentThread().getName();
}
output:
func: ForkJoinPool-1-worker-3
main: main
stub: ForkJoinPool-1-worker-3
accept: ForkJoinPool-1-worker-3
It seems that Future.get() and stub are using the same threead.
A: Executors.newWorkStealingPool(1); uses ForkJoinPool which has an undocumented feature called compensation threads.
From http://www.coopsoft.com/ar/CalamityArticle.html (emphasis mine):
Introduced with JDK1.8 is the CompletableFuture Class. To quote the JavaDoc:
"A Future that may be explicitly completed (setting its value and status), and may include dependent functions and actions that trigger upon its completion."
Not mentioned in the JavaDoc is that when using a large number of dependent functions with a get() method, the framework creates "compensation threads" to continue fetching application tasks from the deques and submission queue.
So when you do future.get(); it blocks but another thread is created in which the task is executed.
When running your code the output I'm getting is:
func: ForkJoinPool-1-worker-1
main: main
stub: ForkJoinPool-1-worker-0
accept: ForkJoinPool-1-worker-1
You haven't showed your threadName() method maybe there is a mistake in it and because of that you're seeing the same thread name (or you use different JVM which uses the same name in such case, check the thread ID)? If not please provide full code which outputs func and stub as the same threads.
A: Future.get() blocks, it means that the code following the Future.get() will not be executed until the Result is obtained. Even in your output you can see stub finishes execution first. Can you try putting a sleep of 5 mins or so in your stub before your return statement and see what happens?
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 7,372 |
CONSUMER_KEY = 'abc123...'#keep the quotes, replace this with your consumer key
CONSUMER_SECRET = 'abc123...'#keep the quotes, replace this with your consumer secret key
ACCESS_KEY = 'abc123...'#keep the quotes, replace this with your access token
ACCESS_SECRET = 'abc123...'#keep the quotes, replace this with your access token secret
| {
"redpajama_set_name": "RedPajamaGithub"
} | 5,873 |
<?php
namespace Molajo\Fieldhandler;
use CommonApi\Exception\UnexpectedValueException;
use CommonApi\Query\MessageInterface;
use stdClass;
/**
* Message Class
*
* @package Molajo
* @copyright 2014-2015 Amy Stephen. All rights reserved.
* @license http://www.opensource.org/licenses/mit-license.html MIT License
* @since 1.0.0
*/
class Message implements MessageInterface
{
/**
* Message Templates
*
* @var array
* @since 1.0.0
*/
protected $message_templates = array();
/**
* Parameter injected tokens
*
* @var array
* @since 1.0.0
*/
protected $parameter_injected_tokens = array();
/**
* Messages
*
* @var array
* @since 1.0.0
*/
public $messages = array();
/**
* Constructor
*
* @param array $message_templates
*
* @since 1.0.0
*/
public function __construct(
array $message_templates = array()
) {
if (count($message_templates) > 0) {
$this->message_templates = $message_templates;
}
$this->messages = array();
$this->parameter_injected_tokens = array();
}
/**
* Set Messages and inject tokens
*
* @param array $message_codes
* @param array $tokens
*
* @return $this
* @since 1.0.0
*/
public function setMessages(array $message_codes, array $tokens)
{
$this->setMessageTokens($tokens);
foreach ($message_codes as $code) {
$template = $this->getMessageTemplate($code);
$message = new stdClass();
$message->code = $code;
$message->message = strtr($template, $this->parameter_injected_tokens);
$this->messages[] = $message;
}
return $this;
}
/**
* Get Messages
*
* @return array
* @since 1.0.0
*/
public function getMessages()
{
return $this->messages;
}
/**
* Get Message Template
*
* @param string $code
*
* @return string
* @since 1.0.0
* @throws \CommonApi\Exception\UnexpectedValueException
*/
protected function getMessageTemplate($code)
{
if (isset($this->message_templates[$code])) {
return $this->message_templates[$code];
}
throw new UnexpectedValueException(
'Fieldhandler Message getMessageTemplate Method: Do not have template: ' . $code
);
}
/**
* Replace tokens in error messages
*
* @param array $tokens
*
* @return $this
* @since 1.0.0
*/
protected function setMessageTokens(array $tokens)
{
$this->parameter_injected_tokens = array();
foreach ($tokens as $token => $value) {
if (is_array($value)) {
$value = implode(',', $value);
} elseif (is_object($value)) {
$value = 'object';
}
$value = (string)$value;
$this->parameter_injected_tokens['{' . $token . '}'] = $value;
}
return $this;
}
}
| {
"redpajama_set_name": "RedPajamaGithub"
} | 1,691 |
BIO (for publicity)
Bob Whitesel (D.Min. and Ph.D. Fuller Theological Seminary) is a sought-after speaker, church health/growth consultant and award-winning writer/scholar on missional leadership, church change and church growth. He is the author of hundreds of articles and 14 books in two decades. He has been called by a national magazine: "the key spokesperson on change theory in the church today." His websites are Leadership.church and ChurchLeadership.university
General Speaking Bio:
Bob Whitesel (D.Min., Ph.D.) is a sought-after speaker, church health/growth consultant and award-winning writer on missional leadership, church change and church growth; who has been called by a national magazine: "the key spokesperson on change theory in the church today." He is the founding professor and former professor of missional leadership at Wesley Seminary at Indiana Wesleyan University. He holds two earned doctorates (D.Min. and Ph.D.) from Fuller Theological Seminary where the faculty awarded him "The Donald McGavran Award for Outstanding Scholarship in Church Growth." He is the author of 14 books in 20 years, including an award-winning series on evangelism titled, Spiritual Waypoints: Helping Other Navigate the Journey (2010). His book ORGANIX: Signs of Leadership in a Changing Church (2011) describes emerging leadership changes that churches must embrace to grow in the 21st Century. Cure for the Common Church: God's Plan to Restore Church Health (2012) has been hailed as "a much needed practical guide for setting a path for renewal and growth" that is "achievable, simple and wise in one stroke." The Healthy Church: Practical Ways to Strengthen a Church's Heart (2013) is the followup to Cure for the Common Church and includes field-tested exercises that will make a church resilient and healthy. In 2016, the watershed book ReMix: Transitioning Your Church to Living Color was released with co-author Mark DeYmaz. And in 2018 his popular book Enthusiast! Finding a Faith that Fills introduces readers to a 30-day devotional to rejuvenate their spiritual lives based on the principles of John Wesley and his "method." And Growing the Post-pandemic Church is "A Leadership.church Guide" to the practices that grow churches in difficult times. His websites are Leadership.church and ChurchLeadership.university
Speaking on Change Bio:
Bob Whitesel (D.Min., Ph.D.) is a sought-after speaker, church health/growth consultant and award-winning writer on missional leadership, church change and church growth; who has been called by a national magazine: "the key spokesperson on change theory in the church today." He is the founding professor and former professor of missional leadership at Wesley Seminary at Indiana Wesleyan University. He holds two earned doctorates (D.Min. and Ph.D.) from Fuller Theological Seminary where the faculty awarded him "The Donald McGavran Award for Outstanding Scholarship in Church Growth." He is the author of 13 books in just 17 years dealing with change and evangelism. He will dissect how to bring about change in a unifying manner. A national magazine called him, "the key spokesperson on change theory in the church today," and for 30+ years he has led a nationally-recognized ministry for coaching churches at Leadership.church and ChurchLeadership.university.
August 4, 2015 by bobwhitesel Categories: bio
← SPEAKER
National & International Tours for leaders & their families, including the popular WesleyTour.com → | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 6,711 |
\section{Introduction}
The concept of phase-space representation of quantum mechanics,
introduced in the pioneering works of Weyl~\cite{Weyl:1928},
Wigner~\cite{Wigner:1932}, and Moyal~\cite{Moyal:1949}, is a very
useful and enlightening approach that sheds light on the
correspondence between quantum and classical worlds.
Numerous applications of the phase-space methods to physical
problems have been extensively discussed in the last
decades~\cite{Lee:1995,Schroek:1996,Schleich:2001,QMPS:2005}.
However, much of this subject is usually illustrated in terms of
continuous variables, most often position and momentum.
For discrete systems, or qudits in the modern parlance of quantum
information, things are less straightforward. Since the dynamical
symmetry group for a qudit is SU($d$), one may be tempted to interpret
its associated phase space as a generalized Bloch
sphere~\cite{Kimura:2003,Schirmer:2004}, which is supported by the
rigorous construction of Kostant~\cite{Kostant:1970} and
Kirilov~\cite{Kirillov:1976} in terms of coadjoint orbits. Even if
this picture is quite popular, especially when applied to qubits, one
can rightly argue that there is a lot of information redundancy there
and that the phase space should be a grid of points, as one could
expect for a truly discrete system.
Indeed, apart from some noteworthy
exceptions~\cite{Hannay:1980,Leonhardt:1995,Miquel:2002}, nowadays
there is a wide consensus in picturing the phase space for a qudit as
a $d \times d$ grid. This can be traced back to the elegant approach
proposed by
Schwinger~\cite{Schwinger:1960a,Schwinger:1960b,Schwinger:1960c}, who
clearly recognized that the expansion of arbitrary operators in terms
of certain operator basis was the crucial mathematical concept in
setting such a grid. These ideas have been rediscovered and developed
further by several authors~\cite{Buot:1973,Cohendet:1988, Kasperkovitz:1994,Opatrny:1995,Rivas:1999,Mukunda:2004,Chaturvedi:2006},
although the contributions of
Wootters~\cite{Wootters:1986,Wootters:1989,Wootters:2006,Wootters:2007}
and Galetti and
coworkers~\cite{Galetti:1988,Galetti:1992,Galetti:1995ly} are worth
stressing.
To equip this grid with properties analogous to the geometry of an
ordinary plane, it turns out
essential~\cite{Klimov:2005,Klimov:2007,Klimov:2009bk} to label the
axes in terms of the elements of a finite field $\Gal{d}$ with $d$
elements. It is well known that such a field exist only when the
dimension is a prime or a power of a prime~\cite{Lidl:1986}. This, of
course, gives a special role to qudits in prime dimensions, but also
is ideally suited to deal with systems of $n$ of these qudits.
Once the natural arena is properly established, the next question is
how to represent states (and operators) on phase space. This is
done through quasidistribution functions, which allow for the
calculation of quantum averages in a way that exactly parallels
classical mechanics. There are, however, important differences with
respect to a classical description: they come from the noncommuting
nature of conjugate quantities (like position and momentum), which
precludes their simultaneous precise measurement and, therefore,
imposes a fundamental limit on the accuracy with which we can
determine a point in phase space. As a distinctive consequence of
this, there is no unique rule by which we can associate a classical
phase-space variable to a quantum operator. Therefore, depending on
the operator ordering, various quasidistributions can be defined. For
continuous variables, the best known are the
Glauber-Sudarshan $P$ function~\cite{Glauber:1963,Sudarshan:1963},
the Husimi $Q$ function~\cite{Husimi:1940}, and the
Wigner $W$ function~\cite{Hillery:1984}, corresponding to normal,
antinormal, and symmetric order, respectively, in the associated
characteristic functions. In fact, all of them are special cases
of the $s$-parametrized quasidistributions introduced by
Cahill and Glauber~\cite{Cahill:1969}.
The problem of generalizing these quasidistributions (mainly the
Wigner function) to finite systems has also a long history. Much of
the previous literature has focused on spin variables, trying to
represent spin states by continuous functions of angle variables.
This idea was initiated by Stratonovich~\cite{Stratonovich:1956},
Berezin~\cite{Berezin:1975}, and Agarwal~\cite{Agarwal:1981}. The
resulting Wigner function, naturally related to the SU(2) dynamical
group, has been further studied by a number of
authors~\cite{Scully:1983,Cohen:1986,Varilly:1989,Heiss:2000}, has
been applied to some problems in quantum
optics~\cite{Dowling:1994,Benedict:1999} and extended to more general
groups~\cite{Brif:1998}.
However, these Wigner functions are not defined in a discrete phase
space. A detailed review of possible solutions can be found in
Ref.~\cite{Bjork:2008}. Perhaps, the most popular one is due to
Wootters and
coworkers~\cite{Wootters:1987,Gibbons:2004,Wootters:2004}, which
imposes a structure by assigning a quantum state to each line in phase
space. Any good assignment of quantum states to lines is called a
``quantum net'' and can be used to define a discrete Wigner
function. In this paper, we show how to introduce a set of
$s$-parametrized functions, in close correspondence with the
continuous case. We emphasize that, although some interesting work has
been done in this direction by using a mod~$d$
invariance~\cite{Ruzzi:2005sd,Marchiolli:2005rm}, our approach works
quite well for many-qudit systems.
Another essential ingredient in any phase-space description is the
notion of coherent states~\cite{Klauder:1999}. This is firmly
established for continuous variables and can easily extended for other
dynamical symmetry groups~\cite{Perelomov:1986}. However, for discrete
systems we have again a big gap waiting to be filled. The reason for
this can be traced back to the fact that, as brightly pointed out in
Ref.~\cite{Ruzzi:2006}, in the continuum we have one, and only one,
harmonic oscillator, while in the discrete there are a lot of
candidates for that role, each one surely with its virtues, but surely
no undisputed champion.
The strategy we adopt to deal with this problem is to look for
eigenstates of the discrete Fourier transform~\cite{Galetti:1996}.
For continuous variables, they have a very distinguishable behavior that is
at the basis of the remarkable properties of coherent states. We explore this
approach, getting a strikingly simple family of discrete coherent states
(even for many qudits) fulfilling all the requirements.
To put the icing on the cake, we also extend the notion of squeezed
states for these systems~\cite{Marchiolli:2007}. For a single qudit,
the resulting states have nice and expected properties. However, when
they really appear as highly interesting is for multipartite systems,
since they present an intriguing relation with entanglement.
In short, the program developed in this paper can be seen as a handy
toolbox for the phase-space analysis of many-qudit systems, which
should be of interest to a large interdisciplinary community working
in these topics.
\section{Phase space for continuous variables}
\label{sec:qpWig}
In this Section we briefly recall the relevant structures needed to
set up a phase-space description of Cartesian quantum mechanics. This
will facilitate comparison with the discrete case later on. For
simplicity, we choose one degree of freedom only, so the associated
phase space is the plane $\mathbb{R}^2$.
The relevant observables are the Hermitian coordinate and momentum
operators $\op{q}$ and $\op{p}$, with canonical commutation relation
(with $\hbar =1$ throughout)
\begin{equation}
[ \op{q},\op{p} ]= i \, \op{\openone} \, ,
\label{eq:HWcom}
\end{equation}
so that they are the generators of the Heisenberg-Weyl
algebra. Ubiquitous and profound, this algebra has become the hallmark
of noncommutativity in quantum theory. To avoid technical problems
with the unbounded operators $\op{q}$ and $\op{p}$, it is convenient
to work with their unitary counterparts~\cite{Putnam:1967}
\begin{equation}
\op{U}(q) = \exp (-iq \,\op{p}) \, ,
\qquad
\op{V}(p)=\exp (ip\,\op{q}) \, ,
\label{UV}
\end{equation}
which generate translations in position and momentum, respectively.
The commutation relations are then expressed in the Weyl form
\begin{equation}
\op{V}(p) \op{U}(q) = e^{iqp} \, \op{U}(q) \op{V}(p) \, .
\label{eq:Weyl}
\end{equation}
Their infinitesimal form immediately gives (\ref{eq:HWcom}), but
(\ref{eq:Weyl}) is more useful in many instances.
In terms of $\op{U}$ and $\op{V}$ a displacement operator can be
introduced as
\begin{equation}
\label{eq:HWDisp1}
\op{D} (q,p) = e^{- i q p/2} \, \op{U} (p) \op{V}(q) \, ,
\end{equation}
which usually is presented in the entangled form $\op{D} (q,p) =
\exp[i(p \op{q} - q \op{p})]$. However, this cannot be done in more
general situations.
The $\op{D} (q, p)$ form a complete orthonormal set (in the trace
sense) in the space of operators acting on $\mathcal{H}$ (the Hilbert
space of square integrable functions on $\mathbb{R}$). The unitarity
imposes that $\op{D}^\dagger (q, p) = \op{D} (-q, - p)$, and
$\op{D}(0,0) = \op{\openone}$. In addition, they obey the simple
composition law
\begin{equation}
\label{eq:com}
\op{D} (\op{q}_{1}, \op{p}_{1}) \op{D} (\op{q}_{2}, \op{p}_{2}) =
e^{i (p_{1} q_{2} - q_{1} p_{2})/2} \,
\op{D} (\op{q}_{1} + \op{q}_{2}, \op{q}_{2} + \op{p}_{2}) \, .
\end{equation}
The displacement operators constitute a basic element for the
notion of coherent states. Indeed, if we choose a fixed normalized
reference state $ |\psi_{0}\rangle $, we can define these coherent
states as~\cite{Perelomov:1986}
\begin{equation}
| q, p \rangle = \op{D} ( q, p) \, | \psi_{0} \rangle \, ,
\label{eq:defCS}
\end{equation}
so they are parametrized by phase-space points. These states
have a number of remarkable properties, inherited from those of
$\op{D} (q, p)$. In particular, $\op{D} (q, p)$ transforms any coherent
state in another coherent state:
\begin{equation}
\op{D} ( \op{q}_{1}, \op{p}_{1} ) \, | q_{2}, p_{2} \rangle
= e^{i (p_{1} q_{2} - q_{1} p_{2})/2} \,
| q_{1} + q_{2}, q_{2} + p_{2} \rangle \, .
\label{eq:comcoh}
\end{equation}
We need to determine the fiducial vector $| \psi_{0}\rangle $. The
standard choice is to take it as the vacuum $|0\rangle $. This has
quite a number of relevant properties, but the one we want to stress
for what follows is that $| 0 \rangle $ is an eigenstate of the
Fourier transform (as they are all the Fock states)~\cite{Peres:1993},
and so is the Gaussian
\begin{equation}
\psi_{0} (q) = \langle q | 0 \rangle =
\frac{1}{\pi^{1/4}} \, \exp ( - q^{2}/2) \, ,
\label{eq:Gauss}
\end{equation}
in appropriate units. In addition, this wavefunction represents a
minimum uncertainty state, namely
\begin{equation}
( \Delta q )^{2} \, ( \Delta p)^{2} = \frac{1}{4} \,,
\label{eq:MUS}
\end{equation}
where $(\Delta q)^{2}$ and $(\Delta p)^{2}$ are the corresponding
variances.
Our next task is to map the density matrix $\op{\varrho}$ into a
function defined on $\mathbb{R}^{2}$. There exists a whole class of
these quasidistribution functions, related to different orderings of
$\op{q}$ and $\op{p}$. The corresponding mappings are generated by an
$s$-ordered kernel
\begin{equation}
W_{\op{\varrho}}^{(s)} ( q, p ) =
\mathop{\mathrm{Tr}}\nolimits [ \op{\varrho} \, \op{w}^{(s)} (q, p) ] \, ,
\label{eq:Wigcan}
\end{equation}
where $\op{w}^{(s)}$ is the double Fourier transform of the
displacement operator with a weight fixed by the operator ordering
\begin{eqnarray}
\op{w}^{(s)} (q , p) & = & \frac{1}{(2\pi)^{2}}
\int_{\mathbb{R}^{2}} \exp [-i (p q^{\prime} - q p^{\prime} ) ] \,
\op{D} ( q^{\prime}, p^{\prime} ) \nonumber \\
& \times &
\langle \psi_{0} | \op{D} (q^{\prime}, p^{\prime} ) | \psi_{0} \rangle^{-s}
\, dq^{\prime} dp^{\prime} \, ,
\label{eq:HWkernelDef}
\end{eqnarray}
and $s \in [-1 , 1]$. The mapping is invertible, so that
\begin{equation}
\op{\varrho} = \frac{1}{(2\pi)^{2}} \int_{\mathbb{R}^{2}}
\op{w}^{(-s)} (q , p) \, W^{(s)}(q,p) \, dq dp \, .
\end{equation}
The Hermitian kernels $\op{w}^{(s)} (q, p)$ are also a complete
trace-orthonormal set and they transform properly under displacements
\begin{equation}
\op{w}^{(s)} ( q, p ) =
\op{D} ( q, p) \, \op{w}^{(s)}(0,0) \,\op{D}^{\dagger} ( q, p) \, .
\label{eq:HWKernelDisp}
\end{equation}
The symmetric ordering ($s = 0$) corresponds to the Wigner function
$W (q, p)$ and the associated kernel $\op{w}^{(0)} (0,0)$ is just $ 2
\op{\mathcal{P}}$, where $\op{\mathcal{P}}$ is the parity operator.
For the antinormal ordering ($s=-1$), which corresponds to the Husimi
$Q$ function, $\op{w}^{(0)}(0,0) = | 0 \rangle \langle 0 |/\pi$.
The quasidistribution functions (\ref{eq:Wigcan}) fulfill all the
basic properties required for any good probabilistic description.
First, due to the Hermiticity of $\op{w}^{(s)}(q,p)$, they are real
for Hermitian operators. Second, on integrating $W^{(s)}(q,p)$ over
one variable, the probability distribution of the conjugate variable
is reproduced. And finally, $W^{(s)}(q,p)$ is covariant, which means
that for the displaced state $\op{\varrho}^{\prime} =
\op{D} (q_{0}, p_{0}) \, \op{\varrho} \, \op{D}^{\dagger} (q_{0}, p_{0})$,
one has
\begin{equation}
W_{\op{\varrho}^\prime}^{(s)} (q, p) =
W_{\op{\varrho}}^{(s)} (q - q_{0}, p - p_{0}) \, ,
\label{eq:HWProps3}
\end{equation}
so that these functions follow displacements rigidly without changing
their form, reflecting the fact that physics should not depend on a
certain choice of the origin.
\section{Single qudit}
\subsection{Discrete phase space}
We consider a system living in a Hilbert space $\mathcal{H}_{d}$,
whose dimension $d$ is assumed from now on to be a prime number.
We choose a computational basis $| \ell \rangle $ in $\mathcal{H}_{d}$
($\ell = 0, \ldots, d-1$) which we arbitrarily interpret as the
``position'' basis, with periodic boundary conditions $| \ell + d
\rangle = | \ell \rangle$. The conjugate ``momentum'' basis can
be introduced by means of the discrete Fourier
transform~\cite{Vourdas:2004}, that is
\begin{equation}
\label{eq:conjBas}
| \tilde{\ell} \rangle = \op{\mathcal{F}} \, | \ell \rangle \, ,
\end{equation}
where
\begin{equation}
\label{FT1}
\op{\mathcal{F}} = \frac{1}{\sqrt{d}}
\sum_{\ell , \ell^\prime = 0}^{d-1} \omega(\ell \, \ell^{\prime}) \,
|\ell \rangle \langle \ell^{\prime}| \, ,
\end{equation}
and we use the notation
\begin{equation}
\omega ( \ell ) = \omega^{\ell} = \exp (i 2\pi \ell/d) \, ,
\end{equation}
$\omega = \exp( i 2\pi/d)$ being a $d$th root of the unity.
Whenever we do not specify the ranges in a sum, we understand
the index running all its natural domain.
Once we have position and momentum basis, the phase space turns
out to be a periodic $d \times d$ grid of points; i.e., the torus
$\mathbb{Z}_{d} \times \mathbb{Z}_{d}$, where $\mathbb{Z}_{d}$
is the field of the integer numbers modulo $d$.
Mimicking the approach in Sec.~\ref{sec:qpWig}, we introduce the
operators $\op{U}$ and $\op{V}$, which generate finite translations
in position and momentum, respectively. In fact, $\op{U}$ generates
cyclic shifts in the position basis, while $ \op{V}$ is diagonal
\begin{equation}
\label{CC}
\op{U}^{n} | \ell \rangle = | \ell + n \rangle \, ,
\qquad
\op{V}^{m} | \ell \rangle = \omega(m \ell) \, | \ell \rangle \, ,
\end{equation}
where addition and multiplication must be understood
modulo~$d$. Conversely, $\op{U}$ is diagonal in the momentum basis and
$\op{V}$ acts as a shift, which is reflected also by the fact that
\begin{equation}
\label{FT2}
\op{V} = \mathcal{F} \, \op{U} \, \mathcal{F}^\dagger \, ,
\end{equation}
much in the spirit of the standard way of looking at complementary
variables in the infinite-dimensional Hilbert space: the position and
momentum eigenstates are Fourier transform one of the other. Note that
the operators $ \op{U}$ and $\op{V}$ are generalizations of the Pauli
matrices $\sigma_{x}$ and $\sigma_{z}$, so many authors use the
notation $\op{X}$ and $\op{Z}$ for them.
One can directly check the identity
\begin{equation}
\label{eq:Weyldis}
\op{V}^{m} \op{U}^{n} = \omega (m n) \, \op{U}^{n} \op{V}^{m} \, ,
\end{equation}
which is the finite-dimensional version of the Weyl form of the
commutation relations and show that they obey a generalized Clifford
algebra~\cite{Chuang:2000}.
One may be tempted to define discrete position and momentum
operators. A possible way to achieve this is to
write~\cite{Vourdas:2005}
\begin{equation}
\op{U} = \exp (- i 2 \pi \op{P} / d ) \, ,
\qquad
\op{V} = \exp ( i 2 \pi \op{Q} / d ) \, ,
\end{equation}
with
\begin{equation}
\op{Q} = \sum_{\ell} \ell \, | \ell \rangle \langle \ell | \, ,
\qquad
\op{P} = \sum_{\tilde{\ell}} \tilde{\ell} \,
| \tilde{\ell} \rangle \langle \tilde{\ell} | \, .
\end{equation}
However, for finite quantum systems the Heisenberg-Weyl group is
discrete, there is no Lie algebra (that is, there are no infinitesimal
displacements) and the role of position and momentum is limited. For
this reason our formalism is mainly based on the operators $\op{U}$
and $\op{V}$.
Next we introduce the displacement operators
\begin{equation}
\op{D}(m,n)= \phi (m,n) \, \op{U}^{n}\op{V}^{m} \, ,
\label{eq:desp}
\end{equation}
where $\phi (m,n)$ is a phase required to avoid plugging extra factors
when acting with $\op{D}$. The conditions of unitarity and periodicity
restrict the possible values of $\phi$. One relevant choice (for $d>2$)
that have been analyzed in the literature is~\cite{Bjork:2008}
\begin{equation}
\phi (m, n) = \omega( 2^{-1} \,m n) \, ,
\label{phi1}
\end{equation}
where $2^{-1}$ is the multiplicative inverse of $2$ in
$\mathbb{Z}_{d}$. For qubits, $\phi(m, n)$ may be taken as $ \phi
(m,n) = \pm i^{mn}$.
Without entering in technical details, this choice guarantees all the
good properties, in particular the analogous to Eq.~(\ref{eq:com}):
\begin{eqnarray}
\label{eq:comsq}
& \op{D}(m_{1},n_{1})\op{D}(m_{2},n_{2}) =
\omega [ 2^{-1}(m_{1}n_{2}-m_{2}n_{1})] & \nonumber \\
& \times \op{D}(m_{1}+m_{2},n_{1}+n_{2}) \, , &
\end{eqnarray}
and the following relation
\begin{equation}
\frac{1}{d} \sum_{m,n} \op{D}(m,n) = \op{\mathcal{P}} \, ,
\label{eq:parity}
\end{equation}
where $\op{\mathcal{P}}$ is the parity operator $\op{\mathcal{P}} |
\ell \rangle = | - \ell \rangle $ modulo $d$. Physically, this is the
basis for translational covariance and this also means that $\op{D}
(m, n)$ translates the standard basis states cyclically $m$ places in
one direction and $n$ places in the orthogonal one, as one would
expect from a displacement operator.
\subsection{Coherent states}
Once a proper displacement operator has been settled, the coherent
states for a single qudit can be defined as
\begin{equation}
\label{eq:defcoh}
| m, n \rangle = \op{D} (m, n) \, | \psi_{0} \rangle \, ,
\end{equation}
where $| \psi_{0} \rangle$ is again a reference state. These states
are also labeled by points of the discrete phase space, as it should be.
A possible choice~\cite{Saraceno:1990,Paz:2004} is to use for
$| \psi_{0} \rangle$ the ground state of the Harper Hamiltonian~\cite{Harper:1955}
\begin{equation}
\label{eq:Harper}
\op{H} = 2 - \frac{\op{U} + \op{U}^\dagger}{2} -
\frac{\op{V} + \op{V}^\dagger}{2} \, ,
\end{equation}
which is considered as the discrete counterpart of the harmonic
oscillator with the proper periodicity conditions. While
such a replacement is interesting, it is by no means unique.
We prefer to take a different route, pioneered by Galetti and
coworkers~\cite{Galetti:1996}. We use again as a guide
the analogy with the continuous case and look for eigenstates $ | f
\rangle $ of the discrete Fourier transform, which play the role of
Fock states for our problem and are determined by
\begin{equation}
\langle \ell | \op{\mathcal{F}} | f \rangle =
i^{\ell} \, \langle \ell | f \rangle \, .
\end{equation}
Obviously, the fact that $\op{\mathcal{F}}^{4} = \op{\openone}$
implies that it has four eigenvalues: 1, $-1$, $i$, and $-i$. The
solutions of this equation were fully studied by
Mehta~\cite{Mehta:1987} (see also Ruzzi~\cite{Ruzzi:2006}). Taking $|
\psi_{0} \rangle $ as the ``ground'' state (i.e., $\ell =0$) one gets
\begin{equation}
| \psi_{0} \rangle = \frac{1}{\sqrt{C}} \sum_{k \in \mathbb{Z}}
\sum_{\ell} \omega (k \ell ) \, e^{-\frac{\pi}{d}k^{2}} \,
| \ell \rangle \, ,
\label{0_cs}
\end{equation}
and the normalization constant $C$ is given by
\begin{equation}
C = \sum_{k \in \mathbb{Z}} e^{-\frac{2\pi}{d}k^{2}} =
\vartheta_{3} \left ( 0 \bigl | e^{-\frac{2\pi}{d}} \right ) \, ,
\end{equation}
$\vartheta _{3}$ being the third Jacobi function~\cite{Mumford:1983}.
Note in passing that this fiducial state can be alternatively
represented as
\begin{equation}
| \psi_{0} \rangle = \frac{1}{\sqrt{C}} \sum_{\ell}
\vartheta_{3} \left ( \frac{\pi \ell}{d} \bigl |
e^{- \frac{\pi}{d}} \right) | \ell \rangle \, .
\label{eq:theta3}
\end{equation}
The appearance of the Jacobi function in the present context
can be directly understood by realizing that this function
is a periodic eigenstate of the discrete Fourier
operator with eigenvalue $+1$ and period $d$. In addition, it plays
the role of the Gaussian for periodic variables, which makes this
approach even more appealing~\cite{Rehacek:2008}.
We also observe that $| \psi_{0} \rangle$ satisfies a ``parity''
condition: if we write it as $| \psi_{0} \rangle = \sum_{\ell}
c_{\ell} \, | \ell \rangle$, then $c_{\ell} = c_{- \ell}$. This
guarantees that the average values of $\op{U}$ and $\op{V}$ in
$| \psi_{0} \rangle$ are the same: $\langle \psi_{0} | \op{U} | \psi_{0}
\rangle = \langle \psi_{0} | \op{V} |\psi_{0} \rangle$.
The coherent states (\ref{eq:defcoh}) have properties fully analogous
to the standard ones for continuous variables, as one can check with
little effort.
The Harper Hamiltonian commutes with the Fourier operator
$[\op{\mathcal{F}},\op{H}]=0$. In fact, the state (\ref{0_cs}) is an
approximate eigenstate of (\ref{eq:Harper}) in the high-dimensional
limit
\begin{equation}
\op{H} | \psi_{0} \rangle \simeq \left ( \frac{\pi}{d} -
\frac{\pi^{2}}{2d^{2}} + \frac{\pi^{3}}{6d^{3}} \right)
| \psi_{0} \rangle \, ,
\qquad d \gg 1 \, ,
\label{HS_07}
\end{equation}
which provides another argument for its use as a reference.
Finally, according to the recent results in Refs.~\cite{Forbes:2003}
and \cite{Massar:2008}, the following uncertainty relation holds
\begin{equation}
\label{eq:DUDV}
(\Delta U)^{2} \, (\Delta V)^{2} \ge \frac{\pi^{2}}{d^{2}} \, ,
\end{equation}
where $(\Delta U)^{2} = 1 - | \langle \psi | \op{U} | \psi \rangle
|^{2}$ [and an analogous expression for $(\Delta V)^{2}$] denotes the
circular dispersion, which is the natural generalization of variance
for unitary operators. One can check that $| \psi_{0} \rangle$
saturates this inequality, confirming that it is also a minimum
uncertainty state.
\subsection{Quasidistribution functions}
The displacement operators lead us to introduce a Hermitian
$s$-ordered kernel
\begin{eqnarray}
\op{w}^{(s)} (m, n) & = & \frac{1}{d} \sum_{k,l} \omega (nk - ml) \,
\op{D}(m,n) \nonumber \\
& \times & \langle \psi_{0} | \op{D} (m, n) | \psi_{0} \rangle^{-s} \, ,
\label{w_s_d}
\end{eqnarray}
which, as $\op{w}^{(s)} (q, p)$ in Eq.~(\ref{eq:HWkernelDef}), appears
as a double Fourier transform of $\op{D}$ with a weight determined by the
operator ordering. However, here the parameter $s$ takes only discrete
values ($s = -1, 0, 1$).
These kernels are normalized and covariant under transformations of the
generalized Pauli group
\begin{equation}
\op{D} (m, n) \, \op{w}^{(s)} (k, l) \,\op{D}^{\dagger} (m, n) =
\op{w}^{(s)} (k + m, l + n) \, .
\label{eq:cov}
\end{equation}
They can be then conveniently represented as
\begin{equation}
\op{w}^{(s)} (m, n) = \op{D} (m, n) \,\op{w}^{(s)} (0,0) \,
\op{D}^{\dagger} (m, n) \, ,
\label{eq:par}
\end{equation}
where, according to Eq.~(\ref{eq:parity}), $\op{w}^{(s)} (0,0)$ coincides
with the parity for $s = 0$ ($d \neq 2$), as in the continuous case.
The $s$-ordered quasidistribution functions $W^{(s)}_{\op{\varrho}}$
are generated through the mapping
\begin{equation}
\label{eq:eds}
W^{(s)}_{\op{\varrho}} ( m, n) = \mathop{\mathrm{Tr}}\nolimits [\op{\varrho} \,
\op{w}^{(s)} (m, n) ] \, ,
\end{equation}
which is invertible, so that
\begin{equation}
\op{\varrho} = \frac{1}{d} \sum_{m,n} \op{w}^{-(s)} (m,n) \,
W^{(s)} (m, n) \, .
\end{equation}
These functions fulfill all the basic properties required for the
probabilistic description we are looking for. Let us apply them to the
reference state $|\psi_{0} \rangle$ (notice that any other coherent
state is just a displaced copy of this one). The corresponding Wigner
function ($s=0$) can be obtained after some algebra. We omit the
details and merely quote the final result:
\begin{eqnarray}
\label{wfcs_1}
W_{|\psi_{0} \rangle} ( m , n ) & = &
\frac{d}{C} \sum_{k} \sum_{p,q \in \mathbb{Z}}
\omega [( 2k - 1 - 2m ) n ] \nonumber \\
& \times & \exp [ - k + 2 m + q d - (d-1)/2]^{2} \nonumber \\
& \times & \exp [ -(\pi / d) ( k + p d - d/2 )^{2}] \, ,
\end{eqnarray}
which, in the limit $d\gg 1$, can be approximated by the compact
expression
\begin{eqnarray}
W^{(0)}_{|\psi_{0} \rangle} ( m , n ) & \simeq &
\frac{\sqrt{2}}{d^{3/2}} \sum_{k,l} (-1)^{kl} \,
\omega (mk-nl) \nonumber \\
& \times & \exp [ - \pi (k^{2}+l^{2})/(2d) ] \, .
\end{eqnarray}
\begin{figure}
\includegraphics[width=0.85\columnwidth]{Fig1.eps}
\caption{$Q$ function (\ref{eq:Qf0}) for the reference state $|
\psi_{0} \rangle$, which plays the role of vacuum for continuous
states.}
\label{Fig1}
\end{figure}
The $Q$ function ($s = -1$) for the same state $| \psi_{0}\rangle$
reduces to
\begin{equation}
Q_{|\psi_{0} \rangle} = | \langle \psi_{0} | \op{D}(m,n) | \psi_{0} \rangle |^{2} \, ,
\label{eq:Qf0}
\end{equation}
which exhibits the additional interesting symmetry
\begin{equation}
Q_{|\psi_{0} \rangle} (m, n) = Q_{|\psi_{0} \rangle} (- n, m ) \, .
\end{equation}
In Fig.~\ref{Fig1} we have plotted this $Q$ function for a
31-dimensional system. The aspect of the figure confirms the issues
one expects from a fairly localized Gaussian state.
\section{Many qudits}
\subsection{Discrete phase space}
Next, we consider a system of $n$ identical qudits, living in the
Hilbert space $\mathcal{H}_{d^{n}}$. Instead of natural numbers, it is
convenient to use elements of the finite field $\Gal{d^{n}}$ to label
states: in this way we can almost directly translate all the
properties studied before for a single qudit and we can endow the
phase-space with many of the geometrical properties of the ordinary
plane~\cite{Gibbons:2004}. In the Appendix we briefly summarize the
basic notions of finite fields needed to proceed.
We denote by $|\lambda \rangle$ [from here on, Greek letters will
label elements in the field $\Gal{d^n}$] an orthonormal basis
in the Hilbert space of the system. Operationally, the elements of the
basis can be labeled by powers of a primitive element using, for
instance, the polynomial or the normal basis.
The generators of the Pauli group act now as
\begin{equation}
\label{eq:XZgf}
\op{U}_{\nu} | \lambda \rangle =
| \lambda + \nu \rangle \, ,
\qquad
\op{V} _{\mu} | \lambda \rangle =
\chi ( \mu \lambda ) \, | \lambda \rangle \, ,
\end{equation}
where $\chi (\lambda )$ is an additive character (defined in the
Appendix) and the Weyl form of the commutation relations reads as
\begin{equation}
\op{V}_{\mu} \op{U}_{\nu} = \chi ( \mu \nu ) \,
\op{U}_{\nu} \op{V}_{\mu} \, .
\end{equation}
The finite Fourier transform~\cite{Vourdas:2007}
\begin{equation}
\label{FTcomp}
\op{\mathcal{F}} = \frac{1}{\sqrt{d^{n}}}
\sum_{\lambda, \lambda^\prime}
\chi (\lambda \, \lambda^\prime) \,
| \lambda \rangle \langle \lambda^\prime |
\end{equation}
allows us to introduce the conjugate basis $| \op{\lambda} \rangle =
\op{\mathcal{F}} | \lambda \rangle$ and also we have
\begin{equation}
\label{FT3}
\op{V}_{\mu} =
\op{\mathcal{F}} \, \op{U}_{\mu} \, \op{\mathcal{F}}^\dagger \, .
\end{equation}
In this way, the concepts delineated in the previous section can
be immediately generalized. For example, the displacement operators are
\begin{equation}
\op{D}(\mu ,\nu ) = \phi (\mu ,\nu ) \, \op{U}_{\nu} \op{V}_{\mu} \, ,
\label{eq:discom}
\end{equation}
where the phase $\phi (\mu ,\nu )$ must satisfy the conditions
\begin{equation}
\label{eq:condphi}
\phi ( \mu , \nu ) \, \phi^{\ast} (\mu ,\nu ) = 1 \, ,
\qquad
\phi ( \mu , \nu ) \, \phi ( - \mu , - \nu ) = \chi ( - \mu \nu ) \, ,
\end{equation}
to guarantee the unitarity and orthogonality of $\op{D}$. We also
impose $\phi (\mu ,0)=1$ and $\phi (0, \nu )=1$, which means that the
displacements along the ``position'' axis $\mu $ and the ``momentum''
axis $\nu $ are not associated with any phase.
For fields of odd characteristics one possible form of this phase is
\begin{equation}
\label{eq:phiodd}
\phi (\mu, \nu) = \chi (- 2^{-1} \, \mu \nu) \, ,
\end{equation}
and we have then the same composition law as in Eq.~(\ref{eq:comsq}),
namely
\begin{eqnarray}
\op{D} (\mu_{1}, \nu_{1}) \op{D} (\mu_{2}, \nu_{2}) & = &
\chi [ 2^{-1} (\mu_{1} \nu_{2} - \mu_{2} \nu_{1})] \nonumber \\
& \times & \op{D} (\mu_{1} + \mu_{2}, \nu_{1} + \nu_{2}) \, .
\end{eqnarray}
\subsection{Coherent states}
Given our previous discussion, it seems reasonable to extend the
coherent states~(\ref{eq:defcoh}) in the form
\begin{equation}
| \mu ,\nu \rangle = \op{D}(\mu , \nu ) \, |\Psi_{0} \rangle \, ,
\label{eq:defcoh_n}
\end{equation}
where $| \Psi_{0} \rangle$ is a reference state to be determined. In
the continuous case, the extension of coherent states (\ref{eq:defCS})
to many degrees of freedom is straightforward: they are simply
obtained by taking the direct product of single-mode coherent
states. To reinterpret (\ref{eq:defcoh_n}) in the same spirit, one
needs first to map the abstract Hilbert space $\mathcal{H}_{d^n}$,
where the $n$-qudit system lives, into $n$ single-qudit Hilbert spaces
$\mathcal{H} _{d} \otimes \cdots \otimes \mathcal{H}_{d}$. This is
achieved by expanding any field element in a convenient basis $\{
\theta_{j} \}$ ($j = 1, \ldots, n$), so that
\begin{equation}
\lambda = \sum_{j} \ell_{j} \, \theta_{j} \, ,
\end{equation}
where $\ell_{j} \in \mathbb{Z}_{d}$. Then, we can represent the states
as
\begin{equation}
| \lambda \rangle =
| \ell_{1} \rangle \otimes \cdots \otimes | \ell_{n} \rangle =
| \ell_{1}, \ldots, \ell_{n} \rangle \, ,
\end{equation}
and the coefficients $\ell_{j}$ play the role of quantum numbers for
each qudit.
The use of the selfdual basis is especially advantageous, since only
then the basic operators (and the Fourier operator) factorize in terms
of single-qudit analogues
\begin{equation}
\op{U}_{\nu} = \op{U}^{n_{1}} \otimes \cdots \otimes
\op{U}^{n_{n}} \, ,
\qquad
\op{V}_{\mu} = \op{V}^{m_{1}} \otimes \cdots \otimes
\op{V}^{m_{n}} \, ,
\end{equation}
and the displacement operators factorize accordingly
\begin{equation}
\op{D} (\mu, \nu) = \op{D}(m_{1}, n_{1}) \otimes \cdots \otimes
\op{D} (m_{n}, n_{n} ) \, ,
\end{equation}
where $m_{j}, n_{j} \in \mathbb{Z}_{d}$ are the coefficients of the
expansion of $\mu$ and $\nu$ in the basis, respectively. In consequence,
the eigenstates of the Fourier transform are direct product of
single-qudit eigenstates and we can write for the reference state
\begin{equation}
| \Psi_{0} \rangle = \bigotimes_{j=1}^{n} |\psi_{0j} \rangle \, ,
\label{0_csmult}
\end{equation}
where $|\psi_{0j}\rangle $ are of the form (\ref{0_cs}) for each qudit
(with $d > 2$). For qubits, we have~\cite{Munoz:2009}
\begin{equation}
| \Psi_{0} \rangle = \bigotimes_{j=1}^{n}
\frac{ ( | 0 \rangle + \xi | 1 \rangle )_{j}}
{( 1 + \xi^{2} )^{1/2}} \, ,
\label{Psi_n2}
\end{equation}
with $\xi =\sqrt{2} - 1$.
Unfortunately, the selfdual basis can be constructed only if either
$d$ is even or both $n$ and $d$ are odd. This means that for such a
simple system as two qutrits, this privileged basis does not exist.
Nevertheless, one can always find an almost selfdual basis and one
can proceed much along the same lines with minor modifications (the
interested reader can consult the comprehensive review~\cite{Bjork:2008}
for a full account of these methods).
It is interesting to stress that for $n$ qubits, the reference
state (\ref{Psi_n2}) can be elegantly written in terms of the field
elements $\Gal{2^{n}}$ as follows
\begin{equation}
| \Psi_{0} \rangle = \frac{1}{( 1 + \xi^{2} )^{n/2}}
\sum_{\alpha \in \Gal{2^{n}}} \xi^{h ( \alpha )} \, | \alpha \rangle \, ,
\label{gen_eq}
\end{equation}
where the function $h ( \alpha )$ counts the number of nonzero
coefficients $a_{j}$ in the expansion of $\alpha$ in the
basis.
The operator transforming from an arbitrary basis
$\{ \theta^{\prime}_{j} \}$ into the selfdual one $\{ \theta_{j} \}$
is given by
\begin{equation}
\op{\mathcal{T}} = \sum_{\mu \in GF(2^{n})}
| m_{1}, \ldots, m_{n} \rangle
\langle m_{1}^{\prime}, \ldots, m_{n}^{\prime} | \, ,
\label{T}
\end{equation}
where
\begin{equation}
\mu = \sum_{j} m_{j}^{\prime} \theta_{j}^{\prime} =
\sum_{j} m_{j} \theta_{j} \, .
\end{equation}
The operator $\op{\mathcal{T}} $ is always a permutation and plays
the role of an entangling (nonlocal) operator.
Let us examine the simple yet illustrative example of a two-qubit
coherent state. According to Eq.~(\ref{gen_eq}), we have
\begin{equation}
| \Psi_{0} \rangle = \frac{1}{1 + \xi^{2}}
( | 0 \rangle + \xi | \sigma \rangle + \xi | \sigma^{2} \rangle +
\xi^{2} | \sigma^{3} \rangle ) \, ,
\label{2q_cs}
\end{equation}
where $\sigma$ is a primitive element. The selfdual basis is
$\{ \sigma , \sigma^{2} \}$, and we have the representation
\begin{eqnarray}
|0 \rangle = |0 0 \rangle =
\left (
\begin{array}{c}
0 \\
0 \\
0 \\
1
\end{array}
\right ) \, ,
\qquad
| \sigma \rangle = |1 0 \rangle =
\left (
\begin{array}{c}
0 \\
0 \\
1 \\
0
\end{array}
\right ) \, , & \nonumber \\
\\
| \sigma^{2} \rangle = |0 1 \rangle =
\left (
\begin{array}{c}
0 \\
1 \\
0 \\
0
\end{array}
\right ) \, ,
\qquad
| \sigma^{3} \rangle = |1 1 \rangle =
\left (
\begin{array}{c}
1 \\
0 \\
0 \\
0
\end{array}
\right ) \, . \nonumber
\label{2q_basis}
\end{eqnarray}
In consequence,
\begin{eqnarray}
| \Psi_{0} \rangle & = &
\frac{1}{1 + \xi^{2}}
\left (
\begin{array}{c}
\xi ^{2} \\
\xi \\
\xi \\
1
\end{array}
\right ) \nonumber \\
& = &
\frac{1}{\sqrt{1 + \xi^{2}}}
\left (
\begin{array}{c}
\xi \\
1
\end{array}
\right )
\otimes
\frac{1}{\sqrt{1 + \xi^{2}}}
\left (
\begin{array}{c}
\xi \\
1
\end{array}
\right ) \, .
\end{eqnarray}
In a non selfdual basis, such as $\{ \sigma , \sigma^{3} \}$, we have
\begin{equation}
| 0 \rangle = |0 0 \rangle \, , \quad
| \sigma \rangle = | 1 0 \rangle \, , \quad
| \sigma^{3} \rangle = | 0 1\rangle \, , \quad
| \sigma^{2}\rangle = | 1 1 \rangle \, ,
\end{equation}
and
\begin{equation}
| \Psi_0 \rangle = \frac{1}{1 + \xi^{2}}
\left (
\begin{array}{c}
\xi \\
\xi^{2} \\
\xi \\
1
\end{array}
\right ) \, .
\end{equation}
The transition operator (\ref{T}) turns out to be
\begin{equation}
\op{\mathcal{T}} =
\left (
\begin{array}{cccc}
0 & 1 & 0 & 0 \\
1 & 0 & 0 & 0 \\
0 & 0 & 1 & 0 \\
0 & 0 & 0 & 1
\end{array}
\right ) \, ,
\end{equation}
which is nothing but a matrix representation of the CNOT operator.
\subsection{Quasidistribution functions}
The displacement operators (\ref{eq:phiodd}) immediately suggest to
introduce an $s$-ordered kernel
\begin{eqnarray}
\op{w}^{(s)} (\mu , \nu ) & = & \frac{1}{d^{n}}
\sum_{\lambda ,\kappa}\chi (\mu \lambda - \nu \kappa ) \, \op{D}(\mu ,\nu )
\nonumber \\
& \times & \langle \Psi_{0} | \op{D} (\mu, \nu ) | \Psi_{0} \rangle^{-s} \, ,
\end{eqnarray}
which, in view of Eq.~(\ref{FTcomp}), can also be interpreted as a
double Fourier transform of $\op{D}(\mu ,\nu )$. We can next introduce $s$-ordered quasidistribution functions through
\begin{equation}
W^{(s)}_{\op{\varrho}} (\mu , \nu )= \mathop{\mathrm{Tr}}\nolimits [ \op{\varrho}\,
\op{w}^{(s)} (\mu , \nu )] \, ,
\label{eq:Wigdn}
\end{equation}
and the inversion relation reads as
\begin{equation}
\op{\varrho} = \frac{1}{d^{n}} \sum_{\mu ,\nu}
\op{w}^{(-s)}(\mu ,\nu ) \, W^{(s)}_{\op{\varrho}}(\mu , \nu ) \, .
\end{equation}
Due to the factorization of the character in the selfdual basis,
the kernels $ \op{w}^{(s)} (\mu, \nu)$ factorize in this basis
\begin{equation}
\op{w}^{(s)} ( \mu ,\nu ) = \prod_{j} \op{w}^{(s)} ( m_{j}, n_{j}) \, ,
\end{equation}
and, consequently, also do the corresponding quasidistributions
\begin{equation}
W_{\op{\varrho}}^{(s)} ( \mu , \nu ) = \prod_{j}
W_{\op{\varrho}_{j}}^{(s)} ( m_{j}, n_{j} ) \, .
\end{equation}
For the particular case of the Wigner function, one can check that
\begin{equation}
\sum_{\mu ,\nu} W_{\op{\varrho}} (\mu ,\nu ) \,\delta_{\nu ,
\alpha \mu + \beta} = \sum_{\mu ,\nu} W_{\op{\varrho}}(\mu ,\nu ) \,
\delta_{\nu ,-\alpha^{-1}\mu
-\beta} \, ,
\end{equation}
that is, the sum over a line of slope $\alpha $ is the same as over a
line of slope $ - \alpha^{-1}$. The sum over the axes $\mu $\ and $\nu
$\ are thus equal
\begin{equation}
\sum_{\mu ,\nu} W_{\op{\varrho}} (\mu ,\nu ) \,\delta_{\nu ,0} =
\sum_{\mu , \nu} W_{\op{\varrho}} (\mu ,\nu ) \,\delta_{\mu ,0} \, .
\end{equation}
Note also, that the $Q$ function reduces to
\begin{equation}
\label{eq:2}
Q_{\op{\varrho}} (\mu , \nu ) =
\langle \mu ,\nu |\op{\varrho} | \mu ,\nu \rangle \, .
\end{equation}
In Fig.~\ref{Fig2} we have plotted this $Q$ function for the reference
state $|\Psi_{0} \rangle$ in a system of three qutrits. The
selfdual basis here is $\{ \sigma, \sigma^{3}, \sigma^{9} \}$
and the primitive element is a solution of the irreducible
polynomial $x^{3} + 2 x^{2} + 1 = 0$.
\begin{figure}
\includegraphics[width=0.85\columnwidth]{Fig2.eps}
\caption{$Q$ function for the reference state $| \Psi_{0} \rangle$,
for a system of three qutrits. The order in the axes is as
follows: $\sigma^{13}$, $\sigma^{17}$, $\sigma^{14}$, $\sigma$,
$\sigma^{2}$, $\sigma^{21}$, $\sigma^{23}$, $\sigma^{7}$, $\sigma^{15}$,
$\sigma^{4}$, $\sigma^{16}$, $\sigma^{6}$, $\sigma^{8}$, 0,
$\sigma^{9}$, $\sigma^{12}$, $\sigma^{25}$, $\sigma^{24}$,
$\sigma^{5}$, $\sigma^{3}$, $\sigma^{19}$, $\sigma^{11}$,
$\sigma^{22}$, $\sigma^{10}$, $\sigma^{20}$, $\sigma^{18}$,
$\sigma^{26}$, with $\sigma$ the primitive element.}
\label{Fig2}
\end{figure}
\section{Squeezed states}
Squeezed states constitute a simple nontrivial enlargement of the
notion of coherent states. In continuous variables, a squeezed state
is a minimum uncertainty state that my have less fluctuations in one
quadrature ($\op{q}$ or $\op{p}$) than a coherent state. They are
generated from the vacuum by using the unitary squeeze operator
\begin{equation}
\op{S} ( \mathfrak{s} ) = \exp [ - i \mathfrak{s} \,
(\op{q} \op{p} + \op{p} \op{q} ) ] \, ,
\end{equation}
with a subsequent displacement to an arbitrary point in the complex
plane
\begin{equation}
\label{eq:SSgen}
| q, p; \mathfrak{s} \rangle = \op{D} (q, p) \op{S} (\mathfrak{s} ) \,
| \psi_{0} \rangle \, .
\end{equation}
It is easy to check that
\begin{equation}
\label{eq:sqact}
\op{S} (\mathfrak{s}) \, \op{q} \, \op{S}^{\dagger} (\mathfrak{s} ) =
\op{q} \, e^{\mathfrak{s}} \, ,
\qquad
\op{S} (\mathfrak{s}) \, \op{p} \, \op{S}^{\dagger} (\mathfrak{s} ) =
\op{p} \, e^{- \mathfrak{s}} \, ,
\end{equation}
so that, the operator $\op{S} (\mathfrak{s})$ attenuates one
quadrature and amplifies the canonical one by the same factor
determined by the squeeze factor $\mathfrak{s}$, which, for
simplicity, we have taken as real. As a simple consequence of
(\ref{eq:sqact}) one can verify the transformations for $\op{U} (q)$
and $\op{V} (p)$:
\begin{equation}
\op{S} (\mathfrak{s} ) \, \op{U} (q) \, \op{S}^{\dagger} (\mathfrak{s} )
= U^{\mathfrak{s}} (q) \, ,
\quad
\op{S} (\mathfrak{s} ) \, \op{V} (p) \, \op{S}^{\dagger} (\mathfrak{s} )
= \op{V}^{- \mathfrak{s}} (p) \, .
\label{SUV}
\end{equation}
For a single qudit, squeezed states have been recently considered in
detail in Ref.~\cite{Marchiolli:2007}, using an extended
Cahill-Glauber formalism. Here, we prefer to follow an alternative
approach and define a squeeze operator as
\begin{equation}
\op{S}_{s} = \sum_{\ell} | \ell \rangle \langle s \ell | \, ,
\qquad
s \in \mathbb{Z}_{d} \, .
\label{S_d}
\end{equation}
At first sight, this can appear as a rather abstract choice. However,
notice that
\begin{equation}
\op{S}_{s}^{\dagger} \, \op{U}^{n} \, \op{S}^{\dagger}_{s} =
\op{U}^{n \, s} \, ,
\qquad
\op{S}_{s} \, \op{V}^{m} \, \op{S}_{s} = \op{V}^{m \, s^{-1}} \, ,
\label{UVS_d}
\end{equation}
which is a direct translation of the action (\ref{SUV}) to this
discrete case. This also means that in the squeezed ``vacuum''
\begin{equation}
\label{eq:sqvac}
| \psi_{0}; s \rangle = \op{S}_{s} | \psi_{0} \rangle \, ,
\end{equation}
the average values of some powers of the displacement operators are
the same
\begin{equation}
\langle \psi_{0}; s | \op{U} | \psi_{0}; s \rangle =
\langle \psi_{0}; s | \op{V}^{s^{2}} | \psi_{0}; s \rangle \, .
\label{psi_S_d}
\end{equation}
Perhaps, the clearest way to visualize this squeezing is to use a
quasidistribution, such as, e.g., the Wigner function. If
$\op{\varrho}_{s} = \op{S}_{s} \, \op{\varrho} \, \op{S}_{s}^{\dagger}$
denotes the density operator of a squeezed state, we have
\begin{equation}
W_{\op{\varrho}_{s}} ( m, n ) = W_{\op{\varrho}} ( s m, s^{-1} n ) \, ,
\label{W_S_d}
\end{equation}
whose geometrical interpretation is obvious and is the phase-space
counterpart of the property (\ref{eq:sqact}). For reasons
that will become evident soon, we refer to this as ``geometrical
squeezing''. We also note the following symmetry property of the
Wigner function
\begin{equation}
\sum_{m,n} W_{\op{\varrho}_{s}} ( m, n ) \, \delta_{n, 0} =
\sum_{m,n} W_{\op{\varrho}_{s^{-1}}} ( m, n ) \, \delta_{m,0} \, .
\end{equation}
For many qudits, our developed intuition suggests a direct translation
of (\ref{S_d}) in terms of the field elements in $\Gal{d^{n}}$, namely
\begin{equation}
\op{S}_{\varsigma} = \sum_{\lambda} |\lambda \rangle
\langle \varsigma \lambda | \, ,
\qquad
\varsigma \in \Gal{d^{n}} \, ,
\label{S_dn}
\end{equation}
in terms of which we can write relations similar to
Eqs.(\ref{UVS_d})-(\ref{W_S_d}). In fact, one can define a squeezed
``vacumm'' as in Eq.~(\ref{eq:sqvac}), i.e., $ | \Psi_{0}; \varsigma
\rangle = \op{S}_{\varsigma} | \Psi_{0} \rangle$. In Fig.~\ref{Fig3}
we plot the Wigner function for this squeezed state in a system of
three qutrits with $\varsigma = \sigma^{7}$.
\begin{figure}
\includegraphics[width=0.85\columnwidth]{Fig3.eps}
\caption{Wigner function for a squeezed ``vacuum'' state $|
\Psi_{0}, \varsigma \rangle$, for a system of three qutrits,
with the same order in the axes as in Fig.~2.}
\label{Fig3}
\end{figure}
Nevertheless, now the squeezing acquires a new physical perspective:
the squeeze operator (\ref{S_dn}) cannot be, in general, factorized
into a product of single qudit squeezing operators. This means that by
applying $\op{S}_{\varsigma}$ to a factorized state we generate
correlations between qudits; i.e., we create entangled states. The
most striking example is of course the $n$ qubit case, since there is
no single qubit squeezing.
To understand these correlations consider a general factorized state
\begin{equation}
| \Psi \rangle =
\sum_{\lambda} \! C_{\lambda} | \lambda \rangle =
\sum_{c_{\ell_{1}}, \ldots, c_{\ell_{n}}} \! \! \!
c_{\ell_{1}} \ldots c_{\ell_{n}} \, | \ell_{1}, \ldots, \ell_{n} \rangle \, , \label{psi_S}
\end{equation}
and apply (\ref{S_dn}). The resulting state turns out to be
\begin{equation}
\op{S}_{\varsigma} | \Psi \rangle = \sum_{\ell_{1}, \cdots, \ell_{n}}
C_{m_{1} \theta_{1} + \ldots + m_{n} \theta_{n}} \,
| \ell_{1}, \ldots, \ell_{n} \rangle \, ,
\end{equation}
where
\begin{eqnarray}
& \displaystyle
m_{i} = \sum_{j,k=0}^{d-1} f_{ijk} \ell_{j} h_{k} \, , & \nonumber \\
& & \\
& f_{ijk}= \mathop{\mathrm{tr}}\nolimits ( \theta_{i} \theta_{j} \theta_{k}) \, ,
\quad
h_{k}= \mathop{\mathrm{tr}}\nolimits (\varsigma^{-1} \theta_{k}) \, , & \nonumber
\end{eqnarray}
$\mathop{\mathrm{tr}}\nolimits$ (written in lower case) is the trace operation in the field
(see the Appendix) and $\{ \theta_{j} \}$ is the basis.
As a example let us consider a three-qubit system. Now the selfdual
basis is $\{ \theta_{1}= \sigma^{3}, \theta_{2} = \sigma^{5},
\theta_{3} = \sigma^{6}\}$, where $\sigma $ is a primitive element,
solution of the irreducible polynomial $x^{3} + x + 1 = 0$. The
result of applying $\op{S}_{\sigma^{k}}$ to the state (\ref{psi_S})
can be expressed in terms of
\begin{eqnarray}
\op{S}_{\sigma} | \Psi \rangle & = &
\sum_{\lambda \in \Gal{2^{3}}} \! \!
C_{\sigma^{6} \lambda} | \lambda \rangle =
\sum_{p, q, r \in \mathbb{Z}_{2}} c_{p+q} c_{p+r} c_{q}
|p, q, r \rangle \, ,
\nonumber \\
& & \\
\op{S}_{\sigma^{3}} | \Psi \rangle & = &
\sum_{\lambda \in \Gal{2^{3}}} \! \!
C_{\sigma^{4} \lambda} | \lambda \rangle =
\sum_{p, q, r \in \mathbb{Z}_{2}} c_{p+q+r} c_{p+r} c_{r}
|p, q, r \rangle \, . \nonumber
\end{eqnarray}
In fact, the transformations $\{ \op{S}_{\sigma^{5}},
\op{S}_{\sigma^{6}}\}$ generate the same entanglement (except
for permutations) as $\op{S}_{\sigma^{3}}$, while $\{ \op{S}_{\theta^{2}}, \op{S}_{\theta^{4}} \}$ generate the same entanglement (again except for permutations) as $\op{S}_{\sigma}$.
\section{Concluding remarks}
In summary, we have provided a handy toolbox for dealing with
many-qudit systems in phase space. The mathematical basis of our
approach is the use algebraic field extensions that produce results
in composite dimensions in a manner very close to the continuous
case.
Another major advantage of our theory relies on the use of the
finite Fourier transform and its eigenstates for the definition of
coherent states. We believe that this makes a clear connection with
the standard coherent states for continuous variables and constitutes
an elegant solution to this problem. The factorization properties of
the resulting coherent states in different bases is also an interesting
question.
We have also established a set of important results that have allowed
us to obtain discrete analogs of squeezed states. While for a single
qudit, these squeezed states have the properties one would expect from
our continuous-variable experience, for many qudits an amazing relation
with entanglement appears.
We think that the techniques presented here are more than a mere
academic curiosity, for they are immediately applicable to a variety
of experiments involving qudit systems.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 2,954 |
\section{Introduction}
Many parts of the brain are organized topographically. Famous examples are the ocular dominance maps and the orientation maps in V1. What is the advantage of such organization and what can we learn from it to develop better inductive biases for deep neural network architectures?
\begin{figure}[h!]
\centering
\includegraphics[width=1.0\linewidth]{figures/comm_diag_alt.png}
\caption{Overview of the Topographic VAE with shifting temporal coherence. The combined color/rotation transformation in input space $\tau_g$ becomes encoded as a $\mathrm{Roll}$ within the capsule dimension. The model is thus able decode unseen sequence elements by encoding a partial sequence and $\mathrm{Roll}$ing activations within the capsules. We see this resembles a commutative diagram.}
\label{fig:traversal}
\vspace{-4mm}
\end{figure}
One potential explanation for the emergence of topographic organization is provided by the principle of redundancy reduction \cite{barlow1961possible}. In the language of Information Theory, redundancy wastes channel capacity, and thus to represent information as efficiently as possible, the brain may strive to transform the input to a neural code where the activations are statistically maximally independent.
In the machine learning literature, this idea resulted in Independent Component Analysis (ICA) which linearly transforms the input to a new basis where the activities are independent and sparse \cite{SejnowskiICA, comon1994independent, hyvarinen2000independent, olshausen1997sparse}. It was soon realized that there are remaining higher order dependencies (such as correlation between absolute values) that can not be transformed away by a linear transformation. For example, along edges of an image, linear-ICA components (e.g. gabor filters) still activate in clusters even though the sign of their activity is unpredictable \cite{Portilla03, Wainwright99b}. This led to new algorithms that explicitly model these remaining dependencies through a topographic organization of feature activations \cite{hyvarinen2001topographic, Osindero2006, osindero2004contrastive, welling2003learning}.
Such topographic features were reminiscent of pinwheel structures observed in V1, encouraging multiple comparisons with topographic organization in the biological visual system \cite{hyvarinen2009natural, HYVARINEN20012413, ma2008overcomplete}.
A second, almost independent body of literature developed the idea of ``equivariance'' of neural network feature maps under symmetry transformations. The idea of equivariance is that symmetry transformations define equivalence classes as the orbits of their transformations, and we wish to maintain this structure in the deeper layers of a neural network. For instance, for images, asserting a rotated image contains the same object for all rotations, the transformation of rotation then defines an orbit where the elements of that orbit can be interpreted as pose or angular orientation. When an image is processed by a neural network, we want features at different orientations to be able to be combined to form new features, but we want to ensure the relative pose information between the features is preserved for all orientations. This has the advantage that the equivalence class of rotations for the complex composite features is guaranteed to be maintained, allowing for the extraction of invariant features, a unified pose, and increased data efficiency. Such ideas are reminiscent of the capsule networks of Hinton et al. \cite{capsules2011, e2018matrix, sabour2017dynamic}, and indeed formal connections to equivariance have been made \cite{lenssen2018group}. Interestingly, by explicitly building neural networks to be equivariant, we additionally see geometric organization of activations into these equivalence classes, and further, the elements within an equivalence class are seen to exhibit higher-order non-Gaussian dependencies \cite{Lyu08b, Lyu08, Wainwright99b, Wainwright00}. The insight of this connection between topographic organization and equivariance hints at a possibility to encourage approximate equivariance from an induced topology in feature space.
To build a model, we need to ask what mechanisms could induce topographic organization of \emph{observed transformations} specifically? We have argued that removing dependencies between latent variables is a possible mechanism; however, to obtain the more structured organisation of equivariant capsule representations, the usual approach is to hard-code this structure into the network, or to encourage it through regularization terms \cite{learninginvariance, Learningtoconvole}. To achieve this same structure \emph{unsupervised}, we propose to incorporate another key inductive bias: ``temporal coherence'' \cite{foldiak, hurri2003TC, stone1996, wiskott2002slow}. The principle of temporal coherence, or ``slowness'', asserts than when processing correlated sequences, we wish for our representations to change smoothly and slowly over space and time. Thinking of time sequences as symmetry transformations on the input, we desire features undergoing such transformations to be grouped into equivariant capsules. We therefore suggest that encouraging slow feature transformations to take place \emph{within a capsule} could induce such grouping from sequences alone.
In the following sections we will explain the details of our Topographic Variational Autoencoder which lies at the intersection of topographic organization, equivariance, and temporal coherence, thereby learning approximately equivariant capsules from sequence data completely unsupervised.
\section{Related Work}
The history of statistical models upon which this work builds is vast, including sparse coding \cite{olshausen1997sparse}, Independant Component Analysis (ICA) \cite{SejnowskiICA, comon1994independent, hyvarinen2000independent}, Slow Feature Analysis (SFA) \cite{probSFA, wiskott2002slow}, and Gaussian scale mixtures \cite{Lyu08, Portilla03, Wainwright99b, Wainwright00}. Most related to this work are topographic generative models including Generative Topographic Maps \cite{GTM}, Bubbles \cite{bubbles}, Topographic ICA \cite{hyvarinen2001topographic}, and the Topographic Product of Student's-t \cite{osindero2004contrastive, welling2003learning}. Prior work on learning equivariant and invariant representations is similarly vast and also has a deep relationship with these generative models. Specifically, Independant Subspace Analysis \cite{hyvarinen2000emergence, stuhmer2019independent}, models involving temporal coherence \cite{foldiak, hurri2003TC, stone1996, wiskott2002slow}, and Adaptive Subspace Self Organizing Maps \cite{kohonen1996emergence} have all demonstrated the ability to learn invariant feature subspaces and even `disentangle' space and time \cite{Grathwohl16, stuhmer2019independent}. Our work assumes a similar generative model to these works while additionally allowing for efficient estimation of the model through variational inference \cite{kingma2013auto, rezende2014stochastic}. Although our work is not the first to combine Student's-t distributions and variational inference \cite{boenninghoff2020variational}, it is the first to provide an efficient method to do so for Topographic Student's-t distributions.
\looseness=-1
Another line of work has focused on constructing neural networks with equivariant representations separate from the framework of generative modeling. Analytically equivariant networks such as Group Equivariant Neural Networks \cite{cohen2016group}, and other extensions \cite{cohen2016steerable, finzi2020generalizing, finzi2021emlp, elise, ravanbakhsh2017equivariance, weiler20183d, scalespaces, worrall2017harmonic} propose to explicitly enforce symmetry to group transformations in neural networks through structured weight sharing. Alternatively, others propose supervised and self-supervised methods for \emph{learning} equivariance or invariance directly from the data itself \cite{learninginvariance, connor2021variational, Learningtoconvole}. One related example in this category uses a group sparsity regularization term to similarly learn topographic features for the purpose of modeling invariance \cite{kavukcuoglu2009learning}. We believe the Topographic Variational Autoencoder presented in this paper is another promising step in the direction of learning approximate equivariance, and may even hint at how such structure could be learned in biological neural networks.
\looseness=-1
Furthermore, the idea of disentangled representations \cite{bengio2013representation} has also been been connected to equivariance and representation theory in multiple recent papers \cite{bouchacourt2021addressing, cohen2015transformation, cohen2014learning, higgins2018definition}. Our work shares a fundamental connection to this distributed operator definition of disentanglement, where the slow roll of capsule activations can be seen as the latent operator. Recently, the authors of \cite{klindt2021nonlinear} demonstrated that incorporating the principle of `slowness' in a variational autoencoder (VAE) yields the ability to learn disentangled representations from natural sequences. While similar in motivation, the generative model proposed in \cite{klindt2021nonlinear} is unrelated to topographic organization and equivariance, and is more aligned with traditional notions of disentanglement.
\looseness=-1
Finally, and importantly, in the neuroscience literature, another popular explanation for topographic organization arises as the solution to the `wiring length' minimization problem \cite{koulakov2001orientation}. Recently, models which attempt to incorporate wiring length constraints have been shown to yield topographic organization of higher level features, ultimately resembling the `face patches' found in primates \cite{keller2021modeling, TDANN}. Interestingly, the model presented in this paper organizes activity based on the same statistical property (local correlation) as the wiring length proxies developed in \cite{TDANN}, but from a generative modeling perspective, demonstrating a computationally principled explanation for the same phenomenon.
\section{Background}
The model in this paper is a first attempt at bridging two yet disjoint classes of models: Topographic Generative Models, and Equivariant Neural Networks. In this section, we will provide a brief background on these two frameworks.
\subsection{Topographic Generative models}
\looseness=-1
Inspired by Topographic ICA, the class of Topographic Generative models can be understood as generative models where the joint distribution over latent variables does not factorize into entirely independent factors, as is commonly done in ICA or VAEs, but instead has a more complex `local' correlation structure. The locality is defined by arranging the latent variables into an n-dimensional lattice or grid, and organizing variables such that those which are closer together on this grid have greater correlation of activities than those which are further apart. In the related literature, activations which are nearby in this grid are defined to have higher-order correlation, e.g. correlations of squared activations (aka `energy'), asserting that all first order correlations are removed by the initial ICA de-mixing matrix.
Such generative models can be seen as hierarchical generative models where there exist higher level independent `variance generating' variables $\mathbf{V}$ which are combined locally to generate the variances $\boldsymbol{\sigma} = \phi(\mathbf{W}\mathbf{V})$ of the lower level topographic variables $\mathbf{T} \sim \mathcal{N}(\mathbf{0}, \boldsymbol{\sigma}^2 \mathbf{I})$, for an appropriate non-linearity $\phi$. The variables $\mathbf{T}$ are thus independent conditioned on $\boldsymbol{\sigma}$. Other related models which can be described under this umbrella include \emph{Independent Subspace Analysis} (ISA) \cite{hyvarinen2000emergence} where all variables within a predefined subspace (or `capsule') share a common variance, and `\emph{temporally coherent}' models \cite{hurri2003TC} where the energy of a given variable between time steps is correlated by extending the topographic neighborhoods over the time dimension \cite{bubbles}. The topographic latent variable $\mathbf{T}$ can additionally be described as an instance of a Gaussian scale mixture (GSM). GSMs have previously been used to model the observed non-Gaussian dependencies between coefficients of steerable wavelet pyramids (interestingly also equivariant to translation \& rotation) \cite{Portilla03, Wainwright99b, Wainwright00}.
\subsection{Group Equivariant Neural Networks}
\looseness=-1
Equivariance is the mathematical notion of symmetry for functions. A function is said to be an equivariant map if the the result of transforming the input and then computing the function is the same as first computing the function and then transforming the output. In other words, the function and the transformation commute. Formally, $f(\tau_\rho [\mathbf{x}]) = \Gamma_\rho [f(\mathbf{x})]$, where $\tau$ and $\Gamma$ denote the (potentially different) operators on the domain and co-domain respectively, but are indexed by the same element $\rho$.
It is well known that convolutional maps in neural networks are translation equivariant, i.e., given a translation $\Gamma_{\rho}$
(applied to each feature map separately) and a convolutional map $f(\cdot)$, we have $f(\Gamma_{\rho}[\mathbf{x}]) = \Gamma_{\rho}[f(\mathbf{x})]$. This can be extended to other transformations (e.g. rotation or mirroring) using Group convolutions ($G$-convolutions)~\cite{cohen2016group}. As a result of the design of $G$-convolutions, feature maps that are related to each other by a rotation of the filter/input are grouped together. Moreover, a rotation of the input results in a transformation (i.e. a permutation and rotation) on the activations of each of these groups in the output. Hence, we can think of these equivalence class groups as capsules where transformations of the input only cause structured transformations $\emph{within}$ a capsule.
As we will demonstrate later, this is indeed analogous to the structure of the representation learned by the Topographic VAE with temporal coherence -- a transformation of the input yields a cyclic permutation of activations \emph{within} each capsule. However, due to the approximate \emph{learned} nature of the equivariant representation, the Topographic VAE does not require the transformations $\tau_{\rho}$ to constitute a group.
\section{The Generative Model}
The generative model proposed in this paper is based on the Topographic Product of Student's-t (TPoT) model as developed in \cite{osindero2004contrastive, welling2003learning}. In the following, we will show how a TPoT random variable can be constructed from a set of independent univariate standard normal random variables, enabling efficient training through variational inference. Subsequently, we will construct a new model where topographic neighborhoods are extended over time, introducing temporal coherence and encouraging the unsupervised learning of approximately equivariant subspaces we call `capsules'.
\subsection{The Product of Student's-t Model}
\looseness=-1
We assume that that our observed data is generated by a latent variable model where the joint distribution over observed and latent variables $\mathbf{x}$ and $\mathbf{t}$ factorizes into the product of the conditional and the prior. The prior distribution $p_{\mathbf{T}}(\mathbf{t})$ is assumed to be a Topographic Product of Student's-t (TPoT) distribution, and we parameterize the conditional distribution with a flexible function approximator:
\begin{equation}
\label{eqn:generative_model}
p_{\mathbf{X}, \mathbf{T}}(\mathbf{x}, \mathbf{t}) = p_{\mathbf{X}| \mathbf{T}}(\mathbf{x}|\mathbf{t})p_{\mathbf{T}}(\mathbf{t}) \ \ \ \ \ \ \ \ p_{\mathbf{X}|\mathbf{T}}(\mathbf{x}|\mathbf{t}) = p_{\theta}(\mathbf{x} | g_{\theta}(\mathbf{t})) \ \ \ \ \ \ \ \ p_{\mathbf{T}}(\mathbf{t}) = \mathrm{TPoT}(\mathbf{t}; \nu)
\end{equation}
The goal of training is thus to learn the parameters $\theta$ such that the marginal distribution of the model $p_{\theta}(\mathbf{x})$ matches that of the observed data. Unfortunately, the marginal likelihood is generally intractable except for all but the simplest choices of $g_{\theta}$ and $p_{\mathbf{T}}$ \cite{Osindero2006}. Prior work has therefore resorted to techniques such as contrastive divergence with Gibbs sampling \cite{welling2003learning} to train TPoT models as energy based models. In the following section, we instead demonstrate how TPoT variables can be constructed as a deterministic function of Gaussian random variables, enabling the use of variational inference and efficient maximization of the likelihood through the evidence lower bound (ELBO).
\subsection{Constructing the Product of Student's-t Distribution}
\looseness=-1
First, note a univariate Student's-t random variable $T$ with $\nu$ degrees of freedom can be defined as:
\begin{equation}
T = \frac{Z}{\sqrt{\frac{1}{\nu}\sum^{\nu}_i U_i^2}} \ \ \ \ \ \mathrm{with}\ \ \ Z, U_i \sim \mathcal{N}(0, 1)\ \ \forall i\ \
\end{equation}
Where $Z$ and $\{U_i\}_{i=1}^\nu$ are independent standard normal random variables. If $\mathbf{T}$ is a multidimensional Student's-t random variable, composed of independent $Z_i$ and $U_i$, then $\mathbf{T} \sim \mathrm{PoT(\nu)}$, i.e.:
\begin{equation}
\mathbf{T} = \left[\frac{Z_1}{\sqrt{\frac{1}{\nu}\sum^{\nu}_{i=1} U_i^2}},\ \ \frac{Z_2}{\sqrt{\frac{1}{\nu}\sum^{2\cdot\nu}_{i=\nu+1} U_i^2}},\ \ \ldots\ \ \frac{Z_n}{\sqrt{\frac{1}{\nu}\sum^{n\cdot\nu}_{i=(n-1)\cdot\nu+1} U_i^2}}\right] \sim \mathrm{PoT(\nu)}
\end{equation}
Note that the Student's-t variable $T$ is large when most of the $\{U_i\}_i$ in its set are small. We can therefore think of the $\{U_i\}_i$ as constraint violations rather then pattern matches: if the input matches all constraints $U_i\approx 0$, the corresponding $T$ variables will activate (see \cite{hinton2013discovering} for further discussion).
\subsection{Introducing Topography}
\looseness=-1
To make the PoT distribution topographic, we strive to correlate the scales of $T_j$ which are `nearby' in our topographic layout. One way to accomplish this is by \emph{sharing} some $U_i$-variables between neighboring $T_j$'s. Formally, we define overlapping neighborhoods $\mathsf{N}(j)$ for each variable $T_j$ and write:
\begin{equation}
\mathbf{T} = \left[\frac{Z_1}{\sqrt{\frac{1}{\nu}\sum_{i \in\mathsf{N}(1)} U_i^2}},\ \ \frac{Z_2}{\sqrt{\frac{1}{\nu}\sum_{i\in\mathsf{N}(2)} U_i^2}},\ \ \ldots\ \ \frac{Z_n}{\sqrt{\frac{1}{\nu}\sum_{i\in\mathsf{N}(n)} U_i^2}}\right] \sim \mathrm{TPoT(\nu)}
\end{equation}
With some abuse of notation, if we define $\mathbf{W}$ to be the adjacency matrix which defines our neighborhood structure, $\mathbf{U}$ and $\mathbf{Z}$ to be the vectors of random variables $U_i$ and $Z_j$, we can write the above succinctly as:
\begin{align}
\label{eqn:full_T}
\mathbf{T} = \left[\frac{Z_1}{\sqrt{\frac{1}{\nu}W_1\mathbf{U}^2}},\ \ \frac{Z_2}{\sqrt{\frac{1}{\nu}W_2\mathbf{U}^2}},\ \ \ldots\ \ \frac{Z_n}{\sqrt{\frac{1}{\nu}W_n\mathbf{U}^2}}\right] = \frac{\mathbf{Z}}{\sqrt{\frac{1}{\nu} \mathbf{W} \mathbf{U}^2}} \sim \mathrm{TPoT(\nu)}
\end{align}
Due to non-linearities such as ReLUs which may alter input distributions, it is beneficial to allow the $Z$ variables to model the mean and scale. We found this can be achieved with the following parameterization: \scalebox{0.7}{$\mathbf{T} = \frac{\mathbf{Z} - \mu}{\sigma\sqrt{1/\nu \mathbf{W} \mathbf{U}^2}}$}. In practice, we found that $\sigma = \sqrt{\nu}$ often works well, finally yielding:
\begin{align}
\label{eqn:final_T}
\mathbf{T} = \frac{\mathbf{Z} - \mu}{\sqrt{\mathbf{W} \mathbf{U}^2}}
\end{align}
Given this construction, we observe that the TPoT generative model can instead be viewed as a latent variable model where all random variables are Gaussian and the construction of $\mathbf{T}$ in Equation \ref{eqn:final_T} is the first layer of the generative `decoder': $g_{\theta}(\mathbf{t}) = g_{\theta}(\mathbf{u}, \mathbf{z})$. In Section \ref{sec:TVAE} we then leverage this interpretation to show how an approximate posterior for the latent variables $\mathbf{Z}$ and $\mathbf{U}$ can be trained through variational inference.
\subsection{Capsules as Disjoint Topologies}
\begin{wrapfigure}{r}{0.4\linewidth}
\vspace{-10mm}
\centering
\includegraphics[width=0.4\textwidth]{figures/Fig2.png}
\caption{An example of a neighborhood structure which induces disjoint topologies (aka capsules). Lines between variables $T_i$ indicate that sharing of $U_i$, and thus correlation.}
\label{fig:depindep}
\end{wrapfigure}
One setting of neighborhood structure $\mathbf{W}$ which is of particular interest is when there exist multiple sets of disjoint neighborhoods. Statistically, the variables of two disjoint topologies are completely independent.
An example of a capsule neighborhood structure is shown in Figure \ref{fig:depindep}. The idea of independant subspaces has previously been shown to learn invariant feature subspaces in the linear setting and is present in early work on Independent Subspace Analysis \cite{hyvarinen2000emergence} and Adaptive Subspace Self Organizing Maps (ASSOM) \cite{kohonen1996emergence}. It is also very reminiscent of the transformed sets of features present in a group equivariant convolutional neural network. In the next section, we will show how temporal coherence can be leveraged to induce the encoding of observed transformations into the internal dimensions of such capsules thereby yielding unsupervised approximately equivariant capsules.
\subsection{Temporal Coherence and Learned Equivariance}
\label{sec:temporal_coherence}
We now describe how the induced topographic organization can be leveraged to learn a basis of approximately equivariant capsules for observed transformation sequences. The resulting representation is composed of a large set of `capsules' where the dimensions inside the capsule are topographically structured, but between the capsules there is independence. To benefit from sequences of input, we encourage topographic structure over time between sequentially permuted activations within a capsule, a property we refer to as \emph{shifting temporal coherence}.
\subsubsection{Temporal Coherence}
Temporal Coherence can be measured as the correlation of squared activation between time steps. One way we can achieve this in our model is by having $T_j$ share $U_i$ between time steps. Formally, the generative model is identical to Equation \ref{eqn:generative_model}, factorizing over timesteps denoted by subscript $l$, i.e. $p_{\mathbf{X}_l, \mathbf{T}_l}(\mathbf{x}_l, \mathbf{t}_l) = p_{\mathbf{X}_l| \mathbf{T}_l}(\mathbf{x}_l|\mathbf{t}_l)p_{\mathbf{T}_l}(\mathbf{t}_l)$. However, $\mathbf{T}_l$ is now a function of a sequence $\{\mathbf{U}_{l+\delta}\}_{\delta=-L}^{L}$:
\begin{align}
\label{eqn:T_seq}
\mathbf{T}_l = \frac{\mathbf{Z}_l - \mu}{\sqrt{\mathbf{W}
\left[
\mathbf{U}_{l+L}^2;
\cdots; \mathbf{U}_{l-L}^2
\right]
}}
\end{align}
Where $\left[\mathbf{U}_{l+L}^2; \cdots; \mathbf{U}_{l-L}^2\right]$ denotes vertical concatenation of the column vectors $\mathbf{U}_l$, and $2L$ can be seen as the window size. We see that the choice of $\mathbf{W}$ now defines correlation structure over time. In prior work on temporal coherence (denoted `Bubbles' \cite{bubbles}), the grouping over time is such that a given variable $T_{l,i}$ has correlated energy with \emph{the same spatial location} $(i)$ at a previous time step $(l-1)$ \big(i.e. $\mathrm{cov}(T_{l,i}^2, T_{l-1,i}^2) > 0$\big). This can be implemented as:
\begin{equation}
\label{eqn:temporal_coherence}
\mathbf{W} \left[\mathbf{U}_{l+L}^2; \cdots; \mathbf{U}_{l-L}^2\right] = \sum_{\delta=-L}^{L} \mathbf{W}_\delta\mathbf{U}_{l+\delta}^2
\end{equation}
Where $\mathbf{W}_\delta$ defines the topography for a single timestep, and is typically the same for all timesteps.
\subsubsection{Learned Equivariance with Shifting Temporal Coherence}
In our model, instead of requiring a single location to have correlated energies over a sequence,
we would like variables at sequentially permuted locations \emph{within a capsule} to have correlated energy between timesteps \big($\mathrm{cov}(T_{l,i}^2, T_{l-1,i-1}^2) > 0$\big). Similarly, this can be implemented as:
\begin{equation}
\label{eqn:eq_roll}
\mathbf{W} \left[\mathbf{U}_{l+L}^2; \cdots; \mathbf{U}_{l-L}^2\right] = \sum_{\delta=-L}^{L} \mathbf{W}_\delta \mathrm{Roll}_{\delta}(\mathbf{U}_{l+\delta}^2)
\end{equation}
Where $\mathrm{Roll}_{\delta}(\mathbf{U}_{l+\delta}^2)$ denotes a cyclic permutation of $\delta$ steps along the capsule dimension.
The exact implementation
of $\mathrm{Roll}$ can be found in Section \ref{sec:roll_def}. As we will show in Section \ref{sec:equivariant_experiments}, TVAE models with such a topographic structure learn to encode observed sequence transformations as $\mathrm{Roll}$s within the capsule dimension, analogous to a group equivariant neural network where
$\tau_\rho$ and $\mathrm{Roll}_1$ can be seen as the action of the transformation $\rho$ on the input and output spaces respectively.
\section{Topographic VAE}
\label{sec:TVAE}
To train the parameters of the generative model $\theta$, we use the above formulation to parameterize an approximate posterior for $\mathbf{t}$ in terms of a deterministic transformation of approximate posteriors over simpler Gaussian latent variables $\mathbf{u}$ and $\mathbf{z}$. Explicitly:
\begin{gather}
\label{eqn:tvae1}
q_{\phi}(\mathbf{z}_l|\mathbf{x}_l) = \mathcal{N}\big(\mathbf{z}_l; \mu_{\phi}(\mathbf{x}_l), \sigma_{\phi}(\mathbf{x}_l) \mathbf{I}\big) \hspace{8mm}
p_{\theta}(\mathbf{x}_l | g_{\theta}(\mathbf{t}_l)) = p_{\theta}(\mathbf{x}_l | g_{\theta}(\mathbf{z}_l, \{\mathbf{u}_l\}))
\\
\label{eqn:tvae2}
q_{\gamma}(\mathbf{u}_l|\mathbf{x}_l) = \mathcal{N}\big(\mathbf{u}_l ; \mu_{\gamma}(\mathbf{x}_l), \sigma_{\gamma}(\mathbf{x}_l) \mathbf{I}\big) \hspace{16mm}
\mathbf{t}_l = \frac{\mathbf{z}_l - \mu}{\sqrt{\mathbf{W} \left[ \mathbf{u}_{l+L}^2; \cdots; \mathbf{u}_{l-L}^2\right]}}
\end{gather}
We denote this model the Topographic VAE (TVAE) and optimize the parameters $\theta, \phi, \gamma$ (and $\mu$) through the ELBO, summed over the sequence length $S$:
\fontsize{9.5}{10}
\begin{equation}
\label{eqn:elbo}
\sum_{l=1}^S \mathbb{E}_{Q_{\phi,\gamma}(\mathbf{z}_l,\mathbf{u}_l|\{\mathbf{x}_l\})}
\left([\log p_{\theta}(\mathbf{x}_l|g_{\theta}(\mathbf{t}_l))] - D_{KL}[q_{\phi}(\mathbf{z}_l|\mathbf{x}_l) || p_{\mathbf{Z}}(\mathbf{z}_l)] - D_{KL}[q_{\gamma}(\mathbf{u}_l|\mathbf{x}_l) || p_{\mathbf{U}}(\mathbf{u}_l)]\right)
\end{equation}
\normalsize
where $Q_{\phi,\gamma}(\mathbf{z}_l,\mathbf{u}_l|\{\mathbf{x}_l\})= q_{\phi}(\mathbf{z}_l|\mathbf{x}_l)\prod_{\delta=-L}^L q_{\gamma}(\mathbf{u}_{l+\delta}|\mathbf{x}_{l+\delta})$, and $\{\cdot\}$ denotes a set over time.
\section{Experiments}
\label{sec:experiments}
In the following experiments, we demonstrate the viability of the Topographic VAE as a novel method for training deep topographic generative models. Additionally, we quantitatively verify that shifting temporal coherence yields approximately equivariant capsules by computing an `equivariance loss' and a correlation metric inspired by the disentanglement literature. We show that equivariant capsule models yield higher likelihood than baselines on test sequences, and qualitatively support these results with visualizations of sequences reconstructed purely from $\mathrm{Roll}$ed capsule activations.
\subsection{Evaluation Methods}
As depicted in Figure \ref{fig:traversal}, we make use of \emph{capsule traversals} to qualitatively visualize the transformations learned by our network. Simply, these are constructed by encoding a partial sequence into a $\mathbf{t}_0$ variable, and decoding sequentially $\mathrm{Roll}$ed copies of this variable. Explicitly, in the top row we show the data sequence $\{\mathbf{x}_l\}_l$, and in the bottom row we show the decoded sequence: $\{g_{\theta}(\mathrm{Roll}_{l}(\mathbf{t_0}))\}_l$.
To measure equivariance quantitatively, we measure an \emph{equivariance error} similar to \cite{Learningtoconvole}. The equivariance error can be seen as the difference between traversing the two distinct paths of the commutative diagram, and provides some measure of how precisely the function and the transform commute. Formally, for a sequence of length $S$, and $\mathbf{\hat{t}}=\mathbf{t} / ||\mathbf{t}||_2$, the error is defined as:
\begin{equation}
\label{eqn:eq_err}
\mathcal{E}_{eq}(\{\mathbf{t}_l\}_{l=1}^{S}) = \sum_{l=1}^{S-1} \sum_{\delta = 1}^{S - l} \left|\left| \mathrm{Roll}_{\delta}(\mathbf{\hat{t}}_{l}) - \mathbf{\hat{t}}_{l+\delta}\right|\right|_1
\end{equation}
Additionally, inspired by existing disentanglement metrics, we measure the degree to which observed transformations in capsule space are correlated with input transformations by introducing a new metric we call $\mathrm{CapCorr}_y$. Simply, this metric computes the correlation between the amount of observed $\mathrm{Roll}$ of a capsule's activation at two timesteps $l$ and $l+\delta$, and the shift of the ground truth generative factors $y_l$ in that same time. Formally, for a correlation coefficient $\mathrm{Corr}$:
\begin{equation}
\mathrm{CapCorr}(\mathbf{t}_{l}, \mathbf{t}_{l+\delta}, y_{l}, y_{l+\delta}) = \mathrm{Corr} \left(\mathrm{argmax}\left[\mathbf{t}_{l} \star \mathbf{t}_{l+\delta}\right], |y_l - y_{l+\delta}|\right)
\label{eqn:capcorr}
\end{equation}
Where $\star$ is discrete periodic cross-correlation across the capsule dimension, and the correlation coefficient is computed across the entire dataset. We see the $\mathrm{argmax}$ of the cross-correlation is an estimate of the degree to which a capsule activation has shifted from time $l$ to $l + \delta$. To extend this to multiple capsules, we can replace the $\mathrm{argmax}$ function with the mode of the $\mathrm{argmax}$ computed for all capsules. We provide additional details and extensions of this metric in Section \ref{sec:appendix_capcorr}. For measuring capsule-metrics on baseline models which do not naturally have capsules, we simply arbitrarily divide the latent space into a fixed set of corresponding capsules and capsule dimensions, and provide such results as equivalent to `random baselines' for these metrics.
\subsection{Topographic VAE without Temporal Coherence}
\label{sec:2d_TVAE}
\begin{wrapfigure}{r}{0.4\linewidth}
\vspace{-10mm}
\includegraphics[width=\linewidth]{figures/max_act.png}
\caption{Maximum activating images for a Topographic VAE trained with a 2D torus topography on MNIST.}
\vspace{-5mm}
\label{fig:TVAE}
\end{wrapfigure}
To validate the TVAE is capable of learning topographically organized representations with deep neural networks, we first perform experiments on a Topographic VAE without Temporal Coherence. The model is constructed as in Equations \ref{eqn:tvae1} and \ref{eqn:tvae2} with $L=0$, and is trained to maximize Equation \ref{eqn:elbo}. We fix $\mathbf{W}$ such that globally the latent variables are arranged in a grid on a 2-dimensional torus (a single capsule), and locally $\mathbf{W}$ sums over 5x5 2D groups of variables. In this setting, $\mathbf{W}$ can be easily implemented as 2D convolution with a 5x5 kernel of $1$'s, stride 1, and cyclic padding. We see that training the model with 3-layer MLP's for the encoders and decoder indeed yields a 2D topographic organization of higher level features. In Figure \ref{fig:TVAE}, we show the maximum activating image for each final layer neuron of the capsule, plotted as a flattened torus. We see that the neurons become arranged according to class, orientation, width, and other learned features.
\subsection{Learning Equivariant Capsules}
\label{sec:equivariant_experiments}
In the remaining experiments, we provide evidence that the Topographic VAE can be leveraged to learn equivariant capsules by incorporating shifting temporal coherence into a 1D baseline topographic model. We compare against two baselines: standard normal VAEs and models that have non-shifting `stationary' temporal coherence as defined in Equation \ref{eqn:temporal_coherence} (denoted `BubbleVAE' \cite{bubbles}).
In all experiments we use a 3-layer MLP with ReLU activations for both encoders and the decoder. We arrange the latent space into 15 circular capsules each of 15-dimensions for dSprites \cite{dSprites17}, and 18 circular capsules each of 18-dimensions for MNIST \cite{lecun2010mnist}. Example sequences $\{\mathbf{x}_l\}_{l=1}^S$ are formed by taking a random initial example, and sequentially transforming it according to one of the available transformations: (X-Pos, Y-Pos, Orientation, Scale) for dSprites, and (Color, Scale, Orientation) for MNIST. All transformation sequences are cyclic such that when the maximum transformation parameter is reached, the subsequent value returns to the minimum. We denote the length of a full transformation sequence by $S$, and the time-extent of the induced temporal coherence (i.e. the length of the input sequence) by $2L$. For simplicity, both datasets are constructed such that the sequence length $S$ equals the capsule dimension (for dSprites this involves taking a subset of the full dataset and looping the scale 3-times for a scale-sequence). Exact details are in Sections \ref{sec:mnist} \& \ref{sec:dsprites}.
In Figure \ref{fig:all_traversals}, we show the capsule traversals for TVAE models with {\small$L \approx \frac{1}{3}S$}. We see that despite the $\mathbf{t}_0$ variable encoding only $\frac{2}{3}$ of the sequence, the remainder of the transformation sequence can be decoded nearly perfectly by permuting the activation through the full capsule -- implying the model has learned to be approximately equivariant to full sequences while only observing partial sequences per training point. Furthermore, we see that the model is able to successfully learn all transformations simultaneously for the respective datasets.
\begin{figure}[h!]
\centering
\includegraphics[width=1.0\linewidth]{figures/all_traversals.png}
\caption{Capsule Traversals for TVAE models on dSprites and MNIST. The top rows show the encoded sequences (with greyed-out images held-out), and the bottom rows show the images generated by decoding sequentially $\mathrm{Roll}$ed copies of the initial activation $\mathbf{t}_0$ (indicated by a grey border).}
\label{fig:all_traversals}
\end{figure} \\
Capsule traversals for the non-equivariant baselines, as well as TVAEs with smaller values of $L$ (which only learn approximate equivariance to partial sequences) are shown in Section \ref{sec:capsule_traversals}. We note that the capsule traversal plotted in Figure \ref{fig:traversal} demonstrates a transformation where color and rotation change simultaneously, differing from how the models in this section are trained. However, as we describe in more detail in Section \ref{sec:generalization_experiments}, we observe that TVAEs trained with individual transformations in isolation (as in this section) are able to generalize, generating sequences of combined transformations when presented with such partial input sequences at test time. We believe this generalization capability to be promising for data efficiency, but leave further exploration to future work. Additional capsule traversals with such unseen combined transformations are shown in Section \ref{sec:generalization_experiments} and further complex learned transformations (such as perspective transforms) are shown at the end of Section \ref{sec:capsule_traversals}.
For a more quantitative evaluation, in Table \ref{table:mnist} we measure the equivariance error and log-likelihood (reported in nats) of the test data under our trained MNIST models as estimated by importance sampling with 10 samples. We observe that models which incorporate temporal coherence (BubbleVAE and TVAE with $L > 0$) achieve low equivariance error, while the TVAE models with shifting temporal coherence achieve the highest likelihood and the lowest equivariance error simultaneously.
\begin{table}[h!]
\centering
\vspace{-4mm}
\caption{Log Likelihood and Equivariance Error on MNIST for different settings of temporal coherence length $L$ relative to sequence length $S$. Mean $\pm$ std. over 3 random initalizations.}
\vspace{2mm}
\begin{tabular}{l r r r r r r}
\toprule
Model & TVAE & TVAE & TVAE & BubbleVAE & VAE \\
$L$ & $L=\frac{1}{2}S$ & $L=\frac{5}{36}S$ & $L=0$ & $L=\frac{5}{36}S$ & $L=0$ \\ \midrule
$\log p(\mathbf{x})$ $\uparrow$ & $\mathbf{-186.8}$ $\pm$ 0.1 & $\mathbf{-186.0}$ $\pm$ 0.7 & -218.5$\pm$ 0.9 & -191.4 $\pm$ 0.5 & -189.0 $\pm$ 0.8 \\
$\mathcal{E}_{eq}$ $\downarrow$ & $\mathbf{574}$ $\pm$ 2 & 3247 $\pm$ 3 & 3217 $\pm$ 105 & 3370 $\pm$ 12 & 13274 $\pm$ 1 \\
\bottomrule
\end{tabular}
\label{table:mnist}
\end{table}
\begin{table}[h!]
\centering
\caption{Equivariance error ($\mathcal{E}_{eq}$ $\downarrow$) and correlation of observed capsule roll with ground truth factor shift ($\mathrm{CapCorr}$ $\uparrow$) for the dSprites dataset. Mean $\pm$ standard deviation over 3 random initalizations.}
\vspace{2mm}
\begin{tabular}{l r r r r r r}
\toprule
Model & TVAE & TVAE & TVAE & TVAE & BubbleVAE & VAE \\
$L$ & $L=\frac{1}{2}S$ & $L = \frac{1}{3}S$ & $L = \frac{1}{6}S$ & $L=0$ & $L=\frac{1}{3}S$ & $L=0$\\
\midrule
$\mathrm{CapCorr}_X$ $\uparrow$ & $\mathbf{1.0}$ $\pm$ 0 & $\mathbf{1.0}$ $\pm$ 0 & 0.67 $\pm$ 0.02 & 0.17 $\pm$ 0.03 & 0.13 $\pm$ 0.01 & 0.18 $\pm$ 0.01 \\
$\mathrm{CapCorr}_Y$ $\uparrow$ & $\mathbf{1.0}$ $\pm$ 0 & $\mathbf{1.0}$ $\pm$ 0 & 0.66 $\pm$ 0.02 & 0.21 $\pm$ 0.02 & 0.12 $\pm$ 0.01 & 0.16 $\pm$ 0.01 \\
$\mathrm{CapCorr}_{O}$ $\uparrow$ & $\mathbf{1.0}$ $\pm$ 0 & $\mathbf{1.0}$ $\pm$ 0 & 0.52 $\pm$ 0.01 & 0.09 $\pm$ 0.01 & 0.10 $\pm$ 0.01 & 0.11 $\pm$ 0.00 \\
$\mathrm{CapCorr}_{S}$ $\uparrow$ & $\mathbf{1.0}$ $\pm$ 0 & $\mathbf{1.0}$ $\pm$ 0 & 0.42 $\pm$ 0.01 & 0.51 $\pm$ 0.01 & 0.50 $\pm$ 0.00 & 0.52 $\pm$ 0.00 \\
\midrule
$\mathcal{E}_{eq}$ $\downarrow$ & $\mathbf{344}$ $\pm$ 5 & 1034 $\pm$ 6 & 2549 $\pm$ 38 & 2971 $\pm$ 9 & 1951 $\pm$ 34 & 6934 $\pm$ 0\\
\bottomrule
\end{tabular}
\label{table:capcorr}
\end{table}
To further understand how capsules transform for observed input transformations, in Table \ref{table:capcorr} we measure $\mathcal{E}_{eq}$ and the $\mathrm{CapCorr}$ metric on the dSprites dataset for the four proposed transformations. We see that the TVAE with $L\geq\frac{1}{3}S$ achieves perfect correlation -- implying the learned representation indeed permutes cyclically within capsules for observed transformation sequences. Further, this correlation gradually decreases as $L$ decreases, eventually reaching the same level as the baselines. We also see that, on both datasets, the equivariance losses for the TVAE with $L=0$ and the BubbleVAE are significantly lower than the baseline VAE, while conversely, the CapCorr metric is not significantly better. We believe this to be due to the fundamental difference between the metrics: $\mathcal{E}_{eq}$ measures continuous L1 similarity which is still low when a representation is locally smooth (even if the change of the representation does not follow the observed transformation), whereas $\mathrm{CapCorr}$ more strictly measures the correspondence between the transformation of the input and the transformation of the representation. In other words, $\mathcal{E}_{eq}$ may be misleadingly low for invariant capsule representations (as with the BubbleVAE), whereas $\mathrm{CapCorr}$ strictly measures equivariance.
\section{Future Work \& Limitations}
\label{sec:limitations}
\looseness=-1
The model presented in this work has a number of limitations in its existing form which we believe to be interesting directions for future research. Foremost, the model is challenging to compare directly with existing disentanglement and equivariance literature since it requires an input sequence which determines the transformations reachable through the capsule roll.
Related to this, we note the temporal coherence proposed in our model is not `causal' (i.e. $\mathbf{t}_{0}$ depends on future $\mathbf{x}_l$).
We believe these limitations could be at least partially alleviated with minor extensions detailed in Section \ref{sec:extensions}.
We additionally note that some model developers may find a priori definition of topographic structure burdensome. While true, we know that the construction of appropriate priors is always a challenging task in latent variable models, and we observe that our proposed TVAE achieves strong performance even with improper specification. Furthermore, in future work, we believe adding learned flexibility to the parameters $\mathbf{W}$ may alleviate some of this burden.
Finally, we note that while this work does demonstrate improved log-likelihood and equivariance error, the study is inherently preliminary and does not examine all important benefits of topographic or approximately equivariant representations. Specifically, further study of the TVAE both with and without temporal coherence in terms of the sample complexity, semi-supervised classification accuracy, and invariance through structured topographic pooling would be enlightening.
\section{Conclusion}
In the above work we introduce the Topographic Variational Autoencoder as a method to train deep topographic generative models, and show how topography can be leveraged to learn approximately equivariant sets of features, a.k.a. capsules, directly from sequences of data with no other supervision. Ultimately, we believe these results may shine some light on how biological systems could hard-wire themselves to more effectively learn representations with equivariant capsule structure.
In terms of broader impact, it is foreseeable our model could be used to generate more realistic transformations of `deepfakes', enhancing disinformation. Given that the model learns \emph{approximate} equivariance, we caution against the over-reliance on equivariant properties as these have no known formal guarantees.
\bibliographystyle{plain}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 5,130 |
<?php
/**
* WPSEO plugin file.
*
* @package WPSEO\Frontend
*/
/**
* Main frontend class for Yoast SEO, responsible for the SEO output as well as removing
* default WordPress output.
*/
class WPSEO_Frontend {
/**
* @var object Instance of this class.
*/
public static $instance;
/**
* @var boolean Boolean indicating whether output buffering has been started.
*/
private $ob_started = false;
/**
* Holds the canonical URL for the current page.
*
* @var string
*/
private $canonical = null;
/**
* Holds the canonical URL for the current page that cannot be overriden by a manual canonical input.
*
* @var string
*/
private $canonical_no_override = null;
/**
* Holds the canonical URL for the current page without pagination.
*
* @var string
*/
private $canonical_unpaged = null;
/**
* Holds the pages meta description.
*
* @var string
*/
private $metadesc = null;
/**
* Holds the generated title for the page.
*
* @var string
*/
private $title = null;
/** @var WPSEO_Frontend_Page_Type */
protected $frontend_page_type;
/** @var WPSEO_WooCommerce_Shop_Page */
protected $woocommerce_shop_page;
/**
* Class constructor.
*
* Adds and removes a lot of filters.
*/
protected function __construct() {
add_action( 'wp_head', array( $this, 'front_page_specific_init' ), 0 );
add_action( 'wp_head', array( $this, 'head' ), 1 );
// The head function here calls action wpseo_head, to which we hook all our functionality.
add_action( 'wpseo_head', array( $this, 'debug_mark' ), 2 );
add_action( 'wpseo_head', array( $this, 'metadesc' ), 6 );
add_action( 'wpseo_head', array( $this, 'robots' ), 10 );
add_action( 'wpseo_head', array( $this, 'canonical' ), 20 );
add_action( 'wpseo_head', array( $this, 'adjacent_rel_links' ), 21 );
add_action( 'wpseo_head', array( $this, 'publisher' ), 22 );
// Remove actions that we will handle through our wpseo_head call, and probably change the output of.
remove_action( 'wp_head', 'rel_canonical' );
remove_action( 'wp_head', 'index_rel_link' );
remove_action( 'wp_head', 'start_post_rel_link' );
remove_action( 'wp_head', 'adjacent_posts_rel_link_wp_head' );
remove_action( 'wp_head', 'noindex', 1 );
// When using WP 4.4, just use the new hook.
add_filter( 'pre_get_document_title', array( $this, 'title' ), 15 );
add_filter( 'wp_title', array( $this, 'title' ), 15, 3 );
add_filter( 'thematic_doctitle', array( $this, 'title' ), 15 );
add_action( 'wp', array( $this, 'page_redirect' ), 99 );
add_action( 'template_redirect', array( $this, 'noindex_feed' ) );
add_filter( 'loginout', array( $this, 'nofollow_link' ) );
add_filter( 'register', array( $this, 'nofollow_link' ) );
// Fix the WooThemes woo_title() output.
add_filter( 'woo_title', array( $this, 'fix_woo_title' ), 99 );
if ( WPSEO_Options::get( 'disable-date', false )
|| WPSEO_Options::get( 'disable-author', false )
|| WPSEO_Options::get( 'disable-post_format', false )
) {
add_action( 'wp', array( $this, 'archive_redirect' ) );
}
add_action( 'template_redirect', array( $this, 'attachment_redirect' ), 1 );
add_filter( 'the_content_feed', array( $this, 'embed_rssfooter' ) );
add_filter( 'the_excerpt_rss', array( $this, 'embed_rssfooter_excerpt' ) );
// For WordPress functions below 4.4.
if ( WPSEO_Options::get( 'forcerewritetitle', false ) && ! current_theme_supports( 'title-tag' ) ) {
add_action( 'template_redirect', array( $this, 'force_rewrite_output_buffer' ), 99999 );
add_action( 'wp_footer', array( $this, 'flush_cache' ), - 1 );
}
if ( WPSEO_Options::get( 'title_test', 0 ) > 0 ) {
add_filter( 'wpseo_title', array( $this, 'title_test_helper' ) );
}
$this->woocommerce_shop_page = new WPSEO_WooCommerce_Shop_Page();
$this->frontend_page_type = new WPSEO_Frontend_Page_Type();
$integrations = array(
new WPSEO_Frontend_Primary_Category(),
new WPSEO_JSON_LD(),
new WPSEO_Remove_Reply_To_Com(),
$this->woocommerce_shop_page,
);
foreach ( $integrations as $integration ) {
$integration->register_hooks();
}
}
/**
* Initialize the functions that only need to run on the frontpage.
*/
public function front_page_specific_init() {
if ( ! is_front_page() ) {
return;
}
add_action( 'wpseo_head', array( $this, 'webmaster_tools_authentication' ), 90 );
}
/**
* Resets the entire class so canonicals, titles etc can be regenerated.
*/
public function reset() {
foreach ( get_class_vars( __CLASS__ ) as $name => $default ) {
switch ( $name ) {
// Clear the class instance to be re-initialized.
case 'instance':
self::$instance = null;
break;
// Exclude these properties from being reset.
case 'woocommerce_shop_page':
case 'frontend_page_type':
break;
// Reset property to the class default.
default:
$this->$name = $default;
break;
}
}
WPSEO_Options::ensure_options_exist();
}
/**
* Get the singleton instance of this class.
*
* @return WPSEO_Frontend
*/
public static function get_instance() {
if ( ! ( self::$instance instanceof self ) ) {
self::$instance = new self();
}
return self::$instance;
}
/**
* Override Woo's title with our own.
*
* @param string $title Title string.
*
* @return string
*/
public function fix_woo_title( $title ) {
return $this->title( $title );
}
/**
* Determine whether this is the homepage and shows posts.
*
* @return bool
*/
public function is_home_posts_page() {
return ( is_home() && 'posts' === get_option( 'show_on_front' ) );
}
/**
* Determine whether the this is the static frontpage.
*
* @return bool
*/
public function is_home_static_page() {
return ( is_front_page() && 'page' === get_option( 'show_on_front' ) && is_page( get_option( 'page_on_front' ) ) );
}
/**
* Determine whether this is the posts page, when it's not the frontpage.
*
* @return bool
*/
public function is_posts_page() {
return ( is_home() && 'page' === get_option( 'show_on_front' ) );
}
/**
* Used for static home and posts pages as well as singular titles.
*
* @param object|null $object If filled, object to get the title for.
*
* @return string
*/
public function get_content_title( $object = null ) {
if ( $object === null ) {
$object = $GLOBALS['wp_query']->get_queried_object();
}
$title = $this->get_seo_title( $object );
if ( $title !== '' ) {
return $title;
}
$post_type = ( isset( $object->post_type ) ? $object->post_type : $object->query_var );
return $this->get_title_from_options( 'title-' . $post_type, $object );
}
/**
* Retrieves the SEO title set in the SEO widget.
*
* @param null $object Object to retrieve the title from.
*
* @return string The SEO title for the specified object, or queried object if not supplied.
*/
public function get_seo_title( $object = null ) {
if ( $object === null ) {
$object = $GLOBALS['wp_query']->get_queried_object();
}
if ( ! is_object( $object ) ) {
return $this->get_title_from_options( 'title-404-wpseo' );
}
$title = $this->get_seo_meta_value( 'title', $object->ID );
if ( $title !== '' ) {
return $this->replace_vars( $title, $object );
}
return $title;
}
/**
* Used for category, tag, and tax titles.
*
* @return string
*/
public function get_taxonomy_title() {
$object = $GLOBALS['wp_query']->get_queried_object();
$title = WPSEO_Taxonomy_Meta::get_term_meta( $object, $object->taxonomy, 'title' );
if ( is_string( $title ) && $title !== '' ) {
return $this->replace_vars( $title, $object );
}
return $this->get_title_from_options( 'title-tax-' . $object->taxonomy, $object );
}
/**
* Used for author titles.
*
* @return string
*/
public function get_author_title() {
$author_id = get_query_var( 'author' );
$title = trim( get_the_author_meta( 'wpseo_title', $author_id ) );
if ( $title !== '' ) {
return $this->replace_vars( $title, array() );
}
return $this->get_title_from_options( 'title-author-wpseo' );
}
/**
* Simple function to use to pull data from $options.
*
* All titles pulled from options will be run through the $this->replace_vars function.
*
* @param string $index Name of the page to get the title from the settings for.
* @param object|array $var_source Possible object to pull variables from.
*
* @return string
*/
public function get_title_from_options( $index, $var_source = array() ) {
$template = WPSEO_Options::get( $index, '' );
if ( $template === '' ) {
if ( is_singular() ) {
return $this->replace_vars( '%%title%% %%sep%% %%sitename%%', $var_source );
}
return '';
}
return $this->replace_vars( $template, $var_source );
}
/**
* Get the default title for the current page.
*
* This is the fallback title generator used when a title hasn't been set for the specific content, taxonomy, author
* details, or in the options. It scrubs off any present prefix before or after the title (based on $seplocation) in
* order to prevent duplicate seperations from appearing in the title (this happens when a prefix is supplied to the
* wp_title call on singular pages).
*
* @param string $sep The separator used between variables.
* @param string $seplocation Whether the separator should be left or right.
* @param string $title Possible title that's already set.
*
* @return string
*/
public function get_default_title( $sep, $seplocation, $title = '' ) {
if ( 'right' === $seplocation ) {
$regex = '`\s*' . preg_quote( trim( $sep ), '`' ) . '\s*`u';
}
else {
$regex = '`^\s*' . preg_quote( trim( $sep ), '`' ) . '\s*`u';
}
$title = preg_replace( $regex, '', $title );
if ( ! is_string( $title ) || ( is_string( $title ) && $title === '' ) ) {
$title = WPSEO_Utils::get_site_name();
$title = $this->add_paging_to_title( $sep, $seplocation, $title );
$title = $this->add_to_title( $sep, $seplocation, $title, wp_strip_all_tags( get_bloginfo( 'description' ), true ) );
return $title;
}
$title = $this->add_paging_to_title( $sep, $seplocation, $title );
$title = $this->add_to_title( $sep, $seplocation, $title, wp_strip_all_tags( get_bloginfo( 'name' ), true ) );
return $title;
}
/**
* This function adds paging details to the title.
*
* @param string $sep Separator used in the title.
* @param string $seplocation Whether the separator should be left or right.
* @param string $title The title to append the paging info to.
*
* @return string
*/
public function add_paging_to_title( $sep, $seplocation, $title ) {
global $wp_query;
if ( ! empty( $wp_query->query_vars['paged'] ) && $wp_query->query_vars['paged'] > 1 ) {
return $this->add_to_title( $sep, $seplocation, $title, $wp_query->query_vars['paged'] . '/' . $wp_query->max_num_pages );
}
return $title;
}
/**
* Add part to title, while ensuring that the $seplocation variable is respected.
*
* @param string $sep Separator used in the title.
* @param string $seplocation Whether the separator should be left or right.
* @param string $title The title to append the title_part to.
* @param string $title_part The part to append to the title.
*
* @return string
*/
public function add_to_title( $sep, $seplocation, $title, $title_part ) {
if ( 'right' === $seplocation ) {
return $title . $sep . $title_part;
}
return $title_part . $sep . $title;
}
/**
* Main title function.
*
* @param string $title Title that might have already been set.
* @param string $separator Separator determined in theme (unused).
* @param string $separator_location Whether the separator should be left or right.
*
* @return string
*/
public function title( $title, $separator = '', $separator_location = '' ) {
if ( is_null( $this->title ) ) {
$this->title = $this->generate_title( $title, $separator_location );
}
return $this->title;
}
/**
* Main title generation function.
*
* @param string $title Title that might have already been set.
* @param string $separator_location Whether the separator should be left or right.
*
* @return string
*/
private function generate_title( $title, $separator_location ) {
if ( is_feed() ) {
return $title;
}
$separator = $this->replace_vars( '%%sep%%', array() );
$separator = ' ' . trim( $separator ) . ' ';
if ( '' === trim( $separator_location ) ) {
$separator_location = ( is_rtl() ) ? 'left' : 'right';
}
// This needs to be kept track of in order to generate
// default titles for singular pages.
$original_title = $title;
// This flag is used to determine if any additional
// processing should be done to the title after the
// main section of title generation completes.
$modified_title = true;
// This variable holds the page-specific title part
// that is used to generate default titles.
$title_part = '';
if ( $this->is_home_static_page() ) {
$title = $this->get_content_title();
}
elseif ( $this->is_home_posts_page() ) {
$title = $this->get_title_from_options( 'title-home-wpseo' );
}
elseif ( $this->woocommerce_shop_page->is_shop_page() ) {
$post = get_post( $this->woocommerce_shop_page->get_shop_page_id() );
$title = $this->get_seo_title( $post );
if ( ! is_string( $title ) || $title === '' ) {
$title = $this->get_post_type_archive_title( $separator, $separator_location );
}
}
elseif ( $this->frontend_page_type->is_simple_page() ) {
$post = get_post( $this->frontend_page_type->get_simple_page_id() );
$title = $this->get_content_title( $post );
if ( ! is_string( $title ) || '' === $title ) {
$title_part = $original_title;
}
}
elseif ( is_search() ) {
$title = $this->get_title_from_options( 'title-search-wpseo' );
if ( ! is_string( $title ) || '' === $title ) {
/* translators: %s expands to the search phrase. */
$title_part = sprintf( __( 'Search for "%s"', 'wordpress-seo' ), esc_html( get_search_query() ) );
}
}
elseif ( is_category() || is_tag() || is_tax() ) {
$title = $this->get_taxonomy_title();
if ( ! is_string( $title ) || '' === $title ) {
if ( is_category() ) {
$title_part = single_cat_title( '', false );
}
elseif ( is_tag() ) {
$title_part = single_tag_title( '', false );
}
else {
$title_part = single_term_title( '', false );
if ( $title_part === '' ) {
$term = $GLOBALS['wp_query']->get_queried_object();
$title_part = $term->name;
}
}
}
}
elseif ( is_author() ) {
$title = $this->get_author_title();
if ( ! is_string( $title ) || '' === $title ) {
$title_part = get_the_author_meta( 'display_name', get_query_var( 'author' ) );
}
}
elseif ( is_post_type_archive() ) {
$title = $this->get_post_type_archive_title( $separator, $separator_location );
}
elseif ( is_archive() ) {
$title = $this->get_title_from_options( 'title-archive-wpseo' );
// @todo [JRF => Yoast] Should these not use the archive default if no title found ?
// WPSEO_Options::get_default( 'wpseo_titles', 'title-archive-wpseo' )
// Replacement would be needed!
if ( empty( $title ) ) {
if ( is_month() ) {
/* translators: %s expands to a time period, i.e. month name, year or specific date. */
$title_part = sprintf( __( '%s Archives', 'wordpress-seo' ), single_month_title( ' ', false ) );
}
elseif ( is_year() ) {
/* translators: %s expands to a time period, i.e. month name, year or specific date. */
$title_part = sprintf( __( '%s Archives', 'wordpress-seo' ), get_query_var( 'year' ) );
}
elseif ( is_day() ) {
/* translators: %s expands to a time period, i.e. month name, year or specific date. */
$title_part = sprintf( __( '%s Archives', 'wordpress-seo' ), get_the_date() );
}
else {
$title_part = __( 'Archives', 'wordpress-seo' );
}
}
}
elseif ( is_404() ) {
$title = $this->get_title_from_options( 'title-404-wpseo' );
// @todo [JRF => Yoast] Should these not use the 404 default if no title found ?
// WPSEO_Options::get_default( 'wpseo_titles', 'title-404-wpseo' )
// Replacement would be needed!
if ( empty( $title ) ) {
$title_part = __( 'Page not found', 'wordpress-seo' );
}
}
else {
// In case the page type is unknown, leave the title alone.
$modified_title = false;
// If you would like to generate a default title instead,
// the following code could be used
// $title_part = $title;
// instead of the line above.
}
if ( ( $modified_title && empty( $title ) ) || ! empty( $title_part ) ) {
$title = $this->get_default_title( $separator, $separator_location, $title_part );
}
if ( defined( 'ICL_LANGUAGE_CODE' ) && false !== strpos( $title, ICL_LANGUAGE_CODE ) ) {
$title = str_replace( ' @' . ICL_LANGUAGE_CODE, '', $title );
}
/**
* Filter: 'wpseo_title' - Allow changing the Yoast SEO <title> output.
*
* @api string $title The page title being put out.
*/
return esc_html( wp_strip_all_tags( stripslashes( apply_filters( 'wpseo_title', $title ) ), true ) );
}
/**
* Function used when title needs to be force overridden.
*
* @return string
*/
public function force_wp_title() {
global $wp_query;
$old_wp_query = null;
if ( ! $wp_query->is_main_query() ) {
$old_wp_query = $wp_query;
wp_reset_query();
}
$title = $this->title( '' );
if ( ! empty( $old_wp_query ) ) {
$GLOBALS['wp_query'] = $old_wp_query;
unset( $old_wp_query );
}
return $title;
}
/**
* Outputs or returns the debug marker, which is also used for title replacement when force rewrite is active.
*
* @param bool $echo Deprecated. Since 5.9. Whether or not to echo the debug marker.
*
* @return string The marker that will be echoed.
*/
public function debug_mark( $echo = true ) {
$marker = $this->get_debug_mark();
if ( $echo === false ) {
_deprecated_argument( 'WPSEO_Frontend::debug_mark', '5.9', 'WPSEO_Frontend::get_debug_mark' );
return $marker;
}
echo "\n${marker}\n";
return '';
}
/**
* Returns the debug marker, which is also used for title replacement when force rewrite is active.
*
* @return string The generated marker.
*/
public function get_debug_mark() {
return sprintf(
'<!-- This site is optimized with the %1$s %2$s - https://yoast.com/wordpress/plugins/seo/ -->',
esc_html( $this->head_product_name() ),
/**
* Filter: 'wpseo_hide_version' - can be used to hide the Yoast SEO version in the debug marker (only available in Yoast SEO Premium).
*
* @api bool
*/
( ( apply_filters( 'wpseo_hide_version', false ) && $this->is_premium() ) ? '' : 'v' . WPSEO_VERSION )
);
}
/**
* Output Webmaster Tools authentication strings.
*/
public function webmaster_tools_authentication() {
// Baidu.
$this->webmaster_tools_helper( 'baiduverify', 'baidu-site-verification' );
// Bing.
$this->webmaster_tools_helper( 'msverify', 'msvalidate.01' );
// Google.
$this->webmaster_tools_helper( 'googleverify', 'google-site-verification' );
// Pinterest.
$this->webmaster_tools_helper( 'pinterestverify', 'p:domain_verify' );
// Yandex.
$this->webmaster_tools_helper( 'yandexverify', 'yandex-verification' );
}
/**
* Helper function for authentication.
*
* @param string $option_key Option key.
* @param string $tag_name The tag name.
*
* @return void
*/
private function webmaster_tools_helper( $option_key, $tag_name ) {
$auth = WPSEO_Options::get( $option_key, '' );
if ( $auth !== '' ) {
printf( '<meta name="%1$s" content="%2$s" />' . "\n", $tag_name, $auth );
}
}
/**
* Main wrapper function attached to wp_head. This combines all the output on the frontend of the Yoast SEO plugin.
*/
public function head() {
global $wp_query;
$old_wp_query = null;
if ( ! $wp_query->is_main_query() ) {
$old_wp_query = $wp_query;
wp_reset_query();
}
/**
* Action: 'wpseo_head' - Allow other plugins to output inside the Yoast SEO section of the head section.
*/
do_action( 'wpseo_head' );
echo $this->show_closing_debug_mark();
if ( ! empty( $old_wp_query ) ) {
$GLOBALS['wp_query'] = $old_wp_query;
unset( $old_wp_query );
}
}
/**
* Output the meta robots value.
*
* @return string
*/
public function robots() {
global $wp_query, $post;
$robots = array();
$robots['index'] = 'index';
$robots['follow'] = 'follow';
$robots['other'] = array();
if ( is_object( $post ) && is_singular() ) {
$private = 'private' === $post->post_status;
$noindex = ! WPSEO_Post_Type::is_post_type_indexable( $post->post_type );
if ( $noindex || $private ) {
$robots['index'] = 'noindex';
}
$robots = $this->robots_for_single_post( $robots );
}
else {
if ( is_search() || is_404() ) {
$robots['index'] = 'noindex';
}
elseif ( is_tax() || is_tag() || is_category() ) {
$term = $wp_query->get_queried_object();
if ( is_object( $term ) && ( WPSEO_Options::get( 'noindex-tax-' . $term->taxonomy, false ) ) ) {
$robots['index'] = 'noindex';
}
// Three possible values, index, noindex and default, do nothing for default.
$term_meta = WPSEO_Taxonomy_Meta::get_term_meta( $term, $term->taxonomy, 'noindex' );
if ( is_string( $term_meta ) && 'default' !== $term_meta ) {
$robots['index'] = $term_meta;
}
if ( $this->is_multiple_terms_query() ) {
$robots['index'] = 'noindex';
}
}
elseif ( is_author() ) {
if ( WPSEO_Options::get( 'noindex-author-wpseo', false ) ) {
$robots['index'] = 'noindex';
}
$curauth = $wp_query->get_queried_object();
if ( WPSEO_Options::get( 'noindex-author-noposts-wpseo', false ) && count_user_posts( $curauth->ID, 'any' ) === 0 ) {
$robots['index'] = 'noindex';
}
if ( get_user_meta( $curauth->ID, 'wpseo_noindex_author', true ) === 'on' ) {
$robots['index'] = 'noindex';
}
}
elseif ( is_date() && WPSEO_Options::get( 'noindex-archive-wpseo', false ) ) {
$robots['index'] = 'noindex';
}
elseif ( is_home() ) {
$page_for_posts = get_option( 'page_for_posts' );
if ( $page_for_posts ) {
$robots = $this->robots_for_single_post( $robots, $page_for_posts );
}
unset( $page_for_posts );
}
elseif ( is_post_type_archive() ) {
$post_type = $this->get_queried_post_type();
if ( WPSEO_Options::get( 'noindex-ptarchive-' . $post_type, false ) ) {
$robots['index'] = 'noindex';
}
}
unset( $robot );
}
// Force override to respect the WP settings.
if ( '0' === (string) get_option( 'blog_public' ) || isset( $_GET['replytocom'] ) ) {
$robots['index'] = 'noindex';
}
$robotsstr = $robots['index'] . ',' . $robots['follow'];
if ( $robots['other'] !== array() ) {
$robots['other'] = array_unique( $robots['other'] ); // @todo Most likely no longer needed, needs testing.
$robotsstr .= ',' . implode( ',', $robots['other'] );
}
$robotsstr = preg_replace( '`^index,follow,?`', '', $robotsstr );
$robotsstr = str_replace( array( 'noodp,', 'noodp' ), '', $robotsstr );
/**
* Filter: 'wpseo_robots' - Allows filtering of the meta robots output of Yoast SEO.
*
* @api string $robotsstr The meta robots directives to be echoed.
*/
$robotsstr = apply_filters( 'wpseo_robots', $robotsstr );
if ( is_string( $robotsstr ) && $robotsstr !== '' ) {
echo '<meta name="robots" content="', esc_attr( $robotsstr ), '"/>', "\n";
}
// If a page has a noindex, it should _not_ have a canonical, as these are opposing indexing directives.
if ( strpos( $robotsstr, 'noindex' ) !== false ) {
remove_action( 'wpseo_head', array( $this, 'canonical' ), 20 );
}
return $robotsstr;
}
/**
* Determine $robots values for a single post.
*
* @param array $robots Robots data array.
* @param int $post_id The post ID for which to determine the $robots values, defaults to current post.
*
* @return array
*/
public function robots_for_single_post( $robots, $post_id = 0 ) {
$noindex = $this->get_seo_meta_value( 'meta-robots-noindex', $post_id );
if ( $noindex === '1' ) {
$robots['index'] = 'noindex';
}
elseif ( $noindex === '2' ) {
$robots['index'] = 'index';
}
if ( $this->get_seo_meta_value( 'meta-robots-nofollow', $post_id ) === '1' ) {
$robots['follow'] = 'nofollow';
}
$meta_robots_adv = $this->get_seo_meta_value( 'meta-robots-adv', $post_id );
if ( $meta_robots_adv !== '' && ( $meta_robots_adv !== '-' && $meta_robots_adv !== 'none' ) ) {
$meta_robots_adv = explode( ',', $meta_robots_adv );
foreach ( $meta_robots_adv as $robot ) {
$robots['other'][] = $robot;
}
unset( $robot );
}
unset( $meta_robots_adv );
return $robots;
}
/**
* This function normally outputs the canonical but is also used in other places to retrieve
* the canonical URL for the current page.
*
* @param bool $echo Whether or not to output the canonical element.
* @param bool $un_paged Whether or not to return the canonical with or without pagination added to the URL.
* @param bool $no_override Whether or not to return a manually overridden canonical.
*
* @return string $canonical
*/
public function canonical( $echo = true, $un_paged = false, $no_override = false ) {
if ( is_null( $this->canonical ) ) {
$this->generate_canonical();
}
$canonical = $this->canonical;
if ( $un_paged ) {
$canonical = $this->canonical_unpaged;
}
elseif ( $no_override ) {
$canonical = $this->canonical_no_override;
}
if ( $echo === false ) {
return $canonical;
}
if ( is_string( $canonical ) && '' !== $canonical ) {
echo '<link rel="canonical" href="' . esc_url( $canonical, null, 'other' ) . '" />' . "\n";
}
}
/**
* This function normally outputs the canonical but is also used in other places to retrieve
* the canonical URL for the current page.
*
* @return void
*/
private function generate_canonical() {
$canonical = false;
$canonical_override = false;
// Set decent canonicals for homepage, singulars and taxonomy pages.
if ( is_singular() ) {
$obj = get_queried_object();
$canonical = get_permalink( $obj->ID );
$this->canonical_unpaged = $canonical;
$canonical_override = $this->get_seo_meta_value( 'canonical' );
// Fix paginated pages canonical, but only if the page is truly paginated.
if ( get_query_var( 'page' ) > 1 ) {
$num_pages = ( substr_count( $obj->post_content, '<!--nextpage-->' ) + 1 );
if ( $num_pages && get_query_var( 'page' ) <= $num_pages ) {
if ( ! $GLOBALS['wp_rewrite']->using_permalinks() ) {
$canonical = add_query_arg( 'page', get_query_var( 'page' ), $canonical );
}
else {
$canonical = user_trailingslashit( trailingslashit( $canonical ) . get_query_var( 'page' ) );
}
}
}
}
else {
if ( is_search() ) {
$search_query = get_search_query();
// Regex catches case when /search/page/N without search term is itself mistaken for search term. R.
if ( ! empty( $search_query ) && ! preg_match( '|^page/\d+$|', $search_query ) ) {
$canonical = get_search_link();
}
}
elseif ( is_front_page() ) {
$canonical = WPSEO_Utils::home_url();
}
elseif ( $this->is_posts_page() ) {
$posts_page_id = get_option( 'page_for_posts' );
$canonical = $this->get_seo_meta_value( 'canonical', $posts_page_id );
if ( empty( $canonical ) ) {
$canonical = get_permalink( $posts_page_id );
}
}
elseif ( is_tax() || is_tag() || is_category() ) {
$term = get_queried_object();
if ( ! empty( $term ) && ! $this->is_multiple_terms_query() ) {
$canonical_override = WPSEO_Taxonomy_Meta::get_term_meta( $term, $term->taxonomy, 'canonical' );
$term_link = get_term_link( $term, $term->taxonomy );
if ( ! is_wp_error( $term_link ) ) {
$canonical = $term_link;
}
}
}
elseif ( is_post_type_archive() ) {
$post_type = $this->get_queried_post_type();
$canonical = get_post_type_archive_link( $post_type );
}
elseif ( is_author() ) {
$canonical = get_author_posts_url( get_query_var( 'author' ), get_query_var( 'author_name' ) );
}
elseif ( is_archive() ) {
if ( is_date() ) {
if ( is_day() ) {
$canonical = get_day_link( get_query_var( 'year' ), get_query_var( 'monthnum' ), get_query_var( 'day' ) );
}
elseif ( is_month() ) {
$canonical = get_month_link( get_query_var( 'year' ), get_query_var( 'monthnum' ) );
}
elseif ( is_year() ) {
$canonical = get_year_link( get_query_var( 'year' ) );
}
}
}
$this->canonical_unpaged = $canonical;
if ( $canonical && get_query_var( 'paged' ) > 1 ) {
global $wp_rewrite;
if ( ! $wp_rewrite->using_permalinks() ) {
if ( is_front_page() ) {
$canonical = trailingslashit( $canonical );
}
$canonical = add_query_arg( 'paged', get_query_var( 'paged' ), $canonical );
}
else {
if ( is_front_page() ) {
$canonical = WPSEO_Sitemaps_Router::get_base_url( '' );
}
$canonical = user_trailingslashit( trailingslashit( $canonical ) . trailingslashit( $wp_rewrite->pagination_base ) . get_query_var( 'paged' ) );
}
}
}
$this->canonical_no_override = $canonical;
if ( is_string( $canonical ) && $canonical !== '' ) {
// Force canonical links to be absolute, relative is NOT an option.
if ( WPSEO_Utils::is_url_relative( $canonical ) === true ) {
$canonical = $this->base_url( $canonical );
}
}
if ( is_string( $canonical_override ) && $canonical_override !== '' ) {
$canonical = $canonical_override;
}
/**
* Filter: 'wpseo_canonical' - Allow filtering of the canonical URL put out by Yoast SEO.
*
* @api string $canonical The canonical URL.
*/
$this->canonical = apply_filters( 'wpseo_canonical', $canonical );
}
/**
* Parse the home URL setting to find the base URL for relative URLs.
*
* @param string $path Optional path string.
*
* @return string
*/
private function base_url( $path = null ) {
$url = get_option( 'home' );
$parts = wp_parse_url( $url );
$base_url = trailingslashit( $parts['scheme'] . '://' . $parts['host'] );
if ( ! is_null( $path ) ) {
$base_url .= ltrim( $path, '/' );
}
return $base_url;
}
/**
* Adds 'prev' and 'next' links to archives.
*
* @link http://googlewebmastercentral.blogspot.com/2011/09/pagination-with-relnext-and-relprev.html
* @since 1.0.3
*/
public function adjacent_rel_links() {
// Don't do this for Genesis, as the way Genesis handles homepage functionality is different and causes issues sometimes.
/**
* Filter 'wpseo_genesis_force_adjacent_rel_home' - Allows devs to allow echoing rel="next" / rel="prev" by Yoast SEO on Genesis installs.
*
* @api bool $unsigned Whether or not to rel=next / rel=prev .
*/
if ( is_home() && function_exists( 'genesis' ) && apply_filters( 'wpseo_genesis_force_adjacent_rel_home', false ) === false ) {
return;
}
/**
* Filter: 'wpseo_disable_adjacent_rel_links' - Allows disabling of Yoast adjacent links if this is being handled by other code.
*
* @api bool $links_generated Indicates if other code has handled adjacent links.
*/
if ( true === apply_filters( 'wpseo_disable_adjacent_rel_links', false ) ) {
return;
}
if ( is_singular() ) {
$this->rel_links_single();
return;
}
$this->rel_links_archive();
}
/**
* Output the rel next/prev links for a single post / page.
*
* @return void
*/
protected function rel_links_single() {
$num_pages = 1;
$queried_object = get_queried_object();
if ( ! empty( $queried_object ) ) {
$num_pages = ( substr_count( $queried_object->post_content, '<!--nextpage-->' ) + 1 );
}
if ( $num_pages === 1 ) {
return;
}
$page = max( 1, (int) get_query_var( 'page' ) );
$url = get_permalink( get_queried_object_id() );
if ( $page > 1 ) {
$this->adjacent_rel_link( 'prev', $url, ( $page - 1 ), 'page' );
}
if ( $page < $num_pages ) {
$this->adjacent_rel_link( 'next', $url, ( $page + 1 ), 'page' );
}
}
/**
* Output the rel next/prev links for an archive page.
*/
protected function rel_links_archive() {
$url = $this->canonical( false, true, true );
if ( ! is_string( $url ) || $url === '' ) {
return;
}
$paged = max( 1, (int) get_query_var( 'paged' ) );
if ( $paged === 2 ) {
$this->adjacent_rel_link( 'prev', $url, ( $paged - 1 ) );
}
// Make sure to use index.php when needed, done after paged == 2 check so the prev links to homepage will not have index.php erroneously.
if ( is_front_page() ) {
$url = WPSEO_Sitemaps_Router::get_base_url( '' );
}
if ( $paged > 2 ) {
$this->adjacent_rel_link( 'prev', $url, ( $paged - 1 ) );
}
if ( $paged < $GLOBALS['wp_query']->max_num_pages ) {
$this->adjacent_rel_link( 'next', $url, ( $paged + 1 ) );
}
}
/**
* Get adjacent pages link for archives.
*
* @since 1.0.2
* @since 7.1 Added $query_arg parameter for single post/page pagination.
*
* @param string $rel Link relationship, prev or next.
* @param string $url The un-paginated URL of the current archive.
* @param string $page The page number to add on to $url for the $link tag.
* @param string $query_arg Optional. The argument to use to set for the page to load.
*
* @return void
*/
private function adjacent_rel_link( $rel, $url, $page, $query_arg = 'paged' ) {
global $wp_rewrite;
if ( ! $wp_rewrite->using_permalinks() ) {
if ( $page > 1 ) {
$url = add_query_arg( $query_arg, $page, $url );
}
}
else {
if ( $page > 1 ) {
$url = user_trailingslashit( trailingslashit( $url ) . $this->get_pagination_base() . $page );
}
}
/**
* Filter: 'wpseo_' . $rel . '_rel_link' - Allow changing link rel output by Yoast SEO.
*
* @api string $unsigned The full `<link` element.
*/
$link = apply_filters( 'wpseo_' . $rel . '_rel_link', '<link rel="' . esc_attr( $rel ) . '" href="' . esc_url( $url ) . "\" />\n" );
if ( is_string( $link ) && $link !== '' ) {
echo $link;
}
}
/**
* Return the base for pagination.
*
* @return string The pagination base.
*/
private function get_pagination_base() {
// If the current page is the frontpage, pagination should use /base/.
$base = '';
if ( ! is_singular() || $this->is_home_static_page() ) {
$base = trailingslashit( $GLOBALS['wp_rewrite']->pagination_base );
}
return $base;
}
/**
* Output the rel=publisher code on every page of the site.
*
* @return boolean Boolean indicating whether the publisher link was printed.
*/
public function publisher() {
$publisher = WPSEO_Options::get( 'plus-publisher', '' );
if ( $publisher !== '' ) {
echo '<link rel="publisher" href="', esc_url( $publisher ), '"/>', "\n";
return true;
}
return false;
}
/**
* Outputs the meta description element or returns the description text.
*
* @param bool $echo Echo or return output flag.
*
* @return string
*/
public function metadesc( $echo = true ) {
if ( is_null( $this->metadesc ) ) {
$this->generate_metadesc();
}
if ( $echo === false ) {
return $this->metadesc;
}
if ( is_string( $this->metadesc ) && $this->metadesc !== '' ) {
echo '<meta name="description" content="', esc_attr( wp_strip_all_tags( stripslashes( $this->metadesc ) ) ), '"/>', "\n";
return '';
}
if ( current_user_can( 'wpseo_manage_options' ) && is_singular() ) {
echo '<!-- ';
printf(
/* Translators: %1$s resolves to the SEO menu item, %2$s resolves to the Search Appearance submenu item. */
esc_html__( 'Admin only notice: this page does not show a meta description because it does not have one, either write it for this page specifically or go into the [%1$s - %2$s] menu and set up a template.', 'wordpress-seo' ),
__( 'SEO', 'wordpress-seo' ),
__( 'Search Appearance', 'wordpress-seo' )
);
echo ' -->' . "\n";
}
}
/**
* Generates the meta description text.
*/
private function generate_metadesc() {
global $post, $wp_query;
$metadesc = '';
$metadesc_override = false;
$post_type = '';
$template = '';
if ( is_object( $post ) && ( isset( $post->post_type ) && $post->post_type !== '' ) ) {
$post_type = $post->post_type;
}
if ( $this->woocommerce_shop_page->is_shop_page() ) {
$post = get_post( $this->woocommerce_shop_page->get_shop_page_id() );
$post_type = $this->get_queried_post_type();
if ( ( $metadesc === '' && $post_type !== '' ) && WPSEO_Options::get( 'metadesc-ptarchive-' . $post_type, '' ) !== '' ) {
$template = WPSEO_Options::get( 'metadesc-ptarchive-' . $post_type );
$term = $post;
}
$metadesc_override = $this->get_seo_meta_value( 'metadesc', $post->ID );
}
elseif ( $this->frontend_page_type->is_simple_page() ) {
$post = get_post( $this->frontend_page_type->get_simple_page_id() );
$post_type = isset( $post->post_type ) ? $post->post_type : '';
if ( ( $metadesc === '' && $post_type !== '' ) && WPSEO_Options::get( 'metadesc-' . $post_type, '' ) !== '' ) {
$template = WPSEO_Options::get( 'metadesc-' . $post_type );
$term = $post;
}
if ( is_object( $post ) ) {
$metadesc_override = $this->get_seo_meta_value( 'metadesc', $post->ID );
}
}
else {
if ( is_search() ) {
$metadesc = '';
}
elseif ( $this->is_home_posts_page() ) {
$template = WPSEO_Options::get( 'metadesc-home-wpseo' );
$term = array();
if ( empty( $template ) ) {
$template = get_bloginfo( 'description' );
}
}
elseif ( $this->is_home_static_page() ) {
$metadesc = $this->get_seo_meta_value( 'metadesc' );
if ( ( $metadesc === '' && $post_type !== '' ) && WPSEO_Options::get( 'metadesc-' . $post_type, '' ) !== '' ) {
$template = WPSEO_Options::get( 'metadesc-' . $post_type );
}
}
elseif ( is_category() || is_tag() || is_tax() ) {
$term = $wp_query->get_queried_object();
$metadesc_override = WPSEO_Taxonomy_Meta::get_term_meta( $term, $term->taxonomy, 'desc' );
if ( is_object( $term ) && isset( $term->taxonomy ) && WPSEO_Options::get( 'metadesc-tax-' . $term->taxonomy, '' ) !== '' ) {
$template = WPSEO_Options::get( 'metadesc-tax-' . $term->taxonomy );
}
}
elseif ( is_author() ) {
$author_id = get_query_var( 'author' );
$metadesc = get_the_author_meta( 'wpseo_metadesc', $author_id );
if ( ( ! is_string( $metadesc ) || $metadesc === '' ) && WPSEO_Options::get( 'metadesc-author-wpseo', '' ) !== '' ) {
$template = WPSEO_Options::get( 'metadesc-author-wpseo' );
}
}
elseif ( is_post_type_archive() ) {
$post_type = $this->get_queried_post_type();
if ( WPSEO_Options::get( 'metadesc-ptarchive-' . $post_type, '' ) !== '' ) {
$template = WPSEO_Options::get( 'metadesc-ptarchive-' . $post_type );
}
}
elseif ( is_archive() ) {
$template = WPSEO_Options::get( 'metadesc-archive-wpseo' );
}
// If we're on a paginated page, and the template doesn't change for paginated pages, bail.
if ( ( ! is_string( $metadesc ) || $metadesc === '' ) && get_query_var( 'paged' ) && get_query_var( 'paged' ) > 1 && $template !== '' ) {
if ( strpos( $template, '%%page' ) === false ) {
$metadesc = '';
}
}
}
$post_data = $post;
if ( is_string( $metadesc_override ) && '' !== $metadesc_override ) {
$metadesc = $metadesc_override;
if ( isset( $term ) ) {
$post_data = $term;
}
}
elseif ( ( ! is_string( $metadesc ) || '' === $metadesc ) && '' !== $template ) {
if ( ! isset( $term ) ) {
$term = $wp_query->get_queried_object();
}
$metadesc = $template;
$post_data = $term;
}
$metadesc = $this->replace_vars( $metadesc, $post_data );
/**
* Filter: 'wpseo_metadesc' - Allow changing the Yoast SEO meta description sentence.
*
* @api string $metadesc The description sentence.
*/
$this->metadesc = apply_filters( 'wpseo_metadesc', trim( $metadesc ) );
}
/**
* Based on the redirect meta value, this function determines whether it should redirect the current post / page.
*
* @return boolean
*/
public function page_redirect() {
if ( is_singular() ) {
global $post;
if ( ! isset( $post ) || ! is_object( $post ) ) {
return false;
}
$redir = $this->get_seo_meta_value( 'redirect', $post->ID );
if ( $redir !== '' ) {
wp_redirect( $redir, 301 );
exit;
}
}
return false;
}
/**
* Outputs noindex values for the current page.
*/
public function noindex_page() {
remove_action( 'wpseo_head', array( $this, 'canonical' ), 20 );
echo '<meta name="robots" content="noindex" />', "\n";
}
/**
* Send a Robots HTTP header preventing URL from being indexed in the search results while allowing search engines
* to follow the links in the object at the URL.
*
* @since 1.1.7
* @return boolean Boolean indicating whether the noindex header was sent.
*/
public function noindex_feed() {
if ( ( is_feed() || is_robots() ) && headers_sent() === false ) {
header( 'X-Robots-Tag: noindex, follow', true );
return true;
}
return false;
}
/**
* Adds rel="nofollow" to a link, only used for login / registration links.
*
* @param string $input The link element as a string.
*
* @return string
*/
public function nofollow_link( $input ) {
return str_replace( '<a ', '<a rel="nofollow" ', $input );
}
/**
* When certain archives are disabled, this redirects those to the homepage.
*
* @return boolean False when no redirect was triggered.
*/
public function archive_redirect() {
global $wp_query;
if (
( WPSEO_Options::get( 'disable-date', false ) && $wp_query->is_date ) ||
( WPSEO_Options::get( 'disable-author', false ) && $wp_query->is_author ) ||
( WPSEO_Options::get( 'disable-post_format', false ) && $wp_query->is_tax( 'post_format' ) )
) {
$this->redirect( get_bloginfo( 'url' ), 301 );
return true;
}
return false;
}
/**
* If the option to disable attachment URLs is checked, this performs the redirect to the attachment.
*
* @return bool Returns succes status.
*/
public function attachment_redirect() {
if ( WPSEO_Options::get( 'disable-attachment', false ) === false ) {
return false;
}
if ( ! is_attachment() ) {
return false;
}
/**
* Allow the developer to change the target redirection URL for attachments.
*
* @api string $attachment_url The attachment URL for the queried object.
* @api object $queried_object The queried object.
*
* @since 7.5.3
*/
$url = apply_filters( 'wpseo_attachment_redirect_url', wp_get_attachment_url( get_queried_object_id() ), get_queried_object() );
if ( ! empty( $url ) ) {
$this->redirect( $url, 301 );
return true;
}
return false;
}
/**
* Replaces the possible RSS variables with their actual values.
*
* @param string $content The RSS content that should have the variables replaced.
*
* @return string
*/
public function rss_replace_vars( $content ) {
global $post;
/**
* Allow the developer to determine whether or not to follow the links in the bits Yoast SEO adds to the RSS feed, defaults to true.
*
* @api bool $unsigned Whether or not to follow the links in RSS feed, defaults to true.
*
* @since 1.4.20
*/
$no_follow = apply_filters( 'nofollow_rss_links', true );
$no_follow_attr = '';
if ( $no_follow === true ) {
$no_follow_attr = 'rel="nofollow" ';
}
$author_link = '';
if ( is_object( $post ) ) {
$author_link = '<a ' . $no_follow_attr . 'href="' . esc_url( get_author_posts_url( $post->post_author ) ) . '">' . esc_html( get_the_author() ) . '</a>';
}
$post_link = '<a ' . $no_follow_attr . 'href="' . esc_url( get_permalink() ) . '">' . esc_html( get_the_title() ) . '</a>';
$blog_link = '<a ' . $no_follow_attr . 'href="' . esc_url( get_bloginfo( 'url' ) ) . '">' . esc_html( get_bloginfo( 'name' ) ) . '</a>';
$blog_desc_link = '<a ' . $no_follow_attr . 'href="' . esc_url( get_bloginfo( 'url' ) ) . '">' . esc_html( get_bloginfo( 'name' ) ) . ' - ' . esc_html( get_bloginfo( 'description' ) ) . '</a>';
$content = stripslashes( trim( $content ) );
$content = str_replace( '%%AUTHORLINK%%', $author_link, $content );
$content = str_replace( '%%POSTLINK%%', $post_link, $content );
$content = str_replace( '%%BLOGLINK%%', $blog_link, $content );
$content = str_replace( '%%BLOGDESCLINK%%', $blog_desc_link, $content );
return $content;
}
/**
* Adds the RSS footer (or header) to the full RSS feed item.
*
* @param string $content Feed item content.
*
* @return string
*/
public function embed_rssfooter( $content ) {
return $this->embed_rss( $content, 'full' );
}
/**
* Adds the RSS footer (or header) to the excerpt RSS feed item.
*
* @param string $content Feed item excerpt.
*
* @return string
*/
public function embed_rssfooter_excerpt( $content ) {
return $this->embed_rss( $content, 'excerpt' );
}
/**
* Adds the RSS footer and/or header to an RSS feed item.
*
* @since 1.4.14
*
* @param string $content Feed item content.
* @param string $context Feed item context, either 'excerpt' or 'full'.
*
* @return string
*/
public function embed_rss( $content, $context = 'full' ) {
/**
* Filter: 'wpseo_include_rss_footer' - Allow the RSS footer to be dynamically shown/hidden.
*
* @api boolean $show_embed Indicates if the RSS footer should be shown or not.
*
* @param string $context The context of the RSS content - 'full' or 'excerpt'.
*/
if ( ! apply_filters( 'wpseo_include_rss_footer', true, $context ) ) {
return $content;
}
if ( is_feed() ) {
$before = '';
$after = '';
if ( WPSEO_Options::get( 'rssbefore', '' ) !== '' ) {
$before = wpautop( $this->rss_replace_vars( WPSEO_Options::get( 'rssbefore' ) ) );
}
if ( WPSEO_Options::get( 'rssafter', '' ) !== '' ) {
$after = wpautop( $this->rss_replace_vars( WPSEO_Options::get( 'rssafter' ) ) );
}
if ( $before !== '' || $after !== '' ) {
if ( ( isset( $context ) && $context === 'excerpt' ) && trim( $content ) !== '' ) {
$content = wpautop( $content );
}
$content = $before . $content . $after;
}
}
return $content;
}
/**
* Used in the force rewrite functionality this retrieves the output, replaces the title with the proper SEO
* title and then flushes the output.
*/
public function flush_cache() {
global $wp_query;
if ( $this->ob_started !== true ) {
return false;
}
$content = ob_get_clean();
$old_wp_query = $wp_query;
wp_reset_query();
// Only replace the debug marker when it is hooked.
if ( $this->show_debug_marker() ) {
$title = $this->title( '' );
$debug_mark = $this->get_debug_mark();
// Find all titles, strip them out and add the new one in within the debug marker, so it's easily identified whether a site uses force rewrite.
$content = preg_replace( '/<title.*?\/title>/i', '', $content );
$content = str_replace( $debug_mark, $debug_mark . "\n" . '<title>' . esc_html( $title ) . '</title>', $content );
}
$GLOBALS['wp_query'] = $old_wp_query;
echo $content;
return true;
}
/**
* Starts the output buffer so it can later be fixed by flush_cache().
*/
public function force_rewrite_output_buffer() {
$this->ob_started = true;
ob_start();
}
/**
* Function used in testing whether the title should be force rewritten or not.
*
* @param string $title Title string.
*
* @return string
*/
public function title_test_helper( $title ) {
WPSEO_Options::set( 'title_test', ( WPSEO_Options::get( 'title_test' ) + 1 ) );
// Prevent this setting from being on forever when something breaks, as it breaks caching.
if ( WPSEO_Options::get( 'title_test' ) > 5 ) {
WPSEO_Options::set( 'title_test', 0 );
remove_filter( 'wpseo_title', array( $this, 'title_test_helper' ) );
return $title;
}
if ( ! defined( 'DONOTCACHEPAGE' ) ) {
define( 'DONOTCACHEPAGE', true );
}
if ( ! defined( 'DONOTCACHCEOBJECT' ) ) {
define( 'DONOTCACHCEOBJECT', true );
}
if ( ! defined( 'DONOTMINIFY' ) ) {
define( 'DONOTMINIFY', true );
}
if ( $_SERVER['HTTP_USER_AGENT'] === "WordPress/{$GLOBALS['wp_version']}; " . get_bloginfo( 'url' ) . ' - Yoast' ) {
return 'This is a Yoast Test Title';
}
return $title;
}
/**
* Get the product name in the head section.
*
* @return string
*/
private function head_product_name() {
if ( $this->is_premium() ) {
return 'Yoast SEO Premium plugin';
}
return 'Yoast SEO plugin';
}
/**
* Check if this plugin is the premium version of WPSEO.
*
* @return bool
*/
protected function is_premium() {
return WPSEO_Utils::is_yoast_seo_premium();
}
/**
* Check if term archive query is for multiple terms (/term-1,term2/ or /term-1+term-2/).
*
* @return bool
*/
protected function is_multiple_terms_query() {
global $wp_query;
if ( ! is_tax() && ! is_tag() && ! is_category() ) {
return false;
}
$term = get_queried_object();
$queried_terms = $wp_query->tax_query->queried_terms;
if ( empty( $queried_terms[ $term->taxonomy ]['terms'] ) ) {
return false;
}
return count( $queried_terms[ $term->taxonomy ]['terms'] ) > 1;
}
/**
* Wraps wp_safe_redirect to allow testing for redirects.
*
* @param string $location The path to redirect to.
* @param int $status Status code to use.
*/
public function redirect( $location, $status = 302 ) {
wp_safe_redirect( $location, $status );
exit;
}
/**
* Checks if the debug mark action has been added.
*
* @return bool True when the action exists.
*/
protected function show_debug_marker() {
return has_action( 'wpseo_head', array( $this, 'debug_mark' ) ) !== false;
}
/**
* Shows the closing debug mark.
*
* @return string The closing debug mark comment.
*/
protected function show_closing_debug_mark() {
if ( ! $this->show_debug_marker() ) {
return '';
}
return sprintf(
"<!-- / %s. -->\n\n",
esc_html( $this->head_product_name() )
);
}
/**
* Builds the title for a post type archive.
*
* @param string $separator The title separator.
* @param string $separator_location The location of the title separator.
*
* @return string The title to use on a post type archive.
*/
protected function get_post_type_archive_title( $separator, $separator_location ) {
$post_type = $this->get_queried_post_type();
$title = $this->get_title_from_options( 'title-ptarchive-' . $post_type );
if ( ! is_string( $title ) || '' === $title ) {
$post_type_obj = get_post_type_object( $post_type );
$title_part = '';
if ( isset( $post_type_obj->labels->menu_name ) ) {
$title_part = $post_type_obj->labels->menu_name;
}
elseif ( isset( $post_type_obj->name ) ) {
$title_part = $post_type_obj->name;
}
$title = $this->get_default_title( $separator, $separator_location, $title_part );
}
return $title;
}
/**
* Retrieves the queried post type.
*
* @return string The queried post type.
*/
protected function get_queried_post_type() {
$post_type = get_query_var( 'post_type' );
if ( is_array( $post_type ) ) {
$post_type = reset( $post_type );
}
return $post_type;
}
/**
* Retrieves the SEO Meta value for the supplied key and optional post.
*
* @param string $key The key to retrieve.
* @param int $post_id Optional. The post to retrieve the key for.
*
* @return string Meta value.
*/
protected function get_seo_meta_value( $key, $post_id = 0 ) {
return WPSEO_Meta::get_value( $key, $post_id );
}
/**
* Replaces the dynamic variables in a string.
*
* @param string $string The string to replace the variables in.
* @param array $args The object some of the replacement values might come from,
* could be a post, taxonomy or term.
* @param array $omit Variables that should not be replaced by this function.
*
* @return string The replaced string.
*/
protected function replace_vars( $string, $args, $omit = array() ) {
$replacer = new WPSEO_Replace_Vars();
return $replacer->replace( $string, $args, $omit );
}
/** Deprecated functions */
// @codeCoverageIgnoreStart
/**
* Outputs or returns the debug marker, which is also used for title replacement when force rewrite is active.
*
* @deprecated 4.4
*
* @param bool $echo Whether or not to echo the debug marker.
*
* @return string
*/
public function debug_marker( $echo = false ) {
if ( function_exists( 'wp_get_current_user' ) && current_user_can( 'manage_options' ) ) {
_deprecated_function( 'WPSEO_Frontend::debug_marker', '4.4', 'WPSEO_Frontend::debug_mark' );
}
return $this->debug_mark( $echo );
}
/**
* Outputs the meta keywords element.
*
* @deprecated 6.3
*
* @return void
*/
public function metakeywords() {
if ( function_exists( 'wp_get_current_user' ) && current_user_can( 'manage_options' ) ) {
_deprecated_function( 'WPSEO_Frontend::metakeywords', '6.3' );
}
}
/**
* Removes unneeded query variables from the URL.
*
* @deprecated 7.0
*
* @return void
*/
public function clean_permalink() {
// As this is a frontend method, we want to make sure it is not displayed for non-logged in users.
if ( function_exists( 'wp_get_current_user' ) && current_user_can( 'manage_options' ) ) {
_deprecated_function( 'WPSEO_Frontend::clean_permalink', '7.0' );
}
}
/**
* Trailing slashes for everything except is_single().
*
* @deprecated 7.0
*/
public function add_trailingslash() {
// As this is a frontend method, we want to make sure it is not displayed for non-logged in users.
if ( function_exists( 'wp_get_current_user' ) && current_user_can( 'manage_options' ) ) {
_deprecated_function( 'WPSEO_Frontend::add_trailingslash', '7.0', null );
}
}
/**
* Removes the ?replytocom variable from the link, replacing it with a #comment-<number> anchor.
*
* @deprecated 7.0
*
* @param string $link The comment link as a string.
*
* @return string The modified link.
*/
public function remove_reply_to_com( $link ) {
_deprecated_function( 'WPSEO_Frontend::remove_reply_to_com', '7.0', 'WPSEO_Remove_Reply_To_Com::remove_reply_to_com' );
$remove_replytocom = new WPSEO_Remove_Reply_To_Com();
return $remove_replytocom->remove_reply_to_com( $link );
}
/**
* Redirects out the ?replytocom variables.
*
* @deprecated 7.0
*
* @return boolean True when redirect has been done.
*/
public function replytocom_redirect() {
_deprecated_function( 'WPSEO_Frontend::replytocom_redirect', '7.0', 'WPSEO_Remove_Reply_To_Com::replytocom_redirect' );
$remove_replytocom = new WPSEO_Remove_Reply_To_Com();
return $remove_replytocom->replytocom_redirect();
}
// @codeCoverageIgnoreEnd
}
| {
"redpajama_set_name": "RedPajamaGithub"
} | 1,277 |
\section{\@ifstar{\origsection*}{\@startsection{section}{1}\z@{.7\linespacing\@plus\linespacing}{.5\linespacing}{\normalfont\scshape\centering\S}}}
\def\@startsection{section}{1}\z@{.7\linespacing\@plus\linespacing}{.5\linespacing}{\normalfont\scshape\centering\S}{\@startsection{section}{1}\z@{.7\linespacing\@plus\linespacing}{.5\linespacing}{\normalfont\scshape\centering\S}}
\makeatother
\usepackage{amsmath,amssymb,amsthm}
\usepackage{mathabx}\usepackage{wasysym}
\usepackage{mathrsfs}
\usepackage{bbm, accents}
\usepackage{xcolor}
\usepackage[backref]{hyperref}
\hypersetup{
colorlinks,
linkcolor={red!60!black},
citecolor={green!60!black},
urlcolor={blue!60!black},
}
\usepackage{tikz}
\usetikzlibrary{calc,decorations.pathmorphing}
\usepackage{bookmark}
\usepackage[abbrev,msc-links,backrefs]{amsrefs}
\usepackage{doi}
\renewcommand{\doitext}{DOI\,}
\renewcommand{\PrintDOI}[1]{\doi{#1}}
\renewcommand{\eprint}[1]{\href{http://arxiv.org/abs/#1}{arXiv:#1}}
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage[english]{babel}
\usepackage{microtype}
\usepackage{graphicx} \usepackage{caption}
\usepackage{subcaption}
\def\geqslant{\geqslant}
\def\leqslant{\leqslant}
\def\EE#1{E(#1)}
\def\VV#1{V(#1)}
\let\setminus=\smallsetminus
\let\sm=\setminus
\newtheorem{fact}{Fact}\newtheorem{thm}[fact]{Theorem}
\newtheorem{cor}[fact]{Corollary}
\linespread{1.3}
\usepackage{geometry}
\geometry{left=27.5mm,right=27.5mm, top=25mm, bottom=25mm}
\numberwithin{equation}{section}
\begin{document}
\title[Decomposing graphs into forests]
{Nash-Williams' theorem on decomposing graphs into forests}
\author{Christian Reiher}
\address{Fachbereich Mathematik, Universit\"at Hamburg, Bundesstra\ss e 55, 20144 Hamburg, Germany}
\email{Christian.Reiher@uni-hamburg.de}
\author{Lisa Sauermann}
\address{Department of Mathematics, Stanford University, 450 Serra Mall,
Building 380, Stanford CA 94305, USA}
\email{lsauerma@stanford.edu}
\begin{abstract}
We give a simple graph-theoretic proof of a classical result due to {\sc C. St. J. A. Nash-Williams} on covering graphs by forests. Moreover we derive a slight generalisation of this statement where some edges are preassigned to distinct forests.
\end{abstract}
\keywords{Nash-William's theorem, forest covering, arboricity}
\maketitle
\section{Introduction}\label{sec:intro}
All graphs considered in this note are finite. The sets of vertices and edges of a graph
$G$ are denoted by $V(G)$ and $E(G)$, respectively.
The {\it restriction} of a graph $G=(V, E)$ to a subset $X$ of $V$, i.e., the graph
on $X$ whose edges are precisely those members of $E$ both of whose ends belong to $X$,
is indicated by $G|X$. When $G$ is clear from the context, we write $e(X)$ for the number
of edges of that graph. For any integer $r\ge 0$ we set $[r]=\{1, 2, \ldots, r\}$.
The following result is due to {\sc C. St. J. A. Nash--Williams} (see \cite{NW2}, and the
related articles \cites{NW1, Tut} as well as \cite{China} for another simple proof).
\begin{thm} \label{thm:1} If $G=(V, E)$ is a graph, and $r\ge 0$ is an integer such that for all
nonempty subsets $X$ of $V$ one has $e(X)\le r(|X|-1)$, then there exists a
partition $E=E_1\cup E_2\cup\ldots\cup E_r$ such that $(V, E_i)$ is a forest for
$i\in[r]$.
\end{thm}
It is plain that the sufficient condition for such a partition to exist given here is
also necessary. Also, the cases $r=0$ and $r=1$ of this statement are immediate, and the
case $r=2$ was recently posed at the All-Russian Mathematical Olympiad, \cite{Russen}.
That case can be dealt with by some peculiar tricks not discussed here and that do not
straightforwardly generalize to $r>2$, but which nevertheless motivated us to reprove
the general case independently. In fact, it turned out that our arguments for the case
$r=2$ yielded slightly more, namely that for any two distinct edges of $G$ there
exists such a partition in which one of the edges belongs to $E_1$ while the other one
belongs to $E_2$. Hence one might guess:
\begin{cor}\label{cor:2}
Given a graph $G=(V, E)$, an integer $r\geqslant 0$ such that for all nonempty
subsets $X$ of $V$ one has $e(X)\leqslant r(|X|-1)$, and moreover a sequence
$e_1, e_2, \ldots, e_{r}$ of distinct edges of $G$, there exists a partition
$E=E_1\cup E_2\cup\ldots\cup E_r$ such that $e_i\in E_i$ for $i\in[r]$ and $(V, E_i)$
is a forest for all $i\in[r]$.
\end{cor}
As we shall see in Section~\ref{sec:3}, this can in fact be derived from Theorem~\ref{thm:1}.
All statements and arguments contained in this article are valid irrespective of whether
multiple edges are allowed to occur in our graphs or not.
\section{Proving Theorem~\ref{thm:1}} \label{sec:2}
In this section we give a simple proof of Theorem~\ref{thm:1} that is, to the best of our
knowledge, new. For this purpose we need some preparation. Let $T$ be a
forest, $k\geq 2$ an integer, and $T_1, T_2, \ldots ,T_k$
mutually vertex disjoint, connected subgraphs of $T$.
Obviously $T_1, T_2, \ldots T_k$ are trees and each of them
is contained in exactly one component of $T$. We call $T_i$
{\it isolated} in $T$ if there is no $T_j$ with $j\neq i$, that is
contained in the same component of $T$ as $T_i$. Furthermore we call
$T_i$ {\it peculiar} in $T$ if there is an edge
$e_i\in E(T)\sm\bigl(E(T_1)\cup \dots \cup E(T_k)\bigr)$ incident with a
vertex of $T_i$, such that $T_i$ is isolated in $T-e_i$
(the reader may notice that one gets an equivalent notion without demanding $e_i$ to be
incident with a vertex of $T_i$).
\begin{fact}\label{fact:3}
At least two of the subgraphs $T_1$, $T_2$, \dots ,$T_k$
are isolated or peculiar in $T$.
\end{fact}
\begin{proof}
Otherwise take a counterexample where $T$ has as few vertices as possible.
If no component of $T$ contains two or more of the $T_i$,
then each them is isolated and we are done. In the remaining cases $T$
is a tree. Consider any leaf $x$ of $T$. As we cannot produce a
smaller counterexample by deleting $x$ thus, there has to be some $T_i$ consisting
solely of~$x$, and the edge of $T$ incident with $x$ witnesses that this
$T_i$ is peculiar. Because of $k\geq 2$ the tree $T$ has at least
two vertices and, consequently, at least two leaves. Applying the foregoing argument to
any two leaves of $T$ we see that at least two of the trees
$T_1$, $T_2$, \dots , and $T_k$ are peculiar.
\end{proof}
\begin{proof}[Proof of Theorem~\ref{thm:1}.]
Arguing indirectly we choose a graph $G=(V,E)$ and an integer $r\geqslant 0$
contradicting Theorem~\ref{thm:1} with $\vert E\vert$ minimal. Then $\vert E\vert\neq 0$.
Let $e=ab$ be an arbitrary edge of $G$.
Because of the choice of $G$, there is at least one partition
$E\sm\{e\}=E_1'\cup E_2'\cup \dots \cup E_r'$ such that $(V, E_i')$ is a forest for
$i=1,2,\dots ,r$. For each of these partitions we consider the component $(C, E_C)$
of $(V, E_1')$ containing the vertex $a$.
From now on let
\[
E\sm\lbrace e\rbrace=E_1'\cup E_2'\cup \dots \cup E_r'
\]
be one of these
partitions with $\vert C\vert$ minimum. Let $\overline{C}=\EE{G\vert C}$.
If $b\not\in C$, the partition $E=(E_1'\cup \lbrace e\rbrace)\cup E_2'\cup \dots \cup E_r'$
would satisfy all conditions of Theorem~\ref{thm:1}, consequently $b\in C$ and $e\in \overline{C}$.
Therefore
\[
\vert E_1'\cap \overline{C}\vert+\dots+\vert E_r'\cap \overline{C}\vert
<\vert \overline{C}\vert=e(C)\leqslant r(\vert C\vert-1)\,.
\]
Thus, there is an $i\in [r]$ with $\vert E_i'\cap \overline{C}\vert<\vert C\vert-1$.
This implies, that $(C, E_i'\cap \overline{C})$ is not connected.
Because of the definition of $(C, E_C)$ we have $i\neq 1$, so we can w.l.o.g. assume~$i=2$.
Let $D_1$, \dots, $D_k$ be the connected components of
$(C, E_2'\cap \overline{C})$, where obviously $k\geq 2$. Thus,
$D_1$, \dots, $D_k$ are mutually vertex disjoint, connected subgraphs
of the forest $(V, E_2')$. We define isolation and peculiarity in $(V, E_2')$
as applying to these subgraphs. By Fact~\ref{fact:3} at least two of the subgraphs
$D_1$, \dots, $D_k$ are isolated or peculiar in $(V, E_2')$, and at
most one of them contains $a$. Let w.l.o.g. $D_1$ be isolated or peculiar
in $(V, E_2')$ and $a\not\in \VV{D_1}$.
If $D_1$ is peculiar in $(V, E_2')$, there is an edge
\[
e_1\in E_2'\sm\bigl(\EE{D_1}\cup \dots \cup \EE{D_k}\bigr)
\]
incident with a vertex $v_1$ of $D_1$, such that $D_1$ is
isolated in $(V, E_2'\sm\lbrace e_1\rbrace)$. Notice that
\[
\EE{D_1}\cup \dots \cup \EE{D_k}=\overline{C}\cap E_2'
\]
yields $e_1\not\in \overline{C}$, wherefore $e_1$ connects $v_1$ and a vertex not in $C$.
If $D_1$ is isolated in $(V, E_2')$, let $v_1$ be an arbitrary vertex of $D_1$.
We consider the uniquely determined path from $a$ to $v_1$ in the tree $(C, E_C)$.
Since $a\not\in \VV{D_1}$ and $v_1\in \VV{D_1}$, this path
contains an edge $e_d$ connecting a vertex of $D_1$ with a vertex of some
$D_i$ with $i\neq 1$.
First, we consider the case when $D_1$ is isolated in $(V, E_2')$.
Then the graph $(V, E_2'\cup \lbrace e_d\rbrace)$ is a forest, because $e_d$
connects different components of $(V, E_2')$. Obviously, the graph
$(V, E_1'\sm \lbrace e_d\rbrace)$ is also a forest and its component including $a$
is a subgraph of $(C, E_C)$. But this subgraph does not contain $v_1$ and has
therefore a number of vertices smaller than $\vert C\vert$. This contradicts
\[
E\sm\lbrace e\rbrace=(E_1'\sm \lbrace e_d\rbrace)\cup (E_2'
\cup \lbrace e_d\rbrace)\cup \dots \cup E_r'
\]
being one of the partitions considered at the beginning.
For the second case let $D_1$ now be peculiar in $(V, E_2')$. Then $D_1$ is isolated in
$(V, E_2'\sm \lbrace e_1\rbrace)$ and the graph
$\bigl(V, (E_2'\sm\lbrace e_1\rbrace)\cup \lbrace e_d\rbrace\bigr)$ is therefore a forest.
On the other hand the graph $(V, E_1'\sm\lbrace e_d\rbrace)$ is also a forest and a subgraph
of the forest $(V, E_1')$. Because $e_d$ belongs to the component $(C, E_C)$ of $(V, E_1')$,
the graph $(V, E_1'\sm\lbrace e_d\rbrace)$ has two components being subgraphs of $(C, E_C)$,
one of which contains $a$ and the other one $v_1$. The edge $e_1$ connects~$v_1$ and a vertex
not in $C$, consequently the graph
$\bigl(V, (E_1'\sm\lbrace e_d\rbrace)\cup \lbrace e_1\rbrace\bigr)$ is
a forest and its component including $a$ is equal to the component of
$(V, E_1'-\lbrace e_d\rbrace)$ including $a$.
Thus, its number of vertices is smaller than $\vert C\vert$.
This contradicts
\[
E\sm\lbrace e\rbrace=\bigl((E_1'\sm \lbrace e_d\rbrace)\cup \lbrace e_1\rbrace\bigr)
\cup \bigl((E_2'\sm\lbrace e_1\rbrace)\cup \lbrace e_d\rbrace\bigr)
\cup \dots \cup E_r'
\]
being one of the partitions considered at the beginning.
\begin{figure}[ht]
\centering
\begin{tikzpicture}[scale=1.2]
\coordinate (x) at (0.1,1);
\coordinate (y) at (0, 5);
\coordinate (r1) at (2, 5.5);
\coordinate (r2) at (2, 4.5);
\coordinate (r3) at (5, 5.5);
\coordinate (r4) at (5.8,4.5);
\coordinate (r5) at (1,1.4);
\coordinate (a) at (1,0.6);
\coordinate (r6) at (2.2,1.5);
\coordinate (r7) at (2.1,0.8);
\coordinate (r8) at (4,0.6);
\coordinate (r9) at (5.6,0.9);
\coordinate (v1) at (5.2,1.5);
\draw[line width=1pt] (y) -- (r2);
\draw[line width=1pt] (r1) -- (r3);
\draw[line width=1pt] (r2) -- (r3);
\draw[line width=1pt] (r4) -- (r2);
\draw[line width=1pt] (x) -- (r5) -- (a) -- (r7) -- (r6);
\draw[line width=1pt] (r7) -- (r8) -- (r9);
\draw[line width=1pt] (r8) -- (v1);
\draw[green, line width=1pt] (y) -- (r1) -- (r2);
\draw[green, line width=1pt] (r1) -- (r4) -- (r3);
\draw[green, line width=1pt] (r9) -- (v1) -- (r6) -- (r8);
\draw[green, line width=1pt] (a) -- (x) -- (y);
\draw[green, line width=1pt] (r5) -- (r6);
\draw[green, line width=1pt] (v1) -- (r4);
\draw[green, line width=1pt] (x) -- (r7);
\foreach \i in {x, y, r1, r2, r3, r4, r5, a, r6, r7, r8, r9, v1}
\fill (\i) circle (2pt);
\node at (5.9,3) {$e_1$};
\node at (3.2, 0.3) {$e_d$};
\node at (1,0.2) {$a$};
\node at (5.6, 1.6) {$v_1$};
\node at (-1.3, 3) {$C$};
\end{tikzpicture}
\caption{$D_1$ is peculiar}
\label{fig:1}
\end{figure}
Since we have obtained a contradiction in each of the two cases, our assumption must
have been wrong and Theorem~\ref{thm:1} is true.
\end{proof}
\section{Deducing Corollary~\ref{cor:2}} \label{sec:3}
The strengthening given by Corollary~\ref{cor:2} will now be deduced from
Theorem~\ref{thm:1} by means of a short argument.
\begin{proof}[Proof of Corollary~\ref{cor:2}] Let $G=(V,E)$, $r\ge 0$, and $e_1, \dots ,e_r$
be as in Corollary~\ref{cor:2}. We call an integer $0\leqslant r'\leqslant r$ {\it restrained}
if there is a partition $E=E_1\cup E_2\cup \dots \cup E_r$ such that
$e_i\in E_i$ for $i\in [r']$ and $(V, E_i)$ is acyclic for all $i\in [r]$.
By Theorem~\ref{thm:1} the integer $0$ is restrained.
Corollary~\ref{cor:2} is equivalent to $r$ being restrained.
It is therefore sufficient to prove the following: if an integer $k$ with $0\leqslant k\leqslant r-1$
is restrained, then the integer $k+1$ is also restrained.
Let $E=E_1\cup E_2\cup \dots \cup E_r$ be a partition such that $e_i\in E_i$ for $i\in [k]$
and $(V, E_i)$ is acyclic for all $i\in [r]$. If $e_{k+1}\in E_{k+1}$, we are done.
We can therefore assume $e_{k+1}\in E_\ell$ with $\ell\neq k+1$. Obviously we can assume that
the two vertices of $e_{k+1}$ belong to the same component of $(V, E_{k+1})$.
The two vertices of $e_{k+1}$ belong to different components of the forest
$(V, E_\ell\sm\lbrace e_{k+1}\rbrace)$. Therefore the uniquely determined path between
these two vertices in the forest $(V, E_{k+1})$ contains vertices of different
components of $(V, E_\ell\sm\lbrace e_{k+1}\rbrace)$. Hence, there is an edge $e\in E_{k+1}$
of this path connecting vertices of different components of $(V, E_\ell\sm\lbrace e_{k+1}\rbrace)$.
Then the graph $\bigl(V, (E_\ell\sm\lbrace e_{k+1}\rbrace)\cup \lbrace e\rbrace\bigr)$ is a forest.
On the other hand the graph
$\bigl(V, (E_{k+1}\sm\lbrace e\rbrace)\cup \lbrace e_{k+1}\rbrace\bigr)$
is by the definition of $e$ also a forest.
\begin{figure}[ht]
\centering
\begin{tikzpicture}[scale=1.4]
\coordinate (r1) at (0, 1);
\coordinate (r2) at (0.6, 2);
\coordinate (r3) at (1, 0);
\coordinate (r4) at (1.9,1.4);
\coordinate (r5) at (1.5,2.8);
\coordinate (r6) at (5.6,0);
\coordinate (r7) at (5.5,1.4);
\coordinate (r8) at (6.2,1.7);
\coordinate (r9) at (4.8,2.6);
\coordinate (r10) at (3,3.2);
\coordinate (r11) at (4.7,3.5);
\draw[line width=1pt] (r3) -- (r4) -- (r5) -- (r10) -- (r9) -- (r7) -- (r6);
\draw[green, line width=1pt] (r9) -- (r8) -- (r6) -- (r3) -- (r1) -- (r2) -- (r4);
\draw[green, line width=1pt] (r2) -- (r5);
\draw[green, line width=1pt] (r10) -- (r11);
\draw[green, line width=1pt] (r7) -- (r8);
\foreach \i in {r1, r2, r3, r4, r5, r6, r7, r8, r9, r10, r11}
\fill (\i) circle (2pt);
\node at (2.4,3.3) {$e$};
\node at (3.7, -0.4) {$e_{k+1}$};
\end{tikzpicture}
\caption{Switching $e$ and $e_{k+1}$}
\label{fig:2}
\end{figure}
Thus, the partition gained from $E=E_1\cup E_2\cup \dots \cup E_r$ by substituting
$E_\ell$ by ${(E_\ell\sm\lbrace e_{k+1}\rbrace )\cup \lbrace e\rbrace}$ and $E_{k+1}$ by
$(E_{k+1}\sm\lbrace e\rbrace)\cup \lbrace e_{k+1}\rbrace$ fulfills all conditions for
$k+1$ being restrained.
\end{proof}
\begin{bibdiv}
\begin{biblist}
\bib{China}{article}{
author={Chen, Boliong},
author={Matsumoto, Makoto},
author={Wang, Jian Fang},
author={Zhang, Zhong Fu},
author={Zhang, Jian Xun},
title={A short proof of Nash-Williams' theorem for the arboricity of a
graph},
journal={Graphs Combin.},
volume={10},
date={1994},
number={1},
pages={27--28},
issn={0911-0119},
review={\MR{1273008}},
doi={10.1007/BF01202467},
}
\bib{NW1}{article}{
author={Nash-Williams, C. St. J. A.},
title={Edge-disjoint spanning trees of finite graphs},
journal={J. London Math. Soc.},
volume={36},
date={1961},
pages={445--450},
issn={0024-6107},
review={\MR{0133253}},
}
\bib{NW2}{article}{
author={Nash-Williams, C. St. J. A.},
title={Decomposition of finite graphs into forests},
journal={J. London Math. Soc.},
volume={39},
date={1964},
pages={12},
issn={0024-6107},
review={\MR{0161333}},
}
\bib{Tut}{article}{
author={Tutte, W. T.},
title={On the problem of decomposing a graph into $n$ connected factors},
journal={J. London Math. Soc.},
volume={36},
date={1961},
pages={221--230},
issn={0024-6107},
review={\MR{0140438}},
}
\bib{Russen}{webpage}{
url={http://www.artofproblemsolving.com/Forum/resources.php?c=143\&cid=61\&year=2007},
}
\end{biblist}
\end{bibdiv}
\end{document} | {
"redpajama_set_name": "RedPajamaArXiv"
} | 3,459 |
Home ► Report
Libya, 60 people killed in air raid by Haftar
Struck a center for migrants next to a military base
At least 60 people died this night in the air raid that the forces of General Khalifa Haftar launched on a detention center for migrants in Tagiura, near Tripoli. Others 80 were injured. The news comes from Libyan emergency sources. The attack was decided after losing of city of Gharian, where Haftar had made its operational base and reconquered by the Tripoli government last week. The center is located near the Dhaman military base, one of the places where Misurata militias and those loyal to the government of President Fayez al-Serraj concentrated their ammunition and vehicle reserves to stop the assault on Tripoli started by the General of Cyrenaica since 4 April.
"I knew with dismay the nightly bombardment in Tagiura, near Tripoli, which hit a center for migrants, causing the death of dozens of people, including women and children. A further tragedy that shows the atrocious impact of the war on civilian population", commented Italian Foreign Minister Enzo Moavero Milanesi. The minister expressed "the clear condemnation of indiscriminate bombardment on civilian areas and the appeal to stop a worsening of hostilities that continually puts in serious danger human lives and destroys essential infrastructures for the population. It is immediately necessary to guarantee serious protection measures for civilians and, in particular, to transfer migrants who are in collection facilities into places that are safe from fighting and under the protection of the United Nations".
DDB - 1222757
Tripoli, Libya, 07/03/2019 09:32
ReportLibya, Haftar launched raid on Mitiga airport
Hit the drone control room. Suspended flights
The battle for Tripoli is continuing unabated. Forces loyal to General Khalifa Haftar, accused of the air raid on a detention center for migrants that caused dozens of victims (see AVIONEWS), bombed Mitiga... more
Science and technologyAirbus demonstrates fully automatic vision-based take-off
The crew comprises of pilots, flight test engineers and test flight engineer
Airbus has successfully performed the first fully automatic vision-based take-off using an its Family test aircraft at Toulouse-Blagnac airport. The test crew comprising of two pilots, two flight test... more | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 1,217 |
Q: Proper settings for plot.estimateEffect in stm package The stm package provides an indispensable set of tools for estimating the effect of covariates on topic prevalence. The plot.estimateEffect() function in particular displays helpful pointrange plots to visualize this effect. However, I find the documentation for plot.estimateEffect() confusing.
This is a silly question, but I want to make absolutely sure I understand this right. Suppose I have a binary covariate ("cov") of {0,1} that I want to use to predict topic prevalence. I input the following arguments:
covariate = "cov"
cov.value1 = 1
cov.value2 = 0
The documentation says to "set the treatment to cov.value1." I do so: in this case, "1" is the treatment, "0" is the control.
The numeric labels are confusing me, since my covariate is also numeric.
When I do this without a custom topic label, the default labels read "Topic X (Covariate Level 1 Compared to 0)", which seems to imply that "1" refers to the left-hand side (and vice versa for "0"). This would seem to confirm that the left-hand side of the x-axis (negative numbers) should represent topic prevalence when cov=1, and the right-hand side (positive numbers) represent cov=0.
However, the sign of the point estimates seems wrong to me with these settings.
When cov.value1 is set to 1 and cov.value2 is set to 0, Topic 1 (which contains a higher proportion of "xxxx" and "yyyy") is right-shifted (toward what the documentation states is the control condition). But given the values of "cov", Topic 1 should appear more often in the treatment condition (when cov=1). ["xxxx" and "yyyy" appear more often when cov=1 (1/2+1/3+1/3) than when cov=0 (1/4+1/2+1/3+1/3).]
I guess the confusion stems from the label names (of what is otherwise a fantastic package). I would find this much more straightforward if the relevant arguments were labeled treatment.value and control.value, for example.
Intuitively I would have expected the plot to show the effect as we move from control to treatment, not treatment to control. Is the documentation right? Does the left-hand side represent treatment (cov.value1), and the right-hand side control (cov.value2)? If so, please help me understand why the sign is in the unexpected direction.
MWE:
library(quanteda)
library(stm)
text_field <- c("xxxx kjbregioberg", "yyyy owbhgoer ergo ergu", "xxxx yyyy fwkbekfbk ew", "xxxx wf ibgfbgr ", "yyyy gf gds", "yyyy frgg fgfg", "yyyy eee ef ")
cov <- c(1, 0, 0, 1, 1, 0, 0)
corp <- as.data.frame(cbind(text_field, cov))
corp$cov <- as.numeric(corp$cov)
corp$text_field <- as.character(corp$text_field)
corp <- corpus(corp, text_field = "text_field")
dfm <- dfm(corp)
dfm <- convert(dfm, to = "stm", docvars = docvars(dfm))
stm_object <- stm(documents = dfm$documents,
vocab = dfm$vocab,
data = dfm$meta,
prevalence = ~ cov,
K = 2, seed = 1, init.type = "Spectral")
topics_outcome <- 1:2
stm_effects <- estimateEffect(topics_outcome ~ cov, # +
stmobj = stm_object,
meta = dfm$meta,
uncertainty = "Global")
plot.estimateEffect(stm_effects,
covariate = "cov",
topics = topics_outcome,
model = stm_object,
method = "difference",
cov.value1 = 0,
cov.value2 = 1,
xlim = c(-1, 1),
cex = 1.5,
main = "",
xlab = "0 1"
)
A: I found this. It may helps
cov.value1 For method "difference", the value or set of values of interest at which to set the covariate. In the case of calculating a treatment/control contrast, set the treatment to cov.value1.
cov.value2 For method "difference", the value or set of values which will be set as the comparison
group. cov.value1 and cov.value2 must be vectors of the same length.
Find more https://cran.r-project.org/web/packages/stm/stm.pdf
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 6,993 |
\section{Introduction}
Most stars form in dense embedded clusters and reside in binary
systems (either primordial or tidal). If both binary components are
massive enough, they end their lives as core-collapsed supernovae
(SNe). Stellar remnants of SN explosions, usually the neutron stars
(NSs), have peculiar velocities at least an order of magnitude
higher than the typical velocities of their progenitors, the OB
stars (e.g. Gunn \& Ostriker \cite{gun70}). It is believed that the
high velocities of NSs are due either to the asymmetry of SN
explosions (e.g. Dewey \& Cordes \cite{dew87}) or to the disruption
of tight massive binaries following the second (symmetric) SN
explosion (e.g. Iben \& Tutukov \cite{ibe96}). The progress in
measuring the proper motions and parallaxes of NSs (pulsars) allows
their peculiar (transverse) velocities to be determined with high
precision and makes it possible to trace their trajectories back to
the parent star clusters (e.g. Hoogerwerf et al. \cite{hoo01};
Chatterjee et al. \cite{cha05}). Recently Vlemmings et al.
(\cite{vle04}; hereafter VCC04) have used the high-precision
astrometric data (proper motions and parallaxes) for two dozen
pulsars to determine their trajectories in the Galactic potential
and to search for pairs with a common origin. They discovered that
two pulsars from their sample (presently separated by $\sim
23^{\degr}$) originated within several parsecs of each other in the
direction of the \object{Cyg OB2} association. VCC04 interpret their
discovery as an indication that the progenitors of both pulsars,
\object{B2020+28} and \object{B2021+51}, were the members of a
common massive binary and suggest that the pulsars were separated at
the birth of the second one following the asymmetric SN explosion.
In this Letter we explore a different scenario for the origin of
B2020+28 and B2021+51. We suggest that these pulsars were separated
before their birth and that they are the remnants of runaway stars
ejected (with velocities similar to those of the pulsars) from the
parent star cluster due to the strong three- or four-body dynamical
encounters. Our scenario does not require any asymmetry in SN
explosions.
\section{Pulsars B2020+28 and B2021+51: origin in a common binary}
The main result presented in VCC04 is that B2020+28 and B2021+51
originated within several parsecs of each other. VCC04 derived the
most likely three-dimensional peculiar velocities of the pulsars at
birth, $\simeq 150$ and $\simeq 500 \, {\rm km} \, {\rm s}^{-1}$
(respectively, for B2021+51 and B2020+28), and the angle between the
velocity vectors $\psi \simeq 160^{\degr}$. These velocities can, in
principle, be produced via disintegration of a tight (semi-detached)
massive binary after the second (symmetric) SN explosion (e.g. Iben
\& Tutukov \cite{ibe96}); in this case, however, $\psi$ is always $<
90^{\degr}$. Moreover, the spin characteristics of B2020+28 and
B2021+51 (typical of non-recycled pulsars) argue against the origin
of these pulsars in a common {\it tight} binary (cf. VCC04).
One possible way to reconcile the kinematic data with the common
binary scenario is to assume that the binary was disrupted either
after the first or the second {\it asymmetric} SN explosion (VCC04).
Note that the similarity between the pulsar's characteristic ages
($\simeq 2.88$ and $\simeq 2.75$ Myr) implies that the mass ratio of
the binary components was $\sim 1$. Therefore, depending on the
initial parameters (binary separation, etc), the binary system at
the moment of the first SN explosion consists of two red supergiant
or Wolf-Rayet stars or of two carbon-oxygen (CO) cores.
The latter situation can be realised if the massive binary evolves
through two common-envelope phases (see Belczy\'{n}ski \& Kalogera
\cite{bel01}). A natural outcome of the evolution of this type of
binaries, provided that the SN explosions were of zero or moderate
asymmetry, is the origin of a binary {\it non-recycled} pulsar
(Belczy\'{n}ski \& Kalogera \cite{bel01}). The CO binary, however,
could be disrupted after the first (or the second) asymmetric SN
explosion if the kick received by the stellar remnant was of proper
magnitude and orientation (see Tauris \& Takens \cite{tau98}).
For illustrative purposes, we consider the disruption of a CO binary
following the first asymmetric SN explosion. For parameters of the
CO binary given in Belczy\'{n}ski \& Kalogera (\cite{bel01}) and
using Eqs.\,(44)--(47) and (51)--(56) given in Tauris \& Takens
(\cite{tau98}), one can show that the pulsar velocities and $\psi$
could be explained if the kick imparted to the first-born pulsar
(B2020+28)
\begin{figure}
\resizebox{8cm}{!}{\includegraphics{7693fig1.eps}}
\caption{The dependence of the velocities of the (first-born)
pulsar and its former companion star (now the runaway
progenitor of the second pulsar) on the angle between the kick vector and
the direction of motion of the exploding star (shown, respectively, by the
solid and the long-dashed lines). The horizontal short-dashed lines indicate
the pulsar velocities suggested by VCC04. See text for details.}
\label{twovel}
\end{figure}
was $\sim 500 \, {\rm km} \, {\rm s}^{-1}$ (for the sake of
simplicity we assume that the second SN explosion was symmetric),
while the angle between the kick vector and the direction of motion
of the exploding star, $\theta$, was $\simeq
40^{\degr}$\footnote{Note that for $64^{\degr} \la \theta \la
290^{\degr}$, the binary system remains bound.} (see Figs.\,1 and 2
and Gvaramadze \cite{gva06}). It is obvious that the kick should be
stronger if, at the moment of the first SN explosion, the binary
consists of red supergiant or Wolf-Rayet stars (cf. VCC04).
Another possibility is that the pulsars attained their velocities in
the course of disintegration of the binary after the second
asymmetric SN explosion. Since both pulsars are not recycled, one
should assume either that the binary separation was sufficiently
large (so that the wind of the secondary star did not affect the
evolution of the first-born pulsar) or that the binary evolved
through a double common-envelope phase (see above). VCC04 suggest
that the pulsars were born in a wide binary, but in their analysis
they draw an erroneous conclusion that the pulsar velocities can be
explained by a kick of only $\simeq 200 \, {\rm km} \, {\rm s}^{-1}$
(see Gvaramadze \cite{gva06}). One can show, however, that in both
the above-mentioned cases the kick imparted by the second SN
explosion should be $\geq 500 \, {\rm km} \, {\rm s}^{-1}$.
\begin{figure}
\resizebox{8cm}{!}{\includegraphics{7693fig2.eps}}
\caption{The angle between the velocity vectors of the first- and
second-born pulsars as a function of the angle between the kick
vector and the direction of motion of the exploding star. The horizontal
dashed line indicates the angle between the pulsar velocity vectors
suggested by VCC04.}
\label{chitheta}
\end{figure}
Thus, the origin of the pulsars in a common binary implies that at
least one of the SN explosions was asymmetric enough to produce a
kick of $\geq 500 \, {\rm km} \, {\rm s}^{-1}$. If, however, SNe can
indeed impart high velocities to NSs, then it is not necessary to
assume that the pulsars originated in the same binary, but instead
one can suggest that they were created by two separate SN explosions
occurred in the same parent star cluster within a few $10^5$ yr. Our
scenario for the origin of the pulsar pair has something in common
with the latter possibility, but we do not require any asymmetry in
the SN explosions.
\section{Pulsars B2020+28 and B2021+51: dynamical ejection from the young
massive star cluster}
The recent discovery of the so-called hypervelocity stars (Brown et
al. \cite{bro05}) and hyperfast pulsars (Chatterjee et al.
\cite{cha05}), the ordinary stars and pulsars moving with extremely
high ($\sim 1\,000 \, {\rm km} \, {\rm s}^{-1}$) peculiar
velocities, suggests a possibility that the hypervelocity stars
could be the progenitors of hyperfast NSs, provided that they are
massive enough (Gvaramadze et al. \cite{gva07}). A strong argument
in support of this possibility comes from the fact that the mass of
one of the hypervelocity stars, HE\,0437$-$5439, is $\ga 8 \,
M_{\odot}$ (Edelmann et al. \cite{ede05}) so that, in principle, it
can end its evolution as a hyperfast NS! The high velocities ($\sim
200-400 \, {\rm km} \, {\rm s}^{-1}$) inferred for some early B-type
stars at high galactic latitudes (Ramspeck et al. \cite{ram01}) also
support the possibility that high-velocity pulsars could originate
from high-velocity runaway stars.
Gvaramadze et al. (\cite{gva07}) suggest that the origin of
hypervelocity stars could be connected not only with scattering
processes involving the supermassive black hole (BH) in the Galactic
centre (the common wisdom; originally suggested by Hills
\cite{hil88}; see also Yu \& Tremaine \cite{yu03}; Gualandris et al.
\cite{gua05}), but also with strong three- or four-body dynamical
encounters in the dense cores of young massive star clusters
(YMSCs), located either in the Galactic disk or near the Galactic
centre. The discovery of a halo population of early B stars, whose
lifetimes are shorter than the times-of-flight from the Galactic
centre (Brown et al. \cite{bro07}; see also Ramspeck et al.
\cite{ram01}), supports this suggestion. We believe, therefore, that
the pulsars B2020+28 and B2021+51 could be the remnants of
high-velocity runaway stars ejected from the same YMSC. The
kinematic and characteristic ages of the pulsars (respectively,
$\sim 2$ and $\sim 3$ Myr; VCC04) imply that by the moment of
ejection the progenitor stars have already become Wolf-Rayet stars
[the short-lived ($< 1$ Myr) helium (He) cores of massive stars; cf.
Gvaramadze et al. \cite{gva07}].
Of the two mechanisms that could be responsible for the origin of
the high-velocity progenitors of B2020+28 and B2021+51, the first
relies on close dynamical encounters between hard (Heggie
\cite{heg75}) massive binary stars in the dense core of a YMSC. The
peculiar velocities of runaway stars produced in this process are
similar to the orbital velocities of the binary components (e.g.
Leonard \& Duncan \cite{leo90}), but occasionally they could be much
higher. Scattering experiments by Leonard (\cite{leo91}) showed that
the maximum velocity attained by the lightest member of the binaries
involved in the interaction (e.g. the He core of a massive star or
an early B-type star) can be as high as the escape velocity, $V_{\rm
esc}$, from the surface of the most massive star in the binaries.
For the main-sequence stars with the mass-radius relationship
(Habets \& Heintze \cite{hab81}), $r_{\rm MS} = 0.8 (m_{\rm MS}
/M_{\odot} )^{0.7} \, R_{\odot}$, where $r_{\rm MS}$ and $m_{\rm
MS}$ are the stellar radius and the mass, the maximum possible
velocity of ejected stars is a weak function of $m_{\rm MS}$,
$V_{\rm ej} ^{\rm max} \simeq V_{\rm esc} \simeq 700 \, {\rm km} \,
{\rm s}^{-1} (m_{\rm MS} /M_{\odot} )^{0.15}$ and could be as high
as $\sim 1\,400 \, {\rm km} \, {\rm s}^{-1}$ (cf. Leonard 1991).
Numerical simulations by Leonard (\cite{leo91}) showed that about $4
\%$ of binary-binary encounters result in the origin of runaway
stars with $V_{\rm ej} \simeq 0.5V_{\rm esc}$, which is enough to
explain the velocity of $\sim 500 \, {\rm km} \, {\rm s}^{-1}$
suggested by VCC04 for one of the pulsars. Note that the results of
Leonard (\cite{leo91}) were used by Tenjes et al. (\cite{ten01}) to
explain the origin of the high-velocity ($\sim 400 \, {\rm km} \,
{\rm s}^{-1}$) runaway star HIP 60350.
Another possible mechanism for producing high-velocity stars is
based on exchange encounters between tight binary stars and a
compact massive object, either a very massive star (VMS), formed
through the runaway stellar collisions and mergers in the mass
segregated core of a YMSC (e.g. Portegies Zwart et al.
\cite{por99}), or its descendant, an intermediate-mass BH (e.g.
Portegies Zwart \& McMillan \cite{por02}). After the close encounter
and tidal breakup of the binary, one of the binary components
(usually the more massive one) becomes bound to the compact object,
while the second one recoils with a high velocity given by $V_{\rm
ej} \sim [M/(m_1 +m_2)]^{1/6} (2Gm_1 /a)^{1/2}$ (Hills \cite{hil88};
see also Gvaramadze et al. \cite{gva07}), where $M$ is the mass of
the compact object, $m_1$ and $m_2$ are the masses of the binary
components ($m_1 >m_2$), and $a$ the binary separation.
In YMSCs of mass $\sim 10^4 \, M_{\odot}$, the mass of the VMS does
not exceed several $100 \, M_{\odot}$, while the thermal
(Kelvin-Helmholtz) time scale of the VMS is shorter than the mean
time between collisions (see Portegies Zwart et al. \cite{por99}).
In this case, the growing VMS rapidly evolves to the thermal
equilibrium (e.g. Suzuki et al. \cite{suz07}), so that one can adopt
the following mass-radius relationship, $R \simeq 1.6 \,
(M/M_{\odot} )^{0.47} R_{\odot}$, where $R$ is the radius of the VMS
(see Freitag et al. \cite{fre06} and references therein). In the
process of an exchange encounter with a binary, the VMS could be
considered as a point mass if the binary tidal radius, $r_{\rm t}
\sim [M/(m_1 +m_2)]^{1/3} a$, is at least several times larger than
$R$. For $M=200-300 \, M_{\odot}, m_1 =30 \, M_{\odot}$ (a
main-sequence star), $m_2 =8\, M_{\odot}$ (a He core), and $a=50 \,
R_{\odot}$, one has $r_{\rm t} \simeq 90-100 \, R_{\odot}$ (i.e.
much larger than $R\simeq 19-23 \, R_{\odot}$) and $V_{\rm ej}
\simeq 630-670 \, {\rm km} \, {\rm s}^{-1}$, that is enough to
explain the pulsar velocities.
In more massive ($\geq 10^5 \, M_{\odot}$) YMSCs, the VMS can
acquire a mass of several $1\,000 \, M_{\odot}$ (Portegies Zwart et
al. \cite{por04}). But in this case, the thermal time scale is
comparable to the collision time and the VMS remains inflated untill
collapsing into an intermediate-mass BH (e.g. Portegies Zwart et al.
\cite{por06}). Exchange encounters with this VMS would not produce
high ejection velocities. The star ejection from the YMSC, however,
would be very effective if the VMS leave behind a BH of mass $\sim
1\,000 \, M_{\odot}$ (e.g. Gualandris \& Portegies Zwart
\cite{gua07}).
\section{Cyg OB2}
The astrometric data on B2020+28 and B2021+51 suggest that these
pulsars (or their progenitors; our preference) were ejected $\sim
1.9$ Myr ago from the same origin at a distance of $\sim 1.9$ kpc in
the direction of the Cyg OB2 association (VCC04). The parent YMSC
(or its descendant) should still be located at about the same
distance since its possible peculiar velocity of $\leq 30 \, {\rm
km} \, {\rm s}^{-1}$ (typical of the OB associations near the Sun;
de Zeeuw et al. \cite{dez99}) would result only in a slight offset
of $\leq 60$ pc (cf. VCC04). To constrain the current age of the
parent cluster, we assume that the initial mass of the progenitor
stars of B2020+28 and B2021+51 could be as high as $\ga 50 \,
M_{\odot}$. (It is believed that stars of this mass can lose most of
their mass via stellar wind or mass transfer on a binary companion
and leave behind NSs; e.g. Vanbeveren et al. \cite{van98}; Wellstein
\& Langer \cite{wel99}; cf. Muno et al. \cite{mun06}.) From this it
follows that the minimum age of the parent YMSC should be $\sim 5$
Myr, that is, $\sim 2$ Myr (the pulsar kinematic age) plus $\sim 3$
Myr (the lifetime of a $\ga 50 \, M_{\odot}$ star). Assuming that
the YMSC initially contained at least 10 stars of mass $> 50
M_{\odot}$, one has the (initial) mass of the cluster of $\geq 10^4
\, M_{\odot}$ (for a $0.2-120 \, M_{\odot}$ Salpeter initial mass
function).
The only likely candidate for the birth cluster of B2020+28 and
B2021+51 in the region suggested by VCC04 is the Cyg OB2
association. Numerous star clusters in its neighbourhood (see, e.g.,
Le Duigou \& Kn\"{o}dlseder \cite{led02}) cannot pretend to play
this role either due to their youth or low masses.
Cyg OB2 is one of the most massive and compact OB associations in
our Galaxy (Kn\"{o}dlseder 2000). The large number ($\sim 100$) of O
stars identified in Cyg OB2 (Kn\"{o}dlseder \cite{kno00}; see also
Comer\'{o}n et al. \cite{com02}) implies that its mass could be as
high as $\sim 10^5 \, M_{\odot}$. The angular radius of Cyg OB2 is
$\sim 1^{\degr}$, while the half light radius is $\sim 13^{'}$
(Kn\"{o}dlseder \cite{kno00}), that at the distance of Cyg OB2 of
$\sim 1.5-1.7$ kpc (Hanson \cite{han03}; Massey \& Thompson
\cite{mas91}) corresponds, respectively, to $\sim 25-30$ pc and
$\sim 5-6$ pc. Note that the centre of Cyg OB2 lies within the
$2\sigma$ likelihood contour of the pulsar birth location and, at
the $2\sigma$ level, the distances to the Cyg OB2 and the birth
location are consistent with each other. Age estimates for Cyg OB2
range from $\sim 1$ to 5 Myr (e.g. Bochkarev \& Sitnik \cite{boc85};
Kn\"{o}dlseder et al. \cite{kno02}). The wide age spread suggests
that the star formation in the Cyg OB2 was non-coeval. The
non-coevality could be understood if the star formation in the
association started initially in the dense core of the parent
molecular cloud and then propagated to its lower density periphery.
It is believed (e.g. Elmegreen \cite{elm00}) that the star formation
occurs on one or two dynamical time scales, $t_{\rm dyn} \sim (G\rho
)^{-1/2}$, where $\rho$ is the gas density in the cloud, so that in
a density-stratified cloud of mass of $\sim 10$ times higher than
the stellar mass of Cyg OB2 and the size similar to that of the
association, the age spread could be comparable with $t_{\rm dyn}
\sim 5$ Myr.
If the progenitor stars of B2020+28 and B2021+51 were ejected from
Cyg OB2, then we suppose that, $\sim 2$ Myr ago (or $\sim 3$ Myr
after the formation of the first massive stars in the centre of the
association), the core of the association was much more compact and
denser. Assuming that the association expands with a velocity equal
to its velocity dispersion ($\simeq 2.4 \, {\rm km} \, {\rm
s}^{-1}$; Kiminki et al. \cite{kim06}), one finds that the stars
located within the current half light radius were originally
concentrated in a region of a radius of $< 1$ pc. It is likely that
the two star clusters projected close to each other ($\sim 6^{'}$)
near the centre of Cyg OB2 (Bica et al. \cite{bic03}) are the
remainders of this dense core. We suggest that the mass of the core
was much higher ($\geq 10^4 \, M_{\odot}$) than the current mass of
the clusters (several $1\,000 M_{\odot}$; Bica et al. \cite{bic03})
and that it was significantly reduced during the last 2 Myr due to
the overall expansion of the association and star ejections
following close dynamical encounters [recent N-body simulations by
Pflamm-Altenburg \& Kroupa (\cite{pfl06}) showed that dynamical
processes in the Trapezium cluster could be responsible for the loss
of at least $75\%$ of its initial content of OB stars]. Thus we
believe that $\sim 2$ Myr ago the conditions in the core of Cyg OB2
were favourable to the dynamical processes discussed in Sect.\,3.
\begin{acknowledgements}
I am grateful to H.Baumgardt, D.Bomans, P.Kroupa, and S.Portegies
Zwart for useful discussions. I am also grateful to D.Bomans and
R.-J.Dettmar for their hospitality during my stay at the
Astronomisches Institut, Ruhr-Universit\"{a}t Bochum, where this
work was partially carried out. This work was partially supported by
the Deutsche Forschungsgemeinschaft.
\end{acknowledgements}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 8,045 |
Q: Running any React-Native project for iOS results in 'Unable to resolve module' error I'm trying out the much hyped React-Native as an alternative to native development. I have installed nvm, react, react-native, and the cli by following this guide.
I have tried downloading and running several projects from github but no project has ever run (navigate to project directory, npm install then react-native run-ios) successfully. I usually get this error: 'Unable to resolve module'.
The error mentions it may be related to https://github.com/facebook/react-native/issues/4968. It also provides some solutions, non of which have ever worked. Reinstalling the node modules never works, neither does resetting the cache. I have also tried reinstalling react, rn, and the cli. The last 'solution' is to recreate the project from scratch, which I haven't tried thus far.
The emperor has no clothes.
P.S Even the example iOS project in the react-native github repo throws errors (albeit different ones: No script URL provided.)
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 2,794 |
{"url":"http:\/\/osl.cs.illinois.edu\/publications\/conf\/hvc\/Sen09.html","text":"# DART: directed automated random testing\n\nBy\n\nFull Text:\nhttp:\/\/dx.doi.org\/10.1007\/978-3-642-19237-1_4\n\n## Abstract\n\nTesting with manually generated test cases is the primary technique used in industry to improve reliability of software\u2013in fact, such testing is reported to account for over half of the typical cost of software development. I will describe directed automated random testing (also known as concolic testing), an efficient approach which combines random and symbolic testing. Concolic testing enables automatic and systematic testing of programs, avoids redundant test cases and does not generate false warnings. Experiments on real-world software show that concolic testing can be used to effectively catch generic errors such as assertion violations, memory leaks, uncaught exceptions, and segmentation faults. From our initial experience with concolic testing we have learned that a primary challenge in scaling concolic testing to larger programs is the combinatorial explosion of the path space. It is likely that sophisticated strategies for searching this path space are needed to generate inputs that effectively test large programs (by, e.g., achieving significant branch coverage). I will present several such heuristic search strategies, including a novel strategy guided by the control flow graph of the program under test.\n\n## BibTeX\n\n@inproceedings{conf\/hvc\/Sen09,\nauthor = \"Sen, Koushik\",\neditor = \"Namjoshi, Kedar S. and Zeller, Andreas and Ziv, Avi\",\ntitle = \"{DART}: Directed Automated Random Testing\",\nbooktitle = \"Haifa Verification Conference\",\ncrossref = \"conf\/hvc\/2009\",\nee = \"http:\/\/dx.doi.org\/10.1007\/978-3-642-19237-1_4\",\nkeywords = \"formal methods, software engineering\",\npages = \"4\",\nyear = \"2009\",\n}\n\n@proceedings{conf\/hvc\/2009,\neditor = \"Namjoshi, Kedar S. and Zeller, Andreas and Ziv, Avi\",\ntitle = \"Hardware and Software: Verification and Testing - 5th\nInternational Haifa Verification Conference, HVC 2009, Haifa,\nIsrael, October 19-22, 2009, Revised Selected Papers\",\nee = \"http:\/\/dx.doi.org\/10.1007\/978-3-642-19237-1\",\nisbn = \"978-3-642-19236-4\",\npublisher = \"Springer\",\nseries = \"Lecture Notes in Computer Science\",\nvolume = \"6405\",\nyear = \"2011\",\n}","date":"2017-03-24 20:00:32","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 1, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.4282726049423218, \"perplexity\": 9356.420393084552}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2017-13\/segments\/1490218188553.49\/warc\/CC-MAIN-20170322212948-00056-ip-10-233-31-227.ec2.internal.warc.gz\"}"} | null | null |
\section{Introduction}
Identifying the particle nature of the cosmological dark matter is a central challenge in modern physics \cite{PDG,dmevidence,Jungman,Snowmass,wimpdetection}. Experiments attempting to directly detect Weakly Interacting Massive Particles (WIMPs) in the laboratory must be sensitive to the very small recoil energies (\SIrange[range-phrase=--,range-units=single]{1}{100}{\keV}) that WIMPs would deposit through elastic scattering on detector target nuclei of comparable mass. These detectors are designed to acquire large background-free exposures by using increasingly massive targets while minimizing all sources of backgrounds to a WIMP signal. The coupling between WIMPs and standard model particles is typically characterized in terms of spin-independent (SI) and spin-dependent (SD) cross sections. As the underlying mechanism for this interaction is unknown, a thorough WIMP-search program must probe both SI and SD couplings.
The superheated liquid detector technology used by the PICO collaboration
affords excellent intrinsic rejection of electron recoils from gamma and beta particles. Alpha decays of U/Th daughter nuclei can be acoustically discriminated against using piezoelectric sensors mounted on the detector surface. Three dimensional optical event reconstruction allows for topological event selection, rejecting multiply scattering neutron events. Materials screening and optimized detector design minimize the sources of single-scatter neutron background, with the goal of acquiring a background-free WIMP-search exposure.
The first blind exposure of the PICO-60 C$_3$F$_8$ detector \cite{30l_16_PRL} achieved this goal, acquiring a 1167-kg-day exposure at a thermodynamic threshold of \SI{3.3}{\keV} with zero single-scatter nuclear recoil candidates in the signal region after unblinding.
Three multi-bubble events were observed during that exposure, while $0.25\pm 0.09$ single- and $0.96\pm 0.34$ multiple-scatter neutron events were predicted by simulation (Sec.~\ref{sec:bg}). This observation indicated that the detector was effectively neutron-limited, unable to attain significant additional WIMP sensitivity simply by acquiring longer exposures.
Following post-run calibrations, an attempt was made to explore the limits of detector stability at higher C$_3$F$_8$ temperatures and lower pressures, reducing the bubble nucleation threshold calculated using Equation 2 of Ref.~\cite{30l_13}. These thermodynamic changes were also expected to increase sensitivity to the environmental gamma background (Sec.~\ref{sec:bg}). The C$_3$F$_8$ temperature was increased from \SI{13.9(1)}{\celsius} to \SI{15.9(1)}{\celsius} and the superheated pressure was progressively reduced from \SI{30.2(3)}{\PSIA} to \SI{21.7(3)}{\PSIA}, effecting a reduction in the nucleation threshold (Sec.~\ref{sec:threshold}) from \SI{3.29(09)}{\keV} to \SI{1.81(08)}{\keV}. The detector continued to operate stably, maintaining a live-time fraction over 75\% during these periods, despite the higher rate of fiducial single-bubble events, as expected with increased sensitivity to the electron-recoil background.
In response, a second blind exposure was acquired between April and June 2017 at a threshold of \SI{2.45}{\keV}, for which the overall background rate was expected to be dominated by the same neutron background rates as at \SI{3.29}{\keV}.
Here we report the results of that efficiency-corrected dark matter exposure of 1404\,kg-days.
Just prior to decommissioning, the temperature was raised to \SI{19.9(1)}{\celsius}, enabling thresholds as low as \SI{1.20(08)}{\keV} to be reached. As expected, the event rate was then dominated by events consistent with electron recoils, but operations remained stable. The higher event rate led to a reduced live-time fraction near 40\% at this lowest threshold. These operating conditions are summarized in Table~\ref{table:opcond}.
\section{Experimental Method}
The PICO-60 apparatus (Fig.~\ref{fig:pico60}) was configured as described in detail in~\cite{30l_13}, with the following changes implemented in 2016 for \cite{30l_16_PRL}.
Rather than CF$_3$I, the bubble chamber was filled with \SI{52.2(5)}{\kg} of C$_3$F$_8$,
as first reported in \cite{30l_16_PRL}. As the superheated operating temperatures for C$_3$F$_8$ are lower than those in CF$_3$I, a new chiller system was used to hold the temperature of the surrounding water tank \cite{30l_13,30l_16_PRL} uniform to approximately \SI{0.1}{\celsius}.
The acoustic transducers, formerly coupled to the inner vessel with epoxy, were changed to a spring-loaded coupling.
The chamber's expansion cycle from the stable, compressed state to the superheated state was identical to the previous run \cite{30l_16_PRL}, with only occasional minor alterations to the maximum cycle period and target pressure. These alterations were a response to temporarily elevated trigger rates observed after a temperature change, when thermal expansion or contraction of the C$_3$F$_8$ caused the position of its interface with the buffer water to shift, and visible water droplets to become localized sources of elevated wall nucleation rates. Relatively rapid cycling of the hydraulic system to an intermediate pressure over a period of approximately one hour was typically observed to return the chamber to stability.
These periods contain only diagnostic information and are neither blinded nor included in the present exposure.
As in \cite{30l_16_PRL}, in order to image the entire C$_3$F$_8$ volume, double the volume used in PICO-60 CF$_3$I \cite{30l_13}, it was necessary to install an upper row of cameras, resulting in a stereoscopic view by each of two vertical pairs. As part of this expansion in scale, the data acquisition hardware and software running the cameras and issuing the primary event trigger was restructured and modularized prior to \cite{30l_16_PRL}. Each column of cameras was controlled by a separate server continuously acquiring images at \SI{340}{\Hz} for this result, improving time resolution compared to the \SI{200}{\Hz} used for the previous exposure \cite{30l_16_PRL}. Each camera filled a ring buffer with incoming images while its control software monitored for the appearance of bubbles by continuously calculating the difference-based spatial temporal entropy image \cite{DSTEI-Jing} $S_I=-\sum_i P_i\log_2 P_i$, where $P_i$ is the fraction of pixels populating intensity bin $i$ of the difference-map histograms generated from consecutive frames.
These camera servers communicated operational state changes and trigger conditions to the primary data acquisition server managing event-level operation of the chamber. The cameras were sent a single digital pulse train to synchronize their exposure timing. This signal was also used to drive the pulse timing of the LEDs illuminating the chamber's inner volume.
\begin{figure}
\includegraphics[width=210 pt,trim=0 0 0 0,clip=true]{PICO60Run2_IV_label.png}
\caption{\label{fig:pico60} The PICO-60 detector as configured for its operation with C$_3$F$_8$. The full volume of target fluid is stereoscopically imaged by two columns of two cameras each.}
\end{figure}
\begin{table*}[ht]
\begin{center}
\begin{tabular}{|c|c|c|c|c|} \hline
T ($^{\circ}$C) & P (psia) & Seitz threshold, $E_T$ (keV) & Livetime (d) & Exposure (kg-d) \\\hline
19.9 & 25.5 & $1.20 \pm 0.1(\mathrm{exp}) \pm 0.1(\mathrm{th})$&0.21 & 8.2 \\
19.9 & 34.3 & $1.58 \pm 0.1(\mathrm{exp}) \pm 0.1(\mathrm{th})$&1.29 & 50.3 \\
15.9 & 21.7 & $1.81 \pm 0.1(\mathrm{exp}) \pm 0.2(\mathrm{th})$&7.04 & 311 \\\hline\hline
15.9 & 30.5 & $2.45 \pm 0.1(\mathrm{exp}) \pm 0.2(\mathrm{th})$&29.95 & 1404 \\\hline
13.9 & 30.2 & $3.29 \pm 0.1(\mathrm{exp}) \pm 0.2(\mathrm{th})$&29.96 & 1167 \cite{30l_16_PRL}\\
\hline
\end{tabular}
\caption{\label{table:opcond}
Details of the four new operating conditions and their associated exposures, as well as the original set of conditions used in \cite{30l_16_PRL}. The two blind exposures are grouped in the lower rows. The experimental uncertainty on the threshold comes from uncertainties on the temperature (\SI{0.1}{\celsius}) and pressure (0.3\,psi), while the theoretical uncertainty comes from the thermodynamic properties of C$_3$F$_8$ including the surface tension, and dominated by uncertainty in the Tolman length \cite{McLure}.}
\end{center}
\end{table*}
The detector was primarily operated at four new sets of thermodynamic conditions, summarized in Table~\ref{table:opcond}. For the \SI{2.45(09)}{\keV} threshold, a second blind analysis \cite{30l_16_PRL} was undertaken by
acquiring a new category of background data with masked acoustics. These acoustic signals allow discrimination between alpha decays and nuclear or electron recoil events with the Acoustic Parameter (AP) analysis variable, optimized to cleanly separate these distributions as in \cite{30l_16_PRL}. Source calibrations and pre-physics background data were used to finalize cuts and efficiencies for bulk single recoil event candidates in an unbiased way.
Unlike the previous blind analysis \cite{30l_16_PRL}, no supplemental neural network was used here to discriminate between alphas and nuclear recoils, though more advanced versions of this machine learning approach are being developed for future PICO detectors \cite{PICO_ML}.
After this analysis was frozen, acoustic information for the physics dataset was processed and the acceptance region was examined.
For the three lowest thresholds (1.20, 1.58, \SI{1.81}{\keV}), acoustic information was never blinded, and a full analysis not performed, as these datasets were always expected to contain many gamma-induced recoils indistinguishable from nuclear recoils by their acoustic signals. Furthermore, these lowest thresholds are not supported by comprehensive nuclear recoil calibrations in C$_3$F$_8$ as introduced for the thresholds of the blind exposures in Sec.~\ref{sec:threshold}.
These datasets thus act primarily
as a confirmation of the ability to operate a bubble chamber stably at very low thresholds, maintaining the superheated state for periods on the order of minutes, and are not included in the WIMP-search analysis.
\section{Bubble Nucleation Threshold}
\label{sec:threshold}
The efficiency with which nuclear recoils nucleate bubbles is measured with a suite of neutron calibration experiments, to which fluorine and carbon efficiency curves at each threshold are fit to monotonically increasing, piecewise linear functions. Well-defined resonances in the $^{51}$V(p,n)$^{51}$Cr reaction are used to produce monoenergetic 50, 61, and \SI{97}{\keV} neutrons directed at a $\sim$30-ml C$_3$F$_8$ bubble chamber at the Tandem Van de Graaff facility at the Universit\'e de Montr\'eal. An SbBe neutron source is also deployed adjacent to the $\sim$30-ml bubble chamber, and an AmBe neutron source adjacent to the PICO-2L chamber~\cite{2l_13}. The initial C$_3$F$_8$ calibration presented in Ref.~\cite{2l_13} and used for the first PICO-60 C$_3$F$_8$ result~\cite{30l_16_PRL} is refined in this analysis with additional calibration data. Datasets have been extended for 61 and \SI{97}{\keV} neutron beams and the \SI{50}{\keV} neutron beam dataset is entirely new, as is the SbBe source, a gamma-induced neutron source that primarily produces monoenergetic \SI{24}{\keV} neutrons. Calibrations were performed at a variety of thermodynamic thresholds, with selected results shown in Figure \ref{fig:multfit}, along with the prediction for the best-fit efficiency model.
\begin{figure*}
\includegraphics[width=\textwidth,trim=0 0 0 0,clip=true]{goodness_pico60_v7c.png}
\caption{\label{fig:multfit}
Green points show the observed rates of single, double, and triple-or-more bubbles for the calibration sources at the listed thermodynamic thresholds. Green error bars indicate statistical uncertainties, and the black error bars at the bottom show the systematic uncertainty on the neutron flux. The blue histograms show the predicted rates from the simulation given the best-fit efficiency model derived from all calibration data. Each dataset is normalized to the observed rate of single bubbles, or double bubbles for SbBe due to gamma background. The normalization of the simulation is constrained by the systematic neutron flux uncertainties shown.}
\end{figure*}
Each of the neutron calibration experiments is simulated in MCNP~\cite{MCNP} or Geant4~\cite{GEANT4}, using differential cross sections for elastic scattering on fluorine from Ref.~\cite{Robinson}. The calibration data is fit using the \texttt{emcee}~\cite{emcee} Markov Chain Monte Carlo (MCMC) python code package. The output of the fitting is a distribution of sets of four efficiency curves (fluorine and carbon curves at each of the 2.45 and 3.29~keV thresholds) with associated likelihoods (Fig.~\ref{fig:efficiency}). The addition of the new lower-energy neutron datasets supports tighter constraints on the low-energy part of the efficiency curves than previously reported, resulting in increased sensitivity to low-mass WIMPs. A detailed paper on the calibration of the bubble nucleation efficiency is in preparation by the collaboration \cite{PICO_NR}.
\begin{figure}
\includegraphics[width=250 pt,trim=0 2cm 0 2cm,clip=true]{efficiency_v15_nb.pdf}
\caption{\label{fig:efficiency} Best-fit fluorine (blue) and carbon (magenta) efficiency curves for \SI{2.45}{\keV} and \SI{3.29}{\keV} data are shown as solid lines. The shaded regions show the band enveloping all efficiency curves fitted within 1$\sigma$. The green dashed lines show the calculated Seitz threshold, with theoretical uncertainties from Table~\ref{table:opcond}.}
\end{figure}
\section{Data Analysis}
The new datasets were processed as in \cite{30l_16_PRL} with independently redefined cuts for each set of operating conditions, and with several improvements and additions.
For this analysis a more sophisticated determination of the fiducial volume was deployed, using better bubble position reconstruction and a tag for wall-originating events based on bubble motion.
The position reconstruction algorithm was modified to have finer, sub-pixel resolution, and to monitor and correct for small shifts in the overall image position on a camera's sensor over time. Each camera's contribution to the reconstruction was individually weighted by the relative quality of the bubble image obtained. Image quality was evaluated as a function of the distance between the bubble's image and the center of the camera's sensor, and included corrections for lighting quality that changed as several LEDs failed during operation.
A new tracking algorithm supplemented information about the bubble position at the time of first appearance with its position across up to nine successive frames, over a total period of \SI{30}{\milli\second}. Bubbles nucleated on the walls were typically observed to follow tracks angled 10$^{\circ}$ or more from vertical over this period and could be rejected. This tagging by zenith angle had 95.6\% acceptance of non-wall events in the annular region \SIrange[range-phrase=--,range-units=single]{3}{7}{\mm} from the wall where it was applied.
Similarly, events near the surface (the top \SI{10}{\mm} of active fluid), where visibility is less favorable, were required to be detected by both cameras within \SI{30}{\milli\second} of each other, to limit uncertainties in position reconstruction. This cut had 100\% acceptance of non-surface events in the cylindrical near-surface region.
Together, these optimizations allow fiducial cut boundaries to be placed closer to the edge of the detector while still classifying zero surface- or wall-originating events as fiducial. The fiducial mass is thus increased relative to \cite{30l_16_PRL} from ($45.7\pm0.5$)\,\si{\kg} to ($48.9\pm 0.8$)\,\si{\kg}.
Together with a higher singles selection efficiency than \cite{30l_16_PRL} due to a slightly wider AP cut (Fig.~\ref{fig:APvsNN}) and lack of a neural network-based acoustic cut, this results in a WIMP-search exposure of \SI{1404}{\kg\day} for the second blind run of PICO-60 C$_3$F$_8$, as detailed in Table~\ref{table:runinfo}.
\begin{figure}
\includegraphics[width=240 pt,trim=0 0 0 0,clip=true]{ZvsR_2_4keV_candidates_postAP_opacity.pdf}
\caption{\label{fig:XYZ} Spatial distribution of single-bubble events in the \SI{2.45}{\keV} WIMP-search data. Z is the reconstructed vertical position of the bubble, and R$^{2}$/R$_{\mathrm{jar}}$ is the distance from the center axis squared, normalized by the nominal jar radius (\SI{145}{\mm}). The outer edge of the fiducial cut is represented by the dashed black line, outside of which all events are excluded. Events reconstructed within the cyan annular region \SIrange[range-phrase=--,range-units=single]{3}{7}{\mm} from the wall were additionally required to satisfy a condition limiting their track's zenith angle, and events in the \SI{10}{\mm} near-surface magenta region were additionally required to have appeared in two cameras with a limited offset in frame index.
Red squares are the 87 single bulk bubbles passing all cuts prior to acoustic unblinding and grey dots are all rejected single-bubble events. Blue circles are the 3 candidate events passing the AP cut.}
\end{figure}
Time-dependent effects over the blind exposure were minimal. As in the past the rate of pressure rise during early stages of bubble growth, measured by a Dytran 2005V fast pressure transducer \cite{dytran}, was used to identify bubble multiplicity. The Dytran signal drifted slowly over time and was renormalized in the analysis. The magnitude of acoustic signals from bubble nucleation is strongly dependent on the temperature and pressure of the superheated liquid. Single-bubble events from $^{252}$Cf neutron and $^{133}$Ba gamma calibration data taken at each set of thermodynamic conditions were used to create separately normalized AP definitions for those conditions.
\section{Background estimates}
\label{sec:bg}
Backgrounds are estimated from a combination of Monte Carlo simulations, measured calibration event rates, and multi-bubble event rates during physics runs.
The total expected background rate is the sum of the following contributions and is summarized in Table~\ref{table:bg_estimates}.
The majority of neutron scatter events induce more than one visible bubble in the detector, unlike single-bubble WIMP-scattering events. Multi-bubble events provide a definitive signature of neutron background and represent the most robust constraint on the rate of single-scatter neutron background events. Geant4 simulations of the detector geometry and composition predict 80\% of neutron events to have multiple bubbles, in agreement with $^{252}$Cf neutron calibration data, and with minimal dependence on the type and location of the neutron source. Since the detector configuration was unchanged between the first and second blind exposures (and since multiplicity was not blinded), the neutron rate is estimated from the overall rate of multi-bubble events from both exposures. Five multi-bubble events were observed, three in the first blind exposure~\cite{30l_16_PRL} and two in the second blind exposure, resulting is a neutron background expectation for the \SI{2.45} {keV} exposure of 0.8\,$\pm$\,0.4 events. The observed multi-bubble rate is modestly higher than predicted from the simulations, which estimate 0.96\,$\pm$\,0.34 and 1.43\,$\pm$\,0.49 multiples in the \SI{3.29} {\keV} and \SI{2.45} {\keV} exposures respectively, and 0.38\,$\pm$\,0.15 single-scatter background events in the \SI{2.45} {\keV} exposure. The discrepancy between the observed and predicted multi-bubble event rate is not significant at the 90\% C.L. The observed multi-bubble event rate is used to calculate a data-driven prediction of the single-bubble neutron background shown in Table~\ref{table:bg_estimates}.
Gamma calibration was performed at \SI{2.45}{\keV} with a \SI{0.1}{\milli\curie} $^{60}$Co source before and after the blinded run. Compared with a Geant4 simulation of the same detector geometry, this produces a measured nucleation efficiency of (2.89 $\pm$ 0.15)$\times 10^{-9}$ for gamma interaction events producing electron recoils above \SI{2.45}{\keV}. Combined with the rate from external gammas as simulated in MCNP, based on measurements from a NaI detector close to PICO-60 at SNOLAB
\cite{robinson_thesis,fustin_thesis}, we estimate a background of
0.12 $\pm$ 0.02 gamma events in this 1404\,kg-day
blind exposure.
More advanced models of the gamma response in superheated fluids are currently under development by the PICO collaboration \cite{PICO_ER}.
The rate of coherent elastic nuclear scattering of $^8$B solar neutrinos on C$_3$F$_8$ is non-negligible for thresholds below approximately \SI{5}{\keV}, so we calculate this contribution to the total background rate.
For the blind exposure acquired at \SI{2.45}{\keV}, this background is projected to contribute (0.10 $\pm$ 0.02) events.
The measured fiducial single-bubble event rate during the second blind run of PICO-60 C$_3$F$_8$, (2.9 $\pm$ 0.3) events/live-day, can be extrapolated to a $^{222}$Rn rate under the assumption that each such event represents one of three alpha decays along the $^{222}$Rn to $^{210}$Po chain. Given the exposure of this dataset, this corresponds to an approximate $^{222}$Rn rate of \SI{2}{\micro\becquerel} in the detector, competitive with DEAP-3600~\cite{deap3600}.
In our \SI{2.45}{\keV} blind exposure, excellent separation of low-AP recoil events from radon chain alphas is maintained.
We therefore assume zero contribution to the total background rate from these events.
\begin{table}
\centering
\begin{tabular}{| c | c | c | c |}
\hline
Neutron & Gamma & CE$\nu$NS & Total \\\toprule\hline
(0.8 $\pm$ 0.4) & (0.12 $\pm$ 0.02) & (0.10 $\pm$ 0.02) & (1.0 $\pm$ 0.4)\\\hline
\end{tabular}
\caption{Summary of estimated single-bubble background contributions for the full 29.95 day livetime of the \SI{2.45}{\keV} blind run of PICO-60 C$_3$F$_8$. ``CE$\nu$NS'' indicates the contribution from coherent elastic neutrino-nucleus scattering.}
\label{table:bg_estimates}
\end{table}
\section{WIMP search Results}
After the decision to unblind the \SI{2.45}{\keV} WIMP-search dataset, the acoustic signals were processed and are presented in Figure~\ref{fig:APvsNN} along with the AP distributions for neutron and gamma calibrations. Three nuclear recoil candidates are observed in the WIMP-search signal region, consistent with the background prediction from Table~\ref{table:bg_estimates} at the 90\% C.L. The total observation of three single-bubble and five multiple-bubble events over the combined exposures is consistent with the expected singles-to-multiples ratio of 1:4 for a neutron dominated background, albeit at somewhat higher rate than predicted by our simulations.
\begin{table*}
\begin{center}
\begin{tabular*}{\textwidth}{ l @{\extracolsep{\fill}} c c c c }
\hline
\hline
\rule{0pt}{2.5ex}Dataset & Efficiency ($\%$) & Fiducial Mass (kg) & Exposure (kg-days) & Number of events \\
\hline
\rule{0pt}{2.5ex}Singles & 95.9 $\substack{+1.9\\-3.4}$ & 48.9 $\pm$ 0.8 & 1404 $\substack{+48\\-75}$ & 3 \\
\rule{0pt}{2.5ex}Multiples & 99.9 $\substack{+0.0\\-0.1}$ & 52.0 $\pm$ 0.1 & 1556 $\substack{+3\\-5}$ & 2 \\ \hline
\hline
\end{tabular*}
\caption{Summary of the final number of events and exposure determination for singles and multiples in the 29.95 live-day WIMP-search dataset of PICO-60 C$_{3}$F$_{8}$ at \SI{2.45}{\keV} thermodynamic threshold. The singles selection efficiency is substantially higher than that of \cite{30l_16_PRL} due to a slightly wider AP acceptance region and the omission of the supplemental neural network-based acoustic cut used in the prior analysis.}
\label{table:runinfo}
\end{center}
\end{table*}
\begin{figure}
\includegraphics[width=240 pt,trim=0.2cm 0.2cm 0.9cm 0.5cm,clip=true]{AP_PRD_log.pdf}
\caption{\label{fig:APvsNN} AP distributions for $^{252}$Cf (blue) and $^{133}$Ba calibration data (combined with $^{252}$Cf, green) and WIMP-search data (red) at \SI{2.45}{\keV} threshold. The acceptance region for nuclear recoil candidates, defined before WIMP-search acoustic data unmasking using neutron and gamma calibration data to span \mbox{(--3$\sigma$,} \mbox{+2$\sigma$)} from the mean, is displayed with dashed blue lines, and reveals 3 candidate events in the WIMP-search data. Alphas from the $^{222}$Rn decay chain can be identified by their time signature and populate the two peaks in the WIMP-search data at high AP. Higher energy alphas from $^{214}$Po produce larger acoustic signals.}
\end{figure}
A Profile Likelihood Ratio (PLR) test~\cite{plr} is used to set WIMP exclusion limits on the combination of 2.45 and 3.29~keV datasets. A test statistic is formed from the ratio of the likelihood for a specific WIMP cross-section to the maximum likelihood over all WIMP cross-sections. The background rate and WIMP detection efficiency in each dataset are treated as nuisance parameters, marginalized over by finding the conditional maximum likelihood for each specific WIMP cross-section.
Given the apparent discrepancy between our predicted and observed neutron background, the background rates are unconstrained in the PLR, with flat likelihood functions for all non-negative values. In future PICO dark matter searches the neutron background rate may be constrained in the PLR by including the multi-bubble event rate, but to be conservative that has not been implemented in this analysis.
For the efficiencies, a likelihood surface is created as a function of WIMP detection efficiency at 2.45 and 3.29~keV. WIMP detection efficiencies, $\Phi$, in units of detected WIMPs per kg-day of exposure per picobarn of WIMP-nucleon scattering cross-section, are derived from the calibration MCMC output by integrating the efficiency curves over the nuclear recoil spectrum from an astrophysical WIMP flux for an array of potential WIMP masses. The two-dimensional WIMP detection efficiency space is divided into bins and within each bin the maximum likelihood set of efficiency curves that fall within that bin is found. The likelihood surface thus created retains any covariance between the efficiency at the two thresholds from the neutron calibration.
The standard halo parametrization~\cite{lewinandsmith} is used, with the following parameters: local dark matter density $\rho_D$=\SI{0.3}{\GeV}$c^{-2}$cm$^{-3}$, galactic escape velocity $v_{\rm{esc}}$ = 544 km/s, velocity of the earth with respect to the halo $v_{\rm Earth}$ = 232 km/s, and characteristic WIMP velocity with respect the halo $v_0$ = 220 km/s. The effective field theory treatment and nuclear form factors described in Refs.~\cite{spindependentcouplings,Anand,Gresham,Gluscevic} are used to determine sensitivity to both spin-dependent and spin-independent dark matter interactions. The $M$ response of Table 1 in Ref.~\cite{spindependentcouplings} is used for SI interactions, and the sum of the $\Sigma'$ and $\Sigma''$ terms from the same table is used for SD interactions. To implement these interactions and form factors, the publicly available \texttt{dmdd} code package~\cite{Gluscevic,Gluscevic2} is used. Figure~\ref{fig:contour} shows examples of the WIMP detection efficiency likelihood surfaces used for 5\,GeV WIMPs with SI coupling and 19\,GeV WIMPs with SD-proton coupling. The likelihood surfaces are marginalized over as nuisance parameters within the PLR, after being convolved with a two-dimensional Gaussian function reflecting experimental uncertainty in the PICO-60 thermodynamic thresholds.
\begin{figure}
\includegraphics[width=280 pt,trim=0 0 0 0,clip=true]
{contour_5_3_SI.pdf}
\includegraphics[width=280 pt,trim=0 0 0 0,clip=true]
{contour_19_SD_cross.pdf}
\caption{\label{fig:contour} Contour plot of integrated efficiency $\Phi$ at \SI{2.45}{\keV} and \SI{3.29}{\keV} with red dot representing best-fit result. Contour layers have been color-coded to represent the difference in $\chi^2$ with respect to the minimum. Details in the outer boundary of the plot are subject to statistical fluctuations.}
\end{figure}
To develop a frequentist WIMP exclusion curve, toy datasets are generated at each point in a grid of WIMP masses and cross-sections. A grid point is then excluded if the observed PLR test statistic for that point is $>$90\% of toy dataset test statistics at that point. A conservative choice is made to generate the toy datasets with no background contribution, but the 90\% exclusion curve is subsequently confirmed to be valid over the range of background rates consistent with the data. The calculated exclusion curves at 90\% C.L. for spin-dependent WIMP-proton and spin-independent WIMP-nucleon elastic scattering cross-sections, as a function of WIMP mass, are shown in Figures~\ref{fig:SD} and~\ref{fig:SI}. The already world-leading limits in the spin-dependent WIMP-proton sector are improved, particularly for WIMP-masses in the \SIrange[range-phrase=--,range-units=single]{3}{5}{\GeV} range.
\begin{figure}
\includegraphics[width=240 pt]{PLR_limit_SDp.pdf}
\caption{\label{fig:SD} The 90\% C.L. limit on the SD WIMP-proton cross section from the profile likelihood analysis of the PICO-60 C$_3$F$_8$ combined blind exposure plotted in thick maroon, along with limits from the first blind exposure of PICO-60 C$_3$F$_8$ (thick blue) \cite{30l_16_PRL}, as well as limits from PICO-60 CF$_3$I (thick red)~\cite{30l_13}, PICO-2L (thick purple)~\cite{2l_15}, PICASSO (green band)~\cite{PICASSOFinallimit}, SIMPLE (orange)~\cite{SIMPLE}, PandaX-II (cyan)~\cite{PANDAX-II}, IceCube (dashed and dotted pink)~\cite{ICECUBElimit}, and SuperK (dashed and dotted black)~\cite{SKlimit,SKlimit2}. The indirect limits from IceCube and SuperK assume annihilation to $\tau$ leptons (dashed) and {\it b} quarks (dotted).
Additional limits, not shown for clarity, are set by LUX~\cite{LUX_SD} and XENON1T~\cite{XENON1T} (comparable to PandaX-II) and by ANTARES~\cite{Ant1,Ant2} (comparable to IceCube).}
\end{figure}
\begin{figure}
\includegraphics[width=240 pt,trim=0 0 25 15,clip=true]{PLR_limit_SI.pdf}
\caption{\label{fig:SI} The 90\% C.L. limit on the SI WIMP-nucleon cross-section from the profile likelihood ratio analysis of the PICO-60 C$_3$F$_8$ combined blind exposure plotted in thick maroon, along with limits from the first blind exposure of PICO-60 C$_3$F$_8$ (thick blue) \cite{30l_16_PRL}, PICO-60 CF$_3$I (thick red)~\cite{30l_13}, PICO-2L (thick purple)~\cite{2l_15}, DarkSide-50 low-mass (gray)~\cite{DarkSide50LM}, LUX (yellow)~\cite{LUX2017}, PandaX-II (cyan)~\cite{PANDAX-II_SI}, XENON1T (green) \cite{XENON1T}, CRESST-II (magenta)~\cite{CRESST}, and CDMS-lite (black)~\cite{CDMSlite}.
Additional limits, not shown for clarity, are set by PICASSO~\cite{PICASSOFinallimit}, XENON100~\cite{XENON100}, SuperCDMS~\cite{SuperCDMS}, CDMS-II~\cite{CDMSII}, and Edelweiss-III~\cite{Edelweiss}.
}
\end{figure}
\section{Acknowledgements}
The PICO collaboration wishes to thank SNOLAB and its staff for support through underground space, logistical and technical services. SNOLAB operations are supported by the Canada Foundation for Innovation and the Province of Ontario Ministry of Research and Innovation, with underground access provided by Vale at the Creighton mine site. We wish to acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canada Foundation for Innovation (CFI) for funding. We acknowledge the support from the National Science Foundation (NSF) (Grant 0919526, 1506337, 1242637 and 1205987). We acknowledge that this work is supported by the U.S.\ Department of Energy (DOE) Office of Science, Office of High Energy Physics (under award DE-SC-0012161), by the DOE Office of Science Graduate Student Research (SCGSR) award, by DGAPA-UNAM (PAPIIT No.\:IA100316 and IA100118) and Consejo Nacional de Ciencia y Tecnolog\'ia (CONACyT, M\'exico, Grant No.\:252167), by the Department of Atomic Energy (DAE), Government of India, under the Centre for AstroParticle Physics II project (CAPP-II) at the Saha Institute of Nuclear Physics (SINP),
European Regional Development Fund-Project ``Engineering applications of microworld physics'' (No.\:CZ.02.1.01/0.0/0.0/16\_019/0000766),
and
the Spanish Ministerio de Ciencia, Innovaci\'on y Universidades (Red Consolider MultiDark, FPA2017$-$90566$-$REDC).
This work is partially supported by the Kavli Institute for Cosmological Physics at the University of Chicago through NSF grant 1125897 and 1806722, and an endowment from the Kavli Foundation and its founder Fred Kavli. We also wish to acknowledge the support from Fermi National Accelerator Laboratory under Contract No.\:DE-AC02-07CH11359, and from Pacific Northwest National Laboratory, which is operated by Battelle for the U.S.\ Department of Energy under Contract No.\:DE-AC05-76RL01830. We also thank Compute Canada (\url{www.computecanada.ca}) and the Centre for Advanced Computing, ACENET, Calcul Qu\'ebec, Compute Ontario and WestGrid for computational support.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 9,197 |
Salaam is a short form of As-salamu alaykum, an Arabic greeting meaning "Peace be upon you". This phrase and the Arabic word 'peace' derive from the Semitic root Š-L-M.
Salaam or Salam may also refer to:
Businesses and organizations
Al-Salam SC, several sports teams
Salaam TV, independent satellite television channel
Salaam Bank, commercial bank headquartered in Bosaso, Somalia
Salaam Somali Bank, commercial bank headquartered in Mogadishu, Somalia
Salam, defunct newspaper in Iran
Salam Zgharta FC, Lebanese association football club
Zee Salaam, Indian Hindi- and Urdu-language Islamic TV channel owned by Zee Network
Music
Salam, film score by Fariborz Lachini
Salam, song by Alabina
"Salaam" (song), peace song by Mosh Ben Ari
Salaam (album), album by Sami Yusuf
Salam (album), album by Irfan Makki
People
Given name
Salaam bin Said Al Shaksy, chief executive
Salaam Remi, American hip hop record producer
Surname
Abd al-Salam (name)
Anbara Salam, Lebanese writer and activist
Ephraim Salaam, American football player
Florin Salam, Romanian manele singer
H. Salam (active from 2021), Indian politician in Kerala
Kawther Salam, Palestinian journalist
Mohammed Ahmed Salam, Yemeni prisoner
Abdul Salam Rocketi, Afghan military personnel
Nawaf Salam, Lebanese diplomat, academic, jurist
Rashaan Salaam, American football player
Salim Ali Salam, Lebanese statesman
Saeb Salam, Lebanese politician
Tammam Salam, Lebanese politician
Waleed Al-Salam, mathematician
Places
Salam, Chaharmahal and Bakhtiari, Iran (also known as Salm)
Salam, Mali
Salam Street, Abu Dhabi
Other uses
Operation Salam, covert operation during World War II
Salaam spasms or epileptic spasms, disorder
Salam leaf or Indonesian bay leaf, leaf of a plant in the family Myrtaceae
See also
El Salam Maritime Transport
Lal Salam, salute, greeting or code word used by communists in South Asia
Salem (disambiguation)
Salim (disambiguation)
Selam (Australopithecus), fossil specimen found in Dikika
Shalom (disambiguation)
Names of God in Islam | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 1,446 |
{"url":"https:\/\/en.wikipedia.org\/wiki\/Partition_matroid","text":"Partition matroid\n\nIn mathematics, a partition matroid or partitional matroid is a matroid formed from a direct sum of uniform matroids.[1]\n\nDefinition\n\nLet $B_i$ be a collection of disjoint sets, and let $d_i$ be integers with $0\\le d_i\\le |B_i|$. Define a set $I$ to be \"independent\" when, for every index $i$, $|I\\cap B_i|\\le d_i$. Then the sets that are independent sets in this way form the independent sets of a matroid, called a partition matroid. The sets $B_i$ are called the blocks of the partition matroid. A basis of the matroid is a set whose intersection with every block $B_i$ has size exactly $d_i$, and a circuit of the matroid is a subset of a single block $B_i$ with size exactly $d_i+1$. The rank of the matroid is $\\sum d_i$.[2]\n\nEvery uniform matroid $U{}^r_n$ is a partition matroid, with a single block $B_1$ of $n$ elements and with $d_1=r$. Every partition matroid is the direct sum of a collection of uniform matroids, one for each of its blocks.\n\nIn some publications, the notion of a partition matroid is defined more restrictively, with every $d_i=1$. The partitions that obey this more restrictive definition are the transversal matroids of the family of disjoint sets given by their blocks.[3]\n\nProperties\n\nAs with the uniform matroids they are formed from, the dual matroid of a partition matroid is also a partition matroid, and every minor of a partition matroid is also a partition matroid. Direct sums of partition matroids are partition matroids as well.\n\nMatching\n\nA maximum matching in a graph is a set of edges that is as large as possible subject to the condition that no two edges share an endpoint. In a bipartite graph with bipartition $(U,V)$, the sets of edges satisfying the condition that no two edges share an endpoint in $U$ are the independent sets of a partition matroid with one block per vertex in $U$ and with each of the numbers $d_i$ equal to one. The sets of edges satisfying the condition that no two edges share an endpoint in $V$ are the independent sets of a second partition matroid. Therefore, the bipartite maximum matching problem can be represented as a matroid intersection of these two matroids.[4]\n\nMore generally the matchings of a graph may be represented as an intersection of two matroids if and only if every odd cycle in the graph is a triangle containing two or more degree-two vertices.[5]\n\nClique complexes\n\nA clique complex is a family of sets of vertices of a graph $G$ that induce complete subgraphs of $G$. A clique complex forms a matroid if and only if $G$ is a complete multipartite graph, and in this case the resulting matroid is a partition matroid. The clique complexes are exactly the set systems that can be formed as intersections of families of partition matroids for which every $d_i=1$.[6]\n\nEnumeration\n\nThe number of distinct partition matroids that can be defined over a set of $n$ labeled elements, for $n=0,1,2,\\dots$, is\n\n1, 2, 5, 16, 62, 276, 1377, 7596, 45789, 298626, 2090910, ... (sequence A005387 in OEIS).\n\nThe exponential generating function of this sequence is $f(x)=\\exp(e^x(x-1)+2x+1)$.[7]\n\nReferences\n\n1. ^ Recski, A. (1975), \"On partitional matroids with applications\", Infinite and finite sets (Colloq., Keszthely, 1973; dedicated to P. Erd\u0151s on his 60th birthday), Vol. III, Colloq. Math. Soc. Jan\u00f3s Bolyai 10, Amsterdam: North-Holland, pp.\u00a01169\u20131179, MR\u00a00389630.\n2. ^ Lawler, Eugene L. (1976), Combinatorial Optimization: Networks and Matroids, Rinehart and Winston, New York: Holt, p.\u00a0272, MR\u00a00439106.\n3. ^ E.g., see Kashiwabara, Okamoto & Uno (2007). Lawler (1976) uses the broader definition but notes that the $d_i=1$ restriction is useful in many applications.\n4. ^ Papadimitriou, Christos H.; Steiglitz, Kenneth (1982), Combinatorial Optimization: Algorithms and Complexity, Englewood Cliffs, N.J.: Prentice-Hall Inc., pp.\u00a0289\u2013290, ISBN\u00a00-13-152462-3, MR\u00a0663728.\n5. ^ Fekete, S\u00e1ndor P.; Firla, Robert T.; Spille, Bianca (2003), \"Characterizing matchings as the intersection of matroids\", Mathematical Methods of Operations Research 58 (2): 319\u2013329, arXiv:math\/0212235, doi:10.1007\/s001860300301, MR\u00a02015015.\n6. ^ Kashiwabara, Kenji; Okamoto, Yoshio; Uno, Takeaki (2007), \"Matroid representation of clique complexes\", Discrete Applied Mathematics 155 (15): 1910\u20131929, doi:10.1016\/j.dam.2007.05.004, MR\u00a02351976. For the same results in a complementary form using independent sets in place of cliques, see Tyshkevich, R. I.; Urbanovich, O. P.; Zverovich, I. \u00c8. (1989), \"Matroidal decomposition of a graph\", Combinatorics and graph theory (Warsaw, 1987), Banach Center Publ. 25, Warsaw: PWN, pp.\u00a0195\u2013205, MR\u00a01097648.\n7. ^ Recski, A. (1974), \"Enumerating partitional matroids\", Studia Scientiarum Mathematicarum Hungarica 9: 247\u2013249 (1975), MR\u00a00379248.","date":"2015-07-08 07:03:34","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 0, \"img_math\": 30, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.8895463943481445, \"perplexity\": 530.5188270530464}, \"config\": {\"markdown_headings\": false, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2015-27\/segments\/1435376093097.69\/warc\/CC-MAIN-20150627033453-00173-ip-10-179-60-89.ec2.internal.warc.gz\"}"} | null | null |
Q: update column values to reflect other column values having two different size dataframes I have the following two data frames.
df_1:
unique_id amount
1 NaN
2 5
df_2:
unique_id amount city email
1 90 Kansas True
2 100 Miami False
3 NaN Kent True
4 123 Newport True
I would like to only update the amount column where unique_id is 1 or 2 or any other that might match on unique_id. The output should be:
unique_id amount city email
1 NaN Kansas True
2 5 Miami False
3 NaN Kent True
4 123 Newport True
I've tried merging and contacting but I am not getting the desired result. I just want an idea of what the best approach is when two data frames are of different sizes and want to update certain column values. Any guidance is greatly appreciated
A: Try with mask:
df_2['amount'] = df_2['amount'].mask(df_2['unique_id'].isin(df_1['unique_id']),
df_2['unique_id'].map(df_1.set_index('unique_id')['amount'])
)
Output:
unique_id amount city email
0 1 NaN Kansas True
1 2 5.0 Miami False
2 3 NaN Kent True
3 4 123.0 Newport True
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 4,983 |
\section{Introduction}\label{sec:into}
In Ref.~\cite{ASW} we have presented a model where the Higgs decay rate to diphotons is increased through loops of mixed vector-like leptons. A vector-like doublet and a vector-like singlet allow for both Yukawa and Dirac masses. The resulting mixing leads to a sign flip of the coupling of the lightest lepton to the Higgs, such that constructive interference with the standard model (SM) amplitude for $h\to\gamma\gamma$ is possible, resulting in a diphoton decay rate that is enhanced relative to the SM. Similar models were presented in~\cite{ArkaniHamed:2012kq,An:2012vp}, and further studies have appeared in~\cite{Kearney:2012zi,Davoudiasl:2012ig,Bae:2012ir,Voloshin:2012tv,McKeen:2012av,Lee:2012wz,Arina:2012aj,Batell:2012zw,Davoudiasl:2012tu,Fan:2013qn,Carmona:2013cq,Cheung:2013bn,Feng:2013mea,Englert:2013tya}
\footnote{Effects of vector-like quark multiplets are discussed e.g. in~\cite{Dawson:2012di,Bonne:2012im,Dawson:2012mk,Moreau:2012da,Bertuzzo:2012bt}.}.
One of the shortcomings of the model~\cite{ASW} is that the additional ${\cal O}(1)$ Yukawa couplings accelerate the downward running of the Higgs quartic coupling, such that vacuum stability is lost at scales around $10$~TeV. This tension between an enhanced di-photon rate and vacuum stability was first noted in~\cite{ASW,ArkaniHamed:2012kq} and further explored in~\cite{Reece:2012gi,Kitahara:2012pb,Chun:2012jw,Carena:2012mw,Huo:2012tw,Kitahara:2013lfa}.
The goal of the present paper is to provide a UV completion of the model~\cite{ASW}, which guarantees vacuum stability up to very high scales and unification of gauge couplings at roughly $10^{16}$~GeV.
In supersymmetric models, the scalar potential is fixed by gauge invariance and supersymmetry relations, so the renormalization group evolution (RGE) of the quartic coupling does not lead to a stability problem, provided that it is well behaved at scales below the SUSY breaking scale. It is therefore natural to try to embed the model~\cite{ASW} into a supersymmetric extension of the SM.
It is well known that within the MSSM the gauge couplings unify at approximately $10^{16}$~GeV with very high accuracy. In order to not destroy the sensible relation between the values of the couplings at the low scale and the beta function coefficients, additional matter fields have to be embedded in complete multiplets of SU(5). A vector-like doublet with leptonic quantum numbers is naturally contained in a $5+\overline{5}$ representation of SU(5), while a corresponding singlet field can emerge from a $10+\overline{10}$. In the language of SO(10) grand unification, these representations can be combined into a $16+\overline{16}$ representation where the additional degrees of freedom have neutrino-like quantum numbers.
Additional matter with couplings to the Higgs sector will also give a positive contribution to the mass of the lightest Higgs boson in the MSSM~\cite{Babu:2008ge,Martin:2009bg,Graham:2009gy,Endo:2012cc,ArkaniHamed:2012gw}, thus might help reducing the fine tuning problem that is present in the Higgs sector of the MSSM.
This paper is organized as follows: In Sec.~\ref{sec:model}, we introduce the model and the particle content at the weak scale. The RGE of the gauge and Yukawa couplings are discussed in Sec.~\ref{sec:rge}. The effects of the new particles on the Higgs mass and di-photon decay rate are calculated in section~\ref{sec:higgs}, while in Sec.~\ref{sec:vacuum} the conditions for vacuum stability are derived. In Sec.~\ref{sec:results} we present numerical results before concluding in Sec.~\ref{sec:conclusions}.
\section{The Model}\label{sec:model}
Our model is the minimal supersymmetric standard model (MSSM) extended by a $16+\overline{16}$ of SO(10), such that the unification of gauge couplings at the GUT scale is guaranteed. Models with such a particle spectrum were previously studied e.g. in~\cite{Babu:2008ge}\cite{Martin:2009bg}. Below $M_{\rm GUT}$ the SO(10) multiplets are split. We assume that all the additional colored states obtain masses above the TeV scale, and therefore are in agreement with current LHC limits.
Below the TeV scale, we assume that the only fields beyond the MSSM particle content are the un-colored components of the original $16+\overline{16}$ supermultiplets. In terms of chiral superfields, these are $SU(2)$ doublets $L'_L$ and $\overline{L''_R}$, the singlet fields $\overline{E_R'}$ and $E''_L$ as well as singlet neutrino superfields $\overline{N'_R}$ and $N''_L$ with the quantum numbers indicated in Tab.~\ref{tab:fields}. Our notation is such that the bar extends over the implicit chiral projector, i.e. $\overline{E'_R}$ is a left-handed superfield.
\begin{table}
\center
\begin{tabular}{| l |c|c|c|c|c|c|}
\hline
Name & $L'_L = \begin{pmatrix} N'_L\\E'_L \end{pmatrix}$ &
$\overline{E'_R}$ & $\overline{N'_R}$ &
$\overline{L''_R} = \begin{pmatrix} \overline{E''_R}\\\overline{N''_R} \end{pmatrix}$ & $E''_L$ & $N''_L$ \\ \hline
Quantum Numbers & (2, -1/2) & (1, 1) & (1, 0) & (2, 1/2) & (1, -1) & (1, 0) \\ \hline
\end{tabular}
\caption{Additional fields below the TeV scale.
All fields are left-handed superfields, and the quantum numbers specify the transformations of the fields under the SU(2)$\times$U(1) gauge group of the SM. Primes indicate fields coming from the $16$ multiplet, while the double primed fields originate from the $\overline{16}$ multiplet. \label{tab:fields}}
\end{table}
The superpotential is
\begin{align}
W =& W_{\rm MSSM} - M_L L'_L \overline{L''_R} + M_E E''_L \overline{E'_R} - y'_c L'_L H_d \overline{E'_R} + y''_c \overline{L''_R} H_u E''_L \label{eqn:superpotential}
\\\notag
-& y_{n}' L'_L H_u \overline{N'_R}+y_{n}'' \overline{L''_R} H_d N''_L - M_{ij} N_i N_j \,,
\end{align}
where we have neglected the colored fields beyond the MSSM, and $(N_1,N_2) = (\overline{N'_R}, N''_L)$ since the neutrinos can have Majorana mass terms in addition to the Dirac mass terms for $L$ and $E$ fields. As in~\cite{ASW}, we impose a parity symmetry under which the vector-like multiplets are odd, such that mixing with MSSM leptons and sleptons is forbidden.
Contraction of SU(2) indices is implicit in~(\ref{eqn:superpotential}). In analogy with the leptonic superfields, the Higgs superfields are defined as $H_u = (H_u^+, H_u^0)^T$ and $H_d = (H_d^0, H_d^-)^T$. Indices are contracted using the anti-symmetric tensor $\epsilon$ with $\epsilon_{12} = 1$, for example $L'_L H_d = (N_L' H_d^- - E_L' H_d^0)$. The sign conventions in~(\ref{eqn:superpotential}) are chosen such that all entries in the charged fermion mass matrix come with a positive sign and all entries in the neutral fermion mass matrix with a negative sign, for convenience.
Explicitly, the charged fermion mass matrix is given by
\begin{align}\label{eqn:lepmass}
\frac{1}{2} \left( e_L' \,e_L'' \, \overline{e'_R}\, \overline{e_R''} \right) {\cal M}_{E}
\begin{pmatrix}
e_L' \\ e_L''\\ \overline{e'_R} \\ \overline{e_R''}
\end{pmatrix} + {\rm h.c.} \quad
& \text{ with }
{\cal M}_E =
\begin{pmatrix}
0&0 & y_c' v_d & M_L \\
0 & 0 & M_E & y_c'' v_u \\
y_c' v_d & M_E & 0 & 0 \\
M_L & y_c'' v_u & 0 & 0
\end{pmatrix},
\end{align}
where we used lower case letters for the fermionic components of the superfields, and we have introduced the scalar vacuum expectation values $v_u = \langle H_u^0 \rangle$ and $v_d = \langle H_d^0 \rangle$. In the supersymmetric limit, the corresponding slepton mass matrix is simply given by ${\cal M}_{\tilde{E}}^2 = {\cal M}_E^\dagger {\cal M}_E$. The $\mu H_u H_d$ term communicates SUSY breaking to the lepton sector, such that, in a non-trivial Higgs background, the charged slepton mass matrix assumes the form
\begin{align}\label{eqn:slepmass}
&{\cal M}^2_{\tilde{E}} =
\\ \notag
& \begin{pmatrix}
|y'_c|^2 v_d^2 + |M_L|^2 + m_{e'_L}^2 & y_c'^{*}v_d M_E + M_L^* y_c'' v_u & -y'_c v_u \mu^* & b_L \\
M_E^* y_c' v_d + y_c''^* v_u M_L & |y''_c|^2 v_u^2 + |M_E|^2 + m_{e''_L}^2 & b_E & -y''_c v_d \mu^* \\
-y'^*_c v_u \mu& b_E & |M_E|^2 + |y_c'|^2 v_d^2 + m_{e_R'}^2 & y_c'^* v_d M_L + M_E^* y_c'' v_u \\
b_L& -y''^*_c v_d \mu & M_L^* y_c' v_d + y_c''^* v_u M_E & |M_L|^2 + |y_c''|^2 v_u^2 + m_{e_R''}^2
\end{pmatrix},
\end{align}
in the basis $\left( \tilde{e}'_L,\, \tilde{e}''_L,\, \tilde{e}'_R,\, \tilde{e}''_R \right)$. The D-term contributions are not explicitly written down since their effects are small, but they are included in our analysis. Note that, in addition to the supersymmetric mass terms, we have allowed for the following bi-linear soft breaking terms in the potential:
\begin{align}
V_{\rm soft,\ell} & = m_{e'_L}^2 |\tilde{\ell}'_L|^2 + m_{e''_R}^2 |\tilde{\ell}''_R|^2 + m_{e''_L}^2 |\tilde{e}''_L|^2 + m_{e'_R}^2 |\tilde{e}'_R|^2
+ {\textstyle \frac{1}{2}}(b_L \tilde\ell_L'^\dagger \tilde\ell_R'' + b_E \tilde{e}_R'^* \tilde{e}''_L + {\rm h.c.})\,.
\end{align}
The structure of the mass matrices and their effects on the $h\to \gamma\gamma$ rate will be further explored in Sec.~\ref{sec:higgs}.
It is worth noting that half of the scalar degrees of freedom can be decoupled by increasing $m_{e_L''}^2$ and $m_{e_R''}^2$ to high values. This limit is similar to the light stau scenario~\cite{Carena:2011aa,Carena:2012gp}, with a $2\times 2$ charged slepton mass matrix
\begin{align}
{\cal M}^2_{\tilde{E},2\times 2} & = \begin{pmatrix}
|y'_c|^2 v_d^2 + |M_L|^2 + m_{e'_L}^2 & -y'_c v_u \mu^* \\
-y'^*_c v_u \mu& |M_E|^2 + |y_c'|^2 v_d^2 + m_{e_R'}^2
\end{pmatrix} .
\end{align}
A similar contribution is obtained when instead the double primed fields $\tilde{e}''_L$ and $\tilde{e}''_R$ are lifted, however in that case the mixing is proportional to $v_d$ and therefore suppressed in the large $\tan\beta$ regime.
\
Before moving to the next section, let us briefly review existing experimental limits on uncolored charged scalars and fermions. The LEP experiments have searched for such particles, and their results are collected in the particle data booklet (PDG)~\cite{PDG}.
Limits on additional charged leptons are given explicitly in the PDG. For a charged lepton that decays to $W^\pm \nu$, a limit of 100.8~GeV is quoted, while a limit of 101.9~GeV is given for a charged lepton that is not degenerate with it's corresponding neutrino. However, as discussed in detail in~\cite{ASW}, if a neutrino-like state is close in mass, these limits become invalidated.
Limits on the scalar partners of standard model leptons strongly depend on the flavor composition of the sleptons, and range from 107~GeV for left-handed selectrons down to 81.9~GeV for staus. As in the fermionic case, most of these bounds are weakened if other neutral states are close by in mass, and none of them can be directly applied to scalar partners of new leptons, as in our model. As absolute lower bound on the mass of the lightest leptons and slepton we therefore impose $m_{\tilde{E}_1} > m_h/2 \approx 62.5$~GeV. To facilitate the discussion in the remainder of the paper, we define two LEP limits:
\begin{itemize}
\item Conservative LEP limit: $m_{E_1} > 100$~GeV, $m_{\tilde{E}_1} > 90$~GeV,
\item Optimistic LEP limit: $m_{E_1} > 62.5$~GeV, $m_{\tilde{E}_1} > 62.5$~GeV,
\end{itemize}
with the understanding that a spectrum that satisfies the conservative LEP limit will certainly be in agreement with current experimental bounds, while a spectrum that satisfies the optimistic LEP limit might require additional invisible neutral states to be close in mass in order to not be in conflict with existing searches. It should be noted that such states are present in our model in the form of new neutrinos and sneutrinos.
The LHC experiments have not yet performed a dedicated search for signatures of vectorlike leptons. Since we assume that both the lightest new lepton and the lightest new slepton are neutral and stable, the leading visible signatures will come from pair production of heavier states that decay to the lightest state emitting a $W$ or $Z$ boson. In particular, for the leptons we can have $p p \to E_1^\pm N_2 \to W^\pm Z N_1 N_1$, and similarly for the sleptons. The resulting trilepton plus missing energy signature is similar to that of MSSM chargino and neutralino searches, for which results are available from both ATLAS~\cite{ATLAS:2013rla} and CMS~\cite{CMS:aro}. These searches exclude chargino masses up to 300~GeV, however only in regions of parameter space where the mass difference between the lightest and the heavier states is larger than the Z-boson mass. For mass differences up to 50~GeV the limit drops to about 170~GeV, and no limit is available if the mass difference is less than 30~GeV.
Due to the different SU(2) quantum numbers, the production cross section for vectorlike leptons is roughly a factor of two smaller than that for charginos and neutralinos, while the cross section for slepton pair production is suppressed even further. It follows that we can always evade the current LHC limits by requiring that the new leptons and sleptons are either close in mass to the lightest new state, or heavier than about 300~GeV. Spectra with exactly these features are suggested by our results from~\cite{ASW}, which is why we only impose the LEP bounds on the lepton and slepton masses. Note however that the 14~TeV LHC with sufficient luminosity will be able to improve upon the LEP bounds in most regions of parameter space~\cite{Carena:2012gp,ArkaniHamed:2012kq}.
Contributions to the electroweak $S$ and $T$ parameters from vectorlike leptons were studied in~\cite{ASW}, and were found to be in agreement with the existing limits even for very light masses for the new fermions. The main reason for this is that while the Yukawa couplings induce some custodial symmetry breaking, the effect is not too large since there is no color factor and a suppression of order $y_c^{'(')}v/M_{L,E}$ from the vectorlike mass terms.
The corresponding slepton contributions were calculated e.g. in~\cite{Martin:2009bg}, and are of the same order as the lepton contributions, such that the overall effect of leptons and sleptons should remain in agreement with the data. Finally we assume that the additional quarks and squarks from the SO(10) multiplets have TeV scale vectorlike masses, such that their contributions decouple. Therefore our model will in general be in agreement with electroweak precision constraints, even with lepton and slepton masses close to the LEP bound.
\section{Running of Couplings}\label{sec:rge}
The RGE of the gauge and Yukawa couplings is strongly affected by the presence of additional matter charged under the strong and weak interactions. To simplify the analysis of this section, we assume a common threshold at the TeV scale for the new vector-like states and for all SUSY partners. While ultimately we will be interested in scenarios where some of the vector-like matter is lighter, the effects on the RGE of the gauge couplings are negligible.
\begin{figure}
\center
\includegraphics{rgeG.pdf}
\caption{Evolution of gauge couplings in the MSSM (dashed lines) and in the MSSM extended by a $16+\overline{16}$ of SO(10) (solid lines). From top to bottom the curves correspond to $\alpha_1$, $\alpha_2$ and $\alpha_3$. }
\label{fig:rgeG}
\end{figure}
The one-loop evolution of gauge couplings is governed by
\begin{align}
\alpha_i^{-1}(\Lambda) = \alpha_i^{-1} (M_Z) - \frac{b_i}{2\pi} \log \frac{\Lambda}{M_Z}\,.
\end{align}
When $\Lambda > 1$~TeV, the one loop beta function coefficients are given by
\begin{align}
\beta_1 & = -\frac{3}{5}-2 N_f = -\frac{53}{5}\,,\notag\\
\beta_2 & = 5 -2 N_f = -5 \,, \notag \\
\beta_3 & = 9 - 2 N_f = -1\,, \notag
\end{align}
with $N_f=5$, while the corresponding values in the MSSM are obtained by taking $N_f=3$. Note in particular that this leads to a sign change in the beta function for $\alpha_3$, which shows that the full particle content of the MSSM $+16+\overline{16}$ is enough to render the strong interactions asymptotically non-free.
Fig.~\ref{fig:rgeG} shows the evolution up to high scales. Unification of gauge couplings occurs roughly around $M_U\sim 1.5\times 10^{16}$~GeV, with coupling strengths of about $\alpha_i \sim 0.15$ corresponding to $g_i \sim 1.4$. At the one loop level this is compatible with perturbative unification. For a discussion of higher order effects see e.g.~\cite{Martin:2009bg}.
Yukawa couplings have a well known fixed point behavior in the infrared, which allows us to determine an upper limit on the magnitude of the couplings at the electroweak scale. At the one-loop level, the RGE equations for the Yukawa couplings take the form
\begin{align}
\frac{d}{d\mu} y_i(\mu) & = \frac{1}{\mu} b_i(\mu)\,,
\end{align}
with beta functions given by~\cite{Auberson:1999kv}
\begin{align}
b_t(\mu) & = \frac{1}{16\pi^2} y_t \left( 6 y_t^2 +y_c''^2 - 4 \pi \left( {\textstyle \frac{13}{15}} \alpha_1 + 3 \alpha_2+ {\textstyle \frac{16}{3} } \alpha_3\right) \right) , \\
b_{y_c'}(\mu) & = \frac{1}{16\pi^2} y_c' \left( 4 y_c'^2 - 4 \pi \left( {\textstyle \frac{9}{5}} \alpha_1 + 3 \alpha_2 \right) \right), \\
b_{y_c''}(\mu) & = \frac{1}{16\pi^2} y_c'' \left( 4 y_c''^2 +3y_t^2 - 4 \pi \left( {\textstyle \frac{9}{5}} \alpha_1+ 3 \alpha_2 \right) \right) ,
\end{align}
where we have suppressed the scale dependence on the right-hand sides. For sufficiently large initial values at the weak scale, $y_c'$ and $y_c''$ will diverge at high scales. The top Yukawa coupling is $y_t(M_Z) = M_t/(v \sin\beta)$, such that the bound on $y_c''$ will be more stringent for small $\tan\beta$, while for larger $\tan\beta$ the bottom and tau Yukawas will lead to a slightly stronger bound on $y_c'$.
\begin{figure}
\center
\includegraphics[width=.48\textwidth]{rgeY1.pdf}\hspace*{.5cm}
\includegraphics[width=.48\textwidth]{rgeY2.pdf}
\caption{The RGE of the Yukawa couplings for $\tan\beta=2$ (left) and $\tan\beta=60$ (right). Shown are the evolution of the new lepton Yukawa couplings $y_c'$ and $y_c''$ for various input values at the high scale, to illustrate the fixed points. }
\label{fig:rgeY}
\end{figure}
For the numerical analysis, we used the $\overline{\text{MS}}$ running masses of the top and bottom quarks at the weak scale, $M_t(M_Z) \approx 165$~GeV and $M_b(M_Z)\approx 2.7$~GeV. Then the top and bottom Yukawas are given by
\begin{align}
y_t(M_Z) & = \frac{M_t(M_Z)}{v \sin\beta}\,, \\
y_b(M_Z) & = \frac{M_b(M_Z)}{v \cos\beta} \frac{1}{1+ \Delta M_b}\,.
\end{align}
The SUSY-QCD correction to the bottom mass, $\Delta M_b$, is relevant in the large $\tan\beta$ regime and was included in our analysis.
The running of the couplings for small and large $\tan\beta$ is shown in Fig.~\ref{fig:rgeY}. At the weak scale, the couplings $y_c'$, $y_c''$ are largely independent of their precise magnitude at the unification scale. In particular, we note that Yukawas of order 0.5-0.9 at the weak scale are natural given order one input values at the high scale. For $\tan\beta \sim 60$ the bottom Yukawa coupling is comparable to $y_t$, and the upper bounds on the new Yukawas are $y_c' \lesssim 0.9$ and $y_c'' \lesssim 0.8$, while for small values of $\tan\beta\sim 2$ the upper bounds are approximately given by $y_c' \lesssim 0.94$ and $y_c'' \lesssim 0.72$.
Comparing with the running of the top and bottom Yukawa coupling, we note that the $\alpha_3$ contribution tends to suppress the quark Yukawas at high scales, or, reversing the argument, leads to larger couplings at low scales. If one would like to obtain larger $y_c'$ and $y_c''$ at the weak scale, one could therefore either give up on perturbativity up to the unification scale or introduce additional interactions at intermediate scales that enhance the leptonic Yukawa couplings.
Before moving to the next section, let us briefly comment on the evolution of the Higgs quartic couplings below the SUSY breaking scale. At the weak scale, the particle content of our model is that of the standard model with an additional Higgs doublet, a set of vector-like leptons as in~\cite{ASW}, and potentially their superpartners. We have seen in~\cite{ASW} that additional order one Yukawa couplings in conjunction with a Higgs mass of 125~GeV lead to vacuum instabilities due to the RGE of the Higgs quartic coupling, but only above the TeV scale. Supersymmetry is expected to be restored at least partially around that scale\footnote{In particular stops and weak gauginos are still allowed to be significantly lighter than one TeV.}, above which the quartic couplings in the Higgs potential are determined through supersymmetric relations and stop being a threat to the stability of the vacuum.
\section{Higgs Properties}\label{sec:higgs}
The MSSM Higgs sector consists of the two Higgs doublets $H_u$, $H_d$ that acquire vacuum expectation values (VEV) $v_u$ and $v_d$, respectively. The VEVs are subject to the constraint $v_u^2 + v_d^2 = (174~{\rm GeV})^2$ and are therefore parametrized as $v_u = v \sin\beta$ and $v_d = v \cos\beta$.
The physical spectrum contains two neutral CP even Higgs bosons $h^0$ and $H^0$, a neutral CP odd boson $A^0$ and a charged scalar $H^\pm$. The neutral mass eigenstates are linear combinations of the neutral doublet components $H_u^0$, $H_d^0$, with the mixing parameterized as:
\begin{align}
\begin{pmatrix} H_u^0\\H_d^0\end{pmatrix} & = \begin{pmatrix} v_u\\v_d\end{pmatrix} + \frac{1}{\sqrt{2}}\begin{pmatrix} \text{cos}\,\alpha & \text{sin}\,\alpha\\-\text{sin}\,\alpha & \text{cos}\,\alpha \end{pmatrix} \begin{pmatrix} h^0\\H^0 \end{pmatrix} + \frac{i}{\sqrt{2}}\begin{pmatrix} \text{cos}\,\beta & \text{sin}\,\beta\\-\text{sin}\,\beta & \text{cos}\,\beta \end{pmatrix} \begin{pmatrix} G^0\\A^0 \end{pmatrix}.
\end{align}
Here $\alpha$ is the CP-even mixing angle, $G^0$ is the neutral Goldstone boson that can be removed by going to unitary gauge, and
by convention $h^0$ is defined to be the lightest of the CP-even mass eigenstates.
A particularly interesting regime is the so called decoupling limit, which is characterized by a large mass scale $m_A$ for the non-standard Higgs bosons $H^0$, $A^0$ and $H^\pm$. In this limit, the lightest CP even Higgs $h^0$ becomes SM like, and the mixing angle $\alpha$ is determined by $\alpha = \beta -\pi/2$. Supersymmetric constraints on the potential constrain its mass to be
\begin{align}
m_{h^0}^2 & = m_Z^2 \cos^2(2\beta) + \text{radiative corrections.}
\end{align}
For the reminder of this paper, we will work in the decoupling limit with $m_A = 1$~TeV, and further assume that the MSSM superpartners do not significantly modify the Higgs properties. This is strongly motivated by the recent discovery of a scalar resonance with a mass of around $125$~GeV and with properties consistent with a SM Higgs boson. While this mass exceeds the lightest Higgs mass in the MSSM at the tree level, it is well known that radiative corrections can bring this mass in agreement with the observation.
The dominant $h^0$ production channel at the LHC is $gg\rightarrow h^0$, where the largest contribution comes from the top loop. As new leptons and sleptons added to the MSSM are not colored, the production channel is not affected significantly. This is consistent with the data. The new particles affect the decay widths of $h^0$ only at the loop level. Hence they can have significant effects only on the loop induced Higgs decays like $h^0\rightarrow\gamma\gamma$ and $h^0 \to Z\gamma$ which are not present at the tree-level in the MSSM.
At the time of the new boson discovery announcement, both ATLAS and CMS experiments at the LHC reported an excess of events in the $p p \to h^0 \to \gamma\gamma$ channel with respect to the SM expectations~\cite{Chatrchyan:julyhiggs}. After including the full 2012 dataset into the analysis, the ATLAS collaboration continues to observe an increased signal strength of $1.65^{+0.34}_{-0.30}$ in the diphoton channel~\cite{ATLASmoriond}, while the signal strength measured by the CMS collaboration has decreased to $0.78^{+0.28}_{-0.26}$ or $1.11^{+0.32}_{-0.30}$, depending on the analysis method~\cite{Chatrchyan:moriond}.
Naively averaging the ATLAS and CMS results, we obtain a signal strength of $1.14\pm 0.21$ or $1.37\pm 0.22$, depending on which of the CMS results is used.
In the following we will discuss the effects of the new leptons and sleptons on the $h^0 \to \gamma\gamma$ decay and on the mass of the Higgs. We focus on conditions to obtain a moderate enhancement of the diphoton rate, which is well in agreement with the data.
\subsection{$h^0\rightarrow\gamma\gamma$ Width}
The matrix of the leptons (sleptons) to $h^0$ couplings is given by the gradients of the lepton (slepton) mass matrix with respect to the Higgs VEVs, projected onto the $h^0$ eigenstate:
\begin{align}
Y_{\tilde{E}h^0}&=\sin\beta \frac{\partial \mathcal{M}^2_{\tilde{E}}}{\partial v_u}+\cos\beta \frac{\partial\mathcal{M}^2_{\tilde{E}}}{\partial v_d}\,,\\
Y_{Eh^0}&=\sin\beta\frac{\partial \mathcal{M}_{E}}{\partial v_u}+\cos\beta \frac{\partial\mathcal{M}_{E}}{\partial v_d}\,,
\end{align}
where ${\cal M}_{\tilde{E}}$ is the slepton mass matrix~(\ref{eqn:slepmass}) while ${\cal M}_E$ is any one of the off-diagonal $2\times 2$ blocks of the lepton mass matrix~(\ref{eqn:lepmass}). We also have used the decoupling relation $\alpha = \beta - \pi/2$.
Therefore, the couplings of $h^0$ to the new charged slepton and lepton mass eigenstates are given as
\begin{align}
C_{\tilde{E}h^0} & = U_S^\dagger Y_{\tilde{E}h^0}U_S\,, \\
C_{Eh^0} &= U_L^\dagger Y_{E h^0} U_R\,,
\end{align}
where $U_S$ is the unitary matrix which diagonalizes $\mathcal{M}_{\tilde{E}}$, while $U_L$ and $U_R$ are the unitary matrices that diagonalize $\mathcal{M}_{E}\mathcal{M}_{E}^\dagger$ and $\mathcal{M}_{E}^\dagger\mathcal{M}_{E}$, respectively.
The new charged leptons and sleptons, $E_i$ and $\tilde{E}_i$, contribute to the $h^0$ decay to two photons at the one loop level.
In the mass basis, the contributions to the decay are given by
\begin{align}\label{eqn:hgg_loop}
\Gamma_{h \to \gamma\gamma} \propto \left| A_1(\tau_w) + \frac{4}{3}A_{1/2}(\tau_t) + \sum_{i=1,2}\frac{C_{Eh^0_{ii}} v}{M_{E_i}}A_{1/2}(\tau_{E_i}) + \frac{1}{2}\sum_{i=1}^4\frac{C_{{\tilde{E}}h^0_{ii}}v}{M^2_{{\tilde{E}}_i}}A_0(\tau_{{\tilde{E}}_i}) \right|^2\,,
\end{align}
where the loop functions for spin 0, spin 1/2 and spin 1 particles are given by~\cite{Djouadi:2005gi}
\begin{align}
A_0(\tau) & = \frac{f\left(\tau\right)-\tau}{\tau^2} & \text{for spin }\; 0 \,,\\
A_{1/2} (\tau) & = \frac{2\left(\tau+\left(\tau-1\right)f\left(\tau\right)\right)}{\tau^2} & \text{for spin }\; \frac{1}{2} \,,\\
A_1(\tau) &=-2-\frac{3}{\tau}-\frac{3\left(2\tau-1\right)f\left(\tau \right) }{\tau^2 } & \text{for spin } \;1\,,
\end{align}
with
\begin{align}
\tau_x&=\frac{m_h^2}{4m_x^2}\\
f\left(\tau\right) &= \begin{cases} \arcsin^2\sqrt{\tau} & \text{for } \tau\leq1\,,\\
\displaystyle
-\frac{1}{4}\left(-i\pi+\log\left[\frac{1+\sqrt{1-\tau^{-1} }}{1-\sqrt{1-\tau^{-1}}}\right]\right)^2\quad \quad &\text{for } \tau>1\,,
\end{cases}
\end{align}
and $m_x$ is the mass of the particle running in the loop.
It is instructive to consider the asymptotic values of the loop functions for $\tau_x\ll 1$, i.e. when the new particle masses are much larger than the half of the lightest Higgs boson mass. Asymptotically $A_{1/2}(\tau \to 0) = 4/3$ and $A_0(\tau \to 0) = 1/3$, while $A_{1/2}(\tau) > 4/3$ and $A_0(\tau) > 1/3$ for $0<\tau<1$. Note in particular that the SM contribution for $m_h=125$~GeV is $A_1(\tau_w)=-8.3$ from the W-boson loop and $\frac 4 3 A_{1/2}(\tau_t) = 1.8$ from the top quark loop.
Since the new leptons don't affect the Higgs production channels, the effect on the di-photon search channel at the LHC is fully described by the ratio
\begin{align}\label{eqn:rgamma}
R_{\gamma \gamma} =\frac{\sigma(pp \to h)}{\sigma_{\rm SM}(pp \to h)} \frac{\Gamma(h \to \gamma\gamma) }{\Gamma(h\to \gamma\gamma)_{\rm SM}} = \frac{\Gamma(h \to \gamma\gamma) }{\Gamma(h\to \gamma\gamma)_{\rm SM}}\,.
\end{align}
In the limit of $M_L\,,M_E\,,\mu\rightarrow0$ and vanishing soft SUSY-breaking mass terms, the prefactors $C_{Eh^0_{ii}} v/M_{E_i}$ and $C_{\tilde{E}h^0_{ii}} v/M^2_{\tilde{E}_i}$ in~(\ref{eqn:hgg_loop}) go to $\tan\beta$ and $\cot\beta$, respectively, which are positive numbers. This leads to destructive interference between the dominant $W$ boson contribution and the charged lepton and slepton loops, thus $R_{\gamma\gamma}< 1$ is expected in this limit.
In order to understand how $R_{\gamma\gamma} > 1$ can be achieved, we note that in the asymptotic limit the charged slepton contribution to the amplitude can be written as
\begin{align}\label{eqn:higgsS_low_energy}
\Delta_{\tilde{E}} & = A_{0}(0) \sum_{i} \frac{v\,C_{\tilde{E}h_{ii}}}{M_{\tilde{E}_i}^2} = \frac{1}{3} v\left(\left(\sin\beta \frac{\partial}{\partial v_u} \log\det{\cal M}^2_{\tilde{E}}\right)+\left(\cos\beta \frac{\partial}{\partial v_d} \log\det{\cal M}^2_{\tilde{E}} \right)\right)\,.
\end{align}
Similarly for leptons,
\begin{align}\label{eqn:higgs_low_energy}
\Delta_E&=A_{1/2}(0) \sum_{i} \frac{v\,C_{Eh_{ii}}}{M_{E_i}} = \frac{4}{3} v\left(\left(\sin\beta\frac{\partial}{\partial v_u} \log\det{\cal M}_E\right)+\left(\cos\beta \frac{\partial}{\partial v_d} \log\det{\cal M}_E \right)\right)\,.
\end{align}
The second equality is a consequence of the Higgs low energy theorem~\cite{Ellis:1975ap,Shifman:1979eb,Falkowski:2007hz,Carena:2012xa}, but can also be understood by noting that the trace of a matrix is basis independent and using $\log\det {\cal M} = {\rm tr}\log{\cal M}$. It is useful to evaluate the derivative partially, which leads to
\begin{align}\label{eqn:enhancementS}
\Delta_{\tilde{E}}&= \frac{v}{3(\det{\cal M}^2_{\tilde{E}})}\left(\sin\beta \frac{\partial \det{\cal M}^2_{\tilde{E}}}{\partial v_u}+ \cos\beta \frac{\partial \det {\cal M}^2_{\tilde{E}}}{\partial v_d}\right),
\end{align}
\begin{align}\label{eqn:enhancementF}
\Delta_E&= \frac{4v}{3(\det{\cal M}_{E})}\left(\sin\beta \frac{\partial \det{\cal M}_E}{\partial v_u}+ \cos\beta \frac{\partial \det {\cal M}_E}{\partial v_d}\right).
\end{align}
To obtain an enhancement of $R_{\gamma\gamma}$, it is clear $\Delta_{\tilde{E}}+ \Delta_E$ must be negative in order to have constructive interference with W boson loop. Furthermore it is evident that a large mass for all new particles will lead to a suppression of the effects due to the determinant in the denominator. This can be used to study the effects of sleptons and leptons separately, since either sector can be decoupled by introducing a sufficient amount of supersymmetry breaking.
Conditions for constructive interference from vector-like leptons were studied in detail in~\cite{ASW}. A new effect that enters in the case of supersymmetry is the dependence on $\tan\beta$, since opposite chirality leptons can not couple to the same Higgs doublet. It follows that
\begin{align}\label{eqn:prefactor2}
\Delta_{E}&=\frac{4}{3}\frac{\tan\beta\,v^2y'_cy''_c}{(1+\text{tan}^2\beta)(-M_LM_E+\frac{\text{tan}\,\beta\,v^2y'_cy''_c}{1+\text{tan}^2\beta})}\,.
\end{align}
Both $\tan\beta = 0$ and $\tan\beta = \infty$ lead to a suppression of $\Delta_E$, while the effect is maximal for $\tan\beta$ of order one. Even then, the effect on $R_{\gamma\gamma}$ is reduced compared to a non-supersymmetric model where leptons of both chiralities are allowed to couple to the same Higgs.
The slepton effects are slightly more involved. Both the vector mass terms $M_L$, $M_E$ and the $\mu$ term appear in off-diagonal elements of the mass matrix, such that in general all four charged slepton fields will mix heavily.
Absence of tachyons implies that $\det {\cal M}_{\tilde{E}}^2$ must be positive, therefore the derivatives in~(\ref{eqn:enhancementS}) must be negative in order to obtain an enhanced $h\to\gamma\gamma$ rate. In order to see how the different terms in the mass matrices influence $\Delta_{\tilde{E}}$, it is instructive to consider the large $\tan\beta$ limit. Then all terms proportional to $v_d$ can be neglected. Further setting the soft breaking parameters to zero, one obtains
\begin{align}\label{eqn:prefactor1}
\Delta_{\tilde{E}}&=-\frac{1}{3}\frac{2 v_u^2 y'^2_c (M_E^2 M_L^2 + 2 (M_E^2 + M_L^2) v^2_u y''^2_c + 3v^4_u y''^4_c)\mu^2}{M_E^4 M_L^4 - v^2_u y'^2_c (M_E^2 + v^2_u y''^2_c) (M_L^2 + v^2_u y''^2_c) \mu^2} \\
&\quad \to
-\frac{1}{3}\frac{2 v_u^2 y'^2_c \mu^2}{M_E^2 M_L^2 - v^2_u y'^2_c \mu^2}\,.
\end{align}
In the last line we have further set $y_c'' \to 0$.
This result highlights the importance of $\mu^2$ for obtaining a large contribution to $R_{\gamma\gamma}$. While the numerator grows linearly with $\mu^2$, the denominator can be held roughly constant by appropriately adjusting $M_E$ and $M_L$, such that the slepton masses are in agreement with the LEP limits. Very large values for $\Delta_{\tilde{E}}$ can therefore be obtained at the expense of some tuning between the mass parameters.
In section~\ref{sec:vacuum} we will see that vacuum stability considerations will lead to an upper bound on $\mu^2$ in the presence of large Yukawa couplings $y_c'$ and $y_c''$ and small slepton masses, which will limit the amount by which $R_{\gamma\gamma}$ can be enhanced. Numerical results that cover the full parameter region of our model and show the maximal possible enhancement are shown in section~\ref{sec:results}.
\subsection{One Loop Corrections to the $h^0$ mass}
To compute the one loop corrections to the $h^0$ mass, we use the one loop effective potential approximation. A supermultiplet $i$ with scalar of mass $M_{S_i}$ and fermion of mass $M_{F_i}$ contributes
\begin{align}
\Delta V_i&=\frac{N_c}{32\pi^2}\left[M_{S_i}^4\left(\log \left(\frac{M_{S_i}^2}{Q^2}\right)-\frac{3}{2}\right)-M_{F_i}^4\left(\log\left(\frac{M_{F_i}^2}{Q^2}\right)-\frac{3}{2}\right)\right]
\end{align}
to the one-loop effective potential. Here $Q$ is the renormalization scale. Then, as given in \cite{Martin:2009bg}, the correction to the tree-level $m_{h^0}^2$ is
\begin{align}\label{eqn:masscorr}
\Delta m_{h^0}^2 &=\left(\frac{\text{sin}^2\beta}{2}\left[\frac{\partial^2}{\partial v_u^2}-\frac{1}{v_u}\frac{\partial}{\partial v_u}\right]+\frac{\text{cos}^2\beta}{2}\left[\frac{\partial^2}{\partial v_d^2}-\frac{1}{v_d}\frac{\partial}{\partial v_d}\right]+\text{sin}\,\beta\,\text{cos}\,\beta\frac{\partial^2}{\partial v_u\partial v_d}\right)\sum_i\Delta V_i \,.
\end{align}
A notable modification of $R_{\gamma\gamma}$ requires at least some of the new leptons or sleptons to be light. In the absence of soft breaking terms, only $\mu$ will induce SUSY breaking in the new lepton sector, and therefore corrections to the Higgs mass will remain small. Larger corrections are possible if soft terms induce a sizable mass splitting between the sleptons and leptons, and it is worth asking whether these corrections can improve upon the fine tuning in the MSSM.
\begin{figure}
\center
\includegraphics[width=.48\textwidth]{HiggsMassLowTan1.pdf}\hspace*{.5cm}
\includegraphics[width=.48\textwidth]{HiggsMassHighTan1.pdf}
\caption{Contributions to the Higgs mass $\Delta m_{h^0}$ in the $m_{\rm soft} - \mu$ plane for $\tan\beta=2$ (left) and $\tan\beta=60$ (right). The red (solid) curves are contours of constant $\Delta m_{h^0}$ while the green (dashed) contours show the mass of the lightest slepton state. The remaining parameters are $M_L = M_E = 200$~GeV, $y_c' = 0.9$, $y_c'' =0.7$, and we assume equal soft masses $m_i^2 = m_{\rm soft}^2$ for all sleptons.
}
\label{fig:higgsmass}
\end{figure}
Compared to the corrections typically obtained from top-stop splitting, we can expect more moderate contributions due to the absence of a color factor and because the Yukawa couplings, that enter in the fourth order in the Higgs mass, are smaller than the top Yukawa coupling.
In Fig.~\ref{fig:higgsmass} we show contours of $\Delta m_{h^0}$ in the $m_{\rm soft}-\mu$ plane. For these plots, we have assumed a base Higgs mass of $m_0=120$~GeV in the MSSM and we show the correction $\Delta m_{h^0} = \sqrt{m_0^2 + \Delta m_{h^0}^2} - m_0$, with $\Delta m_{h^0}^2$ coming from~(\ref{eqn:masscorr}). The sleptons are taken to have equal soft masses $m_i^2 = m_{\rm soft}^2$, while the soft breaking $a$ and $b$ terms are set to zero. The fermionic mass scale is set to $M_L = M_E = 200$~GeV and we use $y_c' = 0.9$ and $y_c''=0.7$, close to their fixed point values.
It is easy to see that even for TeV scale slepton masses, the Higgs mass is lifted by at most one GeV, while the corrections are negligible when both leptons and sleptons are close to the weak scale. It is possible to obtain larger corrections by adding large $a$ terms to the SUSY breaking Lagrangian. However these soft terms can potentially destabilize the vacuum beyond what will be discussed in the next section, so we refrain from introducing them.
Clearly the new lepton sector can not alleviate the fine tuning problem in the MSSM, and furthermore will not be sufficient to lift the Higgs mass to $125$~GeV in the low $\tan\beta$ case. Instead the necessary contributions could arise from SUSY breaking in the vector-like quark sector that is not being discussed in this paper (see e.g.~\cite{Ajaib:2012eb} for a recent discussion).
\section{Vacuum Stability}\label{sec:vacuum}
To analyze the vacuum structure of our model, we need the full scalar potential for the Higgs scalars and the new sleptons. We assume that all other scalar fields have masses large enough to avoid any stability problems. The scalar potential is obtained from the superpotential as
\begin{align}\label{eqn:potential}
V & = \sum_{\phi_i} \left| \frac{\partial W }{\partial \phi_i} \right|^2 + \frac{1}{2}\sum_{a}g_a^2(\sum_{\phi_i} \phi_i^*T^a\phi_i)^2+ V_{\rm soft} + \text{radiative corrections}\,.
\end{align}
where the sum runs over $\phi_i \in \left\{ h_u^0, h_d^0, \tilde{e}_L',\bar{\tilde{e}}_R',\tilde{e}_L'',\bar{\tilde{e}}_R''\right\}$, while fields that are not in danger of acquiring a VEV, like the MSSM superpartners or the sneutrinos, are omitted.
For the different contributions to the potential, we obtain
\begin{align}
V_F =& \,\left| y_c' \tilde{e}_L' \bar{\tilde{e}}_R' - \mu h_u^0 \right|^2 + \left| y_c'' \bar{\tilde{e}}_R'' \tilde{e}_L'' - \mu h_d^0 \right|^2 + \left| M_L \bar{\tilde{e}}_R'' + y_c' h_d^0 \bar{\tilde{e}}_R' \right|^2 \label{eqn:Fterm} \\
& + \left| M_E\tilde{e}_L'' + y'_c \tilde{e}_L' h_d^0 \right|^2 + \left| M_L \tilde{e}'_L + y_c'' h_u^0 \tilde{e}_L'' \right|^2 +
\left| M_E \bar{\tilde{e}}_R' + y_c'' \bar{\tilde{e}}_R'' h_u^0 \right|^2,
\notag\\
%
V_D =& \,\frac{1}{8} g_2^2 \left( (h_u^0)^2 - (h_d^0)^2 +\tilde{e}'^2_L - \bar{\tilde{e}}''^2_R \right)^2 + \frac{1}{8} g_1^2 \left((h_u^0)^2 - (h_d^0)^2 -\tilde{e}'^2_L + \bar{\tilde{e}}''^2_R + 2 \bar{\tilde{e}}'^2_R - 2\tilde{e}''^2_L \right)^2,\label{eqn:potentialD} \\
%
V_{\rm rad} =&\, \frac{1}{8} (g_1^2 + g_2^2) \delta_H (h_u^0)^4, \\
%
V_{\rm soft} =&\, m_{H_u}^2 (h_u^0)^2 + m_{H_d}^2 (h_d^0)^2 + B h_u^0 h_d^0+ V_{\rm soft,\ell}\,.\label{eqn:potential1}
\end{align}
For simplicity we have assumed that all parameters and all fields are real. The Higgs soft mass parameters $m_{H_u}^2$ and $m_{H_d}^2$ are usually replaced by the parameters $v$ and $\tan\beta$ that characterize the electroweak symmetry breaking minimum,
while $B$ is related to the CP-odd Higgs mass through $m_A^2 = 2 B/\sin(2\beta)$ at the tree level.
For large $\tan\beta$, the correction $\delta_H$ must be roughly $\delta_H \sim 1$ in order to obtain the correct Higgs mass for the SM-like Higgs in the decoupling limit, while for $\tan\beta \sim 2$ a value of $\delta_H \sim 2.5$ is necessary.
As long as the slepton mass matrix has only positive eigenvalues, the above potential will have a local minimum characterized by $v=174$~GeV and $\tan\beta$, with a value for the potential $V(v,\tan\beta) \equiv V_0$. Additional minima with non-zero VEVs for some of the charged slepton fields can be induced by the trilinear terms in~(\ref{eqn:Fterm})~\cite{Rattazzi}.
When the sleptons are heavier than the electroweak scale $v=174$~GeV, these additional minima typically have a potential energy larger than the electroweak minimum, and therefore do not pose a problem. However when the slepton masses are of order $v$ or below, some of the charge breaking minima might be lower than the electroweak minimum. Since relatively light sleptons are required to obtain a large enhancement of $R_{\gamma\gamma}$, it is clear that the vacuum structure of the model must be analyzed carefully.
Let us first consider the absolute stability condition, namely the condition that $(v, \tan\beta)$ is the global minimum of the potential. In order to see the effects of the various parameters on the vacuum stability, it is instructive to derive an analytical result in the limit of $\tan\beta \to \infty$ ($v_d \to 0$) and $\tilde{e}''^2_L,\,\bar{\tilde{e}}''^2_R\rightarrow0$, corresponding to a scenario where large soft mass terms for the mirror sleptons prevent them from acquiring a VEV. Further neglecting soft $a$ and $b$ terms, the following term is added to the MSSM Higgs potential:
\begin{align}\label{eqn:abscond}
V'&=(y_c' \tilde{e}_L' \bar{\tilde{e}}_R')^2-2\mu h_u^0y_c' \tilde{e}_L' \bar{\tilde{e}}_R'+M^2_L\tilde{e}'^2_L+M^2_E\bar{\tilde{e}}_R'^2+\text{D-terms}\,.
\end{align}
\begin{figure}
\center
\includegraphics[width=.45\textwidth]{stabilityMy.pdf}\hspace*{.5cm}
\includegraphics[width=.45\textwidth]{stabilityMMU.pdf}
\caption{Vacuum stability constraints on the $\mu$ parameter. Left: Allowed values for $\mu$ in the $M-y_c'$ plane in the two slepton approximation~(\ref{eqn:mubound}).
Right: Constraints in the $\mu-M$ plane for $y_c'=0.9, y_c''=0.7$ and $\tan\beta=60$. Regions below the solid blue (red) lines are consistent with absolute vacuum stability with four (two) light sleptons. The dashed red line indicates the approximation (\ref{eqn:mubound}), while the orange region is excluded by the LEP limits on lepton and slepton masses.
}
\label{fig:analyticResult}
\end{figure}
The instability is induced by the trilinear term proportional to $y_c'\mu$, whereas all other terms are strictly positive. A necessary (but not sufficient) condition for a deeper charge breaking minimum is $V'<0$. Therefore $V'>0$ is sufficient to guarantee stability of the Higgs potential in this limit. After some manipulations, we find
\begin{align}\label{eqn:mubound}
y'^2_c\mu^2&<\left(\frac{k}{2}+\sqrt{(1+\delta_H)\,k\left(y'^2_c+\frac{k}{4}\right)}\right)\left(M_L^2+M_E^2-v^2\sqrt{(1+\delta_H)\,k\left(y'^2_c+\frac{k}{4}\right)}\right)\,,
\end{align}
where $k=\frac{g^2+g^2_Y}{2}$. If the Yukawa coupling is close to the RGE fixed point, this bound constrains $\mu$ to be at most of the same order as the vector masses. Allowed values for $\mu$ as a function of the Yukawa coupling $y_c'$ and of the vector mass scale $M= M_L = M_E$ are shown in Fig.~\ref{fig:analyticResult} (left).
An sizable enhancement of the di-photon rate from the fermion sector requires $y_c'\sim 0.9$ and $M \sim 200$~GeV, which roughly translates to $\mu \lesssim 250$~GeV.
\
When all slepton fields are included, analytic bounds on the model parameters become impossible to derive. The absolute stability condition can however easily be implemented using numerical minimization routines. An additional constraint on $\mu$ comes from the LEP limit on charged sleptons.
The right panel of Fig.~\ref{fig:analyticResult} shows the different constraints in the $\mu-M$ plane, for $\tan\beta=60$, $y_c'=0.9$, $y_c''=0.7$ and $M_L = M_E = M$. The lighter (darker) orange region is excluded by the conservative (optimistic) LEP limits. Regions below the solid red line are consistent with absolute stability when only two of the sleptons are light, and we see that the analytic approximation (dashed) agrees well with the numerical solution. When all four slepton fields are light, the stability constraint on $\mu$ becomes stronger, as indicated by the solid blue line. Overall it is clear that both the LEP limits and the stability constraints have to be taken into account when values of $\mu\sim M$ are being considered.
\
Since the stability limit on $\mu$ is more constraining than the LEP bound in many cases, it is worth noting that absolute stability is not a necessary condition for the model to be viable phenomenologically. After all, the measured Higgs mass of $125$~GeV implies that the standard model itself is only meta-stable~\cite{Degrassi:2012ry,Bezrukov:2012sa}. This means that while the electroweak minimum is not the global minimum of the radiatively corrected Higgs potential, the tunneling rate to the true global minimum is suppressed enough to guarantee a lifetime larger than the age of the universe.
Imposing the weaker meta-stability bound on our model could allow larger values of $\mu$ which in turn will lead to higher attainable diphoton rates, as was discussed in Sec.~\ref{sec:higgs}. If the electroweak minimum is not the global minimum of the potential, the probability to transition into the true charge breaking vacuum per unit volume and time depends on the decay rate~\cite{Coleman:1977py}
\begin{align}
\frac{\Gamma}{V}&=Ae^{-S_E}\;.
\end{align}
Here $A$ is a dimensionful parameter that is expected to be of fourth order in the electroweak scale, $A\sim v^4$, and $S_E$ is the Euclidean action of the bounce solution corresponding to the transition from Higgs vacuum to the charge symmetry breaking true vacuum. The age of the universe is about $1.37\times10^{10}$ years, which implies that $S_E\gtrsim 400$ is a necessary condition for our existence.
The metastability bound computation is performed using the package CosmoTransitions~\cite{cosmotrans}. It finds the bounce solution for the transition from the false vacuum to the real vacuum of a multi-dimensional scalar potential by the method of path deformation. The results of this computation and its impact on $h\rightarrow\gamma\gamma$ enhancement are presented in Sec.~\ref{sec:results}.
\section{Results}\label{sec:results}
The model has several distinct regions of parameter space that can lead to an enhanced $R_{\gamma\gamma}$. In the following we will discuss the most interesting scenarios and their strengths and weaknesses.
\subsection{Vector-like leptons}\label{sec:veclep}
First we will consider the scenario where only the new leptons are light, while the sleptons are lifted to the TeV scale by the diagonal soft mass terms in~(\ref{eqn:slepmass}). TeV scale superpartners are sufficient to protect the Higgs quartic coupling from running to negative values below the scale where supersymmetry is restored. Furthermore large mass terms for the sleptons protect this scenario from vacuum instabilities and ensure that the electroweak symmetry breaking vacuum is a global minimum.
The conditions for obtaining a non-negligible, positive contribution to $R_{\gamma\gamma}$ are summarized in {Eq.~(\ref{eqn:prefactor2}).} First, we require that the combination $M_L M_E - v^2 y_c' y_c'' \sin\beta\cos\beta$ is positive but not too large, in order to obtain the correct sign for $\Delta_E$. Furthermore $\tan\beta$ must be of order one, otherwise the contribution will be suppressed by $1/\tan\beta$. The latter condition emerges because the effective Yukawa couplings of both leptons and mirror leptons to the lightest CP even Higgs boson are rescaled by $\sin\beta$ and $\cos\beta$ respectively, so their ratio should not be too small.
One immediate concern is that a mass of 125~GeV for the lightest Higgs boson is difficult to obtain with $\tan\beta$ of order one within the MSSM. While we have seen in Sec.~\ref{sec:higgs} that the new slepton sector does not significantly improve the situation, there are other ways to lift the Higgs mass without affecting its low energy phenomenology. The most straightforward solution is to assume that additional one loop contributions beyond those of the top quark come from the vector-like quark sector that accompanies the leptons if they are implemented in complete SO(10) multiplets. Other possibilities include scalar singlet extensions of the MSSM or additional gauge interactions~\cite{Huo:2012tw}. For the remainder of this section, we will assume that one of these mechanisms is at work, but does not otherwise affect the phenomenology of the lightest Higgs boson.
\begin{figure}
\center
\includegraphics[width=7cm]{gagavl1x.pdf}\hspace*{.4cm}
\includegraphics[width=7cm]{gagavl2x.pdf}
\caption{Enhancement of the $h\to\gamma\gamma$ rate from vector leptons only, with charged Yukawa couplings {$y'_c = 0.9$ and $y''_c=0.7$.} Left: Contours of $R_{\gamma\gamma}$ in the $M_L-M_E$ plane (blue, dashed), for $\tan\beta=2$. Also shown are contours for the mass of the lightest charged lepton state (orange, solid). Right: Enhancement $R_{\gamma\gamma}$ as function of the lightest charged lepton mass $m_{E_1}$ for different values of $\tan\beta$.
}
\label{fig:vl1}
\end{figure}
Compared with the results of~\cite{ASW}, we expect that the enhancement of $R_{\gamma\gamma}$ is suppressed roughly by $1/\tan\beta$. Note that $\tan\beta \gtrsim 1.5$ is required to ensure perturbativity of the top Yukawa coupling in the presence of sizable Yukawa couplings for the new leptons. In Fig.~\ref{fig:vl1} left we show the enhancement that can be obtained for $\tan\beta=2$ and $y_c'=0.9$, $y_c''=0.7$, close to their respective fixed point values, in the $M_L-M_E$ plane. While the features of the plot are similar, the maximal enhancement that can be obtained for a lightest lepton mass of order 100~GeV is $R_{\gamma\gamma} \lesssim 1.3$, compared with $R_{\gamma\gamma} \lesssim 1.6$ in the non-supersymmetric case.
To further illustrate the importance of $\tan\beta$, in the right panel of Fig.~\ref{fig:vl1}, we show the contours of $R_{\gamma\gamma}$ in the plane of $\tan\beta$ and $m_{E_1}$, the mass of the lightest new lepton. For masses around 100~GeV the enhancement is reduced from around 30\% at $\tan\beta=2$ to below 10\% for $\tan\beta=10$.
Overall one can see that the cost of imposing GUT scale stability cuts the enhancement of $R_{\gamma\gamma}$ in half. If these new fermions are the only particles that have significants effects on the Higgs phenomenology, only a modest enhancement of $R_{\gamma\gamma}$ of around 30\% is compatible with vacuum stability, perturbativity, and grand unification.
\subsection{Supersymmetric leptons}
Here, we will first assume that at the tree level the only source of SUSY breaking in the new lepton sector is through the $\mu$ term, such that a minimal number of parameters are added to the MSSM. When $\mu=0$ each lepton is accompanied by two sleptons with the same mass. These degenerate states are split when $\mu$ is nonzero. Therefore, in the absence of soft breaking parameters, the mass of the lightest slepton is always lower than the corresponding lepton mass.
\begin{figure}
\center
\includegraphics[width=.47\textwidth]{MLME51.pdf}
\includegraphics[width=.47\textwidth]{MLMu51.pdf}\hspace*{.5cm}
\caption{Diphoton enhancement rate with light leptons and sleptons, for $\tan\beta=60$. Regions with stable, meta-stable or unstable vacuum are shaded white, yellow and red, respectively. Regions excluded by LEP are shaded orange.
The blue (dashed) lines are contours of constant $R_{\gamma\gamma}$. The brown (solid) lines are the slepton mass contours of 90, 150 GeV for the left and 62.5, 90, 150 GeV for the right panel reading away from the unstable region. The left panel uses $\mu=280$~GeV. }
\label{fig:hightan}
\end{figure}
Let us first consider the large $\tan\beta$ limit, where only the sleptons contribute significantly to $R_{\gamma\gamma}$. As before, the Yukawa couplings are taken to be $y_c'=0.9$ and $y_c''=0.7$, and we will usually take $M_L = M_E = M$. Since all sleptons are now light, in addition to the LEP bounds we have to impose vacuum (meta-) stability for the model to be viable. Fig.~\ref{fig:hightan} shows the diphoton rate enhancement that can be obtained for $\tan\beta=60$. Stable, meta-stable and unstable regions of parameter space are indicated in white, yellow, and red, respectively. The orange shaded region is excluded since we required $m_{\tilde{E}_1} > m_h/2$.
Contours of constant $R_{\gamma\gamma}$ are shown (blue, dashed) as well as contours of constant $m_{\tilde{E}_1}$ (brown, solid). The left plot, where $\mu=280$~GeV, demonstrates the usual result that $M_L=M_E$ maximizes the enhancement, and we can see that an enhancement of $R_{\gamma\gamma}$ by 40\% is possible for $m_{\tilde{E}_1} \gtrsim 90$~GeV and meta-stability. It is evident from the structure of the lepton mass matrix given in Eq.~(\ref{eqn:lepmass}) that the lepton masses are well above the conservative LEP bound of $100\,\text{GeV}$ for the region of the parameter space shown here.
The right plot of Fig.~\ref{fig:hightan} shows the interplay of the different constraints as $\mu$ is increased. Initially, the LEP bound is more constraining, and enhancements of at most 40\% are possible with $m_{\tilde{E}_1}>90$~GeV. Following the $m_{\tilde{E}_1}=90$~GeV contour to higher values of $\mu$ we note that $R_{\gamma\gamma}$ increases up to 1.5 before we eventually hit the stability bound. Lowering the lightest slepton mass is only possible for $\mu\lesssim 280$~GeV. In that region, enhancements of 100\% or more seem possible when $m_{\tilde{E}_1}$ is very close to the absolute lower bound of 62.5~GeV.
In Fig.~\ref{fig:lowtan} we show the diphoton rates for $\tan\beta = 2$, while all other parameters are the same as before. Here we expect that the fermions contribute up to 30\% enhancement to the diphoton rate, such that higher total rates should be possible. However it turns out that the vacuum stability constraint forces us to consider lower values of $\mu$, such that the overall enhancement is not significantly higher than in the large $\tan\beta$ case.
\begin{figure}
\center
\includegraphics[width=.47\textwidth]{MLME61.pdf}
\includegraphics[width=.47\textwidth]{MLMu61.pdf}\hspace*{.5cm}
\caption{Same as Fig.~\ref{fig:hightan} for $\tan\beta=2$. In the left panel $\mu=175$~GeV.
}
\label{fig:lowtan}
\end{figure}
More precisely, it can be seen in the left panel that an enhancement of over 50\% is possible with $\mu$ as low as 175~GeV and $m_{\tilde{E}_1}> 90$~GeV, which is not possible in the large $\tan\beta$ case. However the right plot clearly shows that larger values of $\mu$ do not further increase $R_{\gamma\gamma}$, since the stability constraint requires larger values of $M$ at the same time. Moving along the $m_{\tilde{E}_1} = 90$~GeV contour, the ratio of fermionic to scalar contributions, $\Delta_{E}/\Delta_{\tilde{E}}$, decreases from about 0.5 at $\mu = 120$~GeV to 0.3 at $\mu=175$~GeV.
Enhancements of order 100\% are again only possible if we lower the slepton masses below the conservative LEP bound.
\
Since the soft mass terms are set to zero, the vector masses $M_L$ and $M_E$ are needed to lift the sleptons above the LEP bound. We have seen in Sec.~\ref{sec:veclep} that the leptonic contribution is maximized around $M_L = M_E = 200$~GeV. However such low values are not allowed since the mass of the lightest slepton would drop below the LEP limit. It is therefore interesting to explore what happens when we add small soft terms to stabilize the slepton masses, such that both the leptonic and the slepton contributions to $R_{\gamma\gamma}$ can be maximized.
It is instructive to qualitatively discuss the impact of the various elements of the slepton mass matrix before proceeding to finding the maximum possible $R_{\gamma\gamma}$ in the presence of soft terms. The main role of the off-diagonal terms is to increase the difference between the eigenvalues, while the sum of the eigenvalues is given by the trace of the mass matrix. Therefore, the parameters that only appear in off-diagonal terms increase the difference between eigenvalues keeping the sum constant, effectively resulting in lowering the lowest eigenvalue.
\begin{figure}
\center
\includegraphics[width=.47\textwidth]{SoftMuHighTan.pdf}
\includegraphics[width=.47\textwidth]{SoftMuLowTan.pdf}\hspace*{.5cm}
\caption{Diphoton rates with nonzero soft mass terms for the sleptons, $m_i^2 =m_{\rm soft}^2$. The vector mass terms $M_L=M_E=M$ are chosen such that the lightest lepton has a mass around 100~GeV. The left panel shows the result for $\tan\beta=60$ and $M=150$~GeV, while the right panel is for $\tan\beta=2$ and $M=190$~GeV.
Colors same as in Fig.~\ref{fig:hightan}.
}
\label{fig:crazy}
\end{figure}
It is clear from the structure of the slepton mass matrix given by Eq.~(\ref{eqn:slepmass}) that the $\mu$ parameter and the bilinear holomorphic soft terms are only present in the off-diagonal terms. Thus both of these can drive the lightest slepton mass below the LEP bound as well as destabilize the vacuum. On the other hand, as illustrated in Eq.~(\ref{eqn:prefactor1}), these off-diagonal terms are essential to produce $R_{\gamma\gamma}$ significantly higher than 1. These two parameters have similar impact on $R_{\gamma\gamma}$, lightest slepton mass and stability. Therefore, in order to find the maximum enhancement, we set the holomorphic soft terms to zero and achieve the $h\rightarrow\gamma\gamma$ enhancement only through $\mu$ as before.
The non-holomorphic soft terms $m_i^2$ that appear on the diagonal of~(\ref{eqn:slepmass}) can then be used to lift the lightest slepton mass above the LEP limit. For simplicity we will take them to be equal, $m_i^2 = m_{\rm soft}^2$. Fig.~\ref{fig:crazy} shows the enhancement rates that can be obtained for $\tan\beta=60$ (left) and $\tan\beta=2$ (right).
The vector mass scale $M$ is chosen such that the lightest lepton mass is about $100$~GeV, corresponding to $M=150$~GeV ($\tan\beta=2$) and $M=190$~GeV ($\tan\beta=60$), respectively.
In the case of large $\tan\beta$, we note that an enhancement of 50\% is now possible without going below the conservative LEP limit on the slepton masses, an improvement of about 10\% compared to the case without soft terms. Similarly, in the low $\tan\beta$ case, we can now get up to 75\% enhancement of the diphoton rate without making the sleptons dangerously light. Furthermore one should note that the absolute stability limit is less constraining now, and values of $R_{\gamma\gamma}$ of 1.5 and 1.6 are compatible with absolute stability for high and low $\tan\beta$, respectively.
\subsection{Split sleptons}\label{sec:ssl}
In Sec.~\ref{sec:vacuum} we have seen that imposing absolute vacuum stability puts modest constraints on the parameter space when only two charged sleptons are allowed to get a VeV, but that the constraints become strong when all four sleptons are taken into consideration.
Here we will consider a scenario where the sleptons are split by TeV scale soft masses $m_{\tilde{E}_i}$ for the double primed fields $\tilde{e}_L''$ and $\tilde{e}_R''$. This will improve the vacuum stability, since non-zero expectation values for these fields are unlikely to give vacua deeper than the EWSB vacuum.
The remaining light charged degrees of freedom are the two sleptons $\tilde{e}_L'$ and $\tilde{e}_L'$ as well as both the charged leptons. Phenomenologically this is a combination of the vector-like lepton model~\cite{ASW} with the light stau scenario~\cite{Carena:2011aa}.
\begin{figure}
\center
\includegraphics[width=7cm]{gagasl1.pdf}\hspace*{.4cm}
\includegraphics[width=7cm]{gagasl2.pdf}
\caption{Enhancement of the $h\to\gamma\gamma$ rate in the split slepton scenario for $\tan\beta = 2$ (left) and $\tan\beta=30$ (right), with charged Yukawa couplings $y'_c = 0.9$ and $y''_c=0.7$. Shown are iso-contours of $R_{\gamma\gamma}$ (blue, dashed), of the lightest slepton mass (green, solid) and of the lightest lepton mass (red, dotted). Note that values of $\mu$ below $100$~GeV are in conflict with limits on chargino masses. The red shaded area indicates either a meta-stable or unstable vacuum. We do not differentiate between the two here since both regions lie outside of the LEP allowed region.
}
\label{fig:sl1}
\end{figure}
Absolute vacuum stability is guaranteed provided that the inequality~(\ref{eqn:mubound}) is satisfied. As can be seen from Fig.~\ref{fig:analyticResult} the conservative LEP limit on the mass of the lightest charged particle, $m_{\tilde{E}_1} \gtrsim 90$~GeV, is in general more constraining, so that the vacuum stability constraint is automatically satisfied for most phenomenologically viable parameter points. Regions of parameter space that are not absolutely stable will be indicated in the plots, but one should keep in mind that they can still be phenomenologically viable if the much weaker meta-stability bound is satisfied.
We can again distinguish two scenarios, low $\tan\beta$, where leptons and the light sleptons contribute to $R_{\gamma\gamma}$, and the large $\tan\beta$ regime, where only the two light sleptons can make a notable contribution.
Let us first consider the small $\tan\beta$ case. For definiteness we take $\tan\beta=2$ and, as in the previous sections, $y_c'=0.9$ and $y_c''=0.7$. In Fig.~\ref{fig:sl1} (left) we show $R_{\gamma\gamma}$ in the $\mu$--$M$ plane, where $M=M_L = M_E$ is the vector mass scale. As can be seen, the enhancement can be increased both by lowering the lightest slepton mass (i.e. increasing $\mu$) or by lowering the lightest lepton mass (decreasing $M$). When both masses are close to 100~GeV an enhancement of about 40\% can be obtained. Values up to 70\% can be reached by lowering the lightest slepton mass further, while still being consistent with limits from the LEP experiments and with absolute vacuum stability.
When $\tan\beta$ is increased, the leptonic contributions are suppressed. The case of $\tan\beta=30$ is shown in Fig.~\ref{fig:sl1} (right). $R_{\gamma\gamma}$ is now independent of the lightest lepton mass, and an enhancement of 30\% or more can only be obtained when the lightest slepton mass is below 100~GeV, and only for sufficiently large values of $\mu$. Imposing absolute vacuum stability is more constraining here, which disfavors the large $\mu$ region where $R_{\gamma\gamma}$ can be of order 1.5 or higher.
\
In Secs.~\ref{sec:veclep}-\ref{sec:ssl}, we have focussed on the maximal possible values of $R_{\gamma\gamma}$ that can be obtained in each scenario. In contrast to many existing models, we find that at least in some scenarios it is possible to obtain enhancements of more than 50\% while at the same time the new particle masses can be kept above the conservative LEP bound, vacuum stability is maintained and all couplings remain perturbative up to high scales. Nevertheless it is evident from our figures that some amount of tuning of the parameters is necessary to obtain such large values for $R_{\gamma\gamma}$, while more modest enhancements of $20\%-40\%$ are possible in larger regions of parameter space and thus appear more natural. Such an enhancement is well in agreement with the signal strength indicated by the combination of the updated ATLAS and CMS results ~\cite{Chatrchyan:moriond,ATLASmoriond}.
\section{Conclusions}\label{sec:conclusions}
The recent discovery of a Higgs-like particle at the LHC opens a new era in particle physics. A very important task is to study
the properties of this particle in detail, and to analyze any possible deviation from the SM predictions that might signal the presence of new physics.
Currently, the measured Higgs-induced diphoton production rate is 2.3~$\sigma$ above the SM
prediction at the ATLAS experiment~\cite{ATLASmoriond}. This provides a motivation for the study of new physics scenarios that can lead to an enhancement of the
diphoton decay width. Vector-like leptons provide an extension
of the SM that leads to such an enhancement.
Quite generally, the presence of new weakly interacting particles with strong couplings to the Higgs boson can provide an enhancement of the loop-induced diphoton rate, but also lead to the presence of new vacua deeper than the physical one. In the case of vector-like leptons such vacua arise at large values of the Higgs fields due to the evolution of the quartic coupling of the Higgs to negative values. Enhancements of the diphoton rate to values larger than 1.5 times the SM one can only be obtained if new physics stabilizes the vacuum at scales smaller than a few TeV. Supersymmetry provides a natural extension of this model in which the vacuum of the Higgs sector is stabilized by the contributions of sleptons.
In this article we study the supersymmetric extension of the vector-like lepton theory introduced in Ref.~\cite{ASW}. We showed the inclusion of a whole vector-like family can lead to a unified theory with values of the gauge couplings that are close to the non-perturbative bound at scales close to the GUT scale. In order to enhance the effects on the Higgs diphoton decay rate, we chose values of the Yukawa couplings leading to large values at the GUT scale, but still consistent with a perturbative treatment of the theory. Vector-like squarks are considered to be heavy and therefore have an impact only in the determination of the Higgs mass.
The phenomenological properties of this supersymmetric extension depends strongly on the values of the soft breaking parameters. For large values of the scalar soft supersymmetry breaking parameters, the theory at low energies reduced to the one studied before. However, for the same values of the Yukawa couplings, the lepton contributions are suppressed by a $\sin 2 \beta/2$ factor and, together with modified values of the perturbativity bounds on these couplings with respect to the SM ones, the enhancements of the diphoton rate remain smaller than about 30 percent for lepton masses above 100~GeV.
For small values of the scalar soft supersymmetry-breaking parameters, the main source of supersymmetry breaking in the Higgs-slepton potential is provided by the Higgsino mass parameter $\mu$. For light sleptons, large values of $\mu$ tend to induce new charge breaking minima in the spectrum, and therefore in the region consistent with vacuum stability light sleptons are associated with relatively light leptons. It follows that both the fermion and the scalar lepton contributions to the Higgs-induced diphoton production rate tend to be important, except for the large $\tan\beta$ regime where the lepton contributions decouple.
We find that for lepton and slepton masses larger than 100~GeV enhancements of order 50 and 40 percent may be obtained for small and larger values of $\tan\beta$ ($\tan\beta = 2$ and 60), respectively. More generally, since the LEP bound depends on the value of the neutrino mass parameters, one can consider leptons and sleptons as light as 62.5~GeV, for which much larger enhancements of the Higgs diphoton rate may be obtained.
Finally, we considered a split slepton scenario, in which the soft supersymmetry-breaking parameters of the new right-handed leptons are considered to be large, while the left-handed ones are kept small. In such a case, the theory is similar to the light-stau scenario, but the lepton contributions remain relevant. We showed that in such a case enhancements of the Higgs-induced diphoton rate of the order of 50 percent and 30 percent can be obtained for small and large values of $\tan\beta$ for lepton and slepton masses above 100~GeV, while as before larger values may be obtained if these bounds were relaxed.
In order to avoid flavor problems, we have introduced a parity symmetry under which the new states are odd. If in addition R-parity is conserved this guarantees three stable particles: the ordinary MSSM LSP as well as the lightest parity odd leptons and sleptons. Assuming that in each sector a neutral particle is the lightest state, this leads to a scenario with multicomponent dark matter with possibly interesting phenomenological consequences.
Even in the light of the recently presented CMS results~\cite{Chatrchyan:moriond}, there are large regions of parameter space in which the vector like leptons and sleptons are present at the weak scale and remain compatible with data. The masses of the new particles in such region can be small enough to make them interesting to study from the perspective of dark matter and low-scale leptogenesis. They can also be probed at the LHC, what serves as another motivation to study supersymmetric vector-like leptons in detail.
\acknowledgments{We would like to thank C.~Wainwright for discussions. Work at ANL is supported in part by the U.S. Department of Energy, Division of High Energy Physics, under grant number DE-AC02-06CH11357, at EFI under grant number DE-FG02-90ER-40560, and at UIC under grant number DE-FG02-12ER41811.}
\paragraph*{Note added:} While finalizing the manuscript for submission, Ref.~\cite{Feng:2013mea} appeared, which also considers a supersymmetric extension for the vector-like lepton scenario. Different from the present work, the authors do not impose perturbativity of the Yukawa couplings at the GUT scale, and therefore allow larger Yukawa couplings at the electroweak scale. We also have shown that the relevance of vacuum stability constraints depends on the choice of soft breaking parameters.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 4,030 |
package com.jetbrains.python;
import com.intellij.ide.plugins.PluginManagerCore;
import com.intellij.openapi.application.PathManager;
import com.intellij.openapi.diagnostic.Logger;
import com.intellij.util.PathUtil;
import org.jetbrains.annotations.NonNls;
import org.jetbrains.annotations.NotNull;
import org.jetbrains.annotations.Nullable;
import java.io.File;
import java.util.List;
public final class PythonHelpersLocator {
private static final Logger LOG = Logger.getInstance(PythonHelpersLocator.class);
private static final String PROPERTY_HELPERS_LOCATION = "idea.python.helpers.path";
private PythonHelpersLocator() {}
/**
* @return the base directory under which various scripts, etc are stored.
*/
@NotNull
public static File getHelpersRoot() {
String property = System.getProperty(PROPERTY_HELPERS_LOCATION);
if (property != null) {
return new File(property);
}
return assertHelpersLayout(getHelperRoot("intellij.python.helpers", "/python/helpers"));
}
@NotNull
public static File getHelpersProRoot() {
return assertHelpersProLayout(getHelperRoot("intellij.python.helpers.pro", "/../python/helpers-pro"));
}
@NotNull
private static File getHelperRoot(@NotNull String moduleName, @NotNull String relativePath) {
@NonNls String jarPath = PathUtil.getJarPathForClass(PythonHelpersLocator.class);
if (PluginManagerCore.isRunningFromSources()) {
return new File(PathManager.getCommunityHomePath() + relativePath);
}
else {
final File pluginBaseDir = getPluginBaseDir(jarPath);
if (pluginBaseDir != null) {
return new File(pluginBaseDir, PathUtil.getFileName(relativePath));
}
else {
return new File(new File(jarPath).getParentFile(), moduleName);
}
}
}
@Nullable
private static File getPluginBaseDir(@NonNls String jarPath) {
if (jarPath.endsWith(".jar")) {
final File jarFile = new File(jarPath);
LOG.assertTrue(jarFile.exists(), "jar file cannot be null");
return jarFile.getParentFile().getParentFile();
}
return null;
}
private static @NotNull File assertHelpersLayout(@NotNull File root) {
final String path = root.getAbsolutePath();
LOG.assertTrue(root.exists(), "Helpers root does not exist " + path);
for (String child : List.of("generator3", "pycharm", "pycodestyle.py", "pydev", "syspath.py", "typeshed")) {
LOG.assertTrue(new File(root, child).exists(), "No '" + child + "' inside " + path);
}
return root;
}
private static @NotNull File assertHelpersProLayout(@NotNull File root) {
final String path = root.getAbsolutePath();
LOG.assertTrue(root.exists(), "Helpers pro root does not exist " + path);
LOG.assertTrue(new File(root, "jupyter_debug").exists(), "No 'jupyter_debug' inside " + path);
return root;
}
/**
* Find a resource by name under helper root.
*
* @param resourceName a path relative to helper root
* @return absolute path of the resource
*/
public static String getHelperPath(@NonNls @NotNull String resourceName) {
return getHelperFile(resourceName).getAbsolutePath();
}
/**
* Finds a resource file by name under helper root.
*
* @param resourceName a path relative to helper root
* @return a file object pointing to that path; existence is not checked.
*/
@NotNull
public static File getHelperFile(@NotNull String resourceName) {
return new File(getHelpersRoot(), resourceName);
}
public static String getPythonCommunityPath() {
File pathFromUltimate = new File(PathManager.getHomePath(), "community/python");
if (pathFromUltimate.exists()) {
return pathFromUltimate.getPath();
}
return new File(PathManager.getHomePath(), "python").getPath();
}
}
| {
"redpajama_set_name": "RedPajamaGithub"
} | 1,662 |
Q: How to set the width of an element equal to device-width? I want the header of my website to take as much width as the monitor's screen. I've tried giving width: 100%, width: vw and width: device-width. The problem with all these is that when I resize the browser window the header takes width equal to that of current viewport not the device-width. The code is as following:
body {
height: 100vh;
width: 100vw;
}
* {
margin: 0;
padding: 0;
font-family: courier;
color: white;
}
header {
display: block;
height: 150px;
width: device-width;
background-color: #666633;
text-align: center;
}
header h1 {
font-size: 2em;
color: white;
}
nav {
width: auto;
margin: auto;
margin-top: 20px
}
<header>
<h1 class="heading"> Syco Scientist Records </h1>
<nav>
<ul>
<li><a href="anupamindex.html" class="active">Home</a>
</li>
<li><a href="ourwork.html">Our Work</a>
</li>
<li><a href="testimonials.html">Testimonials</a>
</li>
<li><a href="projects.html">Projects</a>
</li>
<li><a href="contact.html">Contact Us</a>
</li>
</ul>
</nav>
</header>
As you can see when the browser window is resized there is no option to scroll. Had I given the width in pixels then the navigation bar wouldn't collapse and could be seen when scrolled to right.
*
*How do set the width of header same as monitor screen width even when the browser window is resized?
Only way I think this is possible is to use @media media queries which set different width for header as per the monitor width.
*
*Is there any analogous of device-width for non-mobile platforms?
A: In order to get the screen size not the browser width you will need javascript. You can use
screen.width
(http://www.w3schools.com/jsref/prop_screen_width.asp)
Then you can use that to set the width of the container with javascript.
Note: you may have some discrepancies between operating systems
A: you can manage screen size using javascript
check this example
<script>
var txt = "";
txt += "<p>Total width/height: " + screen.width + "*" + screen.height + "</p>";
txt += "<p>Available width/height: " + screen.availWidth + "*" + screen.availHeight + "</p>";
txt += "<p>Color depth: " + screen.colorDepth + "</p>";
txt += "<p>Color resolution: " + screen.pixelDepth + "</p>";
document.getElementById("demo").innerHTML = txt;
</script>
A: Just use width: auto. It will adjust your width according to the padding.
A: please try this without using javascript
div
{
width: 100vw;
height: 56.25vw;
background: pink;
max-height: 100vh;
max-width: 177.78vh; /* 16/9 = 1.778 */
margin: auto;
position: absolute;
top:0;bottom:0; /* vertical center */
left:0;right:0; /* horizontal center */
}
or
var width = (window.innerWidth > 0) ? window.innerWidth : screen.width;
A: Use the meta tag first for the hole html body then you can use with : 100%.
A: I will use:
width: 100vw;
display: inline-block;
or I will trying with display: flex; Check you don't have anywhere (on parent) position: relative.
A: Won't work without javascript, but it's only three lines...
Here's a fiddle: https://jsfiddle.net/0yL3fbyf/
I get the screen width, put it in a variable and add 'px' to that number:
var x = screen.width + 'px';
Then I assign that variable as width to the header. I gave the header an ID since for some reason getElementsByTagName("header") didn't work:
var y = document.getElementById('my_header');
y.style.width = x;
A: without jquery/javascript it may be hard to match up the width/height of an element you need. options are either setting 100% or hard coding upon the device screen width/height. Thanks.
A: Relative values like vw(viewport width), vh(viewport height), vm(viewport minimum) are well suited for devices like screens, where the device size and resolution is not known or could change.
px,em & % are the values can be used for screen designs. Try different values and see.
Thanks.
A: There's no way you can have an element occupy the entire screen size just using CSS. However, you can use javascript to circumvent this CSS limitation.
<script>
//Here, "header" is the id assigned to the element you want to have this effect on
var header = document.getElementById("header");
header.style.width = screen.width + "px"
header.style.height = screen.height + "px"
</script>
You can either use height or width or use both for an element, depending on your requirements.
A: Add property height: auto; to header, as it will give your code the flexibility to adjust height according to the screen on which it is used. So, now u use any device, your header height will adjust automatically accordingly.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 4,535 |
Псебайское городское поселение — муниципальное образование в составе Мостовского района Краснодарского края России.
Административный центр — посёлок городского типа Псебай.
Городское поселение образовано законом от 16 сентября 2004 года. В рамках административно-территориального устройства Краснодарского края ему соответствует поселковый округ (пгт с подчинёнными ему 4 сельскими населёнными пунктами).
География
Является самым южным в районе. Граничит на востоке с Карачаево-Черкесией, а через обширные горные местности на юго-западе — с городским округом Сочи, на юге — с Абхазией.
На западе примыкает к Баговскому, на севере — к Бесленеевскому и Шедокскому, на северо-востоке — к Андрюковскому сельским поселениям (сельским округам) Мостовского района.
Население
Населённые пункты
В состав городского поселения (поселкового округа) входят 5 населённых пунктов:
Примечания
Ссылки
Генеральный план Псебайского городского поселения
Городские поселения Краснодарского края
Муниципальные образования Мостовского района (Краснодарский край) | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 1,562 |
{"url":"https:\/\/math.stackexchange.com\/questions\/3315857\/the-cdot-p-norm-will-become-the-maximum-norm-when-p-to-infty","text":"# The $|\\cdot|_{p}$ norm will become the maximum norm when $p \\to \\infty$\n\nI'm trying to prove the $$|\\cdot|_{p}$$ norm will become the maximum norm when $$p \\to \\infty$$.\n\nLet $$\\mathbb K$$ denote $$\\mathbb R$$ or $$\\mathbb C$$, and $$x= (x_1, \\ldots, x_m) \\in \\mathbb K^m$$. Then $$\\lim_{p \\to \\infty} \\left ( \\sum_{i=1}^m |x_i|^p \\right )^{1\/p} = \\max _{1 \\leq i\\leq m} |x_{i}|$$\n\nCould you please verify whether my attempt is fine or contains logical gaps\/errors? Any suggestion is greatly appreciated!\n\nMy attempt:\n\nIt suffices to prove the statement in case $$x \\in \\mathbb {(R^+)}^{m}$$, where it becomes $$\\lim_{p \\to \\infty} \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} = \\max _{1 \\leq i\\leq m} x_{i}$$\n\nLet $$l:= \\max _{1 \\leq i\\leq m} x_{i}$$. We have $$l = (l^p)^{1\/p} \\le \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} \\le (ml^p)^{1\/p} = m^{1\/p}l$$\n\nThen $$l = \\lim_{p \\to \\infty} l \\le \\lim_{p \\to \\infty} \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} \\le \\lim_{p \\to \\infty} (m^{1\/p}l) = l$$\n\nand thus by squeeze theorem $$\\lim_{p \\to \\infty} \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} = l$$\n\nThis completes the proof.\n\n\u2022 Hi @AlexR., Please elaborate on why we should subscript $l_m$. I think I well defined $l$ by $l:= \\max _{1 \\leq i\\leq m} x_{i}$. \u2013\u00a0LAD Aug 7 '19 at 4:56\n\u2022 @LeAnhDung I think Alex's comment is bad and that part of your proof is fine. I do, however, think your proof is flawed in that it works with limits before establishing the limit exists. You should be working with liminf's and limsup's \u2013\u00a0mathworker21 Aug 7 '19 at 5:02\n\u2022 @LeAnhDung I mean, it comes down to how you phrase it. Once you have $l \\le (\\sum x_i^p)^{1\/p} \\le m^{1\/p}l$, you can say: since $\\lim_{p \\to \\infty} l = l$ and $\\lim_{p \\to \\infty} m^{1\/p}l = l$, the squeeze theorem implies $\\lim_{p \\to \\infty} (\\sum x_i^p)^{1\/p} = l$. That's what the squeeze theorem says. My objection in my previous comment was that you can't immediately go from$l \\le (\\sum x_i^p)^{1\/p} \\le m^{1\/p}l$ to $\\lim_{p \\to \\infty} l \\le \\lim_{p \\to \\infty} (\\sum x_i^p)^{1\/p} \\le \\lim_{p \\to \\infty} m^{1\/p}l$, since you don't know if the limits exist (yet). \u2013\u00a0mathworker21 Aug 7 '19 at 9:37\n\u2022 Yes it's correct. Good job! However, that is not what I suggested. I clearly suggested to use liminf's and limsup's. You should go from $l \\le (\\sum x_i^p)^{1\/p} \\le m^{1\/p}l$ to $l = \\liminf_p l \\le \\liminf_p (\\sum x_i^p)^{1\/p} \\le \\limsup_p (\\sum x_i^p)^{1\/p} \\le \\limsup_p m^{1\/p}l = l$, so $\\liminf_p (\\sum x_i^p)^{1\/p} = \\limsup_p (\\sum x_i^p)^{1\/p}$ and thus the limit exists (and is then equal to $l$). \u2013\u00a0mathworker21 Aug 8 '19 at 13:41\n\u2022 @LeAnhDung no problem. upvoted your answer :) \u2013\u00a0mathworker21 Aug 8 '19 at 14:53\n\nOn the basis of @mathworker21's suggestion, I added a proof that $$\\lim_{p \\to \\infty} \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p}$$ exists here. It would be great if someone helps me verify my attempt.\nLemma: $$\\left( \\sum_{i=1}^m (x_{i})^{p} \\right)^{q} \\ge \\left( \\sum_{i=1}^m (x_{i})^{q} \\right)^{p}, \\quad (x_1, \\ldots,x_m) \\in {(\\mathbb R^+)}^m, \\quad p,q \\in \\mathbb R^+, \\quad p \\le q$$\n$$\\text{Let } l:= \\max _{1 \\leq i\\leq m} x_{i}.\\ \\text{Then } \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} \\ge (l^p)^{1\/p} = l.\\ \\text{As such, the sequence}$$\n$$\\left \\langle \\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} \\right \\rangle_{p \\in \\mathbb N}$$ is bounded from below. Next we prove that this sequence is decreasing by showing $$\\left ( \\sum_{i=1}^m (x_i)^p \\right )^{1\/p} \\ge \\left ( \\sum_{i=1}^m (x_i)^{p+1} \\right )^{1\/(p+1)}$$ or equivalently $$\\left ( \\sum_{i=1}^m (x_i)^p \\right )^{p+1} \\ge \\left ( \\sum_{i=1}^m (x_i)^{p+1} \\right )^{p}$$, which is true by our Lemma. This completes the proof.","date":"2020-07-02 07:11:44","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 1, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 0, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 19, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.9468549489974976, \"perplexity\": 173.96815984835115}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2020-29\/segments\/1593655878519.27\/warc\/CC-MAIN-20200702045758-20200702075758-00578.warc.gz\"}"} | null | null |
This page has moved to:
- [JLBP-11 Stay up to date with compatible dependencies](https://googlecloudplatform.github.io/cloud-opensource-java/JLBP-11.html)
| {
"redpajama_set_name": "RedPajamaGithub"
} | 8,137 |
{"url":"https:\/\/shiyuzhao.wordpress.com\/2011\/04\/04\/no-positive-definite-matrix\/","text":"It is well known that: Let $A$ be a symmetric matrix, if $x^TAx>0, \\forall x\\in R^n$, then $A$ is positive definite and all this eigenvalues are positive.\nNow my question is: if there exists at least a vector $x\\in R^n$, $x^TAx>0$; and there exists at least a vector $y\\in R^n$, $y^TAy<0$, does $A$ has at least one positive eigenvalue and at least one negative eigenvalue?","date":"2018-07-17 07:35:27","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 0, \"img_math\": 8, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.9799917936325073, \"perplexity\": 43.2435445184989}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 20, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2018-30\/segments\/1531676589618.52\/warc\/CC-MAIN-20180717070721-20180717090721-00213.warc.gz\"}"} | null | null |
Algo.ai of Troy, Michigan advertises itself as the world's first supply chain analyst work bot. What does that mean exactly? It means their technology combines artificial intelligence (AI), augmented reality (AR), and workflow automation to provide unique solutions for retail and distribution industries.
The startup's founder and CEO Amjad Hussain has more than two decades of experience in building AI. He's now been running Algo.ai for more than three years and says its important to realize that the next great startup can be built in a place like Troy, Michigan.
Algo.ai has created an international division based in London.
The company has named Aodan Coburn, formerly EVP International at Sony Pictures Home Entertainment, as president, Algo.ai International. Miguel Geli, previously division CIO for Sony Pictures and European MD of Sony Retail Services Europe, has been appointed COO, Algo.ai International.
"Algo.ai is a company with significant global potential which I am thrilled now to be part of," Geli added in a statement.
Algo.ai (Algomus) is featured in the updated 2018-2019 edition of TopBots' Enterprise AI landscape which highlights the most interesting companies currently using AI to tackle business challenges.
Since TopBots first started tracking applied AI companies in 2017, the industry has grown tremendously and we're thrilled to be included in the list of innovators continuing to enable novel business applications of machine learning.
Artificial Intelligence (AI) and Augmented or Virtual Reality (AR and VR) are quite easily some of the most promising forms of technology within our modern world, to be rivaled only with blockchain. It may come as no surprise that many brands and industries are scrambling to implement these technologies in their products and services. This year, it is expected that 31% of enterprises will start using Artificial Intelligence. What's more, according to a survey done by ISACA, 64% of US consumers believe that augmented reality enhancements would benefit the workplace.
Benzinga reporter and editor Dustin Blitchok visited our Detroit HQ and found out all about what we are building.
Formerly known as Algomus Inc., Algo.ai allows users, including those with little knowledge in data science, to analyze data and find new insights.
Troy, MI, September 26 2018 – Algo.ai, the company connecting Artificial Intelligence (AI), Augmented Reality (AR) and automation to retail and entertainment industries, today announced that it has joined the NVIDIA Inception program. The program is designed to nurture startups aiming to revolutionize industries with advancements in AI and data sciences.
Formerly known as Algomus, Inc., Algo.ai provides an easy-to-use conversational Business Intelligence (BI) and Workflow Automation platform that enables people to work to their full potential by learning to take over tedious and redundant tasks and collaborate using a single source of truth.
The platform uses a wide range of AI and Machine Learning (ML) architectures — including prediction, prescription, time series forecasting, natural language processing (NLP), and robotic process automation (RPA) — to allow for seamless and robust smart human-in-the-loop workflows. With an easy-to-use conversational interface, Algo ensures that users without any background in data science can make data-driven decisions and become more strategic and analytic in their day-to-day tasks.
NVIDIA's Inception program is a virtual accelerator that helps startups during critical stages of product development, prototyping and deployment. It will provide Algo.ai with expanded computing and development resources to accelerate R&D of specialized deep learning architectures and implementing new ML and AR features into production as they continue to onboard enterprise clients in the retail and entertainment industries in the US and Europe.
Machine Learning is transforming retail and other industries by allowing companies to discover patterns in supply chain data. Algo.ai's latest customers include powerhouses in the movie and music industries, which bring unprecedented opportunities to develop AI-powered supply chain solutions tailored to the unique and complex supply chain of entertainment and consumer products. Joining the NVIDIA Inception program will further the development of the Algo platform, revolutionizing the way businesses handle inventory levels, supplier quality, and more.
Algo.ai has created Algo, the world's first supply chain analyst bot. Algo connects Artificial Intelligence, Augmented Reality, and Automation in a unique way to a specific set of industries: retailers, distributors, and manufacturers. The company's deep domain knowledge of both technology and supply chain allows the team to rapidly train Algo on analytically rich workflows and business functions like Demand and Inventory Planning, Sales Forecasting, New Product Lifecycle Management, Assortment Optimization and more. Algo.ai is headquartered in metro Detroit with operations in the US and Europe.
But in the home entertainment business, as in the broader entertainment industry, data is increasingly regarded as the crown, throne and scepter.
Data analytics, quite simply, is predicting future consumer behavior based on as much information about your target audience as you can get. Data comes into play when deciding windows, store allocations, distribution platforms, marketing campaigns, even greenlighting movies.
At the AI in Industrial Automation Summit this June 26 – 27, team Algo will be joining RE•WORK to exhibit and chat to attendees about their current work. In advance of the summit, Ellie Lucy, Global Summit Creator at RE•WORK spoke with Nikki to hear about her current work and their use of AI and deep learning. | {
"redpajama_set_name": "RedPajamaC4"
} | 1,658 |
You are here: Home / Featured / 4 running to fill Wilma Chan's seat on Alameda County board of supervisors
4 running to fill Wilma Chan's seat on Alameda County board of supervisors
May 12, 2022 by Keith Burbank, Bay City News Service Leave a Comment
Four people are running to fill a seat on the Alameda County Board of Supervisors that became vacant following the death of Supervisor Wilma Chan.
Chan's former chief of staff David Brown is serving as supervisor for her district until the election is held. The Alameda County Taxpayer's Association has mounted a challenge to the legality to Brown's appointment to the board.
Running in the June 7 election to take over for Brown are Oakland's Vice Mayor Rebecca Kaplan; Surlene Grant, a former vice mayor of San Leandro; David Kakishiba, executive director of the East Bay Asian Youth Center, and Lena Tam, a former vice mayor in the city of Alameda.
Kaplan was elected in 2008 to the at-large seat on the Oakland City Council. She is a graduate of MIT and Stanford Law School.
Grant was the first Black city councilmember in San Leandro and has a meeting space in the city named after her. Currently, Grant is a small business owner.
Kakishiba has been the executive director of the East Bay Asian Youth Center for 42 years. The center provides education, employment, counseling and support to families of 2,500 children in Oakland.
Kakishiba is also the author and co-founder of Oakland's Kids First! initiative, which requires the city to allocate 2.5 percent of its General Fund revenue annually to youth and children.
Tam is currently the manager of the Water Planning Department at the East Bay Municipal Utility District, which provides the East Bay with drinking water and wastewater service.
Filed Under: Featured, Featured Top Story, National News, News Tagged With: politics | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 4,692 |
On Psychological & Influence Ops in the Info Age: Q in the Crosshairs
March 23, 2021 – There have been a lot of discussions recently surrounding "Q" or "Qanon" and the possible identity of Q, especially since the recent HBO documentary by Cullen Hoback, "Q: Into The Storm." One of the things I find fascinating while mystifying is the lack of research and/or consideration of a very real possibility which reviewers, critics, researchers and mainstream media (MSM) journalists somehow miss. Supposedly for three years the MSM has spent thousands of hours of time and invested massive resources to investigate the phenomena, and in all this time, no one has considered the possibility of a psyop? It almost seems as though they are actually avoiding this topic like the plague. Could there be a reason for the media to be intent on covering up certain connections between the world of intelligence and information warfare and the Q phenomena?
After spending three years as a dedicated Q researcher since the very early days of the Q posts, I believe this is an important line of inquiry that can and must be investigated. Considering that we live in the information age, where our data is syphoned and sold, and we know constant manipulation, nudging and micro-targeting will follow, this topic must be explored fully.
So, to begin … what is a Psychological Operation?
Our military employs psychological operations to support field operations and strategic goals. Psyops are no longer limited to foreign audiences, and the paper MindWar, by Gen. Paul E. Vallely and Col. Michael Aquino (the Satanist) make this very clear.
One of the things that is discussed in this paper is how the Vietnam War was lost in the hearts and minds of the American people at home. The logo at the top says "support by truth" and indeed the MindWar paper discusses using truth to gain the trust of the target audience and to develop a perception as being trustworthy.
Aquino explains that before the Vietnam war, the use of psychological operations were limited and almost disinterested. He believed that by adopting MindWar, a battle could be won without any bullets ever being fired. This new battlefield would be within the minds of men.
Aquino here states that after the first publication of the MindWar doctrine, he began to see it being increasingly employed and adopted by the U.S. Military. He rationalizes this as being better than kinetic war, because it is "non lethal." However I believe he minimized the profound impact it has on the minds of its victims. He wrote that MindWar was utilized on U.S. journalists by embedding them with military units, thus changing their perceptions and perspectives. Those journalists would then influence the minds of the American public with their reporting from the "field" and effect a slow adoption of the military mindset. Aquino claimed that MindWar must be using truth; it must "deliver the goods" as he states, in order to be well received by the target audience. Pay close attention to what he says about the powerful and intense emotions and commitment that MindWar evokes in the minds of the target audience, and that the affects of a failed MindWar campaign could be devastating because of this. He says it could be "socially shattering" and I find it interesting that his co-author of this paper, Gen. Vallely has spoken about 'Q' and claimed that Q was in fact a group from the 'army of northern Virginia.'
How is it then that the media and the researchers who claim to have spent so much time looking into the Q phenomena missed this big red flag? The co-author of the MindWar doctrine confirms that Q is military, and said this on a Canadian Radio program and not one of these MSM journalists thought it was worthy of even a mention? This is important because if Q is part of a MindWar campaign, then they would be using the truth in their operation. When I talk to people who follow Q, one of the main reasons they support it so much is because it tells the truth. But if Q is a MindWar campaign evoking intense emotions and commitments in its target audience, a characteristic I definitely see in the Q movement, then the effects of a failed MindWar campaign could be truly bad. If you have primed people to expect total victory for Trump, for example and he doesn't win in the end, then the people who pinned all their hopes onto that are going to be heartbroken, angry, confused and devastated.
This part of the paper described a "deliberate and aggressive convincing of the participants that we will win this war." Substitute the word war for 'election' and things become quite disturbing.
The paper states that MindWar always speaks the truth and that its greatest effectiveness is its skillful use of communications media, an obviously expert talent of Q. They were able to amplify tweets by "Qing" people, having all the followers then boost the tweet by widely sharing it, and then comment that "Q sent me", etc.
Aquino stated, unlike PSYOP, MindWar does not utilize deception or "selected" and/or "misleading" truth, it uses the WHOLE truth and the MindWar operative must be personally committed to it.
Now, to be clear, I am not stating for a fact that Q is a MindWar operation. I am saying that it is certainly a possibility and one we should be considering when we research the information. It appears very strange that the mainstream media totally ignores this, and explains away Q with their favorite "dude in his parents' basement" canard. The truth of the matter is that Q clearly is well versed in military terminology, appears to be close to the Trump administration and has an obvious understanding of information warfare. All this lends credence to the idea that this is or could be a real military information operation, which does not make it automatically "good" or "bad."
It seems apparent that Q knows how to utilize anons as a "force multiplier" amplifying certain information and researching certain topics. It would appear that Q was using parallel deconstruction of the SpyGate scandal, and detailing how this was run by the FBI against the Trump camp, in coordination with Hillary Clinton's campaign. Q seems tied to the SpyGate saga. Look at the first post Q made — it was October 28, 2017 just one day after Rick Gates and Paul Manafort were indicted. Again, this is ignored by the mainstream media who seem intent on obfuscating these sorts of connections.
One Theory Emerges
I have come up with a theory for IF 'Q' was some sort of psychological operation, what was the purpose, and why. Here are those bullet points, and please keep in mind this is hypothetical and has not yet been proven, but I believe it all these unanswered queries warrant more investigation:
1. UAE, Saudi Arabia, and Israel got in touch with Trump because of bad blood with Obama and a desire to shift the balance of power in the Middle East. There are religious & economic forces driving this alliance. Energy was a major economic driver. Think about the Middle East Abraham Accords agreements achieved, and perhaps the creation of a gulf coalition against Iran.
2. The Mercer Network acted as a go-between for a lot of these meetings. Prince, Papadopoulos, etc. See Cambridge Analytica and SCL Group for more information on this network. Think back channels, Kushner was probably involved with this as well.
3. A shadow media team was built using SCL Group assets; Alex Jones, Project Veritas, OAN etc. Remember, these media outlets are not required to be aware that they are being used for support operations.
4. Roger Stone was sent out to make relationships with alt-media and read people in. He's been encouraging Trump to run for years and is an old fashioned dirty political trickster; but also a Zionist. He's probably compromised via blackmail as he's a swinger. Notice the pattern of sexual degenerates? This is a tactic that Kushner's father used to control people.
5. PsyGroup ran the initial campaign, probably using Alexsandr Dugin's Internet Research Agency (IRA) as an independent contractor. PsyGroup was small & not profitable; they probably outsourced most of the actual work. This created the Russia Collusion narrative, which may have been intentional; a plausible deniability play. The backlash highlighted the need for a new tactic that looked more organic. Remember, when Mueller's team began their investigation into PsyGroup, the company dissolved and was shut down. Why? We will come back to PsyGroup later.
6. SCL either partnered with PsyGroup on Q or managed it themselves. This is where things get tricky because we don't have a lot of data & there are so many shell companies. However it is interesting that WhiteKnight, a PsyGroup subsidiary, was located in the Philippines (enter potential Watkins family connections). The people who run 8Chan, now 8Kun, are also located in the Philippines. Were they somehow recruited by WhiteKnight?
7. George Nader using UAE money paid PsyGroup for the first campaign (they were paid $2 million). Erik Prince has been a routine go-between, among the campaign & these shadow networks. He's part of the Mercer network & may be the guy in charge of managing the different assets. We know Prince lied to a congressional subcommittee. Why? What was he hiding? We also know that the Q posts have discussed Prince and Blackwater in positive terms.
8. Using Cambridge Analytica data & the expertise of any of these private intel agencies, Q was launched to create an organic, grass roots troll farm. It tapped into several conspiracy tropes & united them into a world view. We cannot discount this possibility knowing that Cambridge has around five thousand unique data points on each U.S. Voter, and they specialize in micro-targeted psychological manipulation. They know your greatest fears, desires, insecurities, what you would respond to, etc. better than you do.
9. Jerome Corsi and other members of the shadow media network were sent in to help amplify it. Think Jack Posobiec, Jerome Corsi, etc. This might even be where Defango & Unirock come in. Then they all distanced themselves from it. Even Tracy Beanz & Paul Furber. Plausible deniability plays. Though the alt-media shadow networks continued at various times to use language that signaled to Q followers, and pointed people who might be unfamiliar with it, to it if they saw it via social media. A form of twilight language. We're seeing A LOT more of that happening right now.
10. The shadow media network crafted narratives from some sources like Tucker, Veritas, etc. and amplified them. Q followers and InfoWars fans were sharing the same narratives from the same sources just in different communities and with different "framing," thus giving the illusion of independence and corroboration, though really it's all coming from one source.
It ties back to the private intelligence networks which hack, blackmail, dumpster dive, plant false narratives, etc.
This is simply one possibility if you are analyzing all the information and looking at information that came out during the Mueller investigation via depositions and discovery.
The Daily Beast published an article about PsyGroup, "Psy Group Speed Read: 6 WTF Bits From The New Yorker Exposé on Mueller Probe's Ex-Mossad Spy Group." It claims:
Even for the world of covert intel operations, this one crosses boundaries in both fake-personae espionage and epic influence failure.
In 2017, the notorious Israeli spy-for-hire outfit known as Psy Group, known for its alleged role in social-media manipulation ahead of Donald Trump's election and its spot on Special Counsel Robert Mueller's radar, got entangled in a local election in an obscure part of central California, after a Bernie Sanders die-hard convinced his immigrant mother to try to unseat one of the town's hospital-board members.
But apparently it wasn't the only time the Mossad-linked intelligence group tried to expand its practice in the U.S. On Monday morning, The New Yorker published 'Private Mossad for Hire' by reporters Adam Entous and Ronan Farrow, which unravels additional disturbing details about the Psy Group's other work in America, including a proposition to sabotage campus opponents of Israel and that out-of-proportion attack on the candidate in Rep. Devin Nunes' home district.
Here, a look at the six most WTF bits in the report about this shadowy group.
The New Yorker reports the group's owner Joel Zamel asked Trump ally and former Speaker of the House Newt Gingrich to get Trump's son-in-law Jared Kushner to sign on for the services that included 'online deception.' In one draft strategy from early 2016, it promised to exploit their powers of sowing deceit to more than 50 political groups including the Republican National Committee, the Democratic National Committee, and major super PACs that were deemed influential among voters. By controlling the messages these groups were sending to potential voters, they promised to easily sway first the Republican primary and then the general election.
Entous and Farrow report Kushner checked around with the team, including Brad Parscale, who was in charge of the Trump campaign's web-based strategies, but concluded that they didn't need Psy Group's expertise.
The company's glossy publicity campaign included printed brochures with a price list for services like 'honey traps,' which they depicted with a cartoon cat casting the shadow of a lion to refer to using a sexy spy to get information from various targets, according to the report. It also used a goldfish with a shark fin attached to its back to back up its motto: 'Reality is a matter of perception.' The cost of these services? An average package price of $350,000, or just $275 an hour. – The Daily Beast
This article claims that Brad Parscale turned down PsyGroup as far as the election goes. But what about after the election, when the RussiaGate scandal began? Why did George Nader end up paying PsyGroup $2 million?
Another article, by Haartez looks into Joel Zamel, the guy behind PsyGroup.
It claims that computers were seized and then the company ended up shutting its doors.
As you can see, PsyGroup boasted of having "Deep Web" capabilities.
An article from Global Research claims that 'Q' could have been an FBI PsyOp, citing certain claims from a Reuters report:
A recent Reuters investigation may indicate that "Q Anon" was in fact an FBI cyber psyop.
The 'Q Anon' phenomenon has generally been regarded as a hoax or prank, originated by online message board users in late October 2017, that got out of control. The 'Q Anon' persona was preceded by similar personae, including 'FBI anon', 'CIA anon' and 'White House insider anon'.
'Q Anon' originally called himself 'Q clearance patriot'. Former CIA counterintelligence operative Kevin M. Shipp explained that an actual 'Q clearance leaker' – i.e. someone possessing the highest security clearance at the US Department of Energy, required to access top secret nuclear weapons information – would have been identified and removed within days.
However, in November 2020 Reuters reported that the very first social media accounts to promote the 'Q Anon' persona were seemingly 'linked to Russia' and even 'backed by the Russian government'. For instance, the very first Twitter account to ever use the term 'Q Anon' on social media had previously 'retweeted obscure Russian officials', according to Reuters.
These alleged 'Russian social media accounts', posing as accounts of American patriots, were in contact with politically conservative US YouTubers and drew their attention to the 'Q Anon' persona. This is how, in early November 2017, the 'Q Anon' movement took off.
But given the recent revelations by British investigator David J. Blake – who for the first time was able to conclusively show, at the technical level, that the 'Russian hacking' operation was a cyber psyop run by the FBI and FBI cyber security contractor CrowdStrike – the Reuters report may in fact indicate that 'Q Anon' was neither a hoax nor 'Russian', but another FBI psychological cyber operation.
Of note, US cyber intelligence firm New Knowledge, founded by former NSA and DARPA employees and tasked by the US Senate Intelligence Committee, in 2018, with investigating alleged 'Russian social media operations' relating to the 2016 US presidential election, was itself caught faking a 'Russian social media botnet' in order to influence the 2017 Alabama senate race.
If the 'Q Anon' persona – similar to the Guccifer2.0 'Russian hacker' persona played by an FBI cyber security contractor – was indeed an FBI psychological operation, its goal may have been to take control of, discredit and ultimately derail the supporter base of US President Trump. In this case, the 'Q Anon' movement may have been a modern version of the original FBI COINTELPRO program. – Global Research
I am not very fond of Kevin Shipp, so I don't trust or believe anything he says; in my opinion, once CIA, always CIA. This theory doesn't really add up, to me it seems far more likely this was an operation using private intelligence groups like PsyGroup, who employ former Mossad and SIGINT operatives.
Another article brings up the LtGen Michael Flynn connection with these private Israeli intelligence firms, "Israeli Spyware Firm Embroiled in Mexico Mobile Hacking Scandal. Flynn Was Its Adviser" claims he advised NSO Group:
An Israeli company's software has been used to infiltrate mobile devices held by human rights lawyers, journalists and activists fighting government corruption in Mexico, according to a report published in the New York Times on Monday.
The highly-advanced software, known as Pegasus, is only sold to governments by the Israeli firm NSO Group on condition that the cyber technology be used in anti-terror or anti-criminal intelligence efforts. The company is known for its cyber expertise and according to the report, 'charges $650,000 on top of a flat $500,000 installation fee,' to spy on 10 iPhone holders. The NSO Group has also run operations under different names like OSY Technologies, which paid former U.S. national security adviser Michael Flynn over $40,000 as an advisory board member for nearly a year, until last January.
According to the New York Times, 'at least three Mexican federal agencies have purchased about $80 million worth of spyware' from the Israeli company, and has used it to fully monitor and control mobile activity, including calls, texts, emails, contacts, calendars, microphones and cameras, against civilians critical of the government, most likely without the proper legal permits from a federal judge.
The revelations in Mexico however, aren't the first time the NSO Group has been identified as the source of malicious software used to spy on human rights activists and other civilians, most likely by their governmental clients. In August, 2016, researchers in the U.S. claimed that the firm's technology was used against a political dissident in the United Arab Emirates, a journalist in Mexico and a minority party politician in Kenya. – Haaretz
When we examine the images Q posted, these seem to be capabilities that potentially Q would have. Could they have been taken from someone else's phone? There was a time Q allegedly told Lynn Forrester Rothschild "we can hear you breathing" and this software could lend credence to the notion that they could.
An article published around November of 2020, by Daniel Morrison that claims that "Qanon" is "propaganda" and that he believed he had figured out who is behind it. I should let everyone know that I think this writer is more left-leaning:
The influence of perception is one of the defining features of modern life. Ever since Thomas Barratt used a nice painting to make Pears Soap more appealing, people have competed to change our minds about things. De Beers created the idea that their product, diamonds, were an essential part of marriage. Edward Bernays put cigarettes in the hands of marching suffragettes to make them 'Torches of Freedom'.
The social cost seems secondary. Millions of lives lost to lung cancer is less important than the pursuit of profit or power to these people. And when government health officials declare that their product is dangerous, these industries will do what they can to discredit them. Ultimately, unregulated capitalism is framed as 'freedom', and any attempt by the state to curb that is framed as 'tyranny'. Therein lies the fundamental tension between progressives and conservatives.
Donald Trump was a nepotistic narcissist with a history of bankruptcies, known for tax fraud and partying with pedophiles, who openly bragged about sexual assault.
His campaign would require the creation of a completely Alternative Reality. One where he was a strong, powerful, righteous hero, on a selfless mission to save the world from the clutches of an evil cabal of democratic deep state operatives. Not only was he rescuing children from satanic sex trafficking rings, he was even going to lead the world in a spiritual ascension to the next galactic realm.
Looming large amongst this mess is a curious character called 'Q', who claimed to be from military intelligence, with inside information on how it was all going to go down.
Let's start by going back to 2016. Roger Stone had finally succeeded in making Donald Trump the Republican candidate for President. Stone's whole career (and partnership with Paul Manafort) had been a long list of shameless dirty tricks, and this job was going to be no exception. And in an election, political propaganda isn't just about building your guy up. Taking the other side down is just as effective. And so he set his sights on the democratic candidate: Hillary Clinton.
Jeff Giesea was a digital businessman, Pro-Trump agitator, and associate of Peter Thiel who, drawing on the ideas of Chuck Johnson, published a paper called 'It's time to embrace memetic warfare'. In context, it was pitched as a way to fight ISIS. In practice, it would become a key part of the Trump Campaign's strategy.
'Psychological Operations' have been around for a while, as retired Lt. Colonel MichaelAquino has explained. While they were usually the domain of the military, there is now an entirecommercial industryserving this space. Psy Group is an Israelicompany founded by Joel Zamel, which specialised in online perception management, through social media manipulation, influence campaigns, opposition research, honey traps, as well as clandestineon-the-groundactivities.
In April 2016, during the Republican primaries, Trump Campaign offical Rick Gates asked Psy Group for a proposal. They responded with a quote for $3,125,000 plus media costs, and promised to make it virtually untraceable.
Three months later, ErikPrincearranged a meeting between Zamel, Donald Trump Jr., and George Nader (an emissary for two wealthy Arab princes) in Trump Tower. Shortly after, Nader paid Zamel $2 million.
While FBIanon was doing the anonymous work in the dark corners of 4Chan, Roger Stone was hard at work out front spinning the story with Alex Jones. It all came together, and Trump went on to win the White House.
The influence campaign obviously didn't stop after the election. There was an administration to sustain, and an eventual re-election to run for, amid mounting investigations into their naked corruption.
In late October 2017, the day after Paul Manafort and Rick Gates were indicted, another character appeared on 4Chan, who would come to be known as 'Q'.
There are a few things that set it apart from other LARPs on 4chan. The first is that, like FBIanon, it is so dedicated in driving support for the candidate, that it is functionally indistinguishable from a professional campaign anyway. The second is that instead of answering questions, it asked them. This pointed the narrative outward, into the world, and allowed the readers to fill in the ambiguous gaps and have a hand in creating the story.
This is an ARG-based propaganda campaign in full flight, created by people using sophisticated psychological manipulation techniques.
The foundation of the strategy is explicitly articulated in a pitch deck from Wikistrat, another digital influence company founded by Joel Zamel.
A campaign like this is powerful of course, but it's not much use if it doesn't reach the right people. And this is where companies like Cambridge Analytica (or Emerdata, or whatever else SCL want to re-brand themselves as), and Gloo, come in to the picture. As well as creatingcampaigns, they use unprecedented volumes of Big Data to aim content at the right market with surgical precision for maximum impact. And at the end of 2016, CA signed a memorandum of understanding with our good friends at Psy Group.
Bijan Kian was a partner of the Flynn Intel group. He also apparently introduced Flynn to Joel Zamel. This, I believe, is the final link. The connection between the The Flynn Intel Group, an Influence Agency, a Puzzle Guy, and a New Age Fascist.
Here we need to have a look at Michael Flynn. He joined the army in '81 and made his way up the ranks until Obama nominated him to be the director of the Defense Intelligence Agency in 2012. In 2013, GRU chief General Igor Sergun invited him inside the Russian military intelligence (GRU) headquarters in Moscow. A first for any U.S officer. Flynn wanted to return the favour and invite high-ranking GRU officials to the U.S, but the Director of National Intelligence, James Clapper, said no. Around the same time he formed a relationship with a young guy called Ezra Cohen-Watnick, who has recently been made Acting Undersecretary of Defence for Intelligence and Security (now reporting directly to the acting Secretary of Defense, on equal footing with the military, which as a civilian is unprecedented) . In February 2014, Flynn's close association with a Russian woman was so alarming that some expressed concerns to American authorities that he may have been compromised by Russian intelligence. His chaotic management style, abuse of staff and insubordination to his superiors led to his ousting by August 2014. – Daniel Morrison
Do I believe everything Morrison posits? Of course not. Hell, my family is even named in his article. I do believe however, that we should be looking at this thing from all possible angles, trying to poke holes in our own beliefs so they we don't remain inside an echo chamber. I recommend everyone go read it, and look at the graphics from PsyGroup and Wikistrat's proposals.
Slats Henry put together an interesting thread linking PsyGroup and SLC/Cambridge with Bibi Netanyahu and Rudy Guiliani. We all know how close Jared Kushner and Bibi are.
The above graphic by Wendi Siegleman of Buzzfeed looks at the connections between Cambridge Analytica, the Mercer family, SCL Group and Erik Prince. She has done some incredible investigative work as it relates to Cambridge, Prince and the network of shell companies they appear to be operating.
She has tied Prince to Bibi's "disgraced" Chief of Staff, so we can see that there is a network of players here who often work together.
Qanon as last tip of Aquino/Vallely initiated psyop going all the way back to and up thru Roswell/Collins Elite, New Age, I AM movement, Satanic Panic/HIV/AIDs, Pedogate/Pizzagate, Project Blue Beam, Tesla, etc. etc. Toward disarmement thru MindWar.
— skitzosk8freak (@screentitty) June 5, 2020
This musician and writer seems to think that "Q" is a continuation of MindWar with a goal of psychological disarmament.
Nader, the one who paid PsyGroup the $2 million was apparently involved with the UAE. He would attend the infamous Seychelles secret meeting with Prince and Kushner, to set up a back channel to Russia. "Back channels are important"? (See Karol's thread here).
VisionsSurreal has done a thread looking at PsyGroup involvement with Trump camp contacts, but I think he missed the real role they played, not in 2016 or the election but after in 2017 with the initiation of 'Q' potentially. This is a possibility that Q researchers should be digging into.
Another article by Forensic News looks into Wikistrat and PsyGroup more, "Israeli private intelligence firm claimed recruitment of Khashoggi prior to murder":
Wikistrat, a hybrid geopolitical analysis/intelligence gathering company connected to Saudi and Israeli leadership, claimed to have 'recruited' slain Washington Post journalist Jamal Khashoggi shortly before his murder, and approximately one year after the firm's owner met with senior advisors to the Saudi government.
Thousands of documents obtained exclusively by Forensic News include correspondences between a senior leader of Wikistrat and a lower-level employee that took place in summer 2018 in which the senior leader wrote that Khashoggi had recently been 'recruited' by the firm and encouraged the employee to solicit similar journalists for an unspecified Wikistrat project.
Khashoggi's 'recruitment' by Wikistrat occurred just one year after a small group of businessmen, including Wikistrat owner Joel Zamel, met with senior advisors to the Saudi government. According to the New York Times report, the men allegedly discussed using private intelligence firms to assassinate Iranian political enemies of the newly-minted Saudi Crown Prince, Mohammad bin Salman.
While Zamel reportedly declined such offers, messages show his firm Wikistrat later recruited Khashoggi for a confidential project with unclear purposes for an unknown client.
Wikistrat was founded in 2009 by Zamel, Daniel Green, and former Israeli military intelligence officer Elad Schaffer. Described as 'crowd sourced consultancy', Wikistrat hires experts who produce reports for clients, often government agencies and major corporations. Analysts run simulations, war game scenarios, and risk-monitoring for a wide variety of international clientele.
Zamel, originally born in Australia, moved to Israel to obtain a Master's degree in Government at the Interdisciplinary Center, Herzliya. After entering the clandestine market of private intelligence, Zamel founded Wikistrat and another private intelligence firm, Psy Group, both in Israel. Psy Group had a convoluted corporate structure which ran through Cyprus and ended up in the British Virgin Islands, obscuring true ownership of the company.
Psy Group achieved some level of notoriety for being investigated by Special Counsel Robert Mueller. Mueller was reportedly probing a plan pitched by the firm which was designed to assist the Trump 2016 presidential campaign with social media manipulation. The plan would allegedly be bankrolled by Saudi and UAE leaders. Zamel and both firms, Wikistrat and Psy Group, have also been the target of the Senate Intelligence Committee's bipartisan investigation into 2016 election interference. – Forensic News
This is really interesting, because it discloses the military intelligence connection to Wikistrat and PsyGroup. This is important because 'Q' seemingly indicates they are connected to military intelligence, but what if its really Israeli military intelligence? Another interesting side note – the "crowd sourced" consultancy. Q often crowd sources the research that anons do.
Why did PsyGroup sign an agreement with Cambridge Analytica AFTER the 2016 campaign? What project were they working on that required their participation after the election?
Once again, Kushner seems to be the man in the shadows, potentially doing things without President Trump's knowledge or approval (see here).
So, there you have it: more questions than answers. My hope is that this article gets people thinking, asking questions and coming up with some possible answers. There is information out there that never makes its way into our circles of the internet and this causes us to have a blind spot. We need to be seeking out alternative information that we wouldn't normally look at, and sources as well.
No one can say with any sort of certainty whether or not Q is a MindWar operation, a Psychological Operation (PSYOP) or if it is a real military intelligence dissemination campaign. Q has repeatedly stated Information Warfare — was it created to counter other information warfare operations? There are so many unanswered questions, we hope you will begin to find these answers and bring them out of the shadows and into the light.
~~Further Reading
For more information about the theories discussed above, see the following articles:
"SCL Group Joins the US State Dept." This article by Firetext looks deeply into SCL Group and their recent projects for the U.S. State Department.
"Did Israeli spy firm help Trump win presidency?" by the Electronic Infitada looks at the connections between Israeli intelligence and members of Trumps campaign.
"HOW AN INFLUENCE OPERATION WORKS" by BCC looks at the influence operations by BlackCube and PsyGroup.
"Psy Group sister company controlled by Russian billionaire" by Scott Steadman looks into the Russian connections of PsyGroups sister company.
Exposing Biden's BIG COVID Lies — Follow the Numbers January 12, 2022 at 9:35 pm | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 6,620 |
Home » Blog » Happy New Year and here is another update to keep you in the loop!
Happy New Year and here is another update to keep you in the loop!
I want to start off by apologizing for the shirt delays. According to USPS , ALL the MTO shirts were in the U.S. on December 22th. Due to holiday traffic and the attacks on Paris, customs slowed everything down.
On the bright side the shirts are with me. They look amazing, as expected. I am starting to ship everyone's shirts today!
I have a lot going on for the New Year. I am looking for investors and trying to ramp up basics.
The weather has been down right nuts and some great product ready for you just in time.
A little back ground the scarves.
The fabric composition is 60% cashmere and 40% wool. The yarn is from Cariaggi. I am a huge fan of Cariaggi. Since 1958 Cariaggi has been making some of finest luxury yarns world.
The scarves are woven on very old wooden shuttle looms. There aren't many looms still operating in Italy like these. It's a true artisan craftsmanship at it's best. The process is slow and laborious.
The scarves are 70cm x 200 cm.
In inches that's 27.5 inches ( 2.3ft ) x 78.75 inches ( 6.5ft). That's a nice long scarf!
The scarves retail for $210.00 respectively , which is a steal in it's own right.
We have also made a patterned scarf for 3 seasons. I really enjoyed designing this scarf and it was a pleasure to watch them being made.
65% cotton 18% cashmere 17% linen using the finest yarns , of course. This scarf is a little smaller than it's winter counter part. It's clocking in at 70cm x 180cm.
The patterned scarf is great for anyone and I am sure your better half would love one!
The patterned scarf retails for $165.00 .
We made a few extra sweaters and I am happy we did. True works of soft , yummy, perfection.
All the sweater are 100% wool made from "cashwool" from Zegna Baruffa. The yarn is incredibly soft even for extra long extra fine merino wool.
The tag on the collar is meant to be removed.
The " HUNTER" is a beefier version of last season. It's offered in Burgundy, Navy, and Black. | {
"redpajama_set_name": "RedPajamaC4"
} | 2,189 |
Q: form doesnt set variables when echoed - php I was trying to echo a form that already worked out when I build up before (without echoing it through php).
Now that I'm trying echoing this form
echo "<form action=\"timeschedulingphp.php\" method=\"POST\" enctype=\"multipart/form-data\">";
while($row = mysqli_fetch_array($result)) {
echo "<input type=\"hidden\" id=\"FName\" name=\"FName\" value=\"" . $row['FName'] . "\">
<input type=\"hidden\" id=\"FName\" name=\"FName\" value=\"" . $row['LName'] . "\">" ;
}
echo "<input id=\"date\" type=\"text\" size=8 class=\"form-control mb-2\" name=\"Date\"> </input>
<script type=\"text/javascript\" src=\"js/jquery.js\"></script>
<script type=\"text/javascript\" src=\"js/jquery-ui.js\"></script>
<script type=\"text/javascript\" src=\"js/ui.js\"></script>
<input id=\"appt-time\" type=\"time\" name=\"FromTime\"
min=\"00:00\" max=\"24:00\" required
pattern=\"[0-9]{2}:[0-9]{2}\" class=\"form-control mb-2\"></input>
<input id=\"appt-time\" type=\"time\" name=\"ToTime\"
min=\"00:00\" max=\"24:00\" required
pattern=\"[0-9]{2}:[0-9]{2}\" class=\"form-control mb-2\"></input>
<select name=\"Class\" class=\"form-control mb-2\">
<option value=\"null\"></option>
<option value=\"Piano Class\">Piano Class</option>
<option value=\"Sing Class\">Sing Class</option>
</select>
<select name=\"Present\" class=\"form-control mb-2\">
<option value=\"WasPresent?\">WasPresent?</option>
<option value=\"Present\">Present</option>
<option value=\"Absent\">Absent</option>
<option value=\"Justified\">Justified</option>
</select>
<input type=\"text\" placeholder=\" Notes \" name=\"Notes\" class=\"form-control mb-3\">
<button class=\"btn btn-success\" name=\"register\">Register</button>
<input class=\"btn btn-primary\" type=\"reset\" value=\"Reset\">
</form>";
When I go to timeschedulingphp.php and it controls
if(isset($_POST['register']))
it is not isset.. why?
Thank you for replying me and I'm sorry if it's a noob question
Luca
A: Try to set it as submit type
<button class=\"btn btn-success\" name=\"register\" value=\"Register\" type=\"submit\">Register</button>
A: Okay According To You're Question
Why if(isset($_POST['register']))is not isset or considered have a null value
Because You don't set The Value of the button
Try This
<button class=\"btn btn-success\" value="save" name=\"register\">Register</button>
This Code Will Give $_POST['register'] a Save Value or same as $_POST['register'] = 'save'
Hope This Answer Help
A: To start with you need some information about the differences you can use.
isset()
isset — Determine if a variable is set and is not NULL
In other words, it returns true only when the variabble is not null.
Documentation:
*
*http://www.php.net/manual/en/function.isset.php
empty()
empty — Determine whether a variable is empty
In other words, it will return true if the variable is an empty string, false, array() (or []), NULL, 0, and an unset variable.
Documentation:
*
*http://www.php.net/manual/en/function.empty.php
Related to your code and your issue, it seems $_POST is not set, so you get false in return.
You can do a couple of things to check what is inside $_POST and what to do it it is empty.
While debugging always check what is inside your data. So for that you can use:
echo '<pre>';
var_dump($_POST);
echo '</pre>';
Documentation:
*
*http://nl1.php.net/manual/en/function.var-dump.php
Hopefully, you will receive an array like:
$array = [
1 => 'one',
2 => 'two',
3 => 'three',
4 => 'four',
5 => 'five',
];
echo '<pre>';
var_dump($array);
echo '</pre>';
die;
Outcome:
array(5) {
[1]=>
string(3) "one"
[2]=>
string(3) "two"
[3]=>
string(5) "three"
[4]=>
string(4) "four"
[5]=>
string(4) "five"
}
Related to isset(), and to catch incase the array/string is empty you can do this:"
$register = (isset($_POST['register']) ? 1 : 0);
In the code above $register will always be set. Read it like this:
*
*(isset($_POST['register'])) = 'If $_POST['register'] is set'
*? = 'THEN DO THIS' (in our example it will become (int) 1).
*: = 'IF NOT THEN DO THAT' (In case $_POST['register'] is not set,
$register becomes (int) 0)
If $register keeps staying empty, please reload your page. By filling in your format and send actually data within $_POST. It is possible if you press F5, the $_POST data disappears which means the data is empty.
Please let me know if this was somewhat useful or not. I rather to give you information so you can practice while doing, instead of getting clear code and copy paste this. If you have questions, let me know.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 4,193 |
\section{Conclusion}
In this paper, we tackled the important natural language processing task of language detection in a code-switching twitter corpus using recurrent neural networks. This is a novel application for RNN as most previous research focused on using machine learning methods, such as chained conditional random fields, to solve this type of problems. In fact, in the 2014 EMNLP Code-Switching Workshop most of the participating team used CRF to build their models, and the best performing two teams used SVM-based model. Previous work has already compared RNNs to CRF-based model for natural language processing, and results show that RNNs can produce better accuracy by 1\% on a named entity recognition task. We also tested RNN and found that by with the two proposed extensions, RNNs can also outperform state-of-the-art SVM-based system by 1\% in accuracy, or a 17\% in error rate reduction, while using simpler features and smaller training data.
\section{Data}
\label{sec:data}
All our experiments are conducted on the Twitter data provided by the EMNLP Code--switching Workshop \citep{solorio2014overview}. Due to Twitter's privacy policy, the organizers were not allowed to provide the tweets themselves. Instead, the participants were provided the unique tweet ids and character offsets. Then, the participants had to crawl the data themselves. With situations like deletion and privacy setting changes made by the users after the initial crawl, we are able to crawl $27,255$ Tweets in total. The language--pair breakdown is listed in Table~\ref{tab:pair-breakdown}.
\begin{table}[ht]
\centering
\begin{tabular}{cc}
\hline
Language pair & number of tweets \\
\hline
English--Spanish & 11,400 \\
English--Nepali & 9,993 \\
Mandarin--English & 995 \\
Modern Standard Arabic--Egyptian Arabic & 5,862 \\
\hline
\end{tabular}
\caption{Number of tweets by language pair.}
\label{tab:pair-breakdown}
\end{table}
Although the main goal of the shared task is to predict the code--switching points, the data is annotated with a finer--grained scheme listed in Table~\ref{tab:scheme}. It should be noted that there is a large variance in each label's frequency. \texttt{mixed} and \texttt{ambiguous}, despite their significance in the linguistic theory of bilinguallism, do not appear to happen often in social media. On the other hand, named entities (labeled with \texttt{ne}) occurs very frequently, to the extent that it was the main deciding factor of the participants' overall performance since \texttt{lang1} and \texttt{lang2} are relatively easy to classify.
\begin{table}[ht]
\centering
\begin{tabular}{cc}
\hline
Label & count \\
\hline
lang1 & 215,014 \\
lang2 & 113,282 \\
ambiguous & 1,536 \\
mixed & 197 \\
other & 72,145 \\
ne & 21,833 \\
\hline
\end{tabular}
\caption{Description of labels and their frequencies.}
\label{tab:scheme}
\end{table}
As stated in Sec.~\ref{sec:intro}, we decide to make the task more realistic (and also harder) by training a classifier that considers all possible languages. Therefore we have a single corpus with (some number) of tweets. The number of tokens by language id is in Table~\ref{tab:lang-breakdown}.
\begin{table}[ht]
\centering
\begin{tabular}{cc}
\hline
Label & count \\
\hline
English (en) & 122,585 \\
Modern Standard Arabic (msa) & 79,484 \\
Nepali (ne) & 60,697 \\
Spanish (es) & 33,099 \\
Egyptian Arabic (arz) & 16,292 \\
\hline
\end{tabular}
\caption{Number of tokens by language id.}
\label{tab:lang-breakdown}
\end{table}
\section{Experiments}
\label{sec:experiment}
\subsection{Preliminary Study}
To test the effectiveness of different neural network structures, we first use a smaller data set for a preliminary study. Besides testing both Elman-type and Jordan-type RNN structures, we also tested the effectiveness of our proposed extensions. Our preliminary study dataset contains only a total of 2,734 tweets, where we use 1,000 for training, 1,000 for validation, and the rest for evaluation.
\subsubsection{Experimental Results}
\begin{figure}[ht]
\centering
\includegraphics[width=0.8\textwidth]{prelim}
\caption{Preliminary results of comparing different architecture and feature types, trained on 1,000 tweets.}
\label{fig:prelim}
\end{figure}
In the preliminary study, we tested the five following configurations:
\begin{itemize}
\item \textbf{Jordan}: The basic Jordan-type RNN with input as 7 1-hot word-context vectors and a supervised embedding layer for projecting words onto a real-valued Euclidean space.
\item \textbf{Elman}: The basic Elman-type RNN with input as 7 1-hot word-context vectors and a supervised embedding layer for projecting words onto a real-valued Euclidean space.
\item \textbf{Jordan+ngram}: The Jordan model with an additional 12 1-hot character ngram vector.
\item \textbf{Elman+ngram}: The Elman model with an additional 12 1-hot character ngram vector.
\item \textbf{Elman+ngram+w2v}: The Elman+ngram configuration with an addtional pre-trained word2vec model feeding directly to the hidden layers.
\end{itemize}
As shown in the figure, with the basic architecture, Jordan-type RNN and Elman-type RNN achieved very similary performance (88.4\% and 88.9\%). Adding character ngrams improves the F1 performance of both architectures by roughly 4\% (92.8\% and 92.6\%). The best performing system that uses Elman architecture and both character ngrams and pre-trained Word2Vec performed a F1 score of 93.7\%. Adding the pre-trained Word2Vec features directly to the hidden layer provided an additional 1.2\% improvement. This suggests that comparing to the basic RNN structures, the proposed two extensions can improve the performance by roughly 5\% on the language identification task.
\subsection{Comparing to Previous Work}
To evaluate our approach, we use the training data of 27,255 tweets obtained from the EMNLP 2014 Code-Switching Workshop. We use 2,734 tweets as our evaluation set, and the rest for training and validation. We acknowledge that we are using a different test set to evaluate our systems, and comparing the results with the systems form the workshop. But we also
separate the tweets into training, validation, and evaluation sets using disjoint sets of authors. Nevertheless, both the test data we use and the test data from the workshop were collected using the exact same fashion, and both are from none overlapping sets of authors than the training set.
\subsubsection{Baseline Systems}
The EMNLP Code--Switching Workshop provided a simple deterministic baseline for all categories. Given a word $w$, it looks it up in the training corpus and pick the more frequent language label $\ell(w) \in \{\texttt{lang1}, \texttt{lang2}\}$ as its label. If $\ell(w)=0$ it returns \texttt{other}. And in the case of a tie it returns the language that is more frequent.
\subsubsection{Best Performing Systems in the Shared Task}
According to the overview paper of the workshop \citep{solorio2014overview}, most teams in the shared task adapted the CRF model for language identification. The best Nepali--English submission \cite{barman:14} used SVM as the classification framework. They used character n--grams, context, capitalization, word length, and dictionaries as their features. The best Spanish--English submission \citep{bar:14} also used SVM as the main classifier. They used character n--grams, context, and dictionaries. Additionally they also take language model probabilities as features.
\subsubsection{Experimental Results}
In Table.~\ref{fig:my_label}, we show the performance of our full systems. To compare with the results reported in the workshop, we evaluate our system on the accuracy on two different categories against the best performing systems reported in the EMNLP'14 Code-Switching Workshop. We used only a set of 18,521 tweets to train the networks, because we need the rest of the training set for validation and testing. This is using a significantly smaller training set comparing to previous systems, which have access to the full training data of 27,255 tweets provided by the workshop.
\begin{figure}[ht]
\centering
\begin{tabular}{r r r}
\hline
Systems & English-Spanish & English-Nepali \\
\hline
Jordan+ngram+w2v & \textbf{95.2}\% & \textbf{96.6}\% \\
Elman+ngram+w2v & \textbf{95.2}\% & 96.4\% \\
\hline
SVM (workshop: dcu-uvt) & 92.5\% & 96.3\% \\
SVM (workshop: TAU) & 94.2\% & not reported \\
\hline
Baseline (LangID) & 75.9\% & 70.0\% \\
Baseline (Lexical) & 72.6\% & 68.5\% \\
\hline
\end{tabular}
\caption{Language Identification Accuracy Comparing to the Baseline and Best Reported Systems from the EMNLP'14 Workshop. The parenthesis indicates the size of the training set in number of tweets.}
\label{fig:my_label}
\end{figure}
Similar to the preliminary experiment results, Elman-type and Jordan-type RNNs performed similarly. Comparing to state-of-the-art systems, our networks are able to perform 1\% higher accuracy for English-Spanish category, a 17\% reduction in error rate. For the English-Nepali category, our Jordan-type network performed 0.2\% higher in accuracy, a 8\% reduction in error rate. This shows that the recurrent neural networks and the embedding layers are able to learn meaningful representations for language detection using only the raw lexical features, and are able to rival against SVM-based systems that depend on sophisticated feature extraction to re-represent the input data.
\subsubsection*{Acknowledgments}
This work is initiated by the Deep Learning course at the Language Technologies Institute at Carnegie Mellon University offered by Dr. Bhiksha Raj. The authors would also like to thank the teaching assistant Zhenzhong Lan for the insightful discussions.
\section{Introduction}
\label{sec:intro}
Code--switching refers to the phenomenon that a speaker changes between different languages in a single utterance or conversation. Sociolinguists believe people code--switch because of sociolinguistic motivations, e.g. to express solidarity and familiarity, and to establish authority or distance.\citep{gumperz:82} It is also shown that code--switching has its own grammatical regularities, that switching points almost never occur at certain points.\citep{berkseligson:86} This implies using sophisticated models to identify structures in a mixed-language sentence can potentially help with the language detection task. Also, computational models of code--switching can thus be used to verify linguistic theories. Nonetheless it can also be used to guide downstream NLP applications to use correct language models, which is a practical problem for current social media processing.
There's recently a workshop and shared task on modeling code--switching in social media. The participants were provided code--switched tweets from four language pairs. They then asked to label the tweets for the language \emph{at token level}. This strongly resembles other established sequential labeling problems such as POS tagging and named-entity recognizer (NER); and almost all participants adapted either the conditional random fields model (CRF) or the support vector machine model (SVM). The results are not bad in general. For Nepali--English and Spanish--English most participants were able to get an F1 score around $0.90$ to $0.95$. However, many participants put a lot of efforts into extensive and careful feature engineering. Features used in the shared task include dictionaries, character and word language models, and even output from other NER systems and language identification systems. Despite all the efforts, the outcomes were not always positive. For example, previous work has found that the use of existing NER systems does not translate into improvement of named entities in one submission \citep{king:14}. In comparison, our proposed RNN--based classifier takes only character n--grams, pre--trained embeddings and the texts. Yet our system are competitive against the best results reported in the shared task.
The paper is structured as follows. In Sec.~\ref{sec:related} we review related work in the literature. In Sec.~\ref{sec:data} we describe the code--switched tweets and the annotation scheme in detail. And in Sec.~\ref{sec:method} we describe our RNN--based model. The experiment results are in Sec.~\ref{sec:experiment}.
\section{Methodology}
\label{sec:method}
The proposed network architecture is based the Elman-type and Jordan-type Recurrent Neural Network. We follow the implementation of Mesnil et. al 2013, and use a 3-word window to capture the immediate context, forming a 7 dimention context vector (1-hot). A supervised word embedding layer is used to project each word onto a 100 dimension real value vector. The second layer is a 100 node hidden layer using the sigmoid function. Finally, we use the softmax function at the output layer. Long-term dependencies are captured using either Elman-type or Jordan-type recurrent structure, where the current output depends on the output of the previous 9 hidden layer or final layer, respectively.
\begin{figure}
\centering
\includegraphics[width=0.55\textwidth]{arch}
\caption{Elman-type RNN with Optional Pre-trained Word2Vec and Character Ngram Features.}
\label{fig:arch}
\end{figure}
Extending this architecture to incorporate character ngram features, we use the before mentioned technique to extract a 12-dimension 1-hot vectors from the current word and append it to the 7-dimension context vector, and use the same embedding layer to project them onto the real value vector space. To incorporate the pre-trained Word2Vec model, we use the current word to look up the model, and feed the projected real value vector directly to the hidden layer. Another alternative approach is to replace the supervised embedding layer completely with a pre-trained Word2Vec model, but we think having both a general embedding model and a task specific embedding model can further generalize our method.
In the following subsection, we will talk about the pre-trained Word2Vec model, and also how we extract simple character ngram features. In the evaluation section, we will show the performance of model trained with different network structures and using different feature combinations.
\subsection{Pre-trained Word2Vector}
We employ skip--gram embeddings as our features. Skip--gram word embeddings \cite{mikolov2013efficient} is a log bilinear model that encourages words with similar contexts to have similar embeddings. We use the implementation from Gensim, and trained on a large Twitter corpus of random samples from the live feed. We randomly sample 10,000 tweets each day, spanning over roughly 2,000 days. We did not filter for specific languages. The idea is that words of different languages tend to share different contexts. So the embeddings should provide good separation between languages. And they proved to provide improvement in the code--switching task \citep{lin:14}.
\subsection{Character NGrams}
Character n--grams prove to be valuable features because languages often have distinct character combinations. For example, a word that starts with ``lle'' is more likely to be Spanish than English. Conversely a word that ends with ``tion'' is more likely to be English. These features are widely used by existing programs, such as \texttt{cld2}\footnote{\url{https://code.google.com/p/cld2/}} and \texttt{ldig}\footnote{\url{https://github.com/shuyo/ldig}}.
In our approach, we extract a fix number of character ngram for each word. We use a window of 3 characters, and extract character bigrams and trigrams from both the begining and end of each word, resulting a 12 dimention character ngram vector. For example, the character ngram vector for the word \emph{architecture} is \emph{[arc, rch, chi, ar, rc, ch, ure, tur, ctu, re, ur, tu]}.
\section{Related work}
\label{sec:related}
In this section, we will review previous methods for language identification, both in monolingual and multilingual texts, and computational models of code switching. We will also review previous work in neural networks for processing natural language and re-representing words in a semantic vector space.
\paragraph{Monolingual language identification.}
Monolingual language identification is generally treated as a text categorization problem, often defined as given a text $t$, assign its label $l_{t} \in L$ where $L = \{l_1 \ldots l_N\}$ is a predefined set of languages in this task. \citet{baldwin:10} is an excellent review. For the sake of completeness, we include some of the most relevant work here.
By featurizing each text $f(t) \in F$, this problem can be treated as a standard supervised learning problem. Indeed, many well known classifiers such as Naive Bayes\citep{lui:12}, SVM,\citep{jalam:01} and kernel methods\citep{kruengkrai:05} have been used in the literature. Featurizing each text is an important subproblem. Existing featurization include per--language character frequency, n--grams\citep{cavnar:94}, and linguistically motivated features, such as stop word lists\citep{johnson:93}. People have also achieved excellent accuracy with lower--level features, such as byte sequence\citep{chew:11}.
\paragraph{Multilingual langID.}
The monolinguality of a document is however not always realistic. For example, researchers has found that people have been posting tweets using multiple languages for some time\citep{ling:13}. \citet{lui:14} used topic modeling to model the languages that occur in a document. People have also used sequential labeling algorithms to identify language segments in a document, such as CRF by \citet{king:13}.
\paragraph{Computational models of code--switching.}
Both monolingual and multilingual language identification seek to assign label(s) to a single document or sentence. Computational models of code--switching on the other hand, takes a step further, and tries to pinpoint where the switching between languages happens. Equivalently, one can say code--switching assigns language label at the token level, stead of the sentence level or document level. Linguists have proposed syntactic theories that predicts the locations of code--switching\citep{sankoff:98,belazi:94,cantone:08}, which at the same time prohibits their happening at certain places. Such theories have been incorporated into computational models, such as \citep{li:14}.
\paragraph{Recurrent-Neural-Network for Natural Language Processing}
Due to the sequential natural of natural language data, many problems of natural language processing in the form of tagging are traditionally solved using the linear Hidden Markov Model \citep{kupiec1992robust}. Other frequently used, state-of-the-art machine learning models include Chained-Conditional Random Fields \citep{lafferty2001conditional}, Maximum-Entropy Markov Model \citep{mccallum2000maximum}, and sometimes Support Vector Machines \citep{kudo2001chunking}. More recently, researchers has also begun to explore using RNN architectures to rival, or even outperform such machine learning approaches. \citet{mesnil2013investigation} project 1-hot word vectors using a task-specific, supervised embedding layer, and experimented using both Elman-type \citep{elman1990finding} \citep{mikolov2010recurrent} and Jordan-type \citep{jordan1997serial} RNNs. Their results show that by using the exact same features, RNNs can outperform the state-of-the-art CRF-based model by 1\% F1 score. In our work, we extend their structure to include additional character ngram features, and a pre-trained, generalized embedding layer. Our results show that the proposed architecture trained on only simple lexical features can outperform state-of-the-art SVM-based systems trained on rich features generated using external tools such as named entity recognizers and dictionaries.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 7,601 |
Jacek Augustyn Łopacki (ur. 28 sierpnia 1690 w Krakowie, zm. 12 lipca 1761 tamże) – lekarz, ksiądz rzymskokatolicki, archiprezbiter kościoła Najświętszej Panny Marii w Krakowie.
Życiorys
Był synem lekarza i profesora Akademii Krakowskiej Jacka Łopackiego i jego drugiej żony Elżbiety Kencówny, przyrodnim bratem Stanisława Antoniego. Studiował na Akademii Krakowskiej uzyskując w 1709 doktorat z filozofii. W latach 1709–1711 studiował medycynę w Padwie uzyskując doktorat. W latach 1711–1719 przebywał w Rzymie, gdzie w Szpitalu św. Ducha odbywał praktykę w zakresie anatomii i chirurgii, następnie powrócił do Krakowa. W 1723 otrzymał godność archiprezbitera kościoła Mariackiego w Krakowie i przyjął święcenia kapłańskie. W 1726 został kanonikiem krakowskim, następnie kanonikiem sandomierskim oraz proboszczem w Popradzie na Spiszu. W 1727 uzyskał stopień doktora teologii jednocześnie prowadząc praktykę lekarską, był lekarzem magnaterii i biedoty, działał w Arcybractwie Miłosierdzia, w 1729 został starszym Arcybractwa. W latach 1731–1760 z ramienia kapituły nadzorował prace w katedrze wawelskiej, prowadzone wg projektu Franciszka Placidiego. W latach 1750-1753 dobudował zachodnią kruchtę kościoła Mariackiego również wg projektu Placidiego. Zlecił Andrzejowi Radwańskiemu wykonanie polichromii do kościoła Mariackiego, ufundował barokowe ołtarze zatrudniając do namalowania obrazów Łukasza Orłowskiego i Dominika Estreichera, epitafia, wzbogacił skarbiec kościoła, planował w 1761 zastąpienie Ołtarza Wita Stwosza nowszym barokowym. W 1757 odbudował Wikarówkę. Z polecenia biskupów krakowskich wizytował krakowskie szpitale, w 1738 Szpital św. Ducha, rozpoczął budowę Szpitala św. Łazarza dla najuboższych. 20 kwietnia 1758 przekazał Radzie Miejskiej swój księgozbiór celem utworzenia w Krakowie biblioteki publicznej.
W 1748 wybito medal na jego cześć. Został pochowany na cmentarzu przy kościele Mariackim, przy wejściu do prezbiterium kościoła Mariackiego zachował się nagrobek z czarnego marmuru dębnickiego.
Bibliografia
Zdzisław Gajda, Józef Lepiarczyk Jacek Augustyn Łopacki (1690-1761) w Polski Słownik Biograficzny tom XVIII, wyd. 1973, s.405–407
Encyklopedia Krakowa, wyd. PWN 2000, s. 571
Absolwenci i studenci Akademii Krakowskiej
Archiprezbiterzy kościoła Wniebowzięcia NMP w Krakowie
Duchowni diecezji krakowskiej (I Rzeczpospolita)
Kanonicy krakowscy
Kanonicy sandomierskiej kapituły kolegiackiej
Lekarze I Rzeczypospolitej
Ludzie urodzeni w Krakowie
Pochowani w Krakowie
Teolodzy katoliccy I Rzeczypospolitej
Urodzeni w 1690
Zmarli w 1761 | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 970 |
describe ManageIQ::Providers::EmbeddedAnsible::AutomationManager::ConfigurationScriptSource do
it_behaves_like 'ansible configuration_script_source'
end
| {
"redpajama_set_name": "RedPajamaGithub"
} | 3,383 |
Российское психологическое общество (РПО) — крупнейшая научная и профессиональная организация психологов в Российской Федерации. Является преемником Общества психологов СССР (1957—1994) и Психологического общества, состоящего при Императорском Московском университете (1885—1922).
История
Общество берёт своё начало в 1885 году и в его истории можно выделить три периода:
Психологическое общество, состоящее при Императорском Московском университете (1885—1922)
года по инициативе профессора М. М. Троицкого при Императорском Московском университете было организовано Психологическое общество, более известное как Московское психологическое общество. М. М. Троицкий стал первым председателем общества (1885—1887).
Общество имело целью разработку психологической науки и распространение психологических знаний; оно проводило регулярные заседания и имело два печатных органа — «Труды Московского психологического общества» и ежемесячный журнал «Вопросы философии и психологии». После смерти Троицкого председателями общества были поочерёдно профессора Н. Я. Грот (1888—1899), Л. М. Лопатин (1899—1920) и И. А. Ильин (1920—1922).
Несмотря на название, Психологическое общество создавалась не только как психологическое, но и как философское, а ключевую роль в его деятельности играли философы-идеалисты. С приходом Советской власти общество стало испытывать материальные и организационные трудности, а после высылки за границу ряда его членов во главе с председателем Ильиным оно прекратило своё существование.
Период деинституционализации Общества (1922—1957)
С 1914 года Психологическое общество при Московском университете получило помещение в стенах Психологического института МГУ, который основал и возглавлял товарищ председателя Общества Г. И. Челпанов. В годы Советской власти многие члены общества продолжили работу в Психологическом институте и после формального прекращения существования общества. С 1922 года в Психологическом институте стал работать А. А. Смирнов, который в 1945 году стал директором института. В 1950 году прошла знаменитая «павловская сессия», в ходе которой звучали призывы «перестроить» психологию исключительно в рамках учения И. П. Павлова, то есть фактически её ликвидировать. В 1952 году по решению «павловской сессии» состоялось первое Всесоюзное совещание по психологии, цель которого заключалось, в частности, в том, чтобы «вскрыть ошибки отдельных психологов». А. А. Смирнов выступил на совещании, пытаясь отстоять само право психологии на существование и уберечь учёных-психологов от репрессий. Мало того, Смирнов предпринял шаги, которые для того времени были исключительны по смелости, он поставил на съезде вопрос о возрождении в СССР такой отрасли как социальная психология. В 1954 году А. А. Смирнов выступил с докладом перед Президиумом Академии педагогических наук РСФСР, после которого было принято решение открыть первый после долгого перерыва научный журнал по психологии «Вопросы психологии».
Общество психологов СССР (1957—1994)
В 1957 году по инициативе А. А. Смирнова решением Президиума АПН РСФСР было возрождено Психологическое общество, которое стало Обществом психологов СССР, а А. А. Смирнов был избран его первым президентом (1957—1963).
На 1 сентября 1958 ᴦода в составе Общества психологов СССР было уже 22 отделения в разных областях и республиках СССР и примерно 1000 членов. Центральным стало московское отделение, которое располагалось в том же Психологическом институте, что и Психологическое общество, состоящее при Императорском Московском университете. Вновь воссозданное общество возобновило свою деятельность, опираясь на сложившиеся направления, формы работы и традиции.
29 июня 1959 года открылся I Съезд Общества психологов. II Всесоюзный Съезд Общества психологов СССР прошёл в 1963 году, а далее съезды проводились уже регулярно примерно каждые пять лет: III-й (1968), IV-й (1973), V-й (1978), VI-й (1983), VII (1989). Во многом именно на этих съездах и определялись направления развития советской психологии.
Российское психологическое общество (РПО) (с 1994)
После распада СССР преемником Общества психологов СССР стало образованное 22 ноября 1994 года Российское психологическое общество. На 2017 год численность членов РПО составляет около 5000 человек В структуру РПО входят 62 региональных отделения и 16 научных секций.
РПО входит в состав IUPsyS — Международный союз психологической науки при ЮНЕСКО и Европейской федерации психологических ассоциаций (EFPA).
Руководители общества
Период 1885—1922
председатели общества
Троицкий, Матвей Михайлович, председатель Психологического общества, состоящего при Императорском Московском университете (1885—1887).
Грот, Николай Яковлевич, председатель Психологического общества, состоящего при Императорском Московском университете (1888—1899).
Лопатин, Лев Михайлович, председатель Психологического общества, состоящего при Императорском Московском университете (1899—1920).
Ильин, Иван Александрович, председатель Психологического общества, состоящего при Императорском Московском университете (1920—1922).
Период 1957—настоящее время
президенты общества
1957—1963: Смирнов, Анатолий Александрович, действ. член АПН СССР, президент Общества психологов СССР.
1963—1968: Леонтьев, Алексей Николаевич, действ. член АПН СССР, президент Общества психологов СССР.
1968—1983: Ломов, Борис Фёдорович, член-корр. АН СССР, президент Общества психологов СССР.
1983—1987: Матюшкин, Алексей Михайлович, действ. член РАО, президент Общества психологов СССР.
1988—1991: Зинченко, Владимир Петрович, действ. член РАО, и. о. президента Общества психологов СССР.
1994—2001: Климов, Евгений Александрович, действ. член РАО, президент Российского психологического общества.
2001—2007: Донцов, Александр Иванович, действ. член РАО, президент Российского психологического общества.
2007—настоящий момент: Зинченко, Юрий Петрович, действ. член РАО, президент Российского психологического общества.
вице-президенты
1963—1968: Шорохова, Екатерина Васильевна
1968—1983: Зинченко, Владимир Петрович
1983—1988: Чернышёв, Алексей Сергеевич
1989—1992: Асмолов, Александр Григорьевич
в настоящий момент: Ермаков Павел Николаевич, первый вице-президент РПО
Сотрудничество РПО с международными организациями
Европейской федерации психологических ассоциаций (EFPA)
Международного союза психологической науки при ЮНЕСКО (IUPsyS)
Канадская психологическая ассоциация (CPA)
Кубинское психологическое общество
Японская ассоциация психологов (JPA)
Психологическое общество Южной Африки
Участие РПО в международных организациях
Российское психологическое общество является официальным членом:
Европейской федерации психологических ассоциаций (EFPA)
Международного союза психологической науки при ЮНЕСКО (IUPsyS)
Печатные издания
Российское психологическое общество издаёт/издавало следующие журналы и сборники:
Psychology in Russia: State of the Art («Психология в России: современное состояние»). Ежегодный сборник научных статей. Издаётся с 2008 года. С 2013 года 4 раза в год.
Национальный психологический журнал(National Psychological Journal). Журнал входит в перечень изданий ВАК РФ.
Российский психологический журнал
Материалы съездов и конференций. Вышло 3 тома.
Почётные члены общества
Почётным членами РПО являются:
Бэн, Александр
Виндельбанд, Вильгельм
Вундт, Вильгельм
Гартман, Эдуард фон
Гельмгольц, Герман Людвиг Фердинанд
Гёффдинг, Харальд
Джемс, Уильям
Дюбуа-Реймон, Эмиль Генрих
Рибо, Теодюль
Рише, Шарль
Спенсер, Герберт
Титченер, Эдвард Брэдфорд
Целлер, Эдуард Готтлоб
Гроф, Станислав
Лунт, Ингрид Сесилия
Садовничий, Виктор Антонович
Ханин, Юрий Львович
Шойгу, Сергей Кужугетович
Примечания
См. также
Американская психологическая ассоциация
Вопросы философии и психологии
Факультет психологии МГУ
Ссылки
Богоявленская Д. Б. Московскому психологическому обществу 115 лет (2000)
Ждан А. Н., Марцинковская Т. Д. Московская психологическая школа: традиции и современность
Российское психологическое общество
https://www.facebook.com/ruspsysociety
Психологические организации
Научные общества СССР
Научные общества России
Московский государственный университет | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 4,094 |
Q: Estimating grader bias/variance and MLE test scores given multiple graders assigned to grade each test Suppose we have $m$ graders and $n$ students, and we want to grade a test so that $k$ graders are assigned to grade to each test, and all graders grade the same number of tests. (I realize $m,n,k$ have to satisfy certain properties to make this "perfect assignment" possible, but I'd rather just skip this point and assume it's true). Also, to make things interesting, let's assume the assignment of graders to tests is random (so it's not like a group of graders all have the same set of tests to grade).
Furthermore, let's assume that each grader $i$ has a mean bias $\mu_i$ and a variance $\sigma_i^2$ for that bias associated with their grading, and the bias they apply to each test they grade is sampled independently from a normal distribution with these parameters. And each test $j$ has a "true grade" $c_j$. So then if grader $i$ is assigned to grade test $j$, then the grade they assign will be $c_j + x_{ij}$ where $x_{ij}$ is the sampled bias from the normal distribution with parameters $\mu_i$ and $\sigma_i^2$.
If the $\mu_i$ and $\sigma_i^2$ are unknown, how do we find the maximum likelihood values for the true grade scores $c_j$? If using a prior for graders' parameters is required I guess I'm ok with that. I would also like to know the MLE (or MAP if we go Bayesian) values for the grader parameters $\mu_i$ and $\sigma_i^2$. The idea being that graders with lower estimated variance should be preferred to those with higher variance, if we want as accurate of assigned grades as possible in the future.
I've phrased this in terms of test grading for clairty, but it's actually for an "active learning" problem in machine learning that we are very interested in in our lab, hence insights on this problem could really help.
A: Ok, here's what I have so far:
For the MLE:
You're right. The MLE formula will not have a closed form solution. But, the outline of the formula seems simple enough. Treating each test score as an observation, we have $T_{ij} = c_i + x_{ij}$ ($i$ refers to students and $j$ markers)
Now lets make some assumptions. Suppose that the "true" test scores are also normally distributed with mean $y$ and variance $\Sigma$. Now, each test score is the sum of independent normally distributed variables. Thus, we have that: $TS_{ij}$~$N(y+\mu_j,\Sigma+\sigma_j^2)$
If we knew the values of the types, then the likelihood would be easily given by $L(TS_{ij} | \mu_j,\sigma_j) = \frac{1}{\sqrt{\Sigma+\sigma_j^2}}\phi(\frac{TS_{ij}-y-\mu_j}{\sqrt{\Sigma+\sigma_j^2}})$ where $\phi$ is the standard normal pdf. But, since we do not know each persons type, they need to be integrated out of the likelihood.
$L_{ij}=\int \frac{1}{\sqrt{\Sigma+\sigma_j^2}}\phi(\frac{TS_{ij}-y-\mu_j}{\sqrt{\Sigma+\sigma_j^2}})dF(\mu_j,\sigma_j)$
In terms of identification. I focus on the means only.
Since each grader is assigned tests at random, then on average the average grader's test score will be $y + \mu_j$ -- the average child's test score plus the grader's average bias. There will be $M$ such equations. Similarly, the average of the same test $j$ for the $k$ graders will be $y_j + E[\mu_i]$ -- the child's average test score plus the average marker bias. There are $N$ such equations here. (Note: For simplicity I've imposed the Law the Large numbers here)
This suggests you might have an identification problem. You have $M+N$ equations with $M+N+2$ unknowns. This would seem to suggest some normalizations will be required. For example, you can normalize the average test score and bias to be $0$. $y=E[u_j]=0$
A: I'm looking at your first paragraph. Depending on values of m, n, and k, you might make grader assignments according to a balanced incomplete block design or a partially balanced incomplete block design. Without some kind of balance it does not seem possible to disentangle so many unknown means and variances. There is rich literature on such issues of balance. Even if not directly applicable to the situation you really care about, you might get some ideas how to proceed.
A Bayesian approach with a Gibbs Sampler might be useful. But there are so many latent variables, that I wonder whether conclusions might be guided by priors (even noninformative ones) to a greater extent
than you would prefer. Again here, some sort of balance may be required.
In either case, it seems that a totally random assignment of tests to graders is not a good design choice.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 7,642 |
{"url":"https:\/\/rdrr.io\/cran\/simulator\/f\/vignettes\/fdr.Rmd","text":"# Benjamini-Hochberg Procedure with the Simulator In simulator: An Engine for Running Simulations\n\nlibrary(knitr)\ncode <- file.path(\"false-discovery-rate\",\nc(\"model_functions.R\",\n\"method_functions.R\",\n\"eval_functions.R\",\n\"main.R\"))\ncode_lastmodified <- max(file.info(code)$mtime) sapply(code, read_chunk) Suppose we wish to test$n$hypotheses,$H_1,\\ldots, H_n$, and we have a p-value$\\hat p_i$for each hypothesis$H_i$. That is, $$\\mathbb{P}_{H_i}(\\hat p_i\\le \\alpha) = \\alpha.$$ Benjamini and Hochberg (1995) design a procedure (for when these p-values are independent) that controls what they call the false discovery rate (FDR), which is the expected proportion of the rejected tests that should not have been rejected: $$\\mathbb{E}{{H_i:i\\in S}}\\left[\\frac{\\sum{i\\in S}1{H_i\\text{ rejected}}}{\\max[1,\\sum_{i=1}^n1{H_i\\text{ rejected}}]}\\right].$$ Given a desired FDR$q$, the BH procedure finds a data-adaptive threshold level$\\hat p(q)$and rejects all$H_i$for which$\\hat p_i\\le \\hat p(q)$. The threshold level is given by comparing the sorted p-values$\\hat p_{(1)}\\le \\cdots \\le \\hat p_{(n)}$to a line of slope$q\/n$and identifying the largest p-value that is below this line. That is,$\\hat p(q)=\\hat p_{(\\hat i)}$where $$\\hat i = \\max{i: \\hat p_i \\le q i \/ n}.$$ In this simulation, we verify that the BH procedure works, and we investigate the effect that correlation has on the FDR control. # Main simulation In the Models section below, we show the code for make_correlated_pvalues, a function that generates a model object given parameters$n$(the number of hypotheses),$\\pi_0$(the fraction of hypotheses that are null), and$\\rho$(the correlation between any pair of test statistics). In the simulation below, we fix$n=20$and vary$\\pi_0$and$\\rho$. library(simulator) <<models>> <<methods>> <<metrics>> <<init>> <<main>> <<init>> sim_lastmodified <- file.info(sprintf(\"files\/sim-%s.Rdata\", name_of_simulation))$mtime\nif (is.na(sim_lastmodified) || code_lastmodified > sim_lastmodified) {\n<<main>>\n<<main2>>\n}\nsim <- load_simulation(name_of_simulation) %>% subset_simulation(index = 1:4)\n\n\nThe variable bh_methods is defined in the Methods section below and corresponds to the BH procedure for four different values of $q$.\n\nWe begin by looking at the results when $\\rho=0$ (i.e., independent tests).\n\nsim %>%\nsubset_simulation(rho == 0) %>%\ntabulate_eval(metric_name = \"fdp\", output_type = \"html\",\nformat_args = list(digits = 0, nsmall = 2))\n\n\nIt appears that the BH procedure does control the FDR at each stated $q$ value. However, we also see that when $\\pi_0$ is less than 1, it tends to be more conservative. Indeed, Benjamini and Hochberg show that the FDR of BH does not exceed $\\pi_0q$.\n\nWe might like to increase the number of simulations.\n\nSuppose we return to this simulation several days later and wish to double the number of random draws used. In the above code, we had 100 draws, which were simulated in 4 chunks of 25 draws each. The simulator allows us to add to a simulation without requiring us to re-run parts of the simulation that have already been run.\n\nIf we had closed the R session without saving the image[^]: for the sake of reproducibility, I like to always start with a fresh workspace, so I can be sure that my functions aren't calling a global variable that I have forgotten about), we can open a new one in the directory that has the files directory in it. We start by loading sim, which is the Simulation object (containing all the necessary pointers to saved files). Loading sim is fast because it only loads the file locations, not the files themselves.\n\nsim <- load_simulation(\"fdr\")\n\n\nWe do so by simply adding 4 more chunks, with index=5:8. Each distinct value of index corresponds to a separate random number generator stream. This is convenient because it means that we do not have to rely on the state of the RNG after completing one chunk to start the next one.\n\n<<main2>>\n\nsim <- load_simulation(\"fdr\") # load the one with index = 1:8\n\n\nWe can look again at the table.\n\nsim %>%\nsubset_simulation(rho == 0) %>%\ntabulate_eval(metric_name = \"fdp\", output_type = \"html\",\nformat_args = list(digits = 0, nsmall = 2))\n\n\n## Some plots\n\nThe FDR is the average of the false discovery proportion (FDP). We can look at these raw values (200 for each method-model pair).\n\nsim %>%\nsubset_simulation(rho == 0) %>%\nplot_eval(metric_name = \"fdp\")\n\n\nWhen $\\pi_0=1$, we see that the FDP is either 0 or 1. This is because if we make any number of discoveries, then they will all be false (but if we do not make any discoveries, we have FDP=0).\n\nWe now investigate the effect of $\\rho$, the correlation between the test statistics. We now fix $\\pi_0=0.8$ and look at how the plots vary with $\\rho$.\n\nsim %>%\nsubset_simulation(pi0 == 0.8) %>%\nplot_eval(metric_name = \"fdp\", varying = \"rho\")\n\n\nSince $\\rho$ is numeric, it might be more informative to look at the FDR as a function of $\\rho$.\n\nsim %>%\nsubset_simulation(pi0 == 0.8) %>%\nplot_eval_by(metric_name = \"fdp\", varying = \"rho\")\n\n\nWe see that the procedure becomes more conservative as the dependence increases, but still does control FDR (which was shown for positive dependence in Benjamini and Yekutieli, 2001). Looking at $\\pi_0=1$, we would like to check whether for negative $\\rho$ the procedure is anti-conservative.\n\nsim %>%\nsubset_simulation(pi0 == 1) %>%\nplot_eval_by(metric_name = \"fdp\", varying = \"rho\")\n\n\n# Creating a simulation based on a preexisting one\n\nTo investigate this particular question in greater depth, we might create a new simulation object based on the previous one. We'd like to increase the number of random draws for this particular setting, but don't care about doing so for the others.\n\n<<main3>>\n\nsim_lastmodified <- file.info(sprintf(\"files\/sim-%s.Rdata\",\n\"negdep\"))\\$mtime\nif (is.na(sim_lastmodified) || code_lastmodified > sim_lastmodified) {\n<<main3>>\n} else{\n}\n\n\nWe remake the table (on the basis of 500 random draws) to check for anti-conservativeness.\n\ntabulate_eval(sim2, metric_name = \"fdp\", output_type = \"html\",\nformat_args = list(digits = 0, nsmall = 2))\n\n\nObserve that at this point, sim and sim2 are two separate simulation objects that refer to some common simulation results. For example, their Model and Draws objects are the same.\n\nm <- model(sim, pi0 == 1 & rho == -0.01)\nm2 <- model(sim2)\nall.equal(m, m2)\nd <- draws(sim, pi0 == 1 & rho == -0.01)\nd2 <- draws(sim2, index = 1:8)\nall.equal(d, d2)\n\n\nWhen model and draws (and likewise output and evals) are called on a simulation object, they load the appropriate files referenced by the Simulation object. The models m and m2 are identical (and likewise for d and d2) because both Simulation objects refer to the same saved files. Thus, having multiple simulation objects does not lead to copies of the (potentially large) results files being made. Instead, only the references themselves are duplicated.\n\n# Components\n\n## Models\n\n<<models>>\n\n\n## Methods\n\n<<methods>>\n\n\n## Metrics\n\n<<metrics>>\n\n\n# Conclusion\n\nTo cite the simulator, please use\n\ncitation(\"simulator\")\n\n\n## Try the simulator package in your browser\n\nAny scripts or data that you put into this service are public.\n\nsimulator documentation built on May 29, 2017, 11:29 a.m.","date":"2018-06-22 03:58:51","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 1, \"mathjax_display_tex\": 1, \"mathjax_asciimath\": 1, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.5911646485328674, \"perplexity\": 2372.6603867495496}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2018-26\/segments\/1529267864343.37\/warc\/CC-MAIN-20180622030142-20180622050142-00591.warc.gz\"}"} | null | null |
{"url":"http:\/\/stackoverflow.com\/questions\/9286846\/runtime-of-binets-formula","text":"# Runtime of Binet's Formula\n\nSo I'm computing the Fibonacci numbers using Binet's formula with the GNU MP library. I'm trying to work out the asymptotic runtime of the algorithm.\n\nFor Fib(n) I'm setting the variables to n bits of precision; thus I believe multiplying two numbers is n Log(n). The exponentiation is, I believe n Log(n) multiplications; so I believe I have n Log(n)Log(n Log(n)). Is this correct, in both in assumptions (multiplying floating point numbers and number of multiplications in exponentiation with integer exponent) and in conclusion?\n\nIf my precision is high, and I use precision g(n); then I think this reduces to g(n) Log(g(n)); however I think g(n) should be g(n)=n Log(phi)+1; which shouldn't have a real impact on the asymptotics.\n\n-\nCalculation involves just a multiplication and a en exponentiation right? exp(x * log(\\phi)) (keep precomputed log(\\phi) ). \u2013\u00a0 ElKamina Feb 15 '12 at 2:07\nYou are using n to represent both a number and its bit-count :| \u2013\u00a0 BlueRaja - Danny Pflughoeft Feb 15 '12 at 16:00","date":"2015-08-04 12:35:52","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 1, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.9171330332756042, \"perplexity\": 2978.56268723999}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2015-32\/segments\/1438042990611.52\/warc\/CC-MAIN-20150728002310-00152-ip-10-236-191-2.ec2.internal.warc.gz\"}"} | null | null |
using System.Collections.Generic;
namespace SteamLib.Models
{
public class SteamReportFilterSet : ISteamReportFilterSet
{
public SteamReportFilterSet()
{
Filters = new List<SteamReportFilter>();
}
public long ID { get; set; }
public IList<SteamReportFilter> Filters { get; set; }
}
} | {
"redpajama_set_name": "RedPajamaGithub"
} | 8,773 |
Home / News / Senior recounts moments of violent assault outside mall
Senior recounts moments of violent assault outside mall
By Sheila Moussavi on March 11, 2007
By Sheila Moussavi 2007
"He started firing shots, and I got hit with one, and for the first time I though, 'Whoa, this is actually serious; he isn't joking around.'"
When CFHS senior Daina Deery was first ordered to get in the car of a man she'd never seen before outside of College Square Mall last Tuesday, she thought it was a joke and turned away. Her mistake was soon made obvious when the attacker fired several shots, one of which hit her in the foot. After several minutes of struggling to get her in the car, the man drove away and police officers arrived.
Since then, the story has been told and retold, but now, a week later, it seems time to hear what Deery was actually thinking from the moment she realized it wasn't a joke.
"It was all kind of a blur," she said. "I can't remember feeling anger or anything at all."
As far as Deery can remember, there was only one concrete thought passing continuously through her mind: "I was thinking to God, 'Don't let me die — I'm not ready to die."
But with the exception of this constant prayer, Deery was relying heavily on instinct to save her.
"I knew from Oprah that if you get in the car, there's a good chance you won't make it, so there was no way I was going in without a fight," she said. "I guess I thought if I stall long enough, somebody has to notice that something is wrong here and come help me."
Sure enough, Deery was faintly aware of a crowd beginning to gather around her, and although she couldn't tell how many people were there, she said she remembers thinking it was only a matter of time before someone took action. Unfortunately, she was quickly discouraged.
"Someone actually came up to us and could have touched me, but he just ran away. When he left, I was thinking; 'You've got to be kidding me,' and I didn't think anyone else would have come for a while."
What Deery couldn't have known at the time was that not everyone was as inactive as that unidentified man. CFHS sophomore Barry Firman happened to be driving to Scheels at the time and witnessed most of the incident. Unsure of the best way to approach the situation, Firman had the presence of mind to take down the attacker's license plate number and call the police.
Although he wanted to physically stop the attack, Firman said, "I didn't know if he had a gun or not, and I didn't want to make it any worse by getting involved because then my life's at risk, her life's at risk and anything could happen."
This decision proved to be a good one, and by the time the police had arrived, Deery had sustained no further injuries than the bullet wound in her foot, which she herself considers very fortunate.
"The shot could have hit me anywhere in my body, so I'm glad it hit me in the foot where there won't be any permanent damage," she said. "Overall, I think it ended up being the best situation possible."
This was not the case for everyone. Michael Hruska, who was later identified as Deery's attacker, was found dead in his car the next morning in a Waterloo park. He died of an apparent suicide. Hruska, who was 33 years old and had a history of bi-polar disorder and mental depression, told Deery that he was going crazy.
Despite any bitterness she may be feeling toward Hruska as a result of their brief but life changing interaction, Deery can still "almost pity him."
"He wasn't thinking clearly — he was going insane like he said, and if he wasn't, I don't think he would be making those sort of actions. That didn't really seem like the type of person he was."
Although she never considered her attacker to be in full possession of his faculties and doesn't believe him to have any other motives in trying to abduct her, many would expect Deery to nonetheless feel much less secure on a daily basis. But she doesn't.
"I'm not as worried as some people would think. I guess I'll always be looking around more when I'm getting out of the car, but I really think it was a fluke, and I happened to be in the wrong place at the wrong time."
In parting, Deery left some advice for others in case they encounter the same situation: "Whatever happens, don't get in the car … and watch Oprah."
Related ItemsDeerygungun shotkidnapmallparking lotSheila Moussavi
← Previous Story Real Encounter assembly sparks controversy among high school
Next Story → Rise in Iraqi violence leads to CFHS reflection
Access sidewalk added to north lot
How to Clean Your Gun
Cambodia school project exceeds expectations | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 7,881 |
Kanton Montbenoît () je francouzský kanton v departementu Doubs v regionu Franche-Comté. Tvoří ho 16 obcí.
Obce kantonu
Les Alliés
Arçon
Arc-sous-Cicon
Aubonne
Bugny
La Chaux
Gilley
Hauterive-la-Fresse
La Longeville
Maisons-du-Bois-Lièvremont
Montbenoît
Montflovin
Ouhans
Renédale
Saint-Gorgon-Main
Ville-du-Pont
Montbenoît | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 3,288 |
{"url":"http:\/\/www.sciforums.com\/threads\/lorentz-invariance-of-certain-zero-angles.111142\/page-4#post-2871187","text":"# Debate: Lorentz invariance of certain zero angles\n\nDiscussion in 'Formal debates' started by Pete, Nov 25, 2011.\n\nNot open for further replies.\n1. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\n2.2 Pete's proposed measurements\n\n$\\vec{T_2}(t,l)$ was defined as the position vectors of the elements of T2 in S.\nYou seem to be using $\\vec{T_2}$ to refer to a displacement vector parallel to T2 in S.\nCan we please stick to consistent labels?\n\nBut that's a side issue. In my methodology post, I'm not proposing any measurements on T2.\n\nI propose to measure the angle between $\\vec{v_p}(t)$ and $\\hat{D_p}(t)$, when t=0, and between $\\vec{v_p}'(t')$ and $\\hat{D_p}'(t')$ at the t' of the transformed event of P at t=0.\n\n$\\hat{D_p}(t)$ and $\\hat{D_p}'(t')$ are unit displacement vectors tangent to the wheel at P in S and S' at times t and t' respectively.\n\nRods and clocks:\nAn inertial reference frame is a conceptual, three-dimensional latticework of measuring rods set at right angles to each other with clocks at every point that are synchronised with each other...\n\nPlease Register or Log in to view the hidden image!\n\nSo, the coordinates of any event can be determined by noting which clock is at the location of the event, and the time on that clock when the event occurs.\n\nThe event coordinates can then be used to determine other measurements.\nFor example, the position and orientation of an object at a particular time can be measured by noting which clocks are coincident with the object when they read that particular time.\n\nPersonally, I prefer the implicit use of the rods and clocks latticework, but I did suggest a more direct (but cumbersome) measurement in my first methodology post:\n\nThe angle between $\\vec{v_p}(t)$ and $\\hat{D_p}(t)$ can be directly measured with a protractor as the angle between:\n\u2022 The path followed by an inertial object that is moving with P at the instant of interest, and\n\u2022 A rod at rest that that is coincident with a straight rod tangent to the wheel at P at the instant of interest\n\nTo explain further:\nConstruct a background canvas, parallel to the XY plane, at rest in S.\n\u2022 Attach a marker to rod T1 at the point that touches P.\nThe marker attached to T1 will mark on the canvas a physical line that matches the direction of $\\vec{v_p}(0)$.\n\u2022 Attach markers along the length of T1 that mark the canvas only at t=0\nThe resulting mark is a straight line that matches the direction of $\\hat{D_p}(0)$\n\u2022 The angle between the two lines is the angle between $\\vec{v_p}(0)$ and $\\hat{D_p}(0)$\n\nThe same process works in S':\nConstruct a background canvas, parallel to the XY plane, at rest in S'.\nLet $t'_0$ be the time in S' when P touches T1.\n\u2022 Attach a marker to rod T1 at the point that touches P.\nThe marker attached to T1 will mark on the canvas a physical line that matches the direction of $\\vec{v_p}'(t'_0)$.\n\u2022 Attach markers along the length of T1 that mark the canvas only at t'=$t'_0$\nThe resulting mark is a straight line that matches the direction of $\\hat{D_p}'(t'_0)$\n\u2022 The angle between the two lines is the angle between $\\vec{v_p}(t'_0)$ and $\\hat{D_p}(t'_0)$\n\nto hide all adverts.\n3. ### TachBannedBanned\n\nMessages:\n5,265\nWhy only at t=0? You should be measuring at any t, there is nothing special about t=0. Why do you keep referring to t=0? As such, t' is the transform fo $t \\ne 0$ .\n\n...where t should be arbitrary, there is nothing special about t=0.\n\nThis is all good and fine in theory but how do you plan to do this practically? No experiment that I know of sets an \"array of rulers and clocks\".\n\nHow do you place this \"protractor\"? Do you plan t attach an observer to the rim of the wheel? Also, how would the remote observer(s) like the one on the ground read that protractor?\n\nThe wheel is rotating wrt S, so, do you plan to stick a \"canvas\" to the wheel, rotating with the wheel or do you plan to stick a \"canvas\" to the axle?\n\n\u2022 Why the fixation with t=0, again?\nThe marker will mark on the \"canvas\" the trajectory of the tip (or just a point on the vector) of the vector $\\vec{v_p}(0)$. Not very interesting. What you want, is the direction of $\\vec{v_p}(t)$\nfor arbitrary t.\n\nThere is nothing special about t=0. Also, it isn't clear how you plan to attach a marker to the velocity vector $\\vec{v_p}$.\n\nSame objections as above. As an aside, using \"markers\" and \"canvases\" is hardly an option ever considered in experimental physics, so there is no way you could ever run your proposed experiment.\n\nLast edited: Dec 7, 2011\n\nto hide all adverts.\n5. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nt=0 is when T1 is tangent to P.\nIt's an easy special case, that I believe will be enough to show that the angle of interest is zero in S, and non zero in S'.\n\nI can't tell what you mean here.\nT1 is tangent to the wheel at P at t=0 in S, and at t'= some non zero value in S'.\n\nGood and fine in theory is exactly what we need.\nWe are not planning to do this in practice.\nThis is a theoretical exercise, conducted solely in the abstract on Sciforums.\n\nExactly as you would normally use a protractor.\n\nNo\nThey would walk over to the canvas and measure the angle between the lines.\n\nThe S canvas is at rest in S. It's simply a backdrop, it doesn't need to be stuck to anything.\n\nI'm not sure what you mean.\nYou seem to imply that $\\vec{v_p}(0)$ has a trajectory, which doesn't make sense to me.\n$\\vec{v_p}(0)$ is the instantaneous velocity of P at t=0.\nIt doesn't have a trajectory.\n$\\vec{v_p}(0)$ is parallel to the y-axis, and the marker draws a line parallel to the y-axis.\n\nThe same process could be done for any t.\nBut the rod T1 is in the scenario specifically for the case of t=0, and sticking to t=0 keeps things easier, so we might be done before next Christmas.\n\nto hide all adverts.\n7. ### TachBannedBanned\n\nMessages:\n5,265\nOk, this is a simplification since it deals only with a very restrictive case but if you want to run it this way, I have no objection.\nI object though to your method since it is unimplementable. By contrast, the method that I suggested is fully implementable, it can be viewed as a direct implementation of the Hasselkamp experiment, with one ion source attached to the rim (pointing in the direction of the tangential speed $\\vec{v_P(0)}$) and the other one representing the tangent to the rim $\\vec{T_1}$ direction moving with velocity $\\vec{v_P(0)}$ in the vertical direction according to your specification . We compare their respective frequencies as observed in frame S (anchored to the wheel axle) at t=0, when their positions are coincident in S and we compare them again in S'. So, I suggest that we combine our methods into a single method. OK?\n\nLast edited: Dec 8, 2011\n8. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nCan you explain your objection more?\nRemember that we're not actually going to implement it.\nIt's enough that we can predict the results.\nI think that we're focusing entirely too much on implementation, and it's bogging down the discussion.\n\nAll we really need to do in this section is sufficiently identify the vectors that define the angles of interest.\n\n...isn't yet well defined, but that's not the current topic of discussion.\n\nNo, I think we're measuring different things.\n\n9. ### TachBannedBanned\n\nMessages:\n5,265\nYou can't read protractors remotely.\nYou cannot rely on plots drawn on paper to measure angles between moving parts.\nThe measurement approach needs to be meaningful, not just some pie in the sky.\n\nWe just did that, we just reached an agreement of what is being measured.\n\nI think it is, if you need to discuss more we can reopen the discussion.\n\nNo we are not measuring different things, I just agreed with your definition of measurement. See the previous post.\n\n10. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nYou can read a protractor by looking at it.\nIf plots drawn on paper accurately capture the angles of interest, then you can use them to measure the angles.\nPost 63 sets out exactly how the vectors of interest are captured as lines than can be measured with a protractor at the observer's leisure.\n\nThen we're done with this section.\nThe measurement I propose is the angle between $\\vec{v_p}(t)$ and $\\hat{D_p}(t)$, when t=0, and between $\\vec{v_p}'(t')$ and $\\hat{D_p}'(t')$ at the t' of the transformed event of P at t=0.\n\nIf we agree that this is sufficient definition to calculate a result, then we can move on.\n\nThat's all I intended the methodology section to address anyway... the detailed implementation of practical measurement technique is only confusing things.\n\nYes, I intend to.\n\n11. ### TachBannedBanned\n\nMessages:\n5,265\nThe wheel is turning wrt the axle at high speed, there is no way of setting remotely the protractor, let alone reading it.\n\nBut they don't. You can't expect any precision of measurement from dragging markers on paper.\n\nI am not buying any of the above. This is not how things are measured experimentally, especially when we have much higher precision methods.\n\nOK.\n\nWe will move on when we have the method of measurement agreed on. So far we agreed on what we measure, not on how we measure.\n\n12. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nThe protractor is used to measure the angle between the lines on the canvas.\n\nIn this abstract discussion, the markers can be as precise as we want them to be.\n\nTach, this is an abstract discussion.\nWe're not actually planning practical experiments.\n\nI maintain that it doesn't matter.\nIf we agree on the vector definition, then we can calculate the angle between them.\n\nBut, if we really must have experimental measurements, and if the protractor measuring lines on canvas is too hard, then for my measurements I'm going to stick with measuring event coordinates using the rods-and-clocks framework.\n\n13. ### TachBannedBanned\n\nMessages:\n5,265\nNo, it doesn't work this way, So far I have agreed to all your formulations, this time you will need to agree to my method of measurement.\n\n14. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nWe need to mutually agree on how we will proceed.\n\nDo you have a problem with using the rods and clocks framework to identify event coordinates?\n\n15. ### TachBannedBanned\n\nMessages:\n5,265\nYes, I do. I insist on using a measuring way that we can really implement experimentally. In the end, it is experiments that decide which theory is correct and which one is false. So, if, in the end we still disagree, my method gives you an easy way to decide, in effect is nothing but Hasselkamp's setup on a turntable. I will consider using my method and your method but not your method to the exclusion of mine.\n\n16. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nI don't think so. Not unless you have some ion beams handy, along with all the other engineering required to set up a relativistic rolling wheel.\n\nIn the case of this debate, we agree that special relativity is a correct theory.\nThe disagreement is over the predictions of that theory in a specific scenario.\n\nYes, that's what we agreed to in the proposal.\n\nSo, are we all clear on my methodology?\n\n17. ### TachBannedBanned\n\nMessages:\n5,265\nNot exactly, the angles were supposed to be measured at t=0 in S as per your latest change. Right?\nAlso, I want to make sure that both measurement methods (mine and yours) are spelled out in the methodology.\n\n18. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nYes that's what I have in my original methodology post.\nThe same to any point on the wheel at any time, of course... but T1 is set up for point P at t=0.\n\nIt is given that in S, they ion guns are aimed parallel to $\\vec{v_P}$ and the rod T2.\nBut I don't accept your general assumption that the direction of the ion beams is the same as the direction in which the guns are aimed.\n\nThe relationship holds for the gun at P in S, but I don't think it holds for the gun on T2, and not for either gun in S'.\n\nI don't think you understand what I'm saying. I think you are confusing the motion of the ion gun with the motion of the ion beam.\n\nThis diagram shows an element of the ion beam in S, a short time after it was emitted:\n\nPlease Register or Log in to view the hidden image!\n\nThis element was emitted from the ion gun at t=0.\nIt is moving parallel to the y-axis.\nIt is emitting light in all directions.\n\nSome of the light emitted from this ion beam element reaches the observer at O.\nThat light will be doppler shifted, depending on the angle between the ion beam element's velocity and the line from O to the ion beam element.\n\nThat angle is not a right angle, except at the instant the ion beam element is emitted from the gun.\n\nI think your measurement technique needs a complete rethink.\nIt does not match up well with the Hasselkamp experiment.\n\nHasselkamp et al were measuring the transverse doppler shift of light emitted from moving ions that were emitted from a stationary gun, ie they were interested in the velocity of the ions.\n\nBut you are trying to measure the velocity of the gun.\n\nIn the Hasselkamp experiment, light was emitted directly from the moving objects of interest.\n\nIn your measurements, the moving objects of interest at the point P and the rod T2, so you should have light emitted directly from them. This is where you started with the lasers. The only problem was that you had them pointing the wrong way.\n\nLast edited: Dec 9, 2011\n19. ### TachBannedBanned\n\nMessages:\n5,265\nWe are not using T2, remember. Besides, the relationship holds, but that is besides the point.\n\nThat remains to be seen. This is part of your belief, for the time being all that we need to agree on is the guns parallel to $\\vec{v_P(0)}$ and $\\vec{T_1}$ at t=0 in S only\n\nActually I do understand very well.\n\nYes, this is the time t=0 that we agreed earlier that is the only time we take the measurements. At t=0, the beam is perpendicular on the line of sight.\n\nI do not think so at all, see above.\n\nThis is incorrect, Hasselkamp was trying to measure a pure TDE and I am not trying to measure the velocity of the gun, I am using the TDE measurements from the two guns to show that the vectors $\\vec{v_P(0)}$ and $\\vec{T_1}$ are parallel at t=0 in S.\n\nIn the Hasselkamp experiment light is emitted by the ions, I simply added a motion to the gun emitting the ions.\n\nI don't think so, see above.\n\n20. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nWait...\nMy measurements are separate to your measurements.\nDecisions we make about one set of measurements don't have to affect the other set.\n\nI'm still having difficulty following your line of thought. Please bear with me. You may need to repeat yourself. I apologize, but I really want to be clear on this before we go forward.\n\nConsider for the moment only the ions emitted at t=0 from the gun on P.\nThose particular ions travel up, parallel to the y-axis.\nLight from those ions begins to arrive at O from t=r\/c.\nDoes the observer measure light emitted from those ions only when it first arrives, or does it continue to measure the light from those particular ions as they travel?\n\nEarlier, you said they would be monitored continually as they travel.\nIf they are only monitored when the first arrive, then why is the ion gun even necessary? Why not have the observer just monitor light emitted directly from P?\n\n$\\vec{v_p}(0)$ is the velocity of a gun, not of the ion beam.\n\nThat's not simple at all. It changes the nature of the experiment.\n\n21. ### TachBannedBanned\n\nMessages:\n5,265\nI am not using T2 either. I don't need it. Though, if I want to, I can use it as well and still prove my point.\n\nOnly when they fist arrive, i.e. only the light from the ions that emitted the light at t=0.\n\n...if I wanted to use a gun mounted on T2. I can still do that and I can still prove my point right but I gave up on this part in order to align my conditions precisely with yours.\n\nBecause T1 is moving on the vertical, the gun is moving with it.\n\nYes, no question. What's the point? The second gun ions describe the direction of $\\vec{v_P}$, exactly the way the laser was doing it before. Since you objected to the laser being observed at right angle, I just replaced the laser with an ion gun.\n\nJust enough to make it very interesting. Much better than playing with protractors and markers.\n\nLast edited: Dec 9, 2011\n22. ### PeteIt's not rocket surgeryRegistered Senior Member\n\nMessages:\n10,167\nOK, just so long as I know what you're doing.\n\nNow I'm not sure of the purpose of the ion guns.\n\nThere are two ion guns, right?\nIs the ion gun on P necessary, since the observer is only observing the light from the ions emitted at P, not anywhere else on the beam?\nWhy not just have light emitted directly from P?\n\nSame for the ion gun on T1. Does the observer measure the ion beam at any point other than at the ion gun?\nIf not, then why not just have light emitted directly from T1?\n\nThe point is that the Hasselkamp experiment was about the velocity of the beam, but this scenario is about the velocity of the guns, or rather their point of attachment.\nSo, why not ditch the guns, and just attach simple light sources to P and T1?\n\n23. ### TachBannedBanned\n\nMessages:\n5,265\nBecause the velocity of the ions is a prototype for the tangent to the wheel in P. Besides, I like the idea of generalizing the hasselkamp experiment to rotating platforms and the second gun just does that.\n\nYes, the observation is done in O, I have already explained that. I have also explained why I am using moving ions as a prototype of the velocity.The observation is, again, of the TDE in O.\n\nThis is false, in my scenario, the experiment is about the velocity of the ions. The fact that the ion gun is also moving in the same exact direction as the beam does nothing else than a speed composition.\n\nI could do that, the guns make it more interesting as explained.","date":"2022-05-26 20:57:18","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 1, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 0, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.6010610461235046, \"perplexity\": 842.9712286332267}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.3, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2022-21\/segments\/1652662625600.87\/warc\/CC-MAIN-20220526193923-20220526223923-00544.warc.gz\"}"} | null | null |
Arthur Hills is sometimes called a rolling course as it was built on series of rolling dunes. Ocean breeze adds to the challenge. 10 holes on the water call for a precision play.
This Golf Course features a historic Leamington Lighthouse. It was built in 1879 and 1880 as part of the of the navigation system of lights guiding ships into Port Royal Sound.
I have read several excellent reviews about this Golf Course. Golfers say that it is a great value for the money, immaculate condition of the facility is complemented as well as friendly and knowledgeable staff. | {
"redpajama_set_name": "RedPajamaC4"
} | 9,686 |
{-# LANGUAGE OverloadedStrings, TypeFamilies, FlexibleContexts, ScopedTypeVariables,
PackageImports #-}
import Control.Applicative
import Control.Monad
import "monads-tf" Control.Monad.Trans
import Control.Monad.Base
import Control.Monad.Trans.Control
import Control.Concurrent hiding (yield)
import Control.Concurrent.STM
import Data.Maybe
import Data.HandleLike
import Data.Pipe
import Data.Pipe.List
import Data.Pipe.TChan
import Data.Pipe.ByteString
import System.IO
import Text.XML.Pipe
import Network
import Network.TigHTTP.Server
import Network.TigHTTP.Types
import qualified Data.ByteString.Lazy as LBS
class XmlPusher xp where
type PusherArg xp
generate :: (HandleLike h, MonadBaseControl IO (HandleMonad h)
) => h -> PusherArg xp -> HandleMonad h (xp h)
readFrom :: HandleLike h => xp h -> Pipe () XmlNode (HandleMonad h) ()
writeTo :: HandleLike h => xp h -> Pipe XmlNode () (HandleMonad h) ()
data HttpPull h = HttpPull
(Pipe () XmlNode (HandleMonad h) ())
(Pipe XmlNode () (HandleMonad h) ())
instance XmlPusher HttpPull where
type PusherArg HttpPull = XmlNode -> Bool
generate = makeHttpPull
readFrom (HttpPull r _) = r
writeTo (HttpPull _ w) = w
makeHttpPull :: (HandleLike h, MonadBaseControl IO (HandleMonad h)) =>
h -> (XmlNode -> Bool) -> HandleMonad h (HttpPull h)
makeHttpPull h ip = do
(inc, otc) <- run h ip
return $ HttpPull (fromTChan inc) (toTChan otc)
isPoll :: XmlNode -> Bool
isPoll (XmlNode (_, "poll") _ _ _) = True
isPoll _ = False
main :: IO ()
main = do
soc <- listenOn $ PortNumber 80
forever $ do
(h, _, _) <- accept soc
void . forkIO $ do
(hp :: HttpPull Handle) <- generate h isPoll
void . forkIO $ runPipe_ $ readFrom hp
=$= convert (xmlString . (: []))
=$= toHandleLn stdout
runPipe_ $ fromHandle stdin
=$= xmlEvent
=$= convert fromJust
=$= xmlNode []
=$= writeTo hp
run :: (HandleLike h, MonadBaseControl IO (HandleMonad h)) =>
h -> (XmlNode -> Bool) -> HandleMonad h (TChan XmlNode, TChan XmlNode)
run h ip = do
inc <- liftBase $ atomically newTChan
otc <- liftBase $ atomically newTChan
void . liftBaseDiscard forkIO . runPipe_ $ talk h ip inc otc
return (inc, otc)
talk :: (HandleLike h, MonadBase IO (HandleMonad h)) => h -> (XmlNode -> Bool) ->
TChan XmlNode -> TChan XmlNode -> Pipe () () (HandleMonad h) ()
talk h ip inc otc = do
r <- lift $ getRequest h
lift . liftBase . print $ requestPath r
rns <- requestBody r
=$= xmlEvent
=$= convert fromJust
=$= xmlNode []
=$= toList
if case rns of [n] -> ip n; _ -> False
then (flushTChan otc =$=) . (await >>=) . maybe (return ()) $ \ns ->
lift . putResponse h . responseP $ LBS.fromChunks [xmlString ns]
else do mapM_ yield rns =$= toTChan inc
(fromTChan otc =$=) . (await >>=) . maybe (return ()) $ \n ->
lift . putResponse h . responseP
$ LBS.fromChunks [xmlString [n]]
talk h ip inc otc
flushTChan :: MonadBase IO m => TChan a -> Pipe () [a] m ()
flushTChan c = lift (liftBase . atomically $ allTChan c) >>= yield
allTChan :: TChan s -> STM [s]
allTChan c = do
e <- isEmptyTChan c
if e then return [] else do
x <- readTChan c
(x :) <$> allTChan c
responseP :: HandleLike h => LBS.ByteString -> Response Pipe h
responseP = response
| {
"redpajama_set_name": "RedPajamaGithub"
} | 6,237 |
\section{Introduction}
The Dirac oscillator represents an example of relativistic exactly
solvable quantum model. It was firstly proposed by It\^{o} and
collaborators to replace the momentum operator ${\bf P}$ in the
free particle's Dirac equation by combination ${\bf
P}-im\omega{\bf X}\beta$ where ${\bf X}$ was the position
operator, $m$ being the particle's mass and $\omega$ the
oscillator frequency. Then unusual accidental degeneracy of the
Dirac oscillator's spectrum was investigated by Cook
\cite{Cook_lett_NC_71}. Supersymmetric approach to the Dirac
oscillator was investigated in
\cite{Ui_PTP84,Balanntekin_AnnPhys85}. We note that the name Dirac
oscillator for this relativistic problem was given by Moshinsky
and Szczepaniak \cite{Moshinsky_JPA89} who rederived it and shown
that in nonrelativistic limit the relativistic hamiltonian becomes
a harmonic oscillator with a strong spin-orbit coupling term. The
last work renewed interest to the Dirac oscillator and it was
examined form different viewpoints, such as covariance properties
\cite{Moreno_JPA89}, complete energy spectrum and wavefunctions
\cite{BenitezPRL90}, Lie algebra symmetry \cite{Quesne_JPA90},
shift operators \cite{deLange_JPA91}, hidden supersymmetry
\cite{BenitezPRL90, Beckers_PRD90,Martinez_PRD91,Quesne_IJMPA91},
conformal invariance \cite{Martinez_JMP92}, completeness of
wavefunctions \cite{Szmytkowski_JPA01}, approach based on Clifford
algebra \cite{deLimaRodrigues_PLA08}. Some generalization of Dirac
oscillator was also considered \cite{Zarrinkamar_AnnPh10}.
The Dirac oscillator model was applied to problems of nuclear and
high energy physics. Relativistic many body systems with
interactions modelled by the Dirac oscillator hamiltonians with
applications to mesons and baryons was considered
\cite{Moshinsky_FoundPhys93}. Thermodynamics of Dirac oscillators
in $1+1$ spacetime was noted to be important in studies of
quark-gluon plasma \cite{Dominguez_EPL90}. It was also utilized
for developing of effective approach for description of
intermediate and short-range components of nucleon-nucleon
interaction \cite{Faessler_annPh05}. Dirac oscillator was used for
modelling photon-nucleus scattering \cite{Grineviciute_PRC09}.
Another area where the Dirac oscillator model was extensively
applied is quantum optics. Relation between the Dirac oscillator
and relativistic Jaynes-Cummings model was investigated
\cite{Rozmej_JPA99}. Mapping of the Dirac oscillator onto
Jaynes-Cummings model in case of different dimensions was examined
in \cite{Torres_AIPproc_2010}. In regard to the Jaynes-Cummings
model chirality quantum phase transition in $2+1$ dimensional
Dirac oscillator subjected to constant magnetic field was
investigated \cite{Bermudez_PRA08}. \textit{Zitterbewegung}
behaviour of the Dirac oscillator and possible realization of such
a system was considered in
\cite{Bermudez_PRA08_01,Romera_PRA_2011,Wang_EPJB_2012}. Several
attempts to get experimental realization of such a model were made
\cite{Longhi_OptLett10, Franco-Villafane}.
Here we consider the Dirac oscillator from a bit different point
of view, namely we solve the Dirac oscillator eigenavalue problem
in space with deformed Heisenberg algebra that lead to appearance
of minimal uncertainties in position and momentum. The interest to
the theories with deformed Heisenberg algebra was inspired by
investigations in string theory and independently by several
approaches to quantum gravity
\cite{GrossNPB88,Maggiore_PLB93,Witten_PhysToday96} where it was
suggested the existence of a finite lower bound for resolution of
length $\Delta X$, so called minimal length. Deformed commutation
relations that leads to existence of minimal uncertainty in
position and momentum was proposed firstly by Kempf and
collaborators \cite{KMM_95} and then were investigated from
different viewpoints. We point out that only a few quantum
mechanical problems are solved exactly, that is to say the
harmonic oscillator in one \cite{KMM_95} and $D$ dimensions
\cite{Chang_PRD02}, the one- \cite{Noucier_JPA06} and
three-dimensional \cite{QuesneJPA05} Dirac oscillator and one
dimensional Coulomb-like problem \cite{Fityo_JPA06}.
Lorentz-covariant deformed algebra with minimal length was
proposed and $1+1$ dimensional Dirac oscillator problem was solved
\cite{Quesne_JPA06}. Minimal uncertainty for momentum can be
treated as a consequence of gravity induced decoherence
\cite{Kay}. Uncertainty relation that gives rise to appearance of
minimal momentum is also possible in theories with position
dependent effective mass \cite{Quesne_JPA_04_effmass}. We note
that deformed commutation relations with minimal length and
momentum were proposed even earlier in context of quantum group
theory \cite{Kempf_JMP94}. Later it was shown that similar
uncertainty principle with minimal length and momentum can be
obtained in a gedanken experiment of measuring of position in de
Sitter space \cite{Bambi_CQG}. Deformed algebra with minimal
length and momentum was also obtained in context of Triply Special
Relativity \cite{Kowalski-glikman_PRD04}. It should be noted that
basic principles of triply special relativity adopt three
fundamental constants and one of them can be identified with a
cosmological constant of de Sitter space. In case of deformed
algebra with minimal length and momentum only the harmonic
oscillator was examined
\cite{Quesne_JPA03,Quesne_JPA04,Mignemi_arxiv}.
Our paper is organized as follows. In the second section an
uncertainty relation obtained from deformed algebra is analyzed
then the Dirac oscillator oscillator is reviewed in given
representation. In the third section we obtain equations for small
and large components of a wavefunction and examine requirements
imposed on the wave function. In the forth section energy spectrum
of Dirac oscillator is obtained. In the fifth section
wavefunctions of the problem are derived. Finally, the sixth
section contains the conclusions.
\section{Dirac oscillator }
We consider stationary Dirac oscillator equation which can be
written in the form:
\begin{equation}\label{dirac_eq}
H\Psi=E\Psi, \quad H=\hat{{\bf\alpha}}({\bf P}-im\omega{\bf
X}\hat{\beta})+m\hat{\beta}
\end{equation}
where
\begin{eqnarray}
\hat{{\bf \alpha}}=
\begin{pmatrix}
0 & {\bf \sigma} \\
{\bf \sigma} & 0\\
\end{pmatrix}
\hat{\beta}=
\begin{pmatrix}
I & 0 \\
0 & -I\\
\end{pmatrix}\\
\end{eqnarray}
and $\sigma_i$ , $i=1,2,3$ are the Pauli matrices. We also put
$\hbar=c=1$. It is supposed that position $X_i$ and momentum $P_i$
operators in the equation \ref{dirac_eq} are obeyed to deformed
commutation relations which take form:
\begin{eqnarray}\label{algebra}
\begin{array}{l}
[X_i,P_j]=i\left(\delta_{ij}+\alpha X_iX_j+\beta
P_jP_i+\sqrt{\alpha\beta}(P_iX_j+X_jP_i)\right),\\
\\
{[X_i, X_j]=i\beta\varepsilon_{ijk}L_{k}}, \quad [P_i,
P_j]=i\alpha\varepsilon_{ijk}L_{k}.
\end{array}
\end{eqnarray}
Here $L_{k}$ are components of angular momentum operator and
parameters $\alpha$ and $\beta$ are supposed to be positive. We
also note, that there is summation over dummy indices. Components
of angular momentum operator are defined as follows:
\begin{equation}\label{ang_mom_def}
J_{ij}=\varepsilon_{ijk}L_{k}=\frac{1}{2}(X_iP_j+P_jX_i-X_jP_i-P_iX_j)
\end{equation}
Components of angular momentum operator fulfil the ordinary
commutation relations:
\begin{equation}
[L_i,X_j]=i\varepsilon_{ijk}X_k, \quad
[L_i,P_j]=i\varepsilon_{ijk}P_k.
\end{equation}
In the one-dimensional case the algebra (\ref{algebra}) takes
simpler form:
\begin{equation}\label{algebra_2}
[X,P]=i(1+\alpha X^2+\beta P^2+\sqrt{\alpha\beta}(PX+XP))
\end{equation}
We note that similar one-dimensional deformed algebra was examined
in the work \cite{Quesne_07sigma} but in their case instead of
factor $\sqrt{\alpha\beta}$ in the forth term in the right-hand
side an independent parameter $\kappa$ was used. It is easy to
show that the algebra (\ref{algebra_2}) gives rise to uncertainty
relation:
\begin{equation}\label{unceratinty}
\Delta X\Delta P\geqslant\frac{1}{2}|1+\gamma+\alpha(\Delta
X)^2+\beta(\Delta P)^2+\sqrt{\alpha\beta}\langle
\hat{X}\hat{P}+\hat{P}\hat{X}\rangle|
\end{equation}
where $\hat{X}=X-\langle X\rangle$, $\hat{P}=P-\langle P\rangle$
and $\gamma=(\sqrt{\alpha}\langle X\rangle+\sqrt{\beta}\langle
P\rangle)^2\geqslant 0$. From the inequality $|\langle
\hat{A}\hat{B}+\hat{B}\hat{A}\rangle|\leqslant 2\sqrt{\langle
A^2\rangle\langle B^2\rangle}$ which is valid for any two
operators $\hat{A}$ and $\hat{B}$ it follows that $|\langle
\hat{X}\hat{P}+\hat{P}\hat{X}\rangle|\leqslant 2\Delta X\Delta P$.
Since parameters $\alpha$ and $\beta$ are positive, it leads to
inequality $1+\gamma+\alpha(\Delta X)^2+\beta(\Delta P)^2>0$.
Using these remarks we can rewrite the uncertainty relation
(\ref{unceratinty}) in the form:
\begin{equation}\label{uncert_2}
\Delta X\Delta P\geqslant\frac{1}{2}(1+\gamma+\alpha(\Delta
X)^2+\beta(\Delta P)^2-2\sqrt{\alpha\beta}\Delta X\Delta P).
\end{equation}
The latter uncertainty relation brings minimal uncertainty in
position as well as in momentum:
\begin{equation}\label{min_uncert}
\Delta X\geqslant(\Delta
X)_{min}=\sqrt{\frac{\beta(1+\gamma)}{1+2\sqrt{\alpha\beta}}};
\quad \Delta P\geqslant(\Delta
P)_{min}=\sqrt{\frac{\alpha(1+\gamma)}{1+2\sqrt{\alpha\beta}}}
\end{equation}
It is important to emphasize that these minimal uncertainties do
not appear if parameters $\alpha$ and $\beta$ are negative. Having
done rescaling of uncertainties and parameters of deformation we
can represent uncertainty relation in the well known form obtained
by Kempf \cite{Kempf_JMP94}:
\begin{equation}\label{uncert_3}
\Delta\bar{X}\Delta\bar{P}\geqslant\frac{1}{2}(1+\bar{\alpha}(\Delta\bar{X})^2+\bar{\beta}(\Delta\bar{P})^2),
\end{equation}
where
\begin{equation}
\Delta\bar{X}=\sqrt{\frac{1+\sqrt{\alpha\beta}}{1+\gamma}}\Delta
X,\quad
\Delta\bar{P}=\sqrt{\frac{1+\sqrt{\alpha\beta}}{1+\gamma}}\Delta
P, \quad \bar{\alpha}=\frac{\alpha}{1+\sqrt{\alpha\beta}}, \quad
\bar{\beta}=\frac{\beta}{1+\sqrt{\alpha\beta}}. \nonumber
\end{equation}
It is no doubt that ``rescaled" uncertainty relation
(\ref{uncert_3}) leads to the same minimal uncertainties
(\ref{min_uncert}) as it should be.
In multidimensional case commutation relations (\ref{algebra})
brings to uncertainty relation:
\begin{equation}
\Delta X_i\Delta
P_j\geqslant\frac{1}{2}\left|\delta_{ij}+\gamma_{ij}+\alpha\langle\hat{X}_i\hat{X}_j\rangle+
\beta\langle\hat{P}_j\hat{P}_j\rangle+\sqrt{\alpha\beta}\langle\hat{P}_i\hat{X}_j+\hat{X}_j\hat{P}_i\rangle\right|
\end{equation}
where similarly as it was used in one dimensional case
$\hat{X}_i=X_i-\langle X_i\rangle$, $\hat{P}_i=P_i-\langle
P_i\rangle$ and $\gamma_{ij}=\alpha\langle X_i\rangle\langle
X_j\rangle+\beta\langle P_i\rangle\langle
P_j\rangle+2\sqrt{\alpha\beta}\langle P_i\rangle\langle
X_j\rangle$. It is easy to see that in case when $i=j$ the last
relation reduces to (\ref{uncert_2}) and as a consequence the
minimal uncertainties for position and momentum are the same as in
the one-dimensional case (\ref{min_uncert})
To solve the Dirac equation (\ref{dirac_eq}) representation of
operators $X_i$, $P_j$ that obeyed to commutation relations
(\ref{algebra}) should be defined. The algebra (\ref{algebra})
does not have position or momentum representations because of
noncommutativity of corresponding operators. To build up
representation for position and momentum operators (\ref{algebra})
it was proposed projective transformation \cite{Mignemi_arxiv}
which introduce a relation between the commutation relations
(\ref{algebra}) and Snyder algebra \cite{snyder}. As it was noted
such a transformation is nonsymplectic. The position and momentum
operators can represented as follows \cite{Mignemi_arxiv}:
\begin{eqnarray}
X_i=i\sqrt{1-\beta p^2}\frac{\partial}{\partial
p_i}+\lambda\sqrt{\frac{\beta}{\alpha}}\frac{p_i}{\sqrt{1-\beta
p^2}},\\
P_i=-i\sqrt{\frac{\alpha}{\beta}}\sqrt{1-\beta
p^2}\frac{\partial}{\partial
p_i}+(1-\lambda)\frac{p_i}{\sqrt{1-\beta p^2}}.
\end{eqnarray}
Here $p^2=p_kp_k$ and parameter $\lambda$ is arbitrary real. Since
$\alpha, \beta >0$ it leads to the restriction for the absolute
value of square of variable $p$: $\beta p^2<1$.
To provide hermicity of position and momentum operators scalar
product should be defined with a weight function. It can be
written in the form:
\begin{equation}\label{inner_product_general}
\langle\psi|\vp\rangle=\int\frac{d{\bf p}}{\sqrt{1-\beta
p^2}}\psi^*(\bf{p})\vp({\bf p})
\end{equation}
We note that according to abovementioned remark the domain of
integration is bounded by the sphere: $p^2\leqslant 1/\beta$. It
is worth emphasizing that the weight function does not depend on
the choice of parameter $\lambda$.
Components of the angular momentum operator defined by formula
(\ref{ang_mom_def}) are represented as follows:
\begin{equation}
J_{ij}=\varepsilon_{ijk}L_k=i\left(p_j\frac{\partial}{\partial
p_i}-p_i\frac{\partial}{\partial p_j}\right).
\end{equation}
So the components of angular momentum operator take the same form
as in momentum representation in ordinary quantum mechanics.
Wave function of the Dirac equation (\ref{dirac_eq}) can be
written as a two-component spinor $\psi=\begin{pmatrix}
\psi_1 \\\psi_2\\
\end{pmatrix}$ where functions $\psi_1$ and $\psi_2$ are called
large and small component respectively. The Dirac equation
(\ref{dirac_eq}) can be rewritten as a system of two coupled
equations:
\begin{eqnarray}
B^+\psi_2=(E-m)\psi_1,\label{eq_b_+}
\\ B^-\psi_1=(E+m)\psi_2\label{eq_b_-}.
\end{eqnarray}
where
\begin{equation}\label{operator_B_big}
B^{\pm}=(\bm{\sigma},{\bf P})\pm im\omega(\bm{\sigma},{\bf X})
\end{equation}
To get a factorized equation for the large component $\psi_1$ one
should apply the operator $B^+$ (\ref{operator_B_big}) to the
equation (\ref{eq_b_-}) and then in the right-hand sight of
obtained equation the action of the operator on the component
$\psi_2$ should be replaced by the right-hand side of the equation
(\ref{eq_b_+}). As a result we arrive at:
\begin{equation}
B^+B^-\psi_1=(E^2-m^2)\psi_1.
\end{equation}
Similarly for the small component $\psi_2$ we have:
\begin{equation}
B^-B^+\psi_2=(E^2-m^2)\psi_2.
\end{equation}
The representation of position and momentum operators $X_i$ and
$P_j$ allows one to get the explicit form for the operators
$B^{\pm}$:
\begin{equation}\label{oper_B_+}
B^+=\left[-i\left(\sqrt{\frac{\alpha}{\beta}}-im\omega\right)\sqrt{1-\beta
P^2}\left(\frac{\partial}{\partial p}+\frac{(\bm{\sigma}, {\bf
L})+2}{p}\right)+\left(1-\lambda+im\omega\lambda\sqrt{\frac{\beta}{\alpha}}\right)\frac{p}{\sqrt{1-\beta
p^2}}\right]\sigma_p
\end{equation}
\begin{equation}\label{oper_B_-}
B^-=\sigma_p\left[-i\left(\sqrt{\frac{\alpha}{\beta}}+im\omega\right)\sqrt{1-\beta
P^2}\left(\frac{\partial}{\partial p}-\frac{(\bm{\sigma}, {\bf
L})}{p}\right)+\left(1-\lambda-im\omega\lambda\sqrt{\frac{\beta}{\alpha}}\right)\frac{p}{\sqrt{1-\beta
p^2}}\right]
\end{equation}
where $\sigma_p=(\bm{\sigma},{\bf p})/p$
The equations (\ref{eq_b_+}), (\ref{eq_b_-}) and as a consequence
the operators $B^{\pm}$ would take simpler form if transformation
of large and small functions is performed:
\begin{equation}\label{subst}
\psi_i=\frac{1}{p}\vp_i
\end{equation}
After that transformation equations (\ref{eq_b_+}) and
(\ref{eq_b_-}) can be rewritten as follows:
\begin{eqnarray}
\tilde{\omega}b^+\sigma_p\vp_2=(E-m)\vp_1,\label{equ_1_b+}\\
\tilde{\omega}^*\sigma_pb^-\vp_1=(E+m)\vp_2.\label{equ_1_b-}
\end{eqnarray}
where
$\tilde{\omega}=\left(m\omega+i\sqrt{{\alpha}/{\beta}}\right)$ and
$\tilde{\omega}^*$ is complex conjugate. Operators $b^{\pm}$ are
obtained from relations (\ref{oper_B_+}), (\ref{oper_B_-}) and
(\ref{subst}). They take form:
\begin{equation}\label{op_b_+}
b^+=-\sqrt{1-\beta p^2}\frac{\partial}{\partial
p}-\frac{\sqrt{1-\beta p^2}}{p}((\bm{\sigma},{\bf
L})+1)+\eta\frac{p}{\sqrt{1-\beta p^2}}
\end{equation}
\begin{equation}\label{op_b_-}
b^-=\sqrt{1-\beta p^2}\frac{\partial}{\partial
p}-\frac{\sqrt{1-\beta p^2}}{p}((\bm{\sigma},{\bf
L})+1)+\eta^*\frac{p}{\sqrt{1-\beta p^2}},
\end{equation}
here
\begin{equation}
\eta=\frac{1-\lambda+im\omega\lambda\sqrt{\frac{\beta}{\alpha}}}{m\omega+i\sqrt{\frac{\alpha}{\beta}}}=
\frac{m\omega+i\left(m^2\omega^2\lambda\sqrt{\beta\over{\alpha}}-
\sqrt{\alpha\over{\beta}}(1-\lambda)\right)}{m^2\omega^2+\frac{\alpha}{\beta}}.\nonumber
\end{equation}
To simplify equations (\ref{equ_1_b-}) and (\ref{equ_1_b-}) one
can introduce function:
\begin{equation}
\tilde{\vp}_2=\sigma_p\vp_2
\end{equation}
As a result we arrive at
\begin{eqnarray}
\tilde{\omega}b^+\tilde{\vp}_2=(E-m)\vp_1,\label{equ_1x_b+}\\
\tilde{\omega}^*b^-\vp_1=(E+m)\tilde{\vp}_2.\label{equ_1x_b-}
\end{eqnarray}
\section{Superpartner hamiltonians and components of radial wave function}
Operators $b^{\pm}$ (\ref{op_b_+}) (\ref{op_b_-}) introduced in
the previous section commute with the total angular momentum ${\bf
J}={\bf L}+{\bf S}$ where ${\bf S}=\frac{1}{2}\bm{\sigma}$ as well
as with ${\bf L}^2$ and ${\bf S}^2$, so the solutions $\vp_1$ and
$\tilde{\vp}_2$ of equations (\ref{equ_1x_b+}) and
(\ref{equ_1x_b-}) can be taken in the form representing the fact
that they are eigenfunctions of operators ${\bf L}^2$, ${\bf
S}^2$, ${\bf J}^2$ and $J_z$ with corresponding eigenvalues
$l(l+1)$, 3/4, $j(j+1)$ and $m$ respectively.
\begin{equation}
\vp_1=\vp_1(p,s,j,m)=R_{1;s,j}(p)\mathcal{Y}_{s,j,m}(\th,\vp,\xi)
\end{equation}
\begin{equation}
\vp_2=\vp_2(p,s,j,m)=R_{2;s,j}(p)\mathcal{Y}_{s,j,m}(\th,\vp,\xi)
\end{equation}
where
\begin{equation}
\mathcal{Y}_{s,j,m}(\th,\vp,\xi)=\sum_{\sigma,\mu}\langle
j-s\mu,\frac{1}{2}\sigma|jm\rangle
Y_{j-s,m}(\th,\vp)\chi_{\sigma}(\xi)
\end{equation}
is a spin spherical harmonic \cite{Edmonds} and $R_{1;s,j}(p)$ and
$R_{2;s,j}(p)$ are radial wavefunctions. It should be noted that
$\chi_{\sigma}(\xi)$ denotes a spinor and $\sigma=\pm\frac{1}{2}$.
The main advantage of introduced function $\tilde{\vp}_2$ is
caused by the fact that it has the same spin-angular part as the
function $\vp_1$. Whereas for the function $\vp_2$ we have:
\begin{equation}
\vp_2=\sigma_p\tilde{\vp}_2=\tilde{R}_{2;s,j}(p)\sigma_p\mathcal{Y}_{s,j,m}(\th,\vp,\xi)=-\tilde{R}_{2;s,j}(p)\mathcal{Y}_{-s,j,m}(\th,\vp,\xi)
\end{equation}
Last relation can be written as follows:
\begin{equation}
\vp_2=\vp_{2;-s,j,m}(p,\th,\vp,\xi)=R_{2;-s,j}(p)\mathcal{Y}_{-s,j,m}(\th,\vp,\xi)
\end{equation}
and here $R_{2;-s,j}(p)=-\tilde{R}_{2;s,j}(p)$. We remark that
wavefunctions $\phi_1$ and $\tilde{\phi}_2$ are characterized by
the same value $l=j-s$.
To make equations (\ref{equ_1x_b+}) and (\ref{equ_1x_b-}) simpler
we consider relation:
\begin{equation}
((\bm{\sigma},{\bf L})+1) \mathcal{Y}_{s,j,m}(\th,\vp,\xi)=({\bf J}^2-{\bf L}^2-{\bf S}^2+1)\mathcal{Y}_{s,j,m}(\th,\vp,\xi)=s(2j+1)\mathcal{Y}_{s,j,m}(\th,\vp,\xi).
\end{equation}
Having used last equation one arrives at a system of coupled
equations for radial wavefunctions:
\begin{equation}\label{equ_3_b+}
\tilde{\omega}b_p^+\tilde{R}_2=(E-m)R_1,
\end{equation}
\begin{equation}\label{equ_3_b-}
\tilde{\omega}^*b_p^-R_1=(E+m)\tilde{R}_2.
\end{equation}
we use notation $b^{\pm}_p$ for the radial part of operators $b^{\pm}$ and they take form
\begin{equation}\label{b_p+}
b^+_p=-\sqrt{1-\beta p^2}\frac{\partial}{\partial p}-\frac{k}{p}\sqrt{1-\beta p^2}+\frac{\eta p}{\sqrt{1-\beta p^2}};
\end{equation}
\begin{equation}\label{b_p-}
b^-_p=\sqrt{1-\beta p^2}\frac{\partial}{\partial p}-\frac{k}{p}\sqrt{1-\beta p^2}+\frac{\eta^* p}{\sqrt{1-\beta p^2}};
\end{equation}
where $k=s(2j+1)$.
In radial momentum space the scalar product
(\ref{inner_product_general}) can be represented as follows:
\begin{equation}\label{inner_prod_radial}
\langle R|R'\rangle=\int^{1/\sqrt{\beta}}_0\frac{dp}{\sqrt{1-\beta p^2}}R^*(p)R'(p)
\end{equation}
It is easy to verify that with respect to the scalar product
(\ref{inner_prod_radial}) the operators $b_p^+$ (\ref{b_p+}) and
$b_p^-$ (\ref{b_p-}) are mutually hermitian conjugates.
From equations (\ref{equ_3_b+}) and (\ref{equ_3_b-}) we obtain:
\begin{eqnarray}
b_p^+b_p^-R_1=\frac{1}{|\tilde{\omega}|^2}(E^2-m^2)R_1;\label{equ_4_b+-}\\
b_p^-b_p^+\tilde{R}_2=\frac{1}{|\tilde{\omega}|^2}(E^2-m^2)\tilde{R}_2\label{equ_4_b-+}
\end{eqnarray}
The radial wavefunctions $R_1$ and $\tilde{R}_2$ can be treated as
eigenfunctions of two superpartner hamiltonians
\cite{Cooper_PhysRept95,Junker_96}.
We consider bound state problem so normalizability condition
should be imposed on the relativistic wavefunction
$\psi=\begin{pmatrix}\psi_1 \\\psi_2\\\end{pmatrix}$. It gives
rise to the following relation:
\begin{equation}\label{normal_wavefunct}
\int_0^{1/\sqrt{\beta}}\frac{dp}{\sqrt{1-\beta
p^2}}\left(|R_1|^2+|\tilde{R}_2|^2\right)=1.
\end{equation}
In the presence of deformed commutation relations additional
requirements are imposed on bound state wavefunctions. In case of
uncertainty principle with minimal length it is demanded that any
``physical" wavefunction belongs to the domain of operator
$\bf{P}$ it means that meanvalue of square of momentum operator is
finite. The deformed commutation relations (\ref{algebra}) impose
stricter requirements. To be acceptable a wavefunction should
belong to the domains of operators $\bf{P}$ and $\bf{X}$. As a
result it leads to finite meanvalues for square of both momentum
and position.
Let us suppose that in the right-hand side of equations
(\ref{equ_4_b+-}) and (\ref{equ_4_b-+}) we have eigenvalue
$E^2=m^2$, so the corresponding wavefunctions are necessarily the
solutions of equations:
\begin{eqnarray}
b_p^-R_{1;0}=0,\label{wavefunct_b_0-}\\
b_p^+\tilde{R}_{2;0}=0.\label{wavefunct_b_0+}
\end{eqnarray}
Having integrated equation (\ref{wavefunct_b_0-}) we obtain:
\begin{equation}\label{wavfunct_r_backgr_zero}
R_{1;0}=C_{1;0}p^k(1-\beta p^2)^{\frac{\tilde{\xi}}{2}+i\frac{\tilde{\zeta}}{2}}
\end{equation}
where $\tilde{\xi}=\frac{m\omega}{\alpha+\beta m^2\omega^2}$ and
$\tilde{\zeta}=\frac{\sqrt{\alpha/\beta}(1-\lambda)-m^2\omega^2\lambda\sqrt{\beta/\alpha}}{\alpha+\beta
m^2\omega^2}$ and $C_{1;0}$ is the normalization constant.
The normalization condition (\ref{normal_wavefunct}) implies that
integral from the square module of the function $R_{1;0}$ must be
finite:
\begin{equation}\label{funct_R^0_1}
\int^{1/\sqrt{\beta}}_0\frac{dp}{\sqrt{1-\beta
p^2}}\left|C_{1;0}\right|^2p^{2k}(1-\beta
p^2)^{\tilde{\xi}}=\left|C_{1;0}\right|^2\int^{1/\sqrt{\beta}}_0dp
p^{2k}(1-\beta p^2)^{\tilde{\xi}-\frac{1}{2}}<\infty
\end{equation}
For $p\rightarrow 0$ function $R_{1;0}$ behaves as $p^k$ and
boundary condition $R_{1;0}=0$ leads to the restriction $k>0$ and
this inequality is satisfied if $s=1/2$. When
$p\rightarrow\frac{1}{\sqrt{\beta}}$ convergence of the integral
(\ref{funct_R^0_1}) gives rise to the condition
$\tilde{\xi}-\frac{1}{2}>-1$ or equivalently
$\tilde{\xi}>-\frac{1}{2}$. But this inequality is always
fulfilled because the parameter $\tilde{\xi}$ is defined as
positive. So we conclude that wavefunction $R_{1;0}$ is
normalizable when $s=\frac{1}{2}$.
As it has been already mentioned additional ``physical" conditions
should be imposed on the wave function $R_{1;0}/p$. Meanvalues of
square of momentum and position operators must be finite:
\begin{equation}
\left\langle
\frac{R_{1;0}}{p}\Big|\hat{P}^2\Big|\frac{R_{1;0}}{p}\right\rangle<\infty,
\quad \left\langle
\frac{R_{1;0}}{p}\Big|\hat{X}^2\Big|\frac{R_{1;0}}{p}\right\rangle<\infty.
\end{equation}
Meanvalue for square of momentum can be represented in the form:
\begin{equation}\label{P_2_integral}
\left\langle
\frac{R_{1;0}}{p}\Big|\hat{P}^2\Big|\frac{R_{1;0}}{p}\right\rangle=\int^{1/\sqrt{\beta}}_0\frac{dp
p^2}{\sqrt{1-\beta
p^2}}\frac{R^*_{1;0}}{p}\hat{P}^2_p\frac{R_{1;0}}{p}<\infty,
\end{equation}
where:
\begin{eqnarray}\label{P_2_operator}
\hat{P}^2_p=-\frac{\alpha}{\beta}\left((1-\beta
p^2)\left(\frac{1}{p^2}\frac{\partial}{\partial
p}p^2\frac{\partial}{\partial p}-\frac{l(l+1)}{p^2}\right)-\beta
p\frac{\partial}{\partial p}\right)-\\\nonumber
2i\sqrt{\frac{\alpha}{\beta}}(1-\lambda)p\frac{\partial}{\partial
p}+(1-\lambda)\left((1-\lambda)-i\sqrt{\frac{\alpha}{\beta}}\right)\frac{p^2}{1-\beta
p^2}-3i\sqrt{\frac{\alpha}{\beta}}(1-\lambda)
\end{eqnarray}
is the ``radial" part of square of momentum operator. It should be
noted that all remarks concerning meanvalue of square of momentum
can be applied to the square of position operator because both of
them have similar structure.
Taking into account the explicit form for operator $\hat{P}^2_p$
(\ref{P_2_operator}) and using requirement (\ref{P_2_integral})
one can obtain following condition for integral:
\begin{equation}\label{int_converg}
\int^{1/\sqrt{\beta}}_0dp\quad p^{2k-2}(1-\beta
p^2)^{\tilde{\xi}-\frac{3}{2}}<\infty
\end{equation}
It is easy to convince oneself that convergence of the latter
integral in the vicinity of the point $p=0$ gives rise to the
condition $k>0$. From the other side convergence of the integral
(\ref{int_converg}) in the vicinity of the point $1/\sqrt{\beta}$
will be provided if ${\tilde{\xi}-\frac{3}{2}}>-1$ from which we
obtain restriction on the parameters of oscillator if it is
supposed that parameters of deformation are held fixed:
\begin{equation}\label{cond_gr_state=0}
\frac{1}{\beta}(1-\sqrt{1-\alpha\beta})<m\omega<\frac{1}{\beta}(1+\sqrt{1-\alpha\beta})
\end{equation}
One can conclude that in order to obtain the eigenvalue $E^2=m^2$
in the equation (\ref{equ_4_b+-}) the condition $s=1/2$ should be
required. We note that in case of two-parametric deformed algebra
with minimal length eigenvalue $E^2=m^2$ exists also for positive
projection of spin $s=1/2$ but an additional demand for values $j$
must be satified \cite{QuesneJPA05}. The mentioned requirement
disappears in the limit case when one of those parameters
corresponding to our parameter $\beta$ is kept.
Having integrated equation (\ref{wavefunct_b_0+}) we obtain:
\begin{equation}\label{wavfunct_B_0+_integr}
\tilde{R}_{2;0}=C_{2;0}p^{-k}(1-\beta
p^2)^{-\frac{\tilde{\xi}}{2}+i\frac{\tilde{\zeta}}{2}}
\end{equation}
Again the boundary conditions are imposed on it. For the first we
require that $\tilde{R}_{2;0}\rightarrow 0$ when $p\rightarrow 0$.
The restriction $k<0$ or equivalently $s=-\frac{1}{2}$ follows
immediately from the last requirement. From the other side one
should demand $\tilde{R}_{2;0}\rightarrow 0$ when
$p\rightarrow\frac{1}{\sqrt{\beta}}$ but this requirement cannot
be fulfilled because $-\frac{\tilde{\xi}}{2}<0$. As a result, the
function $\tilde{R}_{2;0}$ is not normalizable. To have physically
acceptable function one should demand $\tilde{R}_{2;0}=0$ and
$R_{1;0}\neq 0$. It is worth mentioning that the same requirement
appears in case of two parametric deformed algebra with minimal
length \cite{QuesneJPA05}. We also remark that the ground state
wavefunction $(R_{1;0}\neq 0, \tilde{R}_{2;0}=0)$ is compatible
with the positive eigenvalue $E=m$ whereas the negative one $E=-m$
will not be compatible with the system (\ref{equ_3_b+}) and
(\ref{equ_3_b-}).
\section{Spectrum of Dirac oscillator}
In this section we will obtain energy spectrum for the Dirac
oscillator. As it was shown in the previous section ground state
with energy $E^2=m^2$ exists only for positive projection of spin
($s=\frac{1}{2}$). In this section we will show that the ground
state with energy $E^2\neq m^2$ can take place for positive
($s=\frac{1}{2}$) as well as for negative ($s=-\frac{1}{2}$)
projection of spin. These two cases that correspond different
ground state energy are considered separately.
\subsection{Case of zero ground state energy}
As it has been already mentioned in the previous section that
whenever $s=\frac{1}{2}$ and the condition (\ref{cond_gr_state=0})
is fulfilled then equation (\ref{equ_4_b+-}) has acceptable
wavefunction corresponding to the ground state energy $E^2-m^2=0$.
To solve eigenvalue problem (\ref{equ_4_b+-}) SUSY QM procedure is
applied \cite{Cooper_PhysRept95,Junker_96}. Operator
$h=b^+_pb^-_p$ is supposed to be the first member of the SUSY QM
hierarchy
\begin{equation}
h_i=b^+_p(k_i,\eta_i)b^-_p(k_i,\eta_i)+\sum^i_{j=0}\varepsilon_j,
\quad i=0,1,2,\ldots
\end{equation}
Imposing shape invariance condition we obtain:
\begin{equation}\label{SI_condition}
b^-_p(k_i,\eta_i)b^+_p(k_i,\eta_i)=b^+_p(k_{i+1},\eta_{i+1})b^-_p(k_{i+1},\eta_{i+1})+\varepsilon_{i+1}
\end{equation}
In explicit form we write:
\begin{equation}\label{Im_eta_cond}
\eta_{i+1}-\eta^*_{i+1}=\eta_i-\eta_i^*
\end{equation}
\begin{equation}\label{k_cond}
k^2_{i+1}-k_{i+1}=k^2_i+k_i
\end{equation}
\begin{equation}\label{Re_eta_cond}
\frac{1}{\beta}|\eta_{i+1}|^2-\eta^*_{i+1}=\frac{1}{\beta}|\eta_{i}|^2+\eta_{i}
\end{equation}
\begin{equation}\label{epsilon_cond}
-k^2_{i+1}\beta-k_{i+1}(\eta_{i+1}+\eta^*_{i+1})-\frac{1}{\beta}|\eta_{i+1}|^2+\varepsilon_{i+1}=
-k^2_{i}\beta-k_{i}(\eta_{i}+\eta^*_{i})-\frac{1}{\beta}|\eta_{i}|^2
\end{equation}
In the following we use notations: ${\rm Re}\eta_i=\xi_i$ and
${\rm Im}\eta_i=\zeta_i$ Having solved the first three equations
we obtain:
\begin{equation}\label{iter_cond_e=0}
\zeta_{i}=\zeta, \quad \xi_{i}=\xi+\beta i, \quad k_i=k+i
\end{equation}
we note that $\xi=\beta\tilde{\xi}$ and $\zeta=\beta\tilde{\zeta}$. It is easy to show that for obtained values $\eta_i$ and $k_i$
the hierarchy hamiltonians $h_i$ have physically acceptable
solutions $R_{1;0}(k_i,\eta_i,p)$ corresponding to the energies
$\Sigma^{i}_{j=0}\varepsilon_j$.
Having used the equation (\ref{epsilon_cond}) we arrive at
following equation for energy eigenvalues:
\begin{equation}\label{eigenvalues_general_e=0}
E^2_n-m^2=\left(m^2\omega^2+\frac{\alpha}{\beta}\right)\sum^n_{j=0}\varepsilon_j=4n\left(m^2\omega^2+\frac{\alpha}{\beta}\right)(\beta(n+k)+\xi)
\end{equation}
Since $k=s(2j+1)$ and
$\xi=\frac{m\omega}{m^2\omega^2+\alpha/\beta}$ the last relation
can be rewritten in the form:
\begin{equation}\label{spectrum_background_zero}
E^2_n-m^2=4n\left[m\omega+(m^2\omega^2\beta+\alpha)\left(n+j+\frac{1}{2}\right)\right]
\end{equation}
We note that in case $\alpha=0$ expression
(\ref{spectrum_background_zero}) is in agreement with
corresponding relation obtained in the work \cite{QuesneJPA05}
when one of their parameters of deformation is set to zero.
The principal quantum number $N=2n+l=2n+j-s$ can be introduced
instead of $n$. Then the relation (\ref{spectrum_background_zero})
can be represented as follows:
\begin{equation}
E^2_n-m^2=2\left(N-j+\frac{1}{2}\right)
\left[m\omega+\frac{1}{2}(m^2\omega^2\beta+\alpha)\left(N+j+\frac{3}{2}\right)\right].
\end{equation}
\subsection{Nonzero ground state energy}
Now we suppose that in the right-hand side of the equations
(\ref{equ_4_b+-}) and (\ref{equ_4_b-+}) we have $E^2-m^2\neq 0$.
It will be shown that in this case the ground state exists for the
following hamiltonian:
\begin{equation}\label{gr_state_0}
h_0=b^+_p(k,\eta)b^-_p(k,\eta)=-\left(\sqrt{1-\beta p^2}\frac{\partial}{\partial p}\right)^2+(\eta-\eta^*)p\frac{\partial}{\partial p}+\frac{k^2-k}{p^2}+
\frac{\frac{1}{\beta}|\eta|^2-\eta^*}{1-\beta p^2}-k(\eta+\eta^*)-k^2\beta-\frac{1}{\beta}|\eta|^2
\end{equation}
In order to obtain ground state energy one should re-factorize hamiltonian $h_0$. It can be represented as follows:
\begin{equation}\label{gr_state_1}
h_0=b^+_p(k',\eta')b^-_p(k',\eta')+\varepsilon',
\end{equation}
where $k'$ and $\eta'$ are new parameters in operators
(\ref{b_p+}) and (\ref{b_p-}) and $\varepsilon'$ defines the
ground state energy. From equations (\ref{gr_state_0}) and
(\ref{gr_state_1}) it follows:
\begin{equation}\label{gr_st_im_eta}
\eta'-\eta'^*=\eta-\eta^*
\end{equation}
\begin{equation}\label{gr_st_k}
k'^2-k'=k^2-k
\end{equation}
\begin{equation}\label{gr_st_re_eta}
\frac{1}{\beta}|\eta'|^2-\eta'=\frac{1}{\beta}|\eta|^2-\eta
\end{equation}
\begin{equation}\label{gr_st_epsilon}
-k'(\eta'+\eta'^*)-\beta k'^2-\frac{1}{\beta}|\eta'|^2+\varepsilon=-k(\eta+\eta^*)-\beta k^2-\frac{1}{\beta}|\eta|^2
\end{equation}
Solving the equations (\ref{gr_st_im_eta})-(\ref{gr_st_re_eta})
we arrive at the relations:
\begin{eqnarray}
k'_1=k, \quad k'_2=1-k;\label{cond_gr_st_k}\\
\zeta'=\zeta,\quad \xi'_1=\xi, \quad
\xi'_2=\beta-\xi.\label{cond_gr_st_eta}
\end{eqnarray}
Since conditions for parameters $k'$ (\ref{cond_gr_st_k}) and
$\eta'$ (\ref{cond_gr_st_eta}) are obtained independently then one
can combine different $k'$ and $\eta'$ to investigate whether
obtained wavefunctions will be physically acceptable.
For the first we consider the case $k'=k$ and $\xi'=\xi$. It
follows immediately that $\varepsilon=0$. So the latter
combination should be left out.
Then if we suppose that $k'=k$ and $\xi'=\beta-\xi$ equation
(\ref{wavefunct_b_0-}) gives us corresponding wavefunction
$R_{1;0}=C_{1;0}p^k(1-\beta
p^2)^{\frac{\beta-\xi}{2\beta}+i\frac{\zeta}{2\beta}}$. The first
requirements imposed on this function are boundary conditions. To
provide the condition $R_{1;0}=0$ at the boundaries one should
demand that $k>0$ and $\beta-\xi>0$. As a consequence condition
$k>0$ leads to requirement $s=\frac{1}{2}$ whereas the demand
$\beta-\xi>0$ gives rise to
$m\omega\in(0,\frac{1-\sqrt{1-4\alpha\beta}}{2\beta})$ or
$m\omega\in(\frac{1+\sqrt{1-4\alpha\beta}}{2\beta}, \infty)$. If
$4\alpha\beta>1$ the last condition can be satisfied for arbitrary
$m\omega$. To make obtained wave function physically acceptable it
must fulfil normalizability condition (\ref{funct_R^0_1}) and even
stronger requirement (\ref{P_2_integral}). From the last
requirement it follows that $-\frac{\xi}{\beta}+\frac{1}{2}>0$ and
as a result it gives rise to the restrictions for the product
$m\omega$:
\begin{equation}\label{cond_nonzero_gr_state}
m\omega\in\left(0,\frac{1-\sqrt{1-\alpha\beta}}{\beta}\right)\bigcup\left(\frac{1+\sqrt{1-\alpha\beta}}{\beta},
\infty\right).
\end{equation}
One can see that obtained restrictions for the product $m\omega$
are opposite to (\ref{cond_gr_state=0}). As a conclusion, if the
relation (\ref{cond_gr_state=0}) is fulfilled then the ground
state has zero energy. If the condition (\ref{cond_gr_state=0}) is
broken the ground state with nonzero energy appears.
From relation (\ref{gr_st_epsilon}) we obtain
\begin{equation}\label{gr_st_energ_s=1/2}
\varepsilon=(\beta-2\xi)(1+2k).
\end{equation}
It is easy to verify that obtained ground state energy is
positive.
To find other eigenvalues of the hamiltonian $h_0$ one should
substitute $\xi'$ instead of $\xi$ into
(\ref{eigenvalues_general_e=0}) and take into account relation
(\ref{gr_st_energ_s=1/2}). After necessary transformations we
arrive at:
\begin{equation}\label{spectrum_s=1/2_E}
E^2_n-m^2=4(n+j+1)\left[-m\omega+(m^2\omega^2\beta+\alpha)\left(n+\frac{1}{2}\right)\right]
\end{equation}
Similarly as in previous case the obtained relation can be
represented in terms of principal quantum number
\begin{equation}
E^2_n-m^2=2\left(N+j+\frac{5}{2}\right)\left[-m\omega+\frac{1}{2}
(m^2\omega^2\beta+\alpha)\left(N-j+\frac{3}{2}\right)\right].
\end{equation}
From relations (\ref{cond_gr_st_k}) and (\ref{cond_gr_st_eta}) it
follows that ground states with nonzero energy other combinations
of $k'$ and $\eta'$ are also possible. We consider combination
$k'=1-k$ and $\xi'=\xi$. Then the ground state wavefunction takes
form: $R_{1;0}=C_{1;0}p^{1-k}(1-\beta
p^2)^{\frac{\xi}{2\beta}+i\frac{\zeta}{2\beta}}$. One of boundary
conditions leads to restriction $k<0$ which can be satisfied if
$s=-\frac{1}{2}$ and then $k'=j+\frac{3}{2}$. Since $\xi>0$ the
second boundary condition is satisfied immediately. It is easy to
persuade oneself that obtained wavefunction is normalizable.
Similarly as in the previous cases to make the obtained
wavefunction physically acceptable we should impose condition
(\ref{P_2_integral}) on it. Having used formula
(\ref{gr_st_epsilon}) we obtain following relation for the ground
state energy:
\begin{equation}
\varepsilon=(\beta+2\xi)(1-2k)
\end{equation}
One can see that ground state energy is also positive. Using the
same procedure as in case with positive $k$ on can obtain energy
spectrum:
\begin{equation}
E^2_n-m^2=4(n+j+1)\left[m\omega+(m^2\omega^2\beta+\alpha)\left(n+\frac{1}{2}\right)\right]
\end{equation}
Having introduced the principal quantum number we rewrite the last
relation in the form:
\begin{equation}
E^2_n-m^2=2\left(N+j+\frac{3}{2}\right)\left[m\omega+\frac{1}{2}
(m^2\omega^2\beta+\alpha)\left(N-j+\frac{1}{2}\right)\right].
\end{equation}
In the end we can choose combination $k'=1-k$ and
$\xi'=\beta-\xi$. This variant leads to the wave function
$R_{1;0}=C_{1;0}p^{1-k}(1-\beta
p^2)^{\frac{\beta-\xi}{2\beta}+i\frac{\zeta}{2\beta}}$. One of the
boundary conditions gives rise to the demand $k<0$ or equivalently
$k'=j+\frac{3}{2}$. Another boundary condition leads to inequality
$\beta-\xi>0$ but as we already know normalizability condition and
boundness of the square of momentum operator should be satisfied
and as a consequence all these demands lead to the condition
(\ref{cond_nonzero_gr_state}).
For the ground state we obtain:
\begin{equation}\label{gr_st_4}
\varepsilon=4(\beta(1-k)-\xi).
\end{equation}
It can be shown that the ground state energy (\ref{gr_st_4}) is
positive if $4\alpha\beta>1$. The same procedure leads us to the
following expression for the spectrum:
\begin{equation}
E^2_n-m^2=4(n+1)\left(-m\omega+(m^2\omega^2\beta+\alpha)\left(n+j+\frac{3}{2}\right)\right)
\end{equation}
Again we rewrite obtained formula replacing the quantum number $n$
by the principal quantum number $N$:
\begin{equation}
E^2_n-m^2=2\left(N-j+\frac{3}{2}\right)\left[-m\omega+(m^2\omega^2\beta+\alpha)
\left(N+j+\frac{5}{2}\right)\right].
\end{equation}
We note that this case does not have ``classical" limit or in
other words when parameters of deformation
$\alpha,\beta\rightarrow 0$ obtained spectrum does not reduce to
any solution of ordinary quantum mechanics \cite{Moshinsky_JPA89}.
Similar situation appears in the case of deformed algebra with
minimal length \cite{QuesneJPA05}.
\section{Radial momentum wavefunctions of Dirac oscillator}
In the previous section the ground state wavefunctions of
hamiltonian $h=b^+_pb^-_p$ have been derived. As it was shown only
the large component of wavefunction can be obeyed to all imposed
requirements. In this section we calculate the remaining large and
small components of radial momentum wavefunction
\subsection{Zero ground state energy}
The large component of radial momentum wavefunction for excited
states can be calculated with help of well-known SUSY QM and SI
technique \cite{Cooper_PhysRept95,Junker_96}. As it is known the
wave functions of the excited states are derived form the ground
state wavefunction with help of recursive procedure which is based
on the relation:
\begin{equation}\label{iter_proc_wf}
R_{1;n} (p;k,\xi)=\frac{1}{\sqrt{e_n-e_0}}b^+_p(k,\xi)R_{1;n-1}(p;k_1,\xi_1).
\end{equation}
Where we used notation $e_i=(E^2_i-m^2)/|\tilde{\omega}|^2$ for
simplicity. According to the conditions (\ref{iter_cond_e=0}) and
(\ref{eigenvalues_general_e=0}) we should impose $e_0=0$,
$k_1=k+1$ and $\xi_1=\xi+\beta$.
Having substituted the explicit form of operator $b^+_p$ into the
relation (\ref{iter_proc_wf}) we arrive at the equation:
\begin{equation}
R_{1;n} (p;k,\xi)=\frac{1}{\sqrt{e_n}}\left(-\sqrt{1-\beta p^2}\frac{\partial}{\partial p}-\frac{k}{p}\sqrt{1-\beta p^2}+\frac{(\xi+i\zeta)p}{\sqrt{1-\beta p^2}}\right)R_{1;n-1}(p;k+1,\xi+\beta).
\end{equation}
Given recursion procedure leads to consequence that the large
component of radial wavefunction takes form:
\begin{equation}\label{wf_e=0}
R_{1;n}(p;k,\xi)=C_{1;n}(k,\xi)p^{b+\frac{1}{2}}(1-\beta p^2)^{\frac{a}{2}+\frac{1}{4}-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z),
\end{equation}
where $C_{1;n}(k,\xi)$ and $P^{(a,b)}_n(z)$ are a normalization
factor a Jacobi polynomial respectively. Here we also denoted:
\begin{equation}
a=\frac{\xi}{\beta}-\frac{1}{2}, \quad b=k-\frac{1}{2}, \quad
z=2\beta p^2-1 \quad (-1<z<1).
\end{equation}
It was argued in the previous section that the small component of
the ground state radial wavefunction vanishes
$\tilde{R}_{2;0}(p;k,\xi)=0$. For excited states small component
can be found by using relation (\ref{equ_3_b-}):
\begin{equation}\label{wf_tilde_R_2}
\tilde{R}_{2;n}(p;k,\xi)=\frac{\tilde{\omega}^*}{E_n+m}b^-_p(k,\xi)R_{1;n}(p;k,\xi)
\end{equation}
Taking into account explicit expressions for the operator
$b^-(k,\xi)$ (\ref{b_p-}) and wave function (\ref{wf_e=0}) one can
rewrite the last relation in form:
\begin{eqnarray}\label{wf_e=0_2}
\tilde{R}_{2;n}(p;k,\xi)=\frac{\tilde{\omega}^*C_{1;n}(k,\xi)}{E_n+m}\sqrt{1-\beta p^2}\left(\frac{\partial}
{\partial p}-\frac{k}{p}+\frac{\eta^* p}{1-\beta
p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{a}{2}+\frac{1}{4}-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial
p}-\frac{b+\frac{1}{2}}{p}+\frac{\left(\beta\left(a+\frac{1}{2}\right)-i\zeta\right)
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=\frac{2\beta
\tilde{\omega}^*C_{1;n}(k,\xi)(n+a+b+1)}{E_n+m}p^{b+\frac{3}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{3}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a+1,b+1)}_{n-1}(z).
\end{eqnarray}
It should be noted that a formula of differentiation of the Jacobi
polynomials was used here \cite{Bateman_1953,Abramowitz}. In the
previous section it was stated that wavefunction
$(R_{1;0}(p;k,\xi)\neq 0,\tilde{R}_{2;0}(p;k,\xi)=0)$ is the
physically acceptable solution of the system of equations
(\ref{equ_3_b+}) and (\ref{equ_3_b-}) only for $ E^2_0=m^2$. At
the same time for excited states: $n=1,2,\ldots$ the solution of
this system of equations is given by
$(R_{1;n}(p;k,\xi),\tilde{R}_{2;n}(p;k,\xi))$. It is necessary to
verify whether these function are physically acceptable or not. It
is easy to persuade oneself that the Jacobi polynomials in
(\ref{wf_e=0}) and (\ref{wf_e=0_2}) do not spoil the convergence
of integral (\ref{normal_wavefunct}) and also meanvalues for
square of momentum and position operators would be finite
similarly as it was given by the condition (\ref{P_2_integral})
for the ground state wavefunction. Finally, the normalization
factor $C_{1;n}$ can be found from the normalization condition
(\ref{normal_wavefunct}):
\begin{equation}
C_{1;n}=\left(\beta^{b+1}(2n+a+b+1)\frac{n!\Gamma(n+a+b+1)}{\Gamma(n+a+1)\Gamma(n+b+1)}\frac{E_n+m}{E_n}\right)^{\frac{1}{2}}
\end{equation}
\subsection{Nonzero groundstate energy}
To find wavefunctions of excited states in remaining cases one
should follow the approach used in the previous section.
Parameters $k$ and $\xi$ in the iteration equation
(\ref{iter_proc_wf}) should be replaced by $k'$ and $\xi'$
correspondingly. It worth noting that at the same time parameters
$k$ and $\xi$ in the equation (\ref{wf_tilde_R_2}) remain
unchanged. As a consequence we can state that equation
(\ref{wf_e=0_2}) remains valid if parameters $a$ and $b$ are
replaced by a new one.
In the case $k'=k$ that corresponds $s=\frac{1}{2}$ and
$\xi'=\beta-\xi$ we obtain:
\begin{equation}
R_{1;n}(p;k,\xi)=C_{1;n}p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)
\end{equation}
where $a=\frac{1}{2}-\frac{\xi}{\beta}$ and $b=k-\frac{1}{2}$.
The relation (\ref{wf_tilde_R_2}) gives rise to:
\begin{eqnarray}
\tilde{R}_{2;n}(p;k,\xi)=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial p}-\frac{k}{p}+\frac{\eta^*
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial
p}-\frac{b+\frac{1}{2}}{p}+\frac{\left(\beta\left(\frac{1}{2}-a\right)-i\zeta\right)
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=-\frac{2\beta\tilde{\omega}^*(a+n)C_{1;n}}{E_n+m}p^{b+\frac{3}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a-\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a-1,b+1)}_{n}(z)
\end{eqnarray}
In the case $k'=1-k$ that corresponds $s=-\frac{1}{2}$ and
$\xi'=\xi$ we arrive at:
\begin{equation}
R_{1;n}(p;k,\xi)=C_{1;n}p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)
\end{equation}
where $a=\frac{\xi}{\beta}-\frac{1}{2}$ and $b=\frac{1}{2}-k$.
Again the relation (\ref{wf_tilde_R_2}) leads to:
\begin{eqnarray}
\tilde{R}_{2;n}(p;k,\xi)=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial p}-\frac{k}{p}+\frac{\eta^*
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial
p}-\frac{\frac{1}{2}-b}{p}+\frac{\left(\beta\left(a+\frac{1}{2}\right)-i\zeta\right)
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=\frac{2\beta\tilde{\omega}^*(b+n)C_{1;n}}{E_n+m}p^{b+\frac{3}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{3}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a+1,b-1)}_{n}(z)
\end{eqnarray}
In the end we consider the case $k'=1-k$ or equivalently as
previously $s=-\frac{1}{2}$ and $\xi'=\beta-\xi$. We arrive at:
\begin{equation}
R_{1;n}(p;k,\xi)=C_{1;n}p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)
\end{equation}
where $a=\frac{1}{2}-\frac{\xi}{\beta}$ and $b=\frac{1}{2}-k$.
Having used relation (\ref{wf_tilde_R_2}) we obtain:
\begin{eqnarray}
\tilde{R}_{2;n}(p;k,\xi)=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial p}-\frac{k}{p}+\frac{\eta^*
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=\frac{\tilde{\omega}^*C_{1;n}}{E_n+m}\sqrt{1-\beta
p^2}\left(\frac{\partial} {\partial
p}-\frac{\frac{1}{2}-b}{p}+\frac{\left(\beta\left(\frac{1}{2}-a\right)-i\zeta\right)
p}{1-\beta p^2}\right)p^{b+\frac{1}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a+\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a,b)}_n(z)\\\nonumber
=-\frac{2\beta\tilde{\omega}^*(n+1)C_{1;n}}{E_n+m}p^{b+\frac{3}{2}}(1-\beta
p^2)^{\frac{1}{2}\left(a-\frac{1}{2}\right)-i\frac{\zeta}{2\beta}}P^{(a-1,b-1)}_{n+1}(z)
\end{eqnarray}
It has been already noted that the last case does not have
``classical" limit when the parameters of deformation $\alpha$,
$\beta$ tend to zero. Similarly in the case of deformed algebra
with minimal length \cite{QuesneJPA05} bounded states which do not
have classical limit appear.
\section{Discussion}
In this work we considered the Dirac oscillator problem in
deformed space given by the commutation relations (\ref{algebra}).
It was shown that deformed commutation relations (\ref{algebra})
give rise to minimal uncertainty in position as well as in
momentum. To find appropriate representation for position and
momentum operators a specific nonsymplectic transformation was
proposed \cite{Mignemi_arxiv}. It allows one to find some relation
between given algebra (\ref{algebra}) and well known Snyder
algebra. Having used proposed representation it has been solved
exactly the Dirac oscillator eigenvalue problem.
It has been shown that the Dirac oscillator in deformed space with
commutation relations (\ref{algebra}) has some common features
with conventional case as well as in case of deformation with
minimal length only. A dissymmetry under the exchange of
$s=\frac{1}{2}$ with $s=-\frac{1}{2}$ that appeared in nondeformed
case due to specific substitution ${\bf P}\rightarrow{\bf
P}-im\omega{\bf X}\hat{\beta}$ takes place in case of Snyder-de
Sitter deformed algebra (\ref{algebra}). The same situation
happens in case of deformed algebra with minimal length
\cite{QuesneJPA05}. If we consider system of equations
(\ref{equ_3_b+}) and (\ref{equ_3_b-}) and make substitution
$\omega\rightarrow -\omega$ the system can be transformed to
equivalent one where $s$ is replaced by $-s$ and $E, R_1,
\tilde{R}_2$ are changed into $-E,-\tilde{R}_2, R_1$ respectively.
This transformation is valid in nondeformed case
\cite{Moshinsky_JPA89} and in the presence of deformed algebra
with minimal length \cite{QuesneJPA05}. In nondeformed situation
it is treated in connection with supersymmetry or, equivalently,
with duality between particles and antiparticles
\cite{Moshinsky_FoundPhys93}. Another similarity with previous
cases lies in the absence of negative energy $E=-m$ ground states
\cite{Moshinsky_JPA89,QuesneJPA05}.
It has been noted above the energy spectrum of the Dirac
oscillator with deformed commutation relations (\ref{algebra})
takes similar form as in case of deformed algebra with minimal
length. \cite{QuesneJPA05}. In particular, the difference
$E^2_n-m^2$ gets terms quadratic in $n$ instead of linear
dependence in nondeformed instance. It should be noted that the
relations for the energy spectrum would be in agreement with each
other if the parameter $\alpha$ in our expressions is set to zero
whereas in relations obtained in \cite{QuesneJPA05} the only
parameter corresponding to our $\beta$ is kept. We also note that
in case of deformed algebra with minimal length ground state with
energy $E^2-m^2=0$ is allowed for small values $j$ only
\cite{QuesneJPA05}. In contrast to it the Snyder-de Sitter algebra
(\ref{algebra}) does not make any restriction for parameter $j$
similarly as it was in ordinary quantum mechanics
\cite{Moshinsky_JPA89}. Ground states with nonvanishing energy
$E^2-m^2\neq 0$ are allowed for both projections of spin:
$s=\frac{1}{2}$ and $s=-\frac{1}{2}$. Here similarly to
nondeformed situation no restriction on value of total angular
momentum quantum number $j$ is imposed. It is worth stressing that
in order to have physically acceptable wavefunctions parameters of
oscillator should fulfil some conditions, namely product $m\omega$
can not take any value but it should satisfy such requirements as
(\ref{cond_gr_state=0}) or (\ref{cond_nonzero_gr_state}).
We also remark that although Dirac oscillator was introduced as a
relativistic problem in our case it is not Lorentz covariant. This
is caused by the fact that chosen algebra of operators
(\ref{algebra}) is not a relativistic one. The algebra
(\ref{algebra_2}) is obtained from the relativistic Snyder-de
Sitter algebra \cite{Mignemi_arxiv,Kowalski-glikman_PRD04} and it
seems that it easy to consider fully relativistic case but
unfortunately some problems appear. The first one is that both
time and energy will be represented by differential operators as
we have here for position and momentum operators. The second
problem is related to the behaviour of minimal uncertainties under
Lorentz transformations. These questions need careful
consideration and will examined elsewhere.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 9,475 |
Q: Cartesian Product of non-rectangular region I know that the Cartesian product $[0,1]\times[0,1]$ represents a unit square in the first quadrant. Is it possible to write the triangular region with vertices $(0,0)$, $(0,1)$ and $(1,1)$, interior included, as a Cartesian product o two sets? I think one can have $A=\{x\in\mathbb{R}\, |\,0\leq x\leq 2\}$ and $B=\{y\in\mathbb{R}\, |\, x\leq y\leq 2, 0\leq x\leq 2\}$. Then one can consider $A\times B$.
However, consider the ordered pair $(1,0.5)$, which is not in the region in question. Is this ordered pair in $A\times B$? Clearly $x=1\in A$, and $y=0.5\in B$ if one picks $x=0$ (which is in $A$), so that $y=0.5$ satisfies $0\leq 0.5\leq 2$. Or does the $x$ that I choose to determine the $y$ have to be $x=1$? So that for $x=1$, the $y$ values in $(1,y)$ can only come from $1\leq y\leq 2$? If so, then it works. Thanks!
A: The problem with $B$ as you have stated it in your question is that it is not well-defined: it depends on the value of $x$ you choose. I think the error you are making is that for an arbitrary element of the cartesian product $(x,y) \in A \times B$, the value $y$ cannot depend on the value $x$.
You can't have such a triangle as the product of two subsets of $\mathbb{R}$. Suppose you could write it as $A \times B$ for some $A,B \subset \mathbb{R}$. Then we would need $(0,0) \in A \times B$ (so that $0 \in A$ and $0 \in B$) and $(1,1) \in A \times B$ (so that $1 \in A$ and $1 \in B$). But then the point $(1,0)$ is in $A \times B$ (since $1 \in A$ and $0 \in B$), which is not in the triangle you described.
Geometrically, you can think of the cartesian product $A \times B$ as putting a (vertical) copy of $B$ at every $x$-coordinate in $A$. So it will always look rectangular (perhaps with some holes).
A: No, you can't, I'm afraid. The $x$ in the definition of $B$ is not the same as the $x$ in the definition of $A$. Indeed, the definition of $B$ is a little ambiguous regarding $x$. Is the condition $x \le y \le 2$ for all $0 \le x \le 2$? Or is it $x \le y \le 2$ for some $0 \le x \le 2$? In the former case, we would have $B = \{2\}$ (and $A \times B$ would be a line segment) or in the latter case, we would have $B = [0, 2]$, in which case $A \times B$ is a square.
In general, $A \times B$ must be a disjoint union of "rectangles", where we use the term in a very loose sense: we include points and open/closed/semi-open rectangles/line segments. It cannot be the triangle you described. Why? If $(0, 0) \in A \times B$, then $0 \in A$ and $0 \in B$. Similarly if $(1, 1) \in A \times B$, then $1 \in A$ and $1 \in B$. This implies that $(1, 0) \in A \times B$, which is not inside the triangle.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 8,840 |
Q: Write a file with the version number during Heroku deployment I want the current code version to be available as a variable at runtime in my express/node.js project running on Heroku.
In the development environment I use a git describe node module and have it give me a semver string, but in the production Heroku environment there is no git repository, and no git binary anyway.
The way I've gone about solving this in the past, though in non-Heroku environments, was to write a version file in the project's root directory with this data as a string, and then have the application load this in rather than invoking git all the time. In most of these cases git did exist and we were in a real repository in production, so those were easy. Where git was not available, or the environment wasn't a full repository (didn't have tags, or was a shallow clone) I did still have the commit's SHA hash at deployment time, provided by the deployment tool, and so wrote an unpleasant script which ran and used it during deployment. It queried Github's API for tags, found a matching one if possible, and wrote the tag name to the version file.
Ideally I want a semver string based on the current tag and offset. In particular, in node with the above-mentioned package I run this:
var version = require('git-describe').gitDescribeSync();
return version.semverString || version.raw;
(In a pinch, git describe --dirty --tags --always would also be acceptable.)
What's a good way to achieve this? Is the current tag name or commit hash or anything like that available (such as in an environment variable) at deployment time on Heroku, which I could then use during the build process to write a file or similar? That'd be messy but it'd work. Or is there something better?
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 789 |
Q: Why won't an alarm handler fire in a process run from a php process? I don't have much experience with PHP/mod_php administration, so I apologize if this is a really simple question.
My question is this - why won't a process that I've spawned from a PHP script via the exec() call receive an alarm interrupt properly?
The long version of my question:
This morning I was handed a bug in an existing php script. After some investigation, I've traced it down to the (apparent) fact that the php was using exec() to run a subprocess, the subprocess was relying on a SIGALRM to escape a loop, and it never received the alarm.
I don't think it matters, but the specific subprocess was /bin/ping. When pinging a device that doesn't return any packets (such as a device w/ a firewall that discards ICMP echo requests instead of returning destination host unreachable), you have to use the -w option to set a timer to allow the program to exit (because -c counts return packets - if the target never returns packets and you don't use -w, you're stuck in and endless loop). When called from the php, the alarm handler that ping -w relies on doesn't fire.
Here're a few interesting lines from using strace to follow the ping call from the command line (where the alarm handler does work):
(snip)
setitimer(ITIMER_REAL, {it_interval={0, 0}, it_value={1, 0}}, NULL) = 0
(snip)
--- SIGALRM (Alarm clock) @ 0 (0) ---
rt_sigreturn(0xe) = -1 EINTR (Interrupted system call)
When I inserted a shell wrapper to allow me to run strace on the ping when called from the web, I found that the setitimer call is present (and appears to run successfully), but that the SIGALRM line and rt_sigreturn() lines aren't present. The ping then continues to run sendmsg() and recvmsg() forever until I kill it by hand.
Trying to reduce variables, I then cut ping out of it and wrote the following perl:
[jj33@g3 t]# cat /tmp/toperl
#!/usr/bin/perl
$SIG{ALRM} = sub { print scalar(localtime()), " ALARM, leaving\n"; exit; };
alarm(5);
print scalar(localtime()), " Starting sleep...\n";
sleep (10);
print scalar(localtime()), " Exiting normally...\n";
It works as expected when run from the command line, the alarm handler fires successfully:
[jj33@g3 t]# /tmp/toperl
Mon May 2 14:49:04 2011 Starting sleep...
Mon May 2 14:49:09 2011 ALARM, leaving
Then I tried running /tmp/toperl via the same PHP page (via both exec() and backticks) that was having problems calling ping. Here's the php wrapper I wrote for the test:
<?
print "Running /tmp/toperl via PHP\n";
$v = `/tmp/toperl`;
print "Output:\n$v\n";
?>
As with ping, /tmp/toperl did not receive its alarm interrupt:
Running /tmp/toperl via PHP
Output:
Mon May 2 14:52:19 2011 Starting sleep...
Mon May 2 14:52:29 2011 Exiting normally...
Then I wrote a quick cgi wrapper in perl to execute in the same Apache, but under mod_cgi instead of mod_php. Here's the wrapper for reference:
[jj33@g3 t]# cat tt.cgi
#!/usr/bin/perl
print "Content-type: text/plain\n\n";
print "Running /tmp/toperl\n";
my $v = `/tmp/toperl`;
print "Output:\n$v\n";
And, lo and behold, the alarm handler worked:
Running /tmp/toperl
Output:
Mon May 2 14:55:34 2011 Starting sleep...
Mon May 2 14:55:39 2011 ALARM, leaving
So, back to my original question - why won't a process I've spawned via exec() in a mod_php controlled PHP script receive an alarm signal when the same spawned process will do so when called from the command line and perl/mod_cgi?
Apache 2.2.17, PHP 5.3.5.
Thanks for any thoughts.
EDIT - DerfK was correct, mod_php is masking out SIGALRM before calling the sub process. I don't have any interest in recompiling ping so I'll end up writing a wrapper for it. Since I already wrote so much text for this question I thought I would also drop in a revision to my toy program /tmp/toperl that tests to see if SIGALRM is being masked out and unblocking it if so.
#!/usr/bin/perl
use POSIX qw(:signal_h);
my $sigset_new = POSIX::SigSet->new();
my $sigset_old = POSIX::SigSet->new();
sigprocmask(SIG_BLOCK, $sigset_new, $sigset_old);
if ($sigset_old->ismember(SIGALRM)) {
print "SIGALRM is being blocked!\n";
$sigset_new->addset(SIGALRM);
sigprocmask(SIG_UNBLOCK, $sigset_new);
} else {
print "SIGALRM NOT being blocked\n";
}
$SIG{ALRM} = sub { print scalar(localtime()), " ALARM, leaving\n"; sigprocmask(SIG_BLOCK, $sigset_new, $sigset_old); exit; };
alarm(5);
print scalar(localtime()), " Starting sleep...\n";
sleep (10);
print scalar(localtime()), " Exiting normally...\n";
Now this test works correctly (meaning it exits after 5 seconds with the "ALARM, leaving" line) in all instances (perl/command line, php/command line, perl/mod_cgi, php/mod_php). In the first three instances it prints the 'SIGALRM NOT being blocked' line, in the latter it prints 'SIGALRM is being blocked!' and correctly unblocks it.
A: Mod_php probably blocks this (using sigprocmask() I presume, masks are maintained through fork() and execve()) in order to prevent signals from mangling Apache (since mod_php runs PHP in apache's process).
If it's due to sigprocmask(), then I think you should be able to use perl's POSIX module to undo it in the exec()'d script, but I'm not exactly sure how it works. The Perl Cookbook has an example of blocking then unblocking SIGINT. I think it should be something like
use POSIX qw(:signal_h);
$sigset=POSIX::SigSet->new(SIGALRM);
sigprocmask(SIG_UNBLOCK,$sigset);
If that doesn't work, then maybe try installing php5-cgi, setting it up as a Handler in Apache for a different extension (say, .phpc) then renaming the script to ping.phpc and updating the links. Since CGI executes in its own process, the CGI version of PHP may not lock down the signals.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 1,347 |
Venilia (pronounced , or as Latin Venīlia) is a Roman deity associated with the winds and the sea. According to Virgil and Ovid, she was a nymph, the sister of Amata and the wife of Janus (or Faunus), with whom she had three children: Turnus, Juturna, and Canens.
She and Salacia are the paredrae of Neptune.
The Venilia Mons, a mountain on Venus, is named for her.
See also
Pantoporia venilia, a butterfly of the family Nymphalidae
Terebra venilia, a species of sea snail
External links
Neptune, Venilia, and Triton Fountain, Library of Congress, Washington DC
Venilia and Horse, Library of Congress, Washington DC
Venilia and Horse, detail
References
Roman goddesses
Sky and weather goddesses
Sea and river goddesses
Neptune (mythology) | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 7,672 |
Facultatea de Litere este o facultate care face parte din cadrul Universității "Alexandru Ioan Cuza" din Iași. Aceasta există din anul 1860, o dată cu înființarea Universității "Alexandru Ioan Cuza".
Istoric
Legea învățământului din 1864 utilizează numele de Facultatea de Filosofie și Litere. Va mai figura în documentele oficiale sub diferite alte denumiri : Facultatea de Litere și Filosofie (până în 1948), Facultatea de Filologie – Istorie – Filosofie (între 1948-1960), Facultatea de Filologie (între 1960-1989), Facultatea de Litere (începând cu 1989). Când a fost înființată facultatea avea doar o singură catedră Literatură clasică (latină) și română. Printre personalitățile care au predat aici sunt: V. A. Urechia, Alexandru I. Philippide, Titu Maiorescu, Garabet Ibrăileanu, Ștefan Vârgolici, Anton Naum, Ilie Bărbulescu, G. Pascu, Traian Bratu, Charles Drouhet, Neculai Șerban, Th. Simenschy, Iorgu Iordan, Constantin Balmuș, Octav Botez, George Călinescu, Dan Simonescu, Șerban Cioculescu, Petru Caraman, Jean Livescu, Alexandru Dima, N. I. Popa, Gheorghe Ivănescu, Șt. Cuciureanu, Ariton Vraciu, Mihai Drăgan.
Departamente
Departamentul de Românistică, Jurnalism și științe ale comunicării și Literatură comparată
Departamentul de Limbi și literaturi străine
Departamentul de Cercetare
Școala Doctorală de Filologie
Legături externe
Pagina principală
Universitatea "Alexandru Ioan Cuza" din Iași
Facultăți din Iași | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 4,472 |
Q: ImageMagick PHP need a good quality thumbnail generation I am using php imagemagick to generate thumbnail image.
*
*If i give bestfit true and size is 50x50 it generating 36x50 size image(good quality)
*if i give bestfit false and size is 50x50 it generating 50x50 size image(bad quality)
I need a solution that whatever the image size(minimum i will upload 200X200) i give
it need to give 50x50 with good quality and techniques there in ImageMAgick ??
A: Use $im->cropThumbnailImage() instead of $im->thumbnailImage().
A: Take a look at PHP Thumb (MIT license).
It supports Adaptive Resizing.
What it does is resize the image to get as close as possible to the desired dimensions, then crops the image down to the proper size from the center.
require_once '/path/to/ThumbLib.inc.php';
$thumb = PhpThumbFactory::create('test.jpg');
$thumb->adaptiveResize(50, 50)->save('/path/to/new_thumb.jpg');
PHP Thumb is a light-weight image
manipulation library aimed at
thumbnail generation. It features the
ability to resize by width, height,
and percentage, create custom crops,
or square crops from the center, and
rotate the image.
PHP Thumb Github WIKI
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 1,277 |
<html xmlns:fb="http://ogp.me/ns/fb#">
<head>
<title>Birthday Responder</title>
<link rel="shortcut icon" href="http://img.1mobile.com/market/i/e/4/e4c683ceaf73da99356372019fa88f75.png" />
<link rel="stylesheet" type="text/css" href="style.css" />
<script type="text/javascript" src="http://connect.facebook.net/es_ES/all.js"></script>
<script type="text/javascript">
window.fbAsyncInit = function() {
FB.init({
appId:'464995143525912',
cookie:true,
status:true,
xfbml:true,
oauth: true
});
FB.getLoginStatus(function(response) {
if (response.authResponse) {
document.location.href="http://bdayresponder.herokuapp.com/comment.php";
}
});
}
</script>
</head>
<body>
<?php
require('./src/facebook.php');
$facebook = new Facebook();
$user=$facebook->getUser();
if($user)
{
$logout=$facebook->getLogoutUrl();
}
else
{
$login=$facebook->getLoginUrl(array(
'redirect_uri' => 'http://bdayresponder.herokuapp.com/comment.php',
'scope' => 'user_birthday,publish_stream,read_stream'
)
);
}
if($user)
{
header("location:http://bdayresponder.herokuapp.com/comment.php");
}
else
{
?>
<a href="<?php echo($login); ?>"><img src="fblogin.png" alt="Login"/></a>
<p align="center">
Your birthday?too many wall posts? Don't worry. This app lets you reply to all wall posts and like them too with a single click.
Hope you like it.<br/>
Please Provide the necessary permissions so that you could be benefitted from the App.<br/>
No request or message would be sent to anyone without your concern.<img src='kopete020.png' alt=':)'/>
</p>
<div class="fb-like" data-href="https://www.facebook.com/pages/Birthday-Responder/423834974322726" data-send="true" data-layout="box_count" data-width="450" data-show-faces="true" data-colorscheme="dark" data-font="verdana"></div>
<?php
}
?>
</body>
</html> | {
"redpajama_set_name": "RedPajamaGithub"
} | 765 |
using System;
using System.Collections.Generic;
using System.Globalization;
using Microsoft.AspNetCore.Authentication.Cookies;
using Microsoft.AspNetCore.Authentication.OpenIdConnect;
using Microsoft.AspNetCore.Builder;
using Microsoft.AspNetCore.Hosting;
using Microsoft.AspNetCore.Http;
using Microsoft.AspNetCore.HttpOverrides;
using Microsoft.AspNetCore.Localization;
using Microsoft.AspNetCore.Mvc;
using Microsoft.AspNetCore.Mvc.Razor;
using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Logging;
using Microsoft.Extensions.Options;
using Login.Api.Infrastructure.Configuration;
using Login.Api.Infrastructure.Middleware;
using Login.Api.Features.Shared.Persistence;
using Login.Api.Features.User;
using Microsoft.Extensions.Primitives;
using Commons.Api.Middleware;
using Commons.Api.Views;
namespace Login.Api.Infrastructure
{
public partial class Startup
{
const string ApplicationName = "login.binggl.net";
const string ApplicationDescription = "login.binggl.net API";
readonly ILogger logger;
private IServiceCollection _serviceColletion;
public Startup(IConfiguration configuration, IHostingEnvironment env, ILogger<Startup> logger)
{
CurrentEnvironment = env;
Configuration = configuration;
this.logger = logger;
logger.LogInformation($"Started application '{ApplicationName}' in mode '{env.EnvironmentName}'.");
}
public IConfiguration Configuration { get; }
private IWebHostEnvironment CurrentEnvironment { get; set; }
// This method gets called by the runtime. Use this method to add services to the container.
public void ConfigureServices(IServiceCollection services)
{
_serviceColletion = services;
services.AddLocalization(options => options.ResourcesPath = "Resources");
services.Configure<ApplicationConfiguration>(Configuration.GetSection("Application"));
services.AddMemoryCache();
services.AddDbContextPool<LoginContext>(options => {
options.UseSqlite(Configuration.GetConnectionString("LoginConnection"));
});
services.AddScoped<ILoginService, LoginService>();
var googleClientId = "";
var googleClientSecret = "";
if (CurrentEnvironment.IsDevelopment())
{
googleClientId = Configuration["GoogleClientId"];
googleClientSecret = Configuration["GoogleClientSecret"];
}
else
{
googleClientId = Configuration["Application:Authentication:GoogleClientId"];
googleClientSecret = Configuration["Application:Authentication:GoogleClientSecret"];
}
// enable authentication; state kept in cookies; using OpenIdConnect - with Google
services.AddAuthentication(options => {
options.DefaultScheme = CookieAuthenticationDefaults.AuthenticationScheme;
})
.AddCookie(options => {
options.Cookie.SecurePolicy = CurrentEnvironment.IsDevelopment() ? CookieSecurePolicy.SameAsRequest : CookieSecurePolicy.Always;
options.Cookie.Name = Configuration["Application:Authentication:CookieName"];
options.Cookie.MaxAge = TimeSpan.FromDays(double.Parse(Configuration["Application:Authentication:CookieExpiryDays"]));
options.SlidingExpiration = true;
options.ReturnUrlParameter = "";
options.AccessDeniedPath = "/error";
options.LoginPath = "/error";
options.LogoutPath = "/logout";
})
.AddOpenIdConnect(options => {
options.Authority = "https://accounts.google.com";
options.ClientId = googleClientId;
options.ClientSecret = googleClientSecret;
options.ResponseType = "code id_token";
options.GetClaimsFromUserInfoEndpoint = true;
options.Scope.Add("email");
options.SaveTokens = true;
options.Events = new OpenIdConnectEvents
{
OnRemoteFailure = OnAuthenticationFailed,
OnTokenValidated = PerformPostTokenValidationAuthorization,
OnRedirectToIdentityProviderForSignOut = OnRedirectToIdentityProviderForSignOut
};
});
services.AddErrorHandling(); // services for the ErrorHandling Middleware below
services.Configure<RazorViewEngineOptions>(options => {
options.ViewLocationExpanders.Add(new FeaturesViewLocationExpander(new[] {
"~/Features/{1}/{0}.cshtml",
"~/Features/Shared/{0}.cshtml"
}));
});
#if BLAZOR
services.AddMvc().AddJsonOptions(options => {
options.SerializerSettings.ContractResolver = new DefaultContractResolver();
}).SetCompatibilityVersion(CompatibilityVersion.Version_2_1)
.AddViewLocalization()
.AddDataAnnotationsLocalization();
services.AddResponseCompression(options => {
options.MimeTypes = ResponseCompressionDefaults.MimeTypes.Concat(new[]
{
MediaTypeNames.Application.Octet,
WasmMediaTypeNames.Application.Wasm,
});
});
#else
services.AddMvc()
.SetCompatibilityVersion(CompatibilityVersion.Version_2_1)
.AddViewLocalization()
.AddDataAnnotationsLocalization();
#endif
}
// This method gets called by the runtime. Use this method to configure the HTTP request pipeline.
public void Configure(IApplicationBuilder app, IHostingEnvironment env, ILoggerFactory loggerFactory,
IOptions<ApplicationConfiguration> appConfig, LoginContext context,
IApplicationLifetime appLifetime)
{
if (env.IsDevelopment())
{
app.UseErrorHandling();
}
else
{
app.UseHsts();
app.UseErrorHandling();
}
app.UseForwardedHeaders(new ForwardedHeadersOptions() {
ForwardedHeaders = ForwardedHeaders.XForwardedFor | ForwardedHeaders.XForwardedProto
});
app.UseSecurityHeaders(options => {
options.ApplicationBaseUrl = appConfig.Value.BaseUrl;
});
var cultures = new List<CultureInfo> { new CultureInfo("en-US"), new CultureInfo("en"), new CultureInfo("de-DE"), new CultureInfo("de") };
app.UseRequestLocalization(new RequestLocalizationOptions
{
DefaultRequestCulture = new RequestCulture("en-US"),
// Formatting numbers, dates, etc.
SupportedCultures = cultures,
// UI strings that we have localized.
SupportedUICultures = cultures
});
// the ForwardedHeaders do not work completely for me
// found similar issues here: https://github.com/aspnet/Docs/issues/2384
// this approach below works for me
app.Use((ctx, next) =>
{
if (ctx.Request.Headers.TryGetValue("X-Forwarded-Proto", out StringValues scheme))
{
logger.LogInformation($"Protocol/scheme forwarded by upstream-proxy: {scheme}");
ctx.Request.Scheme = scheme;
}
return next();
});
// enable authentication; state kept in cookies; using OpenIdConnect - with AAD
app.UseAuthentication();
app.UseJwtProcessor();
//app.UseHttpsRedirection();
app.UseStaticFiles();
app.UseMvc(routes =>
{
routes.MapRoute(
name: "default",
template: "{controller=Home}/{action=Index}/{id?}");
});
#if BLAZOR
app.UseBlazor<Blazor.Program>();
#endif
if (env.IsDevelopment())
{
ContextInitializer.InitialData(context);
}
}
}
}
| {
"redpajama_set_name": "RedPajamaGithub"
} | 7,652 |
Q: Windows 10 - Clearing Event Logs in CMD batch script with all output running on same line? I have a batch script which part of it is clearing the Event logs of Windows 10.
I would like to see the output of the following wevtutil.exe command, but output just on a single line (overwriting each line) instead of many multiple lines.
I know about the ANSI Escape Sequences ESC[1FESC[0J and ESC[1A which can overwrite a previous line (The ESC char is ALT+027 and can be seen in Notepad++), but I haven't been able to figure out how to do that with the wevtutil.exe el command. It still outputs each line one after the other.
Here's what I tried (running in CMD with admin rights):
@echo off
SET OverwriteLine=ESC[1FESC[0J
echo Making sure the variable OverwriteLine actually works. This is line 1.
echo %OverwriteLine%If you see this line instead of line 1, OverwriteLine works.
echo Great, now let^'s see if it works for the "wevtutil.exe cl" command
pause
echo Clearing Event logs...
@echo on
for /F "tokens=*" %%E in ('wevtutil.exe el') DO echo %OverwriteLine% | wevtutil.exe cl "%%E"
pause
I know this can be done via Powershell's $host.ui.RawUI.CursorPosition, but I need a solution for CMD/BAT.
A: As we deal with a single specific issue per question, and your main one appears to be with the implementation of the VT100 sequences, here is a commented example using a completely different for loop just for demonstration of the technique.
@Echo Off
SetLocal EnableExtensions
Rem Create a variable to use as the escape character
For /F %%G In ('Echo Prompt $E ^| %SystemRoot%\System32\cmd.exe') Do Set "\x1b=%%G"
Rem This is a static line included to demonstrate that it is unaffected
Echo(Output
Rem Turns off the cursor and removes a stray new line caused by Echo
Echo(%\x1b%[?25l%\x1b%[1A
Rem For loop example
For /L %%G In (1,1,8) Do (
Rem Clears to the beginning of the line, prints the output, and moves up a line
Echo(%\x1b%[1KLine %%G%\x1b%[1A
Rem Creates a short delay between output lines to see the effect
%SystemRoot%\System32\PATHPING.EXE 127.0.0.1 -n -q 1 -p 650 1>NUL
)
Rem Turns on the cursor again
Echo(%\x1b%[?25h
Pause
A: OK, found the solution:
for /F "tokens=*" %%E in ('wevtutil.exe el') DO (wevtutil.exe cl "%%E" | echo <ESC>[1F<ESC>[0KClearing %%E)
Explanation/Notes:
*
*< ESC> means the special ESC escape code sequence. You can generate this char by typing ALT+027 in Notepad++ if you're editing your code in there, or generate it at runtime using the FOR loop that Compo mentioned.
*We move the cursor to the beginning of the previous line with ESC[1F.
*We then clear from the cursor to the end of the line with ESC[0K.
*Clearing the Windows event logs requires running the CMD script with Administrator rights.
*Expect some event logs to fail. The failed ones will remain on the screen, each on a new line (which might become handy). If you don't want to see any failures, just add 2>nul : DO (wevtutil.exe cl "%%E" 2>nul | echo <ESC>[1F<ESC>[0KClearing %%E)
You can learn more about escape codes here.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 5,623 |
Le Cercle étudiant Endre-Ady (en hongrois : Ady Endre Diákkör, ) est une organisation étudiante tchèque créée en 1957. Elle rassemble les étudiants pragois de la minorité magyare de Slovaquie. Elle est membre du Réseau étudiant rassemblant les organisations étudiantes de la diaspora magyare de Slovaquie. Elle porte le nom du poète hongrois Endre Ady.
Voir aussi
Minorité magyare de Slovaquie
Mouvement de jeunesse en Hongrie
Enseignement à Prague
Prague 1 | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 3,933 |
\section{Introduction and preliminaries}
Let $p$ be a fixed prime number. Throughout this paper, the symbol
$\Bbb Z$, $\Bbb Z_p$, $\Bbb Q_p$, and $\Bbb C_p$ denote the ring of
rational integers, the ring of $p$-adic integers, the field of
$p$-adic rational numbers, and the completion of algebraic closure
of $\Bbb Q_p$, respectively. Let $\mathbb{N}$ be the set of natural
numbers and $\Bbb Z_+ = \Bbb N \cup \{ 0 \}$ . Let $\nu_p $ be the
normalized exponential valuation of $\Bbb C_p$ with $|p|_p=
p^{-\nu_{p}(p)}=p^{-1}$.
Let $UD(\Bbb Z_p)$ be the space of uniformly differentiable function
on $\Bbb Z_p$. For $f \in UD(\Bbb Z_p)$, the $p$-adic invariant
integral on $\Bbb Z_p$ is defined as
\begin{eqnarray} I(f)=\int_{\Bbb Z_p
} f(x) dx= \lim_{N \rightarrow \infty} \frac{1}{p^N} \sum_{x=0}^{p^N
-1} f(x).
\end{eqnarray}
(see [4-5]). From $(1.1)$, we note that
\begin{eqnarray} I(f_1)=I(f)+f'(0),
\end{eqnarray} where $ f'(0)=\frac{df(x)}{dx} |_{x=0}$ and $f_1(x)=f(x+1)$.
For $n \in \mathbb{N}$, let $f_n(x)=f(x+n)$. Then we can derive the
following equation from (1.2).
\begin{eqnarray} I(f_n)=I(f)+ \sum_{i=0}^{n-1}f'(i), \quad (\text{see [4-5]}). \label{1.3}
\end{eqnarray}
Let $d$ be a fixed positive integer. For $n \in \mathbb{N}$, let
\begin{eqnarray*}
X&=& X_d = \lim_{\overleftarrow{N} } \Bbb Z/ dp^N \Bbb Z ,\ X_1 =
\Bbb Z_p , \\ X^\ast &=&\underset{{0<a<d p}\atop {(a,p)=1}} {\cup}
(a+ dp \,\Bbb Z_p ), \\ a+d p^N \Bbb Z_p &=& \{ x\in X | \, x
\equiv a \pmod{dp^N}\},
\end{eqnarray*}
where $a\in \Bbb Z$ lies in $0\leq a < d p^N$. It is easy to see
that
\begin{eqnarray}\int_X f(x)dx = \int_{\Bbb Z_p} f(x) dx, \quad
\text{for} \quad f \in UD(\Bbb Z_p). \label{1.4} \end{eqnarray}
The ordinary Bernoulli polynomials $B_n (x)$ are defined as
\begin{eqnarray*} \frac{t}{ e^t -1 }e^{xt} =\sum_{n=0}^{\infty} B_n (x)
\frac{t^n}{n!}, \end{eqnarray*} and the Bernoulli numbers $B_n$ are
defined as $B_n=B_n(0)$ (see [1-19]).
For $n \in \Bbb N$, let $T_p$ be the p-adic locally constant space
defined by
$$T_p =\underset{n \ge 1} {\cup} \Bbb C_{p^n}= \underset{n \to
\infty} {\lim} \Bbb C_{p^n},$$ where $\Bbb C_{p^n}= \{ \omega |
\omega^{p^n}=1 \}$ is the cyclic group of order $p^n$. It is well
known that the twisted Bernoulli polynomials are defined as
\begin{eqnarray*} \frac{t}{\xi e^t -1 }e^{xt} =\sum_{n=0}^{\infty} B_{n, \xi}(x)
\frac{t^n}{n!}, \quad \xi \in T_p ,
\end{eqnarray*}
and the twisted Bernoulli numbers $B_{n, \xi}$ are defined as $B_{n,
\xi}=B_{n, \xi}(0)$ (see [14-18]).
Let $\chi$ be the Dirichlet's character with conductor $d \in
\mathbb{N}$. Then we have
\begin{eqnarray}
\int_{X}\chi(x) \xi^x e^{xt}dx=
\frac{t\underset{a=0}{\overset{d-1}{\sum}}\chi(a)\xi^a e^{at}}{\xi^d
e^{dt}-1} . \label{1.5}
\end{eqnarray}
It is known that the generalized twisted Bernoulli numbers attached
to $\chi$, $B_{n,\chi, \xi}$, are defined as
\begin{eqnarray}
\frac{t\underset{a=0}{\overset{d-1}{\sum}}\chi(a)\,\xi^a e^{at}
}{\xi^d e^{dt} -1}= \underset{n=0}{\overset{\infty}{\sum}}
B_{n,\chi, \xi} \frac{t^n}{n!}, \quad \xi \in T_p. \label{1.6}
\end{eqnarray}
The generalized twisted Bernoulli polynomials attached to $\chi$,
$B_{n,\chi, \xi} (x)$, are defined as
\begin{eqnarray}
\frac{t\underset{a=0}{\overset{d-1}{\sum}}\chi(a)\,\xi^a e^{at}
}{\xi^d e^{dt} -1}e^{xt}= \underset{n=0}{\overset{\infty}{\sum}}
B_{n,\chi, \xi} (x)\frac{t^n}{n!}, \quad \xi \in T_p, \label{1.7}
\end{eqnarray}
(see [13], [16]). From ($\ref{1.5}$), ($\ref{1.6}$) and
($\ref{1.7}$), we derive that
\begin{eqnarray}
\int_{X}\chi(x) \xi^x x^ndx= B_{n,\chi, \xi} \quad \text{and} \quad
\int_{X}\chi(y) \xi^y (x+y)^n dy= B_{n,\chi, \xi}(x).
\label{1.8}\end{eqnarray}
By ($\ref{1.3}$) and ($\ref{1.4}$), it is easy to see that for $n
\in \mathbb{N}$,
\begin{eqnarray}
\int_{X}f(x+n)dx=\int_{X}f(x)dx+\sum_{i=0}^{n-1}f\,'(i), \label{1.9}
\end{eqnarray}
where $f\,'(i)=\frac{df(x)}{dx}|_{x=i}$. From ($\ref{1.9}$), it
follows that
\begin{eqnarray}
&&\frac{1}{t}(\int_{X}\chi(x)\,\xi^{nd+x}e^{(nd+x)t}dx-\int_{X}\chi(x)\,\xi^{x}e^{xt}dx)
\label{1.10}\\&& = \frac{nd\int_{X}\chi(x)\,\xi^x
e^{xt}dx}{\int_{X}\xi^{ndx}e^{ndxt}
dx}=\frac{\xi^{nd}e^{ndt}-1}{\xi^d
e^{dt}-1}(\sum_{i=0}^{d-1}\chi(i)\,\xi^i e^{it})
=\sum_{k=0}^{\infty}(\sum_{l=0}^{nd-1}\chi(l)\, \xi^l l^{k})
\frac{t^k}{k!}. \quad \notag
\end{eqnarray}
For $k \in \Bbb Z_+$, let us define the $p$-adic functional $T_{k,
\chi, \xi}(n)$ as follows:
\begin{eqnarray}
T_{k, \chi, \xi}(n)=\sum_{l=0}^{n}\chi(l)\xi^l l^{k}. \label{1.11}
\end{eqnarray}
Let $k,n,d \in \mathbb{N}$. By $(\ref{1.10})$ and $(\ref{1.11})$, we
see that
\begin{eqnarray}
\int_{X} \chi(x)\xi^{nd+x}(nd+x)^k dx-\int_{X}\chi(x) \xi^x x^k dx=k
\,T_{k-1, \chi, \xi}(nd-1).\label{1.12}
\end{eqnarray}
From $(\ref{1.8})$ and $(\ref{1.12})$, we have that
\begin{eqnarray}
\frac{\xi^{nd} B_{k,\chi, \xi}(nd)-B_{k,\chi, \xi}}{k}=T_{k-1, \chi,
\xi}(nd-1). \label{1.13}
\end{eqnarray}
For $w_1, w_2, d \in \mathbb{N}$, we note that
\begin{eqnarray}
& &\frac{d\int_{X}\int_{X}\chi(x_1)\chi(x_2)\,\xi^{w_1 x_1+w_2 x_2}
e^{(w_1 x_1+w_2 x_2)t}
dx_1
dx_2 }{\int_{X} \xi^{dw_1 w_2 x }e^{dw_1 w_2 xt} x}\label{1.14}\\
& &\quad =\frac{t(\xi^{dw_1 w_2}e^{dw_1 w_2 t} -1)}{(\xi^{w_1
d}e^{w_1d t}-1)(\xi^{w_2 d}e^{w_2 d t}-1)}
(\sum_{a=0}^{d-1}\chi(a)\xi^{w_1 a}e^{w_1 at}
)(\sum_{b=0}^{d-1}\chi(b)\xi^{w_2 b}e^{w_2 bt} ). \quad \notag
\end{eqnarray}
In the next section, we will consider the extension of
$(\ref{1.14})$ related to the generalized twisted Bernoulli numbers
and polynomials of higher order attached to $\chi$ .
The generalized twisted Bernoulli polynomials of order $k$ attached
to $\chi$, $B_{n,\chi, \xi}^{(k)}(x)$, are defined as
\begin{eqnarray}
\left( \frac{t\underset{a=0}{\overset{d-1}{\sum}}\chi(a)\,\xi^a
e^{at} }{\xi^d e^{dt} -1} \right)^k e^{xt}=
\underset{n=0}{\overset{\infty}{\sum}} B_{n,\chi, \xi}^{(k)}
(x)\frac{t^n}{n!}, \quad \xi \in T_p, \label{1.15}
\end{eqnarray}
and $B_{n,\chi, \xi}^{(k)}=B_{n,\chi, \xi}^{(k)}(0)$ are called the
generalized twisted Bernoulli numbers of order $k$ attached to
$\chi$. When $k=1$, the polynomials and numbers are called the
generalized twisted Bernoulli polynomials and numbers attached to
$\chi$, respectively (see [12]).
The authors of this paper have studied various identities for the
Bernoulli and the Euler polynomials by the symmetric properties of
the $p$-adic invariant integrals (see [6-8], [10]). T. Kim [6]
established interesting identities by the symmetric properties of
the $p$-adic invariant integrals and some relationships between the
power sums and the Bernoulli polynomials. In [8], Kim et al. gave
some identities of symmetry for the generalized Bernoulli
polynomials. The twisted Bernoulli polynomials and numbers are very
important in several field of mathematics and physics, and so have
been studied by many authors (cf. [9-18]). Recently, Kim-Hwang [10]
obtained some relations between the power sum polynomials and
twisted Bernoulli polynomials.
In this paper, we extend our results to the generalized twisted
Bernoulli numbers and polynomials of higher order attached to
$\chi$. The purpose of this paper is to derive some identities of
the higher order generalized twisted Bernoulli numbers and
polynomials attached to $\chi$ from the properties of the $p$-adic
invariant integral. In Section 2, we give interesting identities for
the power sums and the generalized twisted Bernoulli numbers and
polynomials of higher order using the symmetric properties for the
$p$-adic invariant integral.
\medskip
\section{Some identities of the generalized twisted Bernoulli numbers and polynomials of higher order}
Let $w_1, w_2, d \in \mathbb{N}$. For $\xi \in T_p$, we set
\begin{eqnarray}
& &Y(m, \chi, \xi | w_1, w_2) \notag \\
&&\quad=\left(\frac{d\int_{X^m}(\underset{i=1}{\overset{m}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m}{\sum}}
x_i)w_1 } e^{(\underset{i=1}{\overset{m}{\sum}}x_i+w_2x)w_1t}dx_1
\cdots dx_m} {\int_{X}\xi^{dw_1 w_2 x}e^{dw_1w_2xt}dx}\right)\label{2.1}\\
& & \qquad\times
\left(\int_{X^m}(\underset{i=1}{\overset{m}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m}{\sum}}
x_i)w_2 }e^{(\underset{i=1}{\overset{m}{\sum}}x_i+w_1y)w_2t} dx_1
\cdots dx_m \right), \notag
\end{eqnarray}
where
\begin{eqnarray*}
\int_{X^m}f(x_1, \cdots, x_m)dx_1 \cdots dx_m= \underbrace{\int_{X}
\cdots \int_{X}}_{m-\text{times}}f(x_1, \cdots, x_m)dx_1 \cdots
dx_m.
\end{eqnarray*}
In $(\ref{2.1})$, we note that $Y(m, \chi, \xi; w_1, w_2)$ is
symmetric in $w_1, w_2$. From $(\ref{2.1})$, we derive that
\begin{eqnarray}
& &Y(m, \chi, \xi | w_1, w_2) \notag \\
&&=\left(\int_{X^m}(\underset{i=1}{\overset{m}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m}{\sum}}
x_i)w_1 }e^{(\underset{i=1}{\overset{m}{\sum}}x_i)w_1t}dx_1 \cdots
dx_m\right)e^{w_1w_2xt}\qquad \label{2.2}\\
& & \quad \times
\left(\frac{d\int_{X}\chi(x_m)\xi^{w_2 x_m}e^{w_2x_{m}t}dx_m}{\int_{X}\xi^{dw_1 w_2
x}e^{dw_1w_2xt}dx}\right) \notag\\
& & \quad \times
\left(\int_{X^{m-1}}(\underset{i=1}{\overset{m-1}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m-1}{\sum}}
x_i)w_2 }e^{(\underset{i=1}{\overset{m-1}{\sum}}x_i)w_2t}dx_1
\cdots dx_{m-1}\right)e^{w_1w_2yt}. \notag
\end{eqnarray}
From $(\ref{1.10})$ and $(\ref{1.11})$, it follows that
\begin{eqnarray}
\frac{dw_1\int_{X}\chi(x) \xi^x e^{xt}dx}{\int_{X}
\xi^{dw_1x}e^{dw_1xt}dx}=\sum_{i=0}^{w_1d-1}\chi(i)\xi^i e^{it}
=\sum_{k=0}^{\infty}T_{k, \chi, \xi}(w_1d-1)\frac{t^k}{k!}.
\label{2.3}
\end{eqnarray}
By $(\ref{1.15})$, we also see that
\begin{eqnarray}
&&e^{w_1w_2xt}\left(\int_{X^m}(\underset{i=1}{\overset{m}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m}{\sum}}
x_i)w_1 }e^{(\underset{i=1}{\overset{m}{\sum}}x_i)w_1t}dx_1
\cdots dx_m \right)\label{2.4}\\ & & \quad
= \left(\frac{w_1t}{\xi^{dw_1}e^{dw_1t}-1}\sum_{a=0}^{d-1}\chi(a)\xi^{w_1 a}e^{aw_1t}\right)^m e^{w_1w_2xt}=
\underset{n=0}{\overset{\infty}{\sum}} B_{n,\chi, \xi^{w_1}}^{(m)} \notag
(w_{2}x)\frac{w_1^nt^n}{n!}.
\end{eqnarray}
By $(\ref{2.2})$, $(\ref{2.3})$ and $(\ref{2.4})$, we have that
\begin{eqnarray}
& &Y(m, \chi, \xi | w_1, w_2) \label{2.5} \\
& &=\left(\underset{l=0}{\overset{\infty}{\sum}} B_{l,\chi,
\xi^{w_1}}^{(m)}
(w_{2}x)\frac{w_1^lt^l}{l!}\right)\left(\frac{1}{w_1}\underset{k=0}{\overset{\infty}{\sum}}T_{k,
\chi,
\xi^{w_2}}(w_1d-1)\frac{w_2^kt^k}{k!}\right)\left(\underset{i=0}{\overset{\infty}{\sum}}
B_{i,\chi,
\xi^{w_2}}^{(m-1)}(w_{1}y)\frac{w_2^it^i}{i!}\right)\notag
\\
&&=\underset{n=0}{\overset{\infty}{\sum}}\
\left(\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_2^{j}
w_1^{n-j-1}B_{n-j, \, \chi,
\xi^{w_1}}^{(m)}(w_2x)\underset{k=0}{\overset{j}{\sum}}\binom{j}{k}T_{k,
\chi, \xi^{w_2}}(w_1d-1)B_{j-k,\chi,
\xi^{w_2}}^{(m-1)}(w_{1}y)\right)\frac{t^n}{n!}.\notag
\end{eqnarray}
From the symmetry of $Y(m, \chi, \xi | w_1, w_2)$ in $w_1$ and
$w_2$, we see that
\begin{eqnarray}
& &Y(m, \chi, \xi | w_1, w_2) \label{2.6}\\
&&=\underset{n=0}{\overset{\infty}{\sum}}\left(\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_1^{j}
w_2^{n-j-1}B_{n-j, \chi,
\xi^{w_2}}^{(m)}(w_1x)\underset{k=0}{\overset{j}{\sum}}\binom{j}{k}T_{k,
\chi, \xi^{w_1} }(w_2d-1)B_{j-k, \chi,
\xi^{w_2}}^{(m-1)}(w_2y)\right)\frac{t^n}{n!}.\notag
\end{eqnarray}
Comparing the coefficients on the both sides of $(\ref{2.5})$ and
$(\ref{2.6})$, we obtain an identity for the generalized twisted
Bernoulli polynomials of higher order as follows.
\begin{theorem}
Let $d, w_1, w_2 \in \mathbb{N}$. For $n \in \mathbb{Z}_+$ and $m
\in \mathbb{N}$, we have
\begin{eqnarray*}
& &\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_2^{j}
w_1^{n-j-1}B_{n-j, \, \chi,
\xi^{w_1}}^{(m)}(w_2x)\underset{k=0}{\overset{j}{\sum}}\binom{j}{k}T_{k,
\chi, \xi^{w_2}}(w_1d-1)B_{j-k,\chi, \xi^{w_2}}^{(m-1)}(w_{1}y)
\\
& &=\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_1^{j}
w_2^{n-j-1}B_{n-j, \chi,
\xi^{w_2}}^{(m)}(w_1x)\underset{k=0}{\overset{j}{\sum}}\binom{j}{k}T_{k,
\chi, \xi^{w_1} }(w_2d-1)B_{j-k, \chi,
\xi^{w_1}}^{(m-1)}(w_2y).\notag
\end{eqnarray*}
\end{theorem}
\begin{remark}
Taking $m=1$ and $y=0$ in $(2.7)$ derives the following identity :
\begin{eqnarray}
& &\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_2^{j}
w_1^{n-j-1}B_{n-j, \chi, \xi^{w_1}}(w_2x)T_{j,
\chi, \xi^{w_2}}(w_1d-1)\\
& &\quad =\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_1^{j}
w_2^{n-j-1}B_{n-j, \chi, \xi^{w_2}}(w_1x)T_{j, \chi, \xi^{w_1}
}(w_2d-1).\notag
\end{eqnarray}
\end{remark}
Moreover, if we take $x=0$ and $y=0$ in Theorem 1, then we have the
following identity for the generalized twisted Bernoulli numbers of
higher order.
\begin{corollary}
Let $d, w_1, w_2 \in \mathbb{N}$. For $n \in \mathbb{Z}_+$ and $m
\in \mathbb{N}$, we have
\begin{eqnarray*}
& &\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_2^{j}
w_1^{n-j-1}B_{n-j, \, \chi,
\xi^{w_1}}^{(m)}\underset{k=0}{\overset{j}{\sum}}\binom{j}{k}T_{k,
\chi, \xi^{w_2}}(w_1d-1)B_{j-k,\chi, \xi^{w_2}}^{(m-1)}
\\
& &=\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_1^{j}
w_2^{n-j-1}B_{n-j, \chi,
\xi^{w_2}}^{(m)}\underset{k=0}{\overset{j}{\sum}}\binom{j}{k}T_{k,
\chi, \xi^{w_1} }(w_2d-1)B_{j-k, \chi, \xi^{w_1}}^{(m-1)}.\notag
\end{eqnarray*}
\end{corollary}
We also note that taking $m=1$ in Corollary 2 shows the following
identity :
\begin{eqnarray}
& &\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_2^{j}
w_1^{n-j-1}B_{n-j, \chi, \xi^{w_1}}T_{j,
\chi, \xi^{w_2}}(w_1d-1)\\
& &\quad =\underset{j=0}{\overset{n}{\sum}}\binom{n}{j}w_1^{j}
w_2^{n-j-1}B_{n-j, \chi, \xi^{w_2}}T_{j, \chi, \xi^{w_1}
}(w_2d-1).\notag
\end{eqnarray}
Now we will derive another interesting identities for the
generalized twisted Bernoulli numbers and polynomials of higher
order. From $(\ref{1.15})$, $(\ref{2.2})$ and $(\ref{2.3})$, we can
derive that
\begin{eqnarray}
& &Y(m, \chi, \xi | w_1, w_2) \notag \\
&&=\frac{1}{w_1} \left(\sum_{i=0}^{w_1
d-1}\chi(i)\,\xi^{w_2i}\int_{X^m}(\underset{i=1}{\overset{m}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m}{\sum}}
x_i)w_1
}e^{(\underset{i=1}{\overset{m}{\sum}}x_i+\frac{w_2}{w_1}i+w_2
x)w_1t}dx_1 \cdots
dx_m\right) \label{2.9}\\
& & \quad \times
\left(\int_{X^{m-1}}(\underset{i=1}{\overset{m-1}{\prod}}\chi(x_i))\xi^{(\underset{i=1}{\overset{m-1}{\sum}}
x_i)w_2 }e^{(\underset{i=1}{\overset{m-1}{\sum}}x_i+w_1 y)w_2t}dx_1
\cdots dx_{m-1}\right) \notag\\
&&=\underset{n=0}{\overset{\infty}{\sum}}\left(\underset{k=0}{\overset{n}{\sum}}\binom{n}{k}w_1^{k-1}
w_2^{n-k}B_{n-k, \chi,
\xi^{w_2}}^{(m-1)}(w_{1}y)\underset{i=0}{\overset{w_1d
-1}{\sum}}\chi(i)\xi^{w_2 i}B_{k, \chi,
\xi^{w_1}}^{(m)}(w_{2}x+\frac{w_2}{w_1}i)\right)\frac{t^n}{n!}.\quad
\notag
\end{eqnarray}
From the symmetry property of $Y(m, \chi, \xi | w_1, w_2)$ in $w_1$
and $w_2$, we see that
\begin{eqnarray}
& &Y(m, \chi, \xi | w_1, w_2) \label{2.10}\\\
&&=\underset{n=0}{\overset{\infty}{\sum}}\left(\underset{k=0}{\overset{n}{\sum}}\binom{n}{k}w_2^{k-1}
w_1^{n-k}B_{n-k,\chi, \xi^{w_1
}}^{(m-1)}(w_{2}y)\underset{i=0}{\overset{w_2d
-1}{\sum}}\chi(i)\xi^{w_1 i}B_{k, \chi, \xi^{w_2
}}^{(m)}(w_{1}x+\frac{w_1}{w_2}i)\right)\frac{t^n}{n!}. \quad \notag
\end{eqnarray}
Comparing the coefficients on the both sides of $(\ref{2.9})$ and
$(\ref{2.10})$, we obtain the following theorem which shows the
relationship between the power sums and the generalized twisted
Bernoulli polynomials.
\begin{theorem}
Let $d, w_1, w_2 \in \mathbb{N}$. For $n \in \mathbb{Z}_+$ and $m
\in \mathbb{N}$, we have
\begin{eqnarray*}
& &\underset{k=0}{\overset{n}{\sum}}\binom{n}{k}w_1^{k-1}
w_2^{n-k}B_{n-k,\chi,
\xi^{w_2}}^{(m-1)}(w_{1}y)\underset{i=0}{\overset{w_1d
-1}{\sum}}\chi(i)\xi^{w_2 i}B_{k,\chi, \xi^{w_1}}^{(m)}(w_{2}x+\frac{w_2}{w_1}i) \\
& &=\underset{k=0}{\overset{n}{\sum}}\binom{n}{k}w_2^{k-1}
w_1^{n-k}B_{n-k, \chi,
\xi^{w_1}}^{(m-1)}(w_{2}y)\underset{i=0}{\overset{w_2d
-1}{\sum}}\chi(i)\xi^{w_1 i}B_{k, \chi,
\xi^{w_2}}^{(m)}(w_{1}x+\frac{w_1}{w_2}i).\notag
\end{eqnarray*}
\end{theorem}
\begin{remark}
Let $m=1$ and $y=0$ in Theorem 3. Then it follows that
\begin{eqnarray}
w_1^{n-1} \underset{i=0}{\overset{w_1d -1}{\sum}}\chi(i)B_{n,\chi,
\xi^{w_1}}(w_{2}x+\frac{w_2}{w_1}i) =w_2^{n-1}
\underset{i=0}{\overset{w_2d -1}{\sum}}\chi(i)B_{n,\chi,
\xi^{w_2}}(w_{1}x+\frac{w_1}{w_2}i).\label{2.11}
\end{eqnarray}
\end{remark}
Moreover, if we take $x=0$ and $y=0$ in Theorem 3, then we have the
following identity for the generalized twisted Bernoulli numbers of
higher order.
\begin{corollary}
Let $d, w_1, w_2 \in \mathbb{N}$. For $n \in \mathbb{Z}_+$ and $m
\in \mathbb{N}$, we have
\begin{eqnarray*}
& &\underset{k=0}{\overset{n}{\sum}}\binom{n}{k}w_1^{k-1}
w_2^{n-k}B_{n-k,\chi, \xi^{w_2}}^{(m-1)}\underset{i=0}{\overset{dw_1
-1}{\sum}}\chi(i)\xi^{w_2 i}B_{k,\chi, \xi^{w_1}}^{(m)}(\frac{w_2}{w_1}i) \\
& &=\underset{k=0}{\overset{n}{\sum}}\binom{n}{k}w_2^{k-1}
w_1^{n-k}B_{n-k, \chi,
\xi^{w_1}}^{(m-1)}\underset{i=0}{\overset{dw_2
-1}{\sum}}\chi(i)\xi^{w_1 i}B_{k, \chi,
\xi^{w_2}}^{(m)}(\frac{w_1}{w_2}i).\notag
\end{eqnarray*}
\end{corollary}
If we take $m=1$ in Corollary 4, we derive the identity for the
generalized twisted Bernoulli numbers : for $d, w_1, w_2 \in
\mathbb{N}$ and $n \in \mathbb{Z}_+$,
\begin{eqnarray}
w_1^{n-1}\sum_{i=0}^{dw_1-1}\chi(i)\xi^{w_2 i} B_{n, \chi,
\xi^{w_1}}( \frac{w_2}{w_1}i)
=w_2^{n-1}\sum_{i=0}^{dw_2-1}\chi(i)\xi^{w_1 i} B_{n, \chi,
\xi^{w_2}}(\frac{w_1}{w_2}i). \label{2.12}
\end{eqnarray}
\noindent \textbf{Acknowledgement.} This research was supported by
Kyungpook National University research fund 2008.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 5,321 |
Charming 3BR 2BA Colonial with chic, beautifully redone 2 bedroom upper level In-Law in Stamford's Cove area. Located in 2-family zone (R5), ripe with rental income producing potential (currently over $3k per month) and with separate entrances for both levels. MAIN LEVEL offers 2 BR, 1 bath, living room, decent size deck, H/W floors throughout, eat-in kitchen w/ granite wraparound countertop, washer/dryer and ample basement storage. UPPER LEVEL features 2 BR, 1 bath, eat-in kitchen w/ breakfast bar, in-unit washer/dryer AND circular staircase leading up to cozy LOFT - perfect for relaxing. Newer roof & gutters (spring 2010), upgraded electrical (2011), recent furnace/boiler install (2016) and SIZABLE backyard (perfect "mechanics" yard) with MULTIPLE garages potential makes this an AWESOME, worry-free income producing and/or live-rent opportunity. Being sold "AS-IS". Up your game, bring ANY offers and discover the potential!! | {
"redpajama_set_name": "RedPajamaC4"
} | 3,160 |
\section{\@startsection {section}{1}{\z@}%
{-3.5ex \@plus -1ex \@minus -.2ex}%
{2.3ex \@plus.2ex}%
{\normalfont\large\bf}}
\makeatother
\makeatletter
\@addtoreset{footnote}{page}
\makeatother
\theoremstyle{plain}
\newtheorem{Que}{Question}
\newcommand{\ensuremath{\mathfrak{e}}}{\ensuremath{\mathfrak{e}}}
\renewcommand{\dagger}{h}
\newcounter{CO}
\newcommand{\Confirst}[1]{\refstepcounter{CO} \label{con: #1}}
\newcommand{\Const}[1]{\ensuremath{C_{\ref{con: #1}}}}
\newcounter{EO}
\newcommand{\Exfirst}[1]{\refstepcounter{EO} \label{ex: #1} \Example{#1}}
\newcommand{\Example}[1]{{\bf (E{\ref{ex: #1}})}}
\newcommand{\Leqno}[2]{
$$
#2
\leqno{\bf (#1)}
$$}
\renewcommand{\thefootnote}{(\arabic{footnote})}
\usepackage{graphicx}
\newcommand{\mail}{
\scalebox{0.6}{\includegraphics{yano-e-mail.eps}}
}
\begin{document}
\begin{center}
{\Large \bf
Penalising symmetric stable L\'evy paths
}
\end{center}
\begin{center}
Kouji \textsc{Yano}\footnote{
Department of Mathematics, Graduate School of Science, Kobe University, Kobe, Japan.\\
E-mail: \mail}\footnote{
The research of this author is supported by KAKENHI (20740060)}, \qquad
Yuko \textsc{Yano}\footnote{
Research Institute for Mathematical Sciences, Kyoto University, Kyoto, Japan.\label{foot: RIMS}}
\qquad and \qquad
Marc \textsc{Yor}\footnote{
Laboratoire de Probabilit\'es et Mod\`eles Al\'eatoires, Universit\'e Paris VI, Paris, France.}\footnote{
Institut Universitaire de France}\footnotemark[3]
\end{center}
\begin{center}
{\small \today}
\end{center}
\bigskip
\begin{abstract}
Limit theorems for the normalized laws with respect to two kinds of weight functionals
are studied for any symmetric stable L\'evy process of index $ 1 < \alpha \le 2 $.
The first kind is a function of the local time at the origin,
and the second kind is the exponential of an occupation time integral.
Special emphasis is put on the role played by a stable L\'evy counterpart
of the universal $ \sigma $-finite measure,
found in \cite{NRY} and \cite{NRY2},
which unifies the corresponding limit theorems in the Brownian setup
for which $ \alpha =2 $.
\end{abstract}
\section{Introduction}
Roynette, Vallois and Yor (\cite{MR2261065},
\cite{MR2229621}
and \cite{MR2253307}
and references therein)
have shown the existence of the limit laws for normalized Wiener measures
with respect to various weight processes;
we call these studies {\em penalisation problems}.
Najnudel, Roynette and Yor (see \cite{RY}, \cite{MR2339327}, \cite{NRY} and \cite{NRY2})
have recently discovered that
these penalisation problems may be unified
with the help of the following ``universal" $ \sigma $-finite measure on the canonical space:
\begin{align}
\ensuremath{\mathscr{W}} = \int_0^{\infty } \frac{\d u}{\sqrt{2 \pi u}} W^{(u)} \bullet P^{\rm 3B}_0
\label{eq: curly W}
\end{align}
where $ W^{(u)} $ stands for the law of the brownian bridge from $ 0 $ to $ 0 $ of length $ u $,
$ P^{\rm 3B}_0 $ for that of the symmetrized 3-dimensional Bessel process starting from 0,
i.e., $ P^{\rm 3B}_0 = (P^{\rm 3B,+}_0 + P^{\rm 3B,-}_0)/2 $,
and the symbol $ \bullet $ for the concatenation between the laws of these two processes.
The purpose of the present paper
is to develop some of these penalisation problems
in the case of any symmetric stable L\'evy process of index $ 1<\alpha \le 2 $.
As an analogue of $ \ensuremath{\mathscr{W}} $,
we introduce the following $ \sigma $-finite measure
\begin{align}
\ensuremath{\mathscr{P}} = \frac{\Gamma(1/\alpha )}{\alpha \pi}
\int_0^{\infty } \frac{\d u}{u^{1/\alpha }} Q^{(u)} \bullet P^{\dagger}_0
\label{eq: sP intro}
\end{align}
where
$ Q^{(u)} $ stands for the law of the bridge from $ 0 $ to $ 0 $ of length $ u $
and
$ P^{\dagger}_0 $ for the $ h $-path process of the killed process
with respect to the function $ |x|^{\alpha -1} $.
We shall put some special emphasis on the role played
by the universal $ \sigma $-finite measure $ \ensuremath{\mathscr{P}} $
which helps to unify our penalisation problems.
Let $ \ensuremath{\mathbb{D}} $ denote the canonical space of c\`adl\`ag paths $ w:[0,\infty ) \to \ensuremath{\mathbb{R}} $.
Let $ (X_t) $ denote the coordinate process,
$ (\ensuremath{\mathcal{F}}_t) $ its natural filtration,
and $ \ensuremath{\mathcal{F}}_{\infty } = \vee_{t \ge 0} \ensuremath{\mathcal{F}}_t $.
Let $ (P_x) $ denote the law on $ \ensuremath{\mathbb{D}} $
of the symmetric stable process of index $ 1<\alpha \le 2 $
such that $ P_0[{\rm e}^{i \lambda X_t}] = {\rm e}^{- t |\lambda|^{\alpha }} $ for $ \lambda \in \ensuremath{\mathbb{R}} $.
Note that, if $ \alpha =2 $, then $ (X_t) $ has the same law
as $ \sqrt{2} $ times the standard brownian motion.
We say that {\em a family of measures $ \{ \ensuremath{\mathscr{M}}_t \}_{t \ge 0} $ on $ \ensuremath{\mathcal{F}}_{\infty } $
converges as $ t \to \infty $ to a measure $ \ensuremath{\mathscr{M}} $ along $ (\ensuremath{\mathcal{F}}_s) $}
if, for each $ s>0 $, we have
$ \ensuremath{\mathscr{M}}_t[Z_s] \to \ensuremath{\mathscr{M}}[Z_s] $ as $ t \to \infty $
for all bounded $ \ensuremath{\mathcal{F}}_s $-measurable functionals $ Z_s $.
For a measure $ \ensuremath{\mathscr{M}} $ on $ \ensuremath{\mathcal{F}}_{\infty } $ and a functional $ F $
measurable with respect to $ \ensuremath{\mathcal{F}}_{\infty } $,
the symbol $ F \cdot \ensuremath{\mathscr{M}} $ stands for the measure $ A \mapsto \ensuremath{\mathscr{M}}[1_A F] $.
Let $ x \in \ensuremath{\mathbb{R}} $ be fixed.
Then penalisation problems are stated as follows:
\begin{quote}
{\bf Question 1.}
Let $ \Gamma = (\Gamma_t:t \ge 0) $ be a given non-negative process
such that $ P_x[\Gamma_t] \neq 0 $ for large enough $ t $.
\\
{\bf (Q1)}
Does there exist a limit probability measure $ P^{\Gamma}_x $
such that
\begin{align}
\frac{\Gamma_t \cdot P_x}{P_x[\Gamma_t]}
\ \stackrel{t \to \infty }{\longrightarrow} \
P^{\Gamma}_x
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $?}
\label{eq: Q1}
\end{align}
{\bf (Q2)}
How can one characterise the limit probability measure $ P^{\Gamma}_x $ assuming it exists?
\end{quote}
For each $ x \in \ensuremath{\mathbb{R}} $, let $ \ensuremath{\mathscr{P}}_x $ denote the law of $ (x+X_t:t \ge 0) $ under $ \ensuremath{\mathscr{P}} $.
We can gain a clear insight into some of these penalisation problems
if we answer the following
\begin{quote}
{\bf Question 2.}
Let $ \Gamma $ as above.
\\
{\bf (Q1$ ' $)}
Can one find a positive function $ \mu(t) $
and a measurable functional $ \Gamma_{\infty } $
such that
\begin{align}
\frac{\Gamma_t \cdot P_x}{\mu(t)}
\ \stackrel{t \to \infty }{\longrightarrow} \
\Gamma_{\infty } \cdot \ensuremath{\mathscr{P}}_x
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $?}
\label{eq: Q1'}
\end{align}
{\bf (Q2$ ' $)}
For any non-negative $ \ensuremath{\mathscr{P}}_x $-integrable functional $ F $,
can one find a non-negative $ (\ensuremath{\mathcal{F}}_t,P_x) $-martingale $ (M_{t,x}(F):t \ge 0) $ such that
\begin{align}
(F \cdot \ensuremath{\mathscr{P}}_x) |_{\ensuremath{\mathcal{F}}_t} = M_{t,x}(F) \cdot P_x|_{\ensuremath{\mathcal{F}}_t}
, \qquad t \ge 0 ?
\label{eq: Q2'}
\end{align}
\end{quote}
If we can find such a function $ \mu(t) $ as in \eqref{eq: Q1'}
and if $ 0 < \ensuremath{\mathscr{P}}_x[\Gamma_{\infty }] < \infty $,
then we obtain the convergence \eqref{eq: Q1} with the limit probability measure
\begin{align}
P^{\Gamma}_x = \frac{\Gamma_{\infty } \cdot \ensuremath{\mathscr{P}}_x}{\ensuremath{\mathscr{P}}_x[\Gamma_{\infty }]} .
\label{}
\end{align}
We shall prove in Theorem \ref{thm: mart op}
that there exist such martingales $ (M_{t,x}(F)) $ as in \eqref{eq: Q2'}.
We shall call $ M_{t,x}(\cdot) $ the {\em martingale generator}
and we shall study its properties in Sections \ref{sec: univ} and \ref{sec: further}.
Then the limit probability measure $ P^{\Gamma}_x $ is characterised by
\begin{align}
P^{\Gamma}_x |_{\ensuremath{\mathcal{F}}_t} = \frac{M_{t,x}(\Gamma_{\infty })}{\ensuremath{\mathscr{P}}_x[\Gamma_{\infty }]} \cdot P_x|_{\ensuremath{\mathcal{F}}_t}
, \qquad t \ge 0 .
\label{}
\end{align}
Therefore, if we answer Question 2, then we have answered Question 1.
Our strategy to answer {\bf (Q1$ ' $)} is as follows.
Since the index $ \alpha $ is supposed to be in $ (1,2] $,
each point of $ \ensuremath{\mathbb{R}} $ is regular and recurrent.
Hence, associated with the process,
there is a jointly continuous local time $ (L(t,x)) $.
We simply write $ L_t = L(t,0) $,
and, associated with this local time,
there is {\em It\^o's measure $ \ensuremath{\mbox{{\boldmath $n$}}} $ of excursions away from the origin}
(see Section \ref{sec: exc}).
Let $ R $ denote the {\em lifetime} of an excursion path.
For $ t>0 $, we define $ M^{(t)} $ as the probability measure on $ \ensuremath{\mathcal{F}}_t $ given by
\begin{align}
M^{(t)} = \frac{1_{\{ R>t \}}}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} \cdot \ensuremath{\mbox{{\boldmath $n$}}} |_{\ensuremath{\mathcal{F}}_t}
\label{eq: Meander}
\end{align}
and here we call $ M^{(t)} $ the distribution of the {\it stable meander}.
We remark that
our meander distribution \eqref{eq: Meander}
is definitely different from that of \cite{MR1465814} etc.
where the meander is defined by conditioning on $ \{ R>t \} $
the excursion process for the {\em reflected} stable L\'evy process
$ (X_t-\min_{s \le t} X_s:t \ge 0) $.
We shall prove the following formula (Theorem \ref{thm: LED})
of disintegration of $ P_0|_{\ensuremath{\mathcal{F}}_t} $ for each $ t>0 $
with respect to last exit time from the origin:
\begin{align}
\frac{1}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} P_0|_{\ensuremath{\mathcal{F}}_t}
=
\frac{\Gamma(1/\alpha )}{\alpha \pi}
\int_0^t \cbra{ 1-\frac{u}{t} }^{\frac{1}{\alpha }-1}
\frac{\d u}{u^{1/\alpha }} Q^{(u)} \bullet M^{(t-u)} .
\label{eq: LED2}
\end{align}
As a check, the total masses of both sides agree,
as we shall show in Proposition \ref{prop: vn Rt}.
Then, we shall establish (in Theorem \ref{thm: meander}) the convergence
\begin{align}
M^{(t)}
\stackrel{t \to \infty }{\longrightarrow}
P^{\dagger}_0
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $.}
\label{eq: meander conv Pdagger2}
\end{align}
Noting that $ (1-\frac{u}{t})^{\frac{1}{\alpha } -1} \to 1 $ as $ t \to \infty $,
we may expect that, in some sense:
\begin{align}
\int_0^t \cbra{ 1-\frac{u}{t} }^{\frac{1}{\alpha }-1}
\frac{\d u}{u^{1/\alpha }} Q^{(u)} \bullet M^{(t-u)}
\stackrel{t \to \infty }{\longrightarrow}
\int_0^{\infty } \frac{\d u}{u^{1/\alpha }} Q^{(u)} \bullet P^{\dagger}_0 .
\label{eq: convergence of P_0/nRt to sP}
\end{align}
We shall prove several analytic lemmas
which justify the convergence \eqref{eq: convergence of P_0/nRt to sP}
and then
we shall establish the convergence \eqref{eq: Q1'}
with the function $ \mu(t) = \ensuremath{\mbox{{\boldmath $n$}}}(R>t) $.
In order to answer Question 2 (and in particular {\bf (Q2$ ' $)}),
we shall establish the convergence \eqref{eq: Q1'}
and compute the martingale generator
by case study.
We confine ourselves to the following two kinds of weight functionals:
\begin{quote}
(i)
$ \Gamma_t = f(L_t) $ for some non-negative Borel functions $ f $
with some integrability property;
\\
(ii)
$ \displaystyle \Gamma_t = \exp \kbra{ - \int L(t,x) V(\d x) } $
for some non-negative Borel measure $ V $.
We call the problems in such a case the {\em Feynman--Kac penalisations}.
\end{quote}
The organisation of the present paper is as follows.
In Section \ref{sec: prel} we recall some preliminary facts
about symmetric stable L\'evy processes.
In Section \ref{sec: exc} we study It\^o's measure of excursions away from the origin
relatively to the symmetric stable process.
In Section \ref{sec: intro2}
we prove several formulae concerning the stable meander and $ h $-path process,
which play important roles in the study of our penalisation problems.
In Section \ref{sec: univ} we make
general observations on the universal $ \sigma $-finite measure $ \ensuremath{\mathscr{P}}_x $
and the martingale generator $ M_{t,x}(\cdot) $.
In Section \ref{sec: imp lem} we prove several convergence lemmas
which play fundamental roles in the proof of our penalisation problems.
Section \ref{sec: LT penal} is devoted to the study of penalisations
with a function of the local time at the origin.
Section \ref{sec: FK penal} is devoted to the study of Feynman--Kac penalisations.
In Section \ref{sec: further} we characterise certain non-negative
$ (P_0,\ensuremath{\mathcal{F}}_t) $-martingales in terms of $ \ensuremath{\mathscr{P}} $.
\section{Preliminaries about the symmetric stable process of index $ 1 < \alpha \le 2 $}
\label{sec: prel}
Recall that $ (X_t,\ensuremath{\mathcal{F}}_t,P_x) $ is the canonical representation
of a one-dimensional symmetric stable L\'evy process of index $ 1<\alpha \le 2 $
such that
\begin{align}
P_0[{\rm e}^{i \lambda X_t}] = {\rm e}^{-t |\lambda|^{\alpha }}
\quad \text{for} \
\lambda \in \ensuremath{\mathbb{R}} .
\label{}
\end{align}
All results presented in this section are well-known;
see, e.g., \cite{MR1406564}.
\noindent
{\bf 1).}
$ (X_t) $ has a transition density
$ P_x(X_t \in \d y) = p_t(y-x) \d y $
where
$ p_t(x) $ is given by
\begin{align}
p_t(x)
= \frac{1}{\pi} \int_0^{\infty } (\cos x \lambda) {\rm e}^{-t \lambda^{\alpha }} \d \lambda
.
\label{}
\end{align}
For $ q>0 $, we set
\begin{align}
u_q(x) = \int_0^{\infty } {\rm e}^{-qt} p_t(x) \d t
= \frac{1}{\pi} \int_0^{\infty } \frac{\cos x \lambda}{q+\lambda^{\alpha }} \d \lambda .
\label{eq: uqx}
\end{align}
In particular, if we take $ x=0 $, we have
\begin{align}
p_t(0) = p_1(0) t^{-\frac{1}{\alpha }}
\quad \text{where} \
p_1(0) = \frac{\Gamma (1/\alpha )}{\alpha \pi}
\label{}
\end{align}
and
\begin{align}
u_q(0) = u_1(0) q^{\frac{1}{\alpha }-1}
\quad \text{where} \
u_1(0) = \frac{\Gamma(1-1/\alpha ) \Gamma(1/\alpha )}{\alpha \pi} .
\label{eq: uq0}
\end{align}
\noindent
{\bf 2).}
Let $ T_{\{ a \}} $ denote the first hitting time of $ a $ for the coordinate process $ (X_t) $:
\begin{align}
T_{\{ a \}} = \inf \{ t>0: X_t=a \} .
\label{}
\end{align}
Then the Laplace transform of the law of $ T_{\{ 0 \}} $ is given by
\begin{align}
P_x[{\rm e}^{-q T_{\{ 0 \}}}] = \frac{u_q(x)}{u_q(0)}
, \qquad x \in \ensuremath{\mathbb{R}} , \ q>0
\label{eq: LT of T0}
\end{align}
(see, e.g., \cite[pp. 64]{MR1406564}).
For further study of the law of $ T_{\{ 0 \}} $, see \cite{YYY2}.
Since $ T_{\{ y \}} $ under $ P_x $ has the same law as $ T_{\{ 0 \}} $ under $ P_{x-y} $,
the formula \eqref{eq: LT of T0} implies the following facts:
\\ \quad
{\rm (i)}
Each point is a recurrent state, i.e.,
$ P_x(T_{\{ y \}}<\infty ) = 1 $ for any $ x,y \in \ensuremath{\mathbb{R}} $ with $ x \neq y $;
\\ \quad
{\rm (ii)}
Each point is regular for itself, i.e.,
$ P_x(T_{\{ x \}}=0)=1 $ for any $ x \in \ensuremath{\mathbb{R}} $.
\noindent
{\bf 3).}
The process admits a jointly continuous local time $ L(t,x) $ such that
\begin{align}
L(t,x) = \lim_{\ensuremath{\varepsilon} \to 0+} \frac{1}{2 \ensuremath{\varepsilon}} \int_0^t 1_{\{ |X_s-x|<\ensuremath{\varepsilon} \}} \d s
\label{}
\end{align}
almost surely.
We simply write $ L_t = L(t,0) $.
Denote the inverse local time at the origin by $ \tau_l = \inf \{ t>0: L_t>l \} $.
Then $ (\tau_l:l \ge 0) $ is a stable subordinator of index $ 1-1/\alpha $
such that
\begin{align}
P_0[{\rm e}^{-q \tau_l}] = {\rm e}^{-l/u_q(0)}
\label{}
\end{align}
(see, e.g., \cite[pp. 131]{MR1406564}),
where $ u_q(0) $ is given explicitly by \eqref{eq: uq0}.
Let $ \theta_t:\ensuremath{\mathbb{D}} \to \ensuremath{\mathbb{D}} $ stand for the shift operator:
$ \theta_t(w) = w(t+\cdot) $.
Since $ \tau_l = T_{\{ 0 \}} + \tau_l \circ \theta_{T_{\{ 0 \}}} $, we have
\begin{align}
P_x \ebra{ \int_0^{\infty } {\rm e}^{-qt} \d L_t }
=& P_x \ebra{ \int_0^{\infty } {\rm e}^{-q \tau_l} \d l }
= P_x[{\rm e}^{-q T_{\{ 0 \}}}] \int_0^{\infty } P_0[{\rm e}^{-q\tau_l}] \d l
\\
=& \frac{u_q(x)}{u_q(0)} \cdot u_q(0)
= \int_0^{\infty } {\rm e}^{-qt} p_t(x) \d t
\label{}
\end{align}
for all $ q>0 $.
Hence we see that
\begin{align}
P_x \ebra{ \int_0^{\infty } f(t) \d L_t }
= \int_0^{\infty } f(t) p_t(x) \d t
\label{}
\end{align}
for any non-negative measurable function $ f $ on $ [0,\infty ) $.
Consequently, we may write
\begin{align}
P_x[\d L_t] = p_t(x) \d t
, \qquad x \in \ensuremath{\mathbb{R}}.
\label{}
\end{align}
\section{It\^o's measure of excursions away from the origin}
\label{sec: exc}
Since the origin is a regular and recurrent state,
we can apply It\^o's excursion theory
(\cite{MR0402949}; see also \cite{MR1406564} and \cite{MR1138461} for details).
We denote by $ \ensuremath{\mathbb{E}} $ the set of c\`adl\`ag paths $ e:[0,\infty ) \to \ensuremath{\mathbb{R}} \cup \{ \Delta \} $
such that
\begin{align}
\begin{cases}
e(t) \in \ensuremath{\mathbb{R}} \setminus \{ 0 \}
\quad & \text{for $ 0<t<R(e) $}, \\
e(t)=\Delta
\quad & \text{for $ t \ge R(e) $}
\end{cases}
\label{}
\end{align}
where
\begin{align}
R = R(e) = \inf \{ t>0: e(t)=\Delta \} .
\label{}
\end{align}
We call $ \ensuremath{\mathbb{E}} $ {\em the set of excursions}
and every element $ e $ of $ \ensuremath{\mathbb{E}} $ an {\em excursion path}.
For an excursion path $ e \in \ensuremath{\mathbb{E}} $,
we call $ R(e) $ the {\em lifetime} of $ e $.
The point $ \Delta $ is called the {\em cemetery}.
We set $ D = \{ l: \tau_l - \tau_{l-}>0 \} $.
For each $ l \in D $,
we set
\begin{align}
e_l(t) =
\begin{cases}
X_{t+\tau_{l-}}
, \quad & \text{for} \ 0 \le t < \tau_l - \tau_{l-} , \\
\Delta
, \quad & \text{for} \ t \ge \tau_l - \tau_{l-}.
\end{cases}
\label{}
\end{align}
Then It\^o's fundamental theorem (\cite{MR0402949}) asserts that
the point process $ (e_l: l \in D) $ taking values on $ \ensuremath{\mathbb{E}} $
is a Poisson point process.
Its characteristic measure will be denoted by $ \ensuremath{\mbox{{\boldmath $n$}}} $
and called {\em It\^o's measure of excursions away from the origin}.
It\^o's measure $ \ensuremath{\mbox{{\boldmath $n$}}} $ is a $ \sigma $-finite measure on any $ \ensuremath{\mathcal{F}}_t $
which has no mass outside the set
\begin{align}
\{ e \in \ensuremath{\mathbb{E}} : X_0(e)=0 , \ 0<R(e)<\infty \} .
\label{}
\end{align}
For the fact that $ \ensuremath{\mbox{{\boldmath $n$}}}(\{ X_0 = 0 \}^c) = 0 $, see \cite{Y}.
For $ x \in \ensuremath{\mathbb{R}} \setminus \{ 0 \} $,
we denote by $ P^0_x $
the law of the killed process, i.e.,
the law on $ \ensuremath{\mathbb{E}} $ of the path $ (X^0_t) $ under $ P_x $ where
\begin{align}
X^0_t =
\begin{cases}
X_t
, \quad & 0 \le t < T_{\{ 0 \}} , \\
\Delta
, \quad & t \ge T_{\{ 0 \}} .
\end{cases}
\label{}
\end{align}
We shall utilise the following formulae.
\begin{Thm}[Markov property of $ \ensuremath{\mbox{{\boldmath $n$}}} $] \label{thm: Markov of vn}
It holds that
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}}[Z_t F(X \circ \theta_t)]
= \int \ensuremath{\mbox{{\boldmath $n$}}}[Z_t;X_t \in \d x] P^0_x[F(X)]
\label{}
\end{align}
for any $ t>0 $,
any non-negative $ \ensuremath{\mathcal{F}}_t $-measurable functional $ Z_t $
and any non-negative measurable functional $ F $ on $ \ensuremath{\mathbb{E}} $.
\end{Thm}
\begin{Thm}[Compensation formula]
\label{thm: compensation}
Let $ F=F(t,\omega,e) $ be a measurable functional on
$ [0,\infty ) \times \ensuremath{\mathbb{D}} \times \ensuremath{\mathbb{E}} $ such that,
for every fixed $ e \in \ensuremath{\mathbb{E}} $,
the process $ (F(t,\cdot,e):t \ge 0) $ is $ (\ensuremath{\mathcal{F}}_t) $-predictable.
Then\footnote{Here the symbol $ \tilde{} $ means independence.}
\begin{align}
P_0 \ebra{ \sum_{l \in D} F(\tau_{l-},X,e_l) }
= P_0 \otimes \tilde{\ensuremath{\mbox{{\boldmath $n$}}}} \ebra{ \int_0^{\infty } \d L_t F(t,X,\tilde{X}) } .
\label{eq: compensation}
\end{align}
\end{Thm}
We omit the proofs of Theorems \ref{thm: Markov of vn} and \ref{thm: compensation}.
For their proofs, see \cite{MR1406564}, \cite{MR1138461} and \cite{MR1725357}.
\subsection{Entrance law}
In order to characterise the entrance law, we need the following
\begin{Thm}[\cite{CFY} and \cite{MR2247835}] \label{thm: master1}
For any non-negative measurable function $ f $ on $ \ensuremath{\mathbb{R}} $, it holds that
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} \ensuremath{\mbox{{\boldmath $n$}}} \ebra{f(X_t)} \d t
= \int f(x) P_x \ebra{ {\rm e}^{-q T_{\{ 0 \}}} } \d x .
\label{eq: master1}
\end{align}
\end{Thm}
We remark that the relation \eqref{eq: master1} can be found
in Chen--Fukushima--Ying \cite[Eq. (2.8)]{CFY}
and Fitzsimmons--Getoor \cite[Eq. (3.22)]{MR2247835}
in a fairly general Markovian framework as
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} \ensuremath{\mbox{{\boldmath $n$}}}[f(X_t)] \d t
= \int f(x) \hat{P}_x \ebra{ {\rm e}^{-q T_{\{ 0 \}}(\hat{X})} } m(\d x)
\label{eq: master CFY and FG}
\end{align}
where
$ (X_t,P_x) $ and $ (\hat{X},\hat{P}_x) $ are
in weak duality with respect to the reference measure $ m $.
In our case, $ (\hat{X},\hat{P}_x) = (-X_t,P_x) $
and $ m(\d x) = \d x $, the Lebesgue measure.
Although \eqref{eq: master1} is a special case of \eqref{eq: master CFY and FG},
we give the proof of Theorem \ref{thm: master1} for completeness of this paper.
\begin{proof}[Proof of Theorem \ref{thm: master1}]
Note that
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} f(X_t) \d t
= \sum_{l \in D} {\rm e}^{-q \tau_{l-}} \int_0^{R(e_l)} {\rm e}^{-qt} f(e_l(t)) \d t .
\label{}
\end{align}
By Theorem \ref{thm: compensation}, we obtain
\begin{align}
P_0 \ebra{ \int_0^{\infty } {\rm e}^{-qt} f(X_t) \d t }
= P_0 \ebra{ \int_0^{\infty } {\rm e}^{-qt} \d L_t } \ensuremath{\mbox{{\boldmath $n$}}} \ebra{ \int_0^R {\rm e}^{-qt} f(X_t) \d t } .
\label{}
\end{align}
Since $ P_0 \ebra{ \int_0^{\infty } {\rm e}^{-qt} \d L_t } = u_q(0) $,
we have
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} \ensuremath{\mbox{{\boldmath $n$}}} \ebra{f(X_t)} \d t
= \int f(x) \frac{u_q(x)}{u_q(0)} \d x .
\label{eq: master1-1}
\end{align}
By the identity \eqref{eq: LT of T0},
we obtain \eqref{eq: master1}.
The proof is complete.
\end{proof}
The following formula holds:
\begin{Prop} \label{prop: vn Rt}
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}}(R>t) = \ensuremath{\mbox{{\boldmath $n$}}}(R>1) t^{\frac{1}{\alpha }-1}
\label{}
\end{align}
where
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}}(R>1)
=
\frac{\alpha \pi}{\Gamma(1-1/\alpha) \Gamma (1/\alpha )^2} .
\label{}
\end{align}
In particular,
\begin{align}
\frac{\ensuremath{\mbox{{\boldmath $n$}}}(R>t-s)}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} = \cbra{ 1-\frac{s}{t} }^{\frac{1}{\alpha }-1}
\qquad \text{for} \
0<s<t .
\label{eq: con R ratio}
\end{align}
\end{Prop}
Although it is well-known, we again give the proof for completeness of this paper.
\begin{proof}
Take $ f = 1 $ in \eqref{eq: master1-1}.
Then we have
$ \ensuremath{\mbox{{\boldmath $n$}}}[f(X_t)] = \ensuremath{\mbox{{\boldmath $n$}}}(R>t) $,
and the identity \eqref{eq: master1-1} implies that
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} \ensuremath{\mbox{{\boldmath $n$}}}(R>t) \d t
= \frac{1}{q u_q(0)} = \frac{1}{u_1(0)} q^{-1/\alpha } .
\label{eq: LT vnR}
\end{align}
This completes the proof.
\end{proof}
The following theorem characterises the entrance law.
\begin{Thm} \label{thm: rho}
There exists a bi-measurable function $ \rho(t,x) $
which is at the same time
a space density of the entrance law
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}}(X_t \in \d x) = \rho(t,x) \d x
\label{eq: ent law density}
\end{align}
and a time density of the first hitting time
\begin{align}
P_x(T_{\{ 0 \}} \in \d t) = \rho(t,x) \d t .
\label{eq: first pas density}
\end{align}
That is,
\begin{align}
\rho(t,x) = \frac{\ensuremath{\mbox{{\boldmath $n$}}}(X_t \in \d x)}{\d x} = \frac{P_x(T_{\{ 0 \}} \in \d t)}{\d t} .
\label{eq: ent law and first pas}
\end{align}
\end{Thm}
\begin{proof}
Note that
$ P^0_x(X_t \in \d y) = p^0_t(x,y) \d y $ where
\begin{align}
p^0_t(x,y) = p_t(y-x) - \int_0^t p_{t-s}(y) P_x(T_{\{ 0 \}} \in \d s) .
\label{}
\end{align}
Now we set
\begin{align}
\rho(t,x) = \int \ensuremath{\mbox{{\boldmath $n$}}}(X_{t/2} \in \d y) p^0_{t/2}(y,x) .
\label{}
\end{align}
Let $ f $ be a non-negative measurable function on $ \ensuremath{\mathbb{R}} $.
By the Markov property,
we see that
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}}[f(X_t)] = \int \ensuremath{\mbox{{\boldmath $n$}}}(X_{t/2} \in \d y) P^0_y[f(X_{t/2})]
= \int f(x) \rho(t,x) \d x .
\label{eq: ent law integral}
\end{align}
Hence we obtain \eqref{eq: ent law density}.
Using the formulae \eqref{eq: ent law integral}
and \eqref{eq: master1},
we see that
\begin{align}
\int \d x f(x) \int_0^{\infty } {\rm e}^{-qt} \rho(t,x) \d t
=&
\int_0^{\infty } {\rm e}^{-qt} \ensuremath{\mbox{{\boldmath $n$}}}[f(X_t)] \d t
\\
=&
\int \d x f(x) P_x[{\rm e}^{-q T_{\{ 0 \}}}] .
\label{}
\end{align}
Hence we obtain \eqref{eq: first pas density}.
\end{proof}
\section{Stable meander and $ h $-path process} \label{sec: intro2}
\subsection{Disintegration with respect to the last exit time}
For $ u>0 $, let $ Q^{(u)} $ denote the law of the bridge
$ P_0(\cdot | X_u=0) $
considered to be a probability measure on $ \ensuremath{\mathcal{F}}_u $.
We denote by $ X^{(u)} = (X_t:0 \le t \le u) $
the coordinate process considered up to time $ u $.
We denote the {\em concatenation} between
the two processes $ X^{(u)} $ and $ \tilde{X}^{(v)}=(\tilde{X}_t:0 \le t \le v) $
by $ X^{(u)} \bullet \tilde{X}^{(v)} = ((X^{(u)} \bullet \tilde{X}^{(v)})_t:0 \le t \le u+v) $:
\begin{align}
\cbra{ X^{(u)} \bullet \tilde{X}^{(v)} }_t =
\begin{cases}
X^{(u)}_t
, \quad & 0 \le t < u , \\
\tilde{X}^{(v)}_{t-u}
, \quad & u \le t \le u+v .
\end{cases}
\label{}
\end{align}
The measure $ Q^{(u)} \bullet M^{(v)} $
is defined as the law of the concatenation $ X^{(u)} \bullet \tilde{X}^{(v)} $
between the two processes $ X^{(u)} $ and $ \tilde{X}^{(v)} $
where $ (X^{(u)},\tilde{X}^{(v)}) $ is considered
under the product measure $ Q^{(u)} \otimes M^{(v)} $.
Here and in what follows, we emphasize independence with the symbol $ \tilde{} $,
unless otherwise stated.
For $ t>0 $, we denote last exit time from the origin before $ t $ by
\begin{align}
g_t = \inf \{ s \le t: X_s = 0 \} .
\label{}
\end{align}
The following formula describes disintegration of $ P_0|_{\ensuremath{\mathcal{F}}_t} $
with respect to $ g_t $:
\begin{Thm} \label{thm: LED}
For each $ t>0 $, it holds that
\begin{align}
P_0|_{\ensuremath{\mathcal{F}}_t} = \int_0^t \ensuremath{\mbox{{\boldmath $n$}}}(R>t-u) P_0[\d L_u] Q^{(u)} \bullet M^{(t-u)} .
\label{eq: LED}
\end{align}
In other words, the following statements hold:
\\ \quad
{\rm (i)}
The distribution of $ g_t $ is given by
$ P_0(g_t \in \d u) = \ensuremath{\mbox{{\boldmath $n$}}}(R>t-u) P_0[\d L_u] $;
\\ \quad
{\rm (ii)}
Given $ g_t=u $, $ (X_t:t \in [0,u]) $ and $ (X_{u+t}:t \in [0,t-u]) $
are independent under $ P_0 $;
\\ \quad
{\rm (iii)}
$ (X_t:t \in [0,u]) $ under $ P_0 $ is distributed as the stable bridge $ Q^{(u)} $;
\\ \quad
{\rm (iv)}
$ (X_{u+t}:t \in [0,t-u]) $ under $ P_0 $ is distributed as the stable meander $ M^{(t-u)} $.
\end{Thm}
\begin{Rem}
We note that the formula \eqref{eq: LED} is the counterpart
of Salminen \cite[Prop. 4]{MR1454113} in his study of last exit decomposition
for linear diffusions.
\end{Rem}
\begin{Rem}
We remark that (i) implies
\begin{align}
P_0(g_t \in \d u) = \frac{(t-u)^{\frac{1}{\alpha }-1} u^{-\frac{1}{\alpha }} \d u }
{\Gamma (1-1/\alpha ) \Gamma (1/\alpha )}
\label{}
\end{align}
for some constant $ C $, which shows that
$ \frac{1}{t} g_t $ has the Beta$ (1-\frac{1}{\alpha },\frac{1}{\alpha }) $ distribution.
For further discussions, see \cite{YYY2}.
\end{Rem}
\begin{proof}[Proof of Theorem \ref{thm: LED}]
Let us prove
\begin{align}
P_0|_{\ensuremath{\mathcal{F}}_t}
= \int_0^t P_0[\d L_u] Q^{(u)} \bullet (\ensuremath{\mbox{{\boldmath $n$}}}|_{\ensuremath{\mathcal{F}}_{t-u}}) ,
\label{eq: P0(t)}
\end{align}
which is equivalent to \eqref{eq: LED}.
Let $ F(t,w) $ be a non-negative continuous functional
on $ [0,\infty ) \times \ensuremath{\mathbb{D}} $.
For each $ t \ge 0 $, we define a measurable functional $ F_t $ on $ \ensuremath{\mathbb{D}}([0,t];\ensuremath{\mathbb{R}}) $
by $ F_t(X^{(t)}) = F(t,X_{t \wedge \cdot}) $.
Then we have
\begin{align}
\int_0^{\infty } \d t F_t(X^{(t)})
= \sum_{l \in D} \int_0^{R(e_l)} \d r
F_{\tau_{l-}+r} \cbra{ X^{(\tau_{l-})} \bullet e_l } .
\label{}
\end{align}
Now we appeal to Theorem \ref{thm: compensation}
and we obtain
\begin{align}
\int_0^{\infty } P_0[F_t(X^{(t)})] \d t
=
\cbra{ P_0 \otimes \tilde{\ensuremath{\mbox{{\boldmath $n$}}}} }
\ebra{ \int_0^{\infty } \d L_t \int_0^{\infty } \d r 1_{ \{ \tilde{R}>r \} }
F_{t+r} \cbra{ X^{(t)} \bullet \tilde{X}^{(r)} } }
.
\label{}
\end{align}
Since $ P_0[ \int_0^{\infty } G(X^{(u)}) \d L_u ]
= \int_0^{\infty } P_0[\d L_u] Q^{(u)}[G(X^{(u)})] $,
we obtain
\begin{align}
\int_0^{\infty } P_0[F_t(X^{(t)})] \d t
=&
\int_0^{\infty } P_0[\d L_u]
\cbra{ Q^{(u)} \otimes \tilde{\ensuremath{\mbox{{\boldmath $n$}}}} }
\ebra{ \int_0^{\infty } \d r 1_{ \{ \tilde{R}>r \} }
F_{u+r} \cbra{ X^{(u)} \bullet \tilde{X}^{(r)} } }
.
\label{}
\end{align}
Changing variables to $ t=r+u $ and the order of integrations, we have
\begin{align}
\int_0^{\infty } P_0[F_t(X^{(t)})] \d t
=
\int_0^{\infty } \d t
\int_0^t P_0[\d L_u]
\cbra{ Q^{(u)} \bullet (\ensuremath{\mbox{{\boldmath $n$}}}|_{\ensuremath{\mathcal{F}}_{t-u}}) }
\ebra{ 1_{ \{ R>t-u \} }
F_t \cbra{ X^{(t)} } }
.
\label{eq: int P0 Ft X(t)}
\end{align}
Since the identity \eqref{eq: int P0 Ft X(t)} holds
with $ F_t $ replaced by $ {\rm e}^{-qt} F_t $ for any $ q>0 $,
we obtain
\begin{align}
P_0[F_t(X^{(t)})]
= \int_0^t P_0[\d L_u] \cbra{ Q^{(u)} \bullet (\ensuremath{\mbox{{\boldmath $n$}}}|_{\ensuremath{\mathcal{F}}_{t-u}}) }
\ebra{ 1_{ \{ R>t-u \} } F_t(X^{(t)}) } .
\label{}
\end{align}
This completes the proof.
\end{proof}
\begin{Rem}
In the above argument, we have proven the following formulae:
\begin{align}
\int_0^{\infty } P_0^{(t)} \d t
=&
\int_0^{\infty } P_0^{(\tau_l)} \d l \bullet
\int_0^{\infty } \ensuremath{\mbox{{\boldmath $n$}}}(R>r) M^{(r)} \d r
\\
=&
\int_0^{\infty } P_0[\d L_u] Q^{(u)} \bullet
\int_0^{\infty } \ensuremath{\mbox{{\boldmath $n$}}}(R>r) M^{(r)} \d r .
\label{}
\end{align}
Here we adopt the notations $ P_0^{(t)} $ and $ P_0^{(\tau_l)} $
which are found in \cite{MR1725357},
but we do not go into details.
\end{Rem}
\subsection{Harmonicity of the function $ |x|^{\alpha -1} $}
Set
\begin{align}
h(x)
= \lim_{q \to 0+} \{ u_q(0)-u_q(x) \}
= \frac{1}{\pi} \int_0^{\infty } \frac{1-\cos x \lambda}{\lambda^{\alpha }} \d \lambda .
\label{}
\end{align}
Then we have
\begin{align}
h(x)
= h(1) |x|^{\alpha -1}
\label{}
\end{align}
where
\begin{align}
h(1) = 2 \cos \frac{(2-\alpha) \pi}{2} .
\label{}
\end{align}
\begin{Thm} \label{thm: harmonic}
The function $ h(x) = h(1) |x|^{\alpha -1} $ is harmonic
for the killed process, i.e.,
\begin{align}
P^0_x[h(X_t)] = P_x[h(X_t);T_{\{ 0 \}}>t] = h(x)
, \qquad x \in \ensuremath{\mathbb{R}} \setminus \{ 0 \} , \ t>0 .
\label{}
\end{align}
Equivalently, $ (h(X_{t \wedge T_{\{ 0 \}}})) $ is a $ (P_x,\ensuremath{\mathcal{F}}_t) $-martingale.
\end{Thm}
We omit the proof,
because Theorem \ref{thm: harmonic} follows immediately from the
\begin{Thm}[Salminen--Yor \cite{SY}] \label{thm: Sal-Yor}
\Confirst{con: Tanaka}
For $ x \in \ensuremath{\mathbb{R}} $,
there exist a square-integrable martingale $ N_t^x $
and some constant $ C $
such that
\begin{align}
|X_t|^{\alpha -1} = |x|^{\alpha -1} + N_t^x + C L(t,x)
\quad \text{under} \
P_x .
\label{}
\end{align}
\end{Thm}
\begin{Thm} \label{thm: harmonic2}
It holds that
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}}[h(X_t)] = 1
, \qquad t>0 .
\label{eq: vn hXt = 1}
\end{align}
\end{Thm}
\begin{proof}[Proof of Theorem \ref{thm: harmonic2}]
Theorem \ref{thm: rho} and the identity \eqref{eq: LT of T0} imply that
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} \ensuremath{\mbox{{\boldmath $n$}}}[h(X_t)] \d t
= \int h(x) P_x[{\rm e}^{-q T_{\{ 0 \}}}] \d x
= \int h(x) \frac{u_q(x)}{u_q(0)} \d x .
\label{}
\end{align}
Hence it suffices to prove that
\begin{align}
\int u_q(x) h(x) \d x = \frac{u_q(0)}{q}
, \qquad x \in \ensuremath{\mathbb{R}}.
\label{eq: uq h}
\end{align}
Let $ r $ be such that $ 0<r<q $.
By the resolvent equation $ U_q U_r = (U_r-U_q)/(q-r) $, we have
\begin{align}
\int u_q(x-y) u_r(y-z) \d y = \frac{1}{q-r} \kbra{ u_r(x-z) - u_q(x-z) } .
\label{}
\end{align}
Letting $ x=z=0 $ and using the symmetry $ u_q(-y)=u_q(y) $, we have
\begin{align}
\int u_q(y) u_r(y) \d y = \frac{1}{q-r} \kbra{ u_r(0) - u_q(0) } .
\label{}
\end{align}
Now we have
\begin{align}
\int u_q(y) \kbra{ u_r(0) - u_r(y) } \d y
= \frac{u_q(0)}{q-r} - \frac{ru_r(0)}{q(q-r)} .
\label{}
\end{align}
If we let $ r $ decrease to 0, then we see that
\begin{align}
u_r(0) - u_r(x)
= \frac{1}{\pi} \int_0^{\infty } \frac{1-\cos x \lambda}{r+\lambda^{\alpha }} \d \lambda
\label{}
\end{align}
increases to $ h(x) $,
and that $ ru_r(0) \to 0 $. Hence we obtain \eqref{eq: uq h}
by the monotone convergence theorem.
\end{proof}
\begin{Rem}
For generalisations of Theorems \ref{thm: harmonic} and \ref{thm: harmonic2}
for symmetric L\'evy processes, see \cite{Y}.
\end{Rem}
\subsection{Convergence of the stable meander to the $ h $-path process}
Let us introduce the {\em $ h $-path process} $ (P^{\dagger}_x:x \in \ensuremath{\mathbb{R}}) $
as
\begin{align}
P^{\dagger}_x |_{\ensuremath{\mathcal{F}}_t}
=&
\frac{h(X_t)}{h(x)} \cdot P^0_x |_{\ensuremath{\mathcal{F}}_t}
, \qquad x \in \ensuremath{\mathbb{R}} \setminus \{ 0 \},
\label{}
\\
P^{\dagger}_0 |_{\ensuremath{\mathcal{F}}_t}
=&
h(X_t) \cdot \ensuremath{\mbox{{\boldmath $n$}}}|_{\ensuremath{\mathcal{F}}_t} .
\label{eq: Imhof}
\end{align}
From Theorem \ref{thm: harmonic} and the Markov properties of $ P^0_x $ and $ \ensuremath{\mbox{{\boldmath $n$}}} $,
it follows that such a process exists uniquely.
Remark that, when $ \alpha =2 $,
the $ h $-path process coincides up to some scale transform
with the symmetrization of three-dimensional Bessel process;
consequently, the identity \eqref{eq: Imhof}
is nothing but the {\em Imhof relation}
(see, e.g., \cite[17, Exercise XII.4.18]{MR1725357}).
The following result asserts
that the meander converges to the $ h $-path process.
\begin{Thm} \label{thm: meander}
It holds that
\begin{align}
M^{(t)}
\ \stackrel{t \to \infty }{\longrightarrow} \
P^{\dagger}_0
\qquad \text{along} \
(\ensuremath{\mathcal{F}}_s) .
\label{eq: meander conv Pdagger}
\end{align}
\end{Thm}
In order to prove Theorem \ref{thm: meander}, we need the
\begin{Lem} \label{lem: Ytx}
For $ t>0 $ and $ x \neq 0 $, set
\begin{align}
Y(t,x) =
\frac{P_x(T_{\{ 0 \}}>t)}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)} .
\label{eq: Ytx}
\end{align}
Then it holds
that $ Y(t,x) \to 1 $ as $ t \to \infty $ for any fixed $ x \neq 0 $,
and that $ Y(t,x) $ is bounded in $ t>0 $ and $ x \neq 0 $.
\end{Lem}
\begin{proof}[Proof of Lemma \ref{lem: Ytx}]
Using \eqref{eq: LT of T0}, we have
\begin{align}
\int_0^{\infty } {\rm e}^{-qt} P_x(T_{\{ 0 \}}>t) \d t
= \frac{u_q(0)-u_q(x)}{q u_q(0)}
\sim h(x) \frac{q^{-1/\alpha } }{u_1(0)}
\qquad \text{as} \
q \to 0+ .
\label{}
\end{align}
Hence we may apply a tauberian theorem.
By Proposition \ref{prop: vn Rt}, we obtain
\begin{align}
P_x(T_{\{ 0 \}}>t) \sim
h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)
\qquad \text{as $ t \to \infty $} .
\label{asymp}
\end{align}
This shows the first assertion.
Since the function $ t \mapsto Y(t,1) $ is continuous
and $ Y(t,1) \to 1 $ as $ t \to \infty $,
we see that $ Y(t,1) $ is bounded in $ t>0 $.
By scaling property $ P_x(T_{\{ 0 \}}>t) = P_1(T_{\{ 0 \}}>|x|^{-\alpha }t) $, we have
$ Y(t,x)= Y(|x|^{-\alpha }t,1) $.
This proves the second assertion.
\end{proof}
Now let us proceed to prove Theorem \ref{thm: meander}.
\begin{proof}[Proof of Theorem \ref{thm: meander}]
Let $ s>0 $ be fixed
and let $ Z_s $ be a bounded $ \ensuremath{\mathcal{F}}_s $-measurable functional.
By the Markov property of $ \ensuremath{\mbox{{\boldmath $n$}}} $, we have
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}} \ebra{ Z_s 1_{\{ R>t \}} }
= \ensuremath{\mbox{{\boldmath $n$}}} \ebra{ Z_s 1_{\{ R>s \}} P_{X_s} (T_{\{ 0 \}}>t-s) } .
\label{}
\end{align}
By the Imhof relation \eqref{eq: Imhof} and by \eqref{eq: Ytx}, we have
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}} \ebra{ Z_s 1_{\{ R>t \}} }
=& P^{\dagger}_0 \ebra{ Z_s P_{X_s} (T_{\{ 0 \}}>t-s) / h(X_s) }
\\
=& P^{\dagger}_0 [Z_s Y(t-s,X_s)] \cdot \ensuremath{\mbox{{\boldmath $n$}}}(R>t-s) .
\label{}
\end{align}
Dividing both sides by $ \ensuremath{\mbox{{\boldmath $n$}}}(R>t) $, using Proposition \ref{prop: vn Rt},
and then applying the bounded convergence theorem,
we obtain
\begin{align}
M^{(t)}[Z_s] = P^{\dagger}_0[Z_s Y(t-s,X_s)] \cdot \cbra{ 1-\frac{s}{t} }^{\frac{1}{\alpha }-1}
\to P^{\dagger}_0[Z_s]
\label{eq: meander M and DP}
\end{align}
as $ t \to \infty $.
This completes the proof.
\end{proof}
\subsection{Convergence of the meander weighed by a multiplicative functional}
Let $ (\ensuremath{\mathcal{E}}_t:t \ge 0) $ be an $ (\ensuremath{\mathcal{F}}_t) $-adapted process
which satisfies $ 0 \le \ensuremath{\mathcal{E}}_t \le 1 $
and enjoys the multiplicativity property:
\begin{align}
\ensuremath{\mathcal{E}}_{t+s} = \ensuremath{\mathcal{E}}_t \cdot (\ensuremath{\mathcal{E}}_s \circ \theta_t) .
\label{}
\end{align}
Such a process is called a {\em multiplicative functional};
see, e.g., \cite{MR0264757}.
Then it necessarily follows that
$ t \mapsto \ensuremath{\mathcal{E}}_t $ is non-increasing.
For later use,
we need the following result
which asserts that the convergence of the meander to the $ h $-path process
is still valid with an extra weighing by a multiplicative functional.
\begin{Thm} \label{thm: Penal P2 meander}
\begin{align}
\ensuremath{\mathcal{E}}_t \cdot M^{(t)}
\ \stackrel{t \to \infty }{\longrightarrow} \
\ensuremath{\mathcal{E}}_{\infty } \cdot P^{\dagger}_0
\qquad \text{along} \
(\ensuremath{\mathcal{F}}_s) .
\label{}
\end{align}
\end{Thm}
To prove Theorem \ref{thm: Penal P2 meander}, we need the following two lemmas.
\begin{Lem} \label{lem: conv in law}
For any $ x \in \ensuremath{\mathbb{R}} $, it holds that
\begin{align}
\begin{split}
& \cbra{ (X_t:t \ge 0),(\lambda^{-1/\alpha } X_{\lambda t}:t \ge 0) }
\ \text{under} \ P^{\dagger}_x
\\
\stackrel{{\rm law}}{\longrightarrow}&
\cbra{ (X_t:t \ge 0),(\tilde{X}_t:t \ge 0) }
\ \text{under} \ P^{\dagger}_x \otimes \tilde{P}
\end{split}
\label{eq: conv of pair in law}
\end{align}
as $ \lambda \to \infty $
where
\begin{align}
\tilde{P} =
\begin{cases}
P^{\dagger}_0 \ & \text{if} \ 1<\alpha <2 \ \text{or if} \ x=0,
\\
P^{\rm 3B,+}_0 \ & \text{if} \ \alpha =2 \ \text{and} \ x>0,
\\
P^{\rm 3B,-}_0 \ & \text{if} \ \alpha =2 \ \text{and} \ x<0.
\end{cases}
\label{}
\end{align}
\end{Lem}
\begin{proof}
We prove the claim only in the case $ 1<\alpha <2 $;
in fact, almost the same argument works in the other cases.
Set $ X^{(\lambda)}_t = \lambda^{-1/\alpha } X_{\lambda t} $.
Let us apply the convergence theorem of \cite[Theorem VI.16]{MR762984}.
First, let $ t \ge 0 $ be fixed
and let $ f:\ensuremath{\mathbb{R}} \to \ensuremath{\mathbb{R}} $ be a continuous function such that
$ \lim_{|x| \to \infty } f(x) = 0 $.
Then we have
\begin{align}
\lim_{\lambda \to \infty } P^{\dagger}_x[f(X^{(\lambda)}_t)]
=
\lim_{\lambda \to \infty } P^{\dagger}_{\lambda^{-1/\alpha } x}[f(X_t)]
=
P^{\dagger}_0[f(X_t)] .
\label{}
\end{align}
In fact, the first identity follows from the scaling property
and the second follows from the Feller property of the $ h $-path process,
which is proved in \cite{Y}.
Hence we obtain
\begin{align}
X^{(\lambda)}_t \ \text{under} \ P^{\dagger}_x
\quad \stackrel{{\rm law}}{\longrightarrow} \quad
X_t \ \text{under} \ P^{\dagger}_0
\label{}
\end{align}
as $ \lambda \to \infty $.
By a standard argument involving the Markov property, we see that
the convergence \eqref{eq: conv of pair in law} holds
in the sense of finite dimensional distributions.
Second, for any sequence $ \{ \lambda_n \} $ with $ \lambda_n \to \infty $,
let us check the {\em Aldous condition}:
For a sequence of positive constants $ \{ \delta_n \} $ converging to zero
and for a bounded sequence of stopping times $ \{ \rho_n \} $,
\begin{align}
\abra{ X_{\rho_n+\delta_n} - X_{\rho_n} }
+
\abra{ X^{(\lambda_n)}_{\rho_n+\delta_n} - X^{(\lambda_n)}_{\rho_n} }
\stackrel{n \to \infty }{\longrightarrow}
0
\qquad \text{in $ P^{\dagger}_x $-probability}.
\label{eq: Aldous}
\end{align}
The convergence \eqref{eq: Aldous} is equivalent to
\begin{align}
X^{(\lambda_n)}_{\rho_n+\delta_n}
- X^{(\lambda_n)}_{\rho_n}
\stackrel{n \to \infty }{\longrightarrow}
0
\qquad \text{in $ P^{\dagger}_x $-probability}.
\label{eq: Aldous2}
\end{align}
To prove \eqref{eq: Aldous2}, it suffices to prove that
\begin{align}
P^{\dagger}_x \ebra{ \abra{ X^{(\lambda_n)}_{\rho_n+\delta_n}
- X^{(\lambda_n)}_{\rho_n} } \wedge 1 }
\stackrel{n \to \infty }{\longrightarrow}
0 .
\label{eq: Aldous3}
\end{align}
By the strong Markov property and by the scaling property, we have
\begin{align}
P^{\dagger}_x \ebra{ \abra{ X^{(\lambda_n)}_{\rho_n+\delta_n}
- X^{(\lambda_n)}_{\rho_n} } \wedge 1 }
=
P^{\dagger}_x \ebra{
P^{\dagger}_{\lambda^{-1/\alpha } X_{\rho_n}}
\ebra{ \abra{ X_{\delta_n} - X_0 } \wedge 1 } } .
\label{}
\end{align}
Hence we can easily obtain the convergence \eqref{eq: Aldous3}
by the Feller property of the $ h $-path process.
\end{proof}
\begin{Lem} \label{lem: Penal P2 meander}
For any $ x \neq 0 $, it holds that
\begin{align}
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>t]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\to P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }]
\qquad \text{as} \
t \to \infty .
\label{}
\end{align}
\end{Lem}
\begin{proof}[Proof of Lemma \ref{lem: Penal P2 meander}]
For $ t>s>0 $, we have $ \ensuremath{\mathcal{E}}_t \le \ensuremath{\mathcal{E}}_s $,
and hence we have
\begin{align}
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>t]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\le&
\frac{P_x[\ensuremath{\mathcal{E}}_s;T_{\{ 0 \}}>t]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\\
=&
P^{\dagger}_x \ebra{ \ensuremath{\mathcal{E}}_s \frac{P_{X_s}(T_{\{ 0 \}}>t-s)}{h(X_s) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)} }
\\
=&
P^{\dagger}_x [ \ensuremath{\mathcal{E}}_s Y(t-s,X_s) ] \cdot \cbra{1-\frac{s}{t}}^{\frac{1}{\alpha }-1} .
\label{}
\end{align}
By Lemma \ref{lem: Ytx} and by the bounded convergence theorem,
we have
\begin{align}
\limsup_{t \to \infty }
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>t]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\le P^{\dagger}_x[\ensuremath{\mathcal{E}}_s] .
\label{}
\end{align}
Since $ P^{\dagger}_x[\ensuremath{\mathcal{E}}_s] \to P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }] $ as $ s \to \infty $,
we obtain the upper estimate:
\begin{align}
\limsup_{t \to \infty }
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>t]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\le P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }] .
\label{}
\end{align}
By Lemma \ref{lem: conv in law}, we have
\begin{align}
\cbra{ \ensuremath{\mathcal{E}}_t,t^{-1/\alpha } |X_t| }
\ \text{under} \ P^{\dagger}_x
\stackrel{{\rm law}}{\longrightarrow}
(\ensuremath{\mathcal{E}}_{\infty },|\tilde{X}_1|)
\ \text{under} \ P^{\dagger}_x \otimes P^{\dagger}_0 .
\label{}
\end{align}
Hence, by Fatou's lemma, we have
\begin{align}
\liminf_{t \to \infty }
\frac{P^{\dagger}_x[\ensuremath{\mathcal{E}}_t / h(X_t)]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
=&
\liminf_{t \to \infty }
\frac{P^{\dagger}_x[\ensuremath{\mathcal{E}}_t / h(t^{-1/\alpha } |X_t|)]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>1)}
\\
\ge&
\frac{P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }] P^{\dagger}_0[1/h(|X_1|)]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>1)}
=
P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }] .
\label{}
\end{align}
Thus we obtain the lower estimate:
\begin{align}
\liminf_{t \to \infty }
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>t]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
= \liminf_{t \to \infty }
\frac{P^{\dagger}_x[\ensuremath{\mathcal{E}}_t / h(X_t)]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ge P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }] .
\label{}
\end{align}
Therefore the proof is now completed.
\end{proof}
Now we prove Theorem \ref{thm: Penal P2 meander}.
\begin{proof}[Proof of Theorem \ref{thm: Penal P2 meander}]
For a bounded $ \ensuremath{\mathcal{F}}_s $-measurable functional $ Z_s $ and for $ t>s>0 $, we have
\begin{align}
M^{(t)}[Z_s \ensuremath{\mathcal{E}}_t]
= P^{\dagger}_0 \ebra{ Z_s \ensuremath{\mathcal{E}}_s \frac{P_{X_s}[\ensuremath{\mathcal{E}}_{t-s};T_{\{ 0 \}}>t-s]}{h(X_s) \ensuremath{\mbox{{\boldmath $n$}}}(R>t-s)} }
\cdot \cbra{ 1-\frac{s}{t} }^{\frac{1}{\alpha }-1} .
\label{}
\end{align}
Note that
\begin{align}
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>r]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>r)}
\le
\frac{P_x(T_{\{ 0 \}}>r)}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>r)}
= Y(r,x) ,
\label{}
\end{align}
which is uniformly bounded in $ r>0 $ and $ x \neq 0 $ by Lemma \ref{lem: Ytx}.
Note also that
\begin{align}
\frac{P_x[\ensuremath{\mathcal{E}}_t;T_{\{ 0 \}}>r]}{h(x) \ensuremath{\mbox{{\boldmath $n$}}}(R>r)}
\ \stackrel{r \to \infty }{\longrightarrow} \
P^{\dagger}_x[\ensuremath{\mathcal{E}}_{\infty }]
, \qquad x \neq 0
\label{}
\end{align}
by Lemma \ref{lem: Penal P2 meander}.
Hence we apply bounded convergence theorem and obtain
\begin{align}
M^{(t)}[Z_s \ensuremath{\mathcal{E}}_t]
\ \stackrel{t \to \infty }{\longrightarrow} \
P^{\dagger}_0 \ebra{ Z_s \ensuremath{\mathcal{E}}_s P^{\dagger}_{X_s}[\ensuremath{\mathcal{E}}_{\infty }] }
=
P^{\dagger}_0 \ebra{ Z_s \ensuremath{\mathcal{E}}_{\infty } } .
\label{}
\end{align}
This completes the proof.
\end{proof}
\section{General observations on the $ \sigma $-finite measure
unifying our penalisation problems
and the martingale generator}
\label{sec: univ}
Following \cite{NRY} and \cite{NRY2}, we make general observations
on the measure $ \ensuremath{\mathscr{P}} $.
\subsection{The $ \sigma $-finite measure unifying our penalisation problems}
Recall the definition of $ \ensuremath{\mathscr{P}} $:
\begin{align}
\ensuremath{\mathscr{P}}
= \int_0^{\infty } P_0[\d L_u] Q^{(u)} \bullet P^{\dagger}_0
\label{eq: cP}
\end{align}
where
\begin{align}
P_0[\d L_u] = \frac{\Gamma(1/\alpha )}{\alpha \pi} \frac{\d u}{u^{1/\alpha }}
\label{}
\end{align}
and where
$ P^{\dagger}_0 $ is defined by
\begin{align}
P^{\dagger}_0 |_{\ensuremath{\mathcal{F}}_t} = h(X_t) \cdot \ensuremath{\mbox{{\boldmath $n$}}} |_{\ensuremath{\mathcal{F}}_t}
, \qquad t>0 .
\label{}
\end{align}
Denote
\begin{align}
g = \sup \{ t \ge 0 : X_t=0 \} .
\label{}
\end{align}
\begin{Thm} \label{thm: sP}
The following statements hold:
\\ \quad
{\rm (i)}
$ \ensuremath{\mathscr{P}}(g \in \d u) = P_0[\d L_u] $;
\\ \quad
{\rm (ii)}
$ \ensuremath{\mathscr{P}} $ is a $ \sigma $-finite measure on $ \ensuremath{\mathcal{F}}_{\infty } $;
\\ \quad
{\rm (iii)}
$ \ensuremath{\mathscr{P}} $ is singular with respect to $ P_0 $ on $ \ensuremath{\mathcal{F}}_{\infty } $;
\\ \quad
{\rm (iv)}
For each $ t>0 $ and for $ A \in \ensuremath{\mathcal{F}}_t $,
one has
\begin{align}
\ensuremath{\mathscr{P}}(A) =& 0 \qquad \text{if} \ P_0(A)=0,
\label{eq: sPA = 0}
\\
\ensuremath{\mathscr{P}}(A) =& \infty \qquad \text{if} \ P_0(A)>0.
\label{eq: sPA = infty}
\end{align}
\end{Thm}
\begin{Rem}
For each $ t>0 $, \eqref{eq: sPA = 0} asserts that
$ \ensuremath{\mathscr{P}} $ is equivalent to $ P_0 $ on $ \ensuremath{\mathcal{F}}_t $,
but \eqref{eq: sPA = infty} asserts that
$ \ensuremath{\mathscr{P}} $ is {\em never} $ \sigma $-finite on $ \ensuremath{\mathcal{F}}_t $.
We insist that,
since $ \ensuremath{\mathscr{P}} $ is not $ \sigma $-finite on $ \ensuremath{\mathcal{F}}_t $,
\eqref{eq: sPA = 0} does not imply the existence
of an $ \ensuremath{\mathcal{F}}_t $-measurable Radon--Nikodym density.
\end{Rem}
\begin{proof}[Proof of Theorem \ref{thm: sP}]
{\rm (i)}
Since $ P^{\dagger}_0 $ is locally equivalent to $ \ensuremath{\mbox{{\boldmath $n$}}} $,
we see that
$ P^{\dagger}_0(X_s \neq 0 \ \text{for any} \ s \le t) = 1 $
for any $ t>0 $.
This shows that
$ P^{\dagger}_0(X_t \neq 0 \ \text{for any} \ t>0) = 1 $.
Hence we see, by the definition \eqref{eq: cP} of $ \ensuremath{\mathscr{P}} $, that
$ g=u $ under the measure $ Q^{(u)} \bullet P^{\dagger}_0 $.
Thus we obtain the desired result.
{\rm (ii)}
It is obvious by (i) that
$ \ensuremath{\mathscr{P}}(g<u) $ is finite for each $ u>0 $.
{\rm (iii)}
On one hand, we have $ \ensuremath{\mathscr{P}}(g=\infty )=0 $.
On the other hand,
since the origin for $ (X_t,P_0) $ is recurrent, we have $ P_0(g<\infty )=0 $.
This implies that $ \ensuremath{\mathscr{P}} $ is singular to $ P_0 $ on $ \ensuremath{\mathcal{F}}_{\infty } $.
{\rm (iv)}
Let $ A \in \ensuremath{\mathcal{F}}_t $
and suppose that $ P_0(A)=0 $.
For $ T>t $, we have
\begin{align}
\int_0^T P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 }(A)
=
\int_0^t P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 }(A)
+ \int_t^T P_0[\d L_u] Q^{(u)}(A) .
\label{}
\end{align}
For $ 0<u<t $, we have
$ (Q^{(u)} \bullet P^{\dagger}_0)(A)
= (Q^{(u)} \bullet \ensuremath{\mbox{{\boldmath $n$}}})[1_A h(X_t)] $,
and hence we obtain
\begin{align}
\int_0^t P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 }(A) = P_0 \ebra{ 1_A h(X_t) } = 0 .
\label{}
\end{align}
For $ t<u<T $, we have
\begin{align}
\int_t^T P_0[\d L_u] Q^{(u)}(A) = P_0 \ebra{ 1_A (L_T-L_t) } = 0 .
\label{}
\end{align}
Letting $ T \to \infty $, we obtain $ \int_t^{\infty } P_0[\d L_u] Q^{(u)}(A) = 0 $.
Therefore we obtain $ \ensuremath{\mathscr{P}}(A)=0 $.
Conversely,
let $ A \in \ensuremath{\mathcal{F}}_t $
and suppose that $ P_0(A)>0 $.
Then
\begin{align}
\ensuremath{\mathscr{P}}(A) \ge \int_t^{\infty } P_0[\d L_u] Q^{(u)}(A) = P_0 \ebra{ 1_A (L_{\infty }-L_t) } .
\label{}
\end{align}
Note that the last quantity is $ \infty $
since $ P_0(L_{\infty }=\infty )=1 $.
Hence we obtain $ \ensuremath{\mathscr{P}}(A)=\infty $.
\end{proof}
\subsection{The martingale generator}
\begin{Thm} \label{thm: mart op}
For each $ x \in \ensuremath{\mathbb{R}} $, $ t \ge 0 $
and for each non-negative measurable or $ \ensuremath{\mathscr{P}}_x $-integrable functional $ F $,
there exists a unique $ \ensuremath{\mathcal{F}}_t $-measurable functional $ M_{t,x}(F) $ (possibly taking infinite values) such that
\begin{align}
(F \cdot \ensuremath{\mathscr{P}}_x)|_{\ensuremath{\mathcal{F}}_t} = M_{t,x}(F) \cdot P_x|_{\ensuremath{\mathcal{F}}_t} .
\label{}
\end{align}
In particular, if $ F $ is $ \ensuremath{\mathscr{P}}_x $-integrable,
then the process $ (M_{t,x}(F):t \ge 0) $ is an $ (\ensuremath{\mathcal{F}}_t,P_x) $-martingale
such that
\begin{align}
M_{0,x}(F)=\ensuremath{\mathscr{P}}_x[F]
\label{}
\end{align}
and that
\begin{align}
\lim_{t \to \infty } M_{t,x}(F) = 0
\qquad \text{$ P_x $-almost surely.}
\label{}
\end{align}
\end{Thm}
In the case $ x=0 $, we write $ M_t(F) $ for $ M_{t,0}(F) $.
For each $ x \in \ensuremath{\mathbb{R}} $,
we call the operator $ L^1(\ensuremath{\mathscr{P}}_x) \ni F \mapsto (M_{t,x}(F):t \ge 0) $
the {\em martingale generator}.
\begin{proof
It is obvious that the uniqueness holds in the sense that,
if $ F=G $ $ \ensuremath{\mathscr{P}} $-almost everywhere, then $ M_{t,x}(F)=M_{t,x}(G) $ $ P_0 $-almost surely.
Without loss of generality,
we may suppose that $ x=0 $ and that $ F $ is non-negative.
Let $ n $ be a positive integer and set $ F_n = F \cdot 1_{\{ g<n \}} $.
By (ii) and (iv) of Theorem \ref{thm: sP},
we see that $ (F_n \cdot \ensuremath{\mathscr{P}})|_{\ensuremath{\mathcal{F}}_t} $ is
a finite measure and is absolutely continuous with respect to $ P_0|_{\ensuremath{\mathcal{F}}_t} $.
Hence we may apply the Radon--Nikodym theorem
to obtain the desired functional $ M_t(F_n) $ as the Radon--Nikodym derivative.
Hence the desired functional $ M_t(F) $ is obtained
as the increasing limit $ \lim_{n \to \infty } M_t(F_n) $
by the monotone convergence theorem.
Suppose that $ F $ is $ \ensuremath{\mathscr{P}} $-integrable.
For $ s \le t $, we have
\begin{align}
P_0[Z_s M_t(F)]
= \ensuremath{\mathscr{P}}[Z_s F]
= P_0[Z_s M_s(F)] .
\label{}
\end{align}
Hence $ (M_t(F):t \ge 0) $ is a $ (\ensuremath{\mathcal{F}}_t,P_0) $-martingale.
It is obvious that
$ M_0(F)=\ensuremath{\mathscr{P}}[F] $.
Since $ (M_t(F):t \ge 0) $ is a non-negative martingale,
$ M_t(F) $ converges $ P_0 $-almost surely
to a non-negative $ \ensuremath{\mathcal{F}}_{\infty } $-measurable functional $ M_{\infty }(F) $.
For $ 0<s<t \le \infty $, set $ A(s,t) = \{ g_t \ge s \} \in \ensuremath{\mathcal{F}}_t $.
Note that $ P_0(A(s,\infty )) = P_0(g \ge s) = 1 $.
Applying Fatou's lemma
and then applying the dominated convergence theorem,
we obtain
\begin{align}
P_0[M_{\infty }(F)]
= P_0[1_{A(s,\infty )} M_{\infty }(F)]
\le&
\liminf_{t \to \infty } P_0[1_{A(s,t)} M_t(F)]
\\
=&
\liminf_{t \to \infty } \ensuremath{\mathscr{P}}[1_{A(s,t)} F]
= \ensuremath{\mathscr{P}}[1_{A(s,\infty )} F] .
\label{}
\end{align}
Since $ \ensuremath{\mathscr{P}}(g=\infty )=0 $, we have $ \lim_{s \to \infty } \ensuremath{\mathscr{P}}[1_{A(s,\infty )} F] = 0 $.
Hence we obtain $ P_0[M_{\infty }(F)]=0 $,
which implies that $ P_0(M_{\infty }(F)=0)=1 $.
Therefore the proof is completed.
\end{proof}
\section{Convergence lemmas}
\label{sec: imp lem}
Let $ 0<\gamma <1 $.
For integrable functions $ \psi_t(u) $
such that $ \psi_t(u) \to \exists \psi(u) $ as $ t \to \infty $,
we may expect that
\begin{align}
\int_0^t \cbra{1-\frac{u}{t}}^{\gamma -1} \psi_t(u) \d u
\to
\int_0^{\infty } \psi(u) \d u
\qquad \text{as} \ t \to \infty .
\label{eq: imp lem asymp}
\end{align}
We need this convergence for several functions $ \psi_t $
in order to solve our penalisation problems,
as we have seen roughly in \eqref{eq: convergence of P_0/nRt to sP}.
In fact, we shall see that
we must be careful in dealing with the convergence \eqref{eq: imp lem asymp}.
In this section
we give some sufficient conditions for the convergence \eqref{eq: imp lem asymp}
as well as a counterexample.
If $ \psi_t $'s satisfy
\begin{align}
\int_0^t \psi_t(u) \d u \to \int_0^{\infty } \psi(u) \d u
\qquad \text{as} \ t \to \infty ,
\label{eq: imp lem asymp2}
\end{align}
then the convergence \eqref{eq: imp lem asymp} is equivalent to
\begin{align}
I(\psi_t,t) \to 0
\qquad \text{as} \
t \to \infty
\label{}
\end{align}
where
\begin{align}
I(\psi,t) = \int_0^t \kbra{ \cbra{ 1-\frac{u}{t} }^{\gamma -1} -1 } \psi(u) \d u .
\label{}
\end{align}
First, we present the following counterexample.
\begin{Ex}
The convergence \eqref{eq: imp lem asymp} {\em fails} if
\begin{align}
\psi_t(u) \equiv \psi(u)
= \sum_{n=1}^{\infty } n^{\frac{2+\gamma }{1-\gamma }}
1_{\cbra{n-n^{-\frac{4-\gamma }{1-\gamma }},n}}(u) .
\label{}
\end{align}
\end{Ex}
\begin{proof}
$ \psi $ is integrable since
$ \int_0^{\infty } \psi(u) \d u = \sum_{n=1}^{\infty } n^{-2} < \infty $.
But $ \limsup_t I(\psi,t) = \infty $ because
\begin{align}
I(\psi,n)
\ge& n^{\frac{2+\gamma }{1-\gamma }} \cdot n^{1-\gamma }
\int_{n-n^{-\frac{4-\gamma }{1-\gamma }}}^n (n-u)^{\gamma -1} \d u
- n^{-2}
\\
=& n^{\frac{2+\gamma }{1-\gamma }} \cdot n^{1-\gamma }
\cdot \gamma ^{-1} n^{-\frac{(4-\gamma )\gamma }{1-\gamma }} - n^{-2}
\\
=& \gamma ^{-1} n^{3-2\gamma } - n^{-2}
\to \infty
\qquad \text{as} \
n \to \infty .
\label{}
\end{align}
This prevents the convergence \eqref{eq: imp lem asymp}.
\end{proof}
On the other hand, we give three sufficient conditions
for the convergence \eqref{eq: imp lem asymp};
the first one is rather theoretical,
but the second and third ones can be readily applied.
\begin{Lem}[Dominated convergence] \label{lem: imp lem1}
Suppose that $ \psi_t $'s are integrable functions
such that $ \int_0^{\infty } \psi_t(u) \d u \to \int_0^{\infty } \psi(u) \d u $
for some integrable function $ \psi $.
Suppose, in addition, that $ |\psi_t| \le \tilde{\psi}_t $
for some integrable function $ \tilde{\psi}_t $
such that $ \lim_{t \to \infty } I(\tilde{\psi}_t,t) = 0 $.
Then
\begin{align}
\int_0^t \cbra{1-\frac{u}{t}}^{\gamma -1} \psi_t(u) \d u
\to
\int_0^{\infty } \psi(u) \d u
\qquad \text{as} \ t \to \infty
\label{}
\end{align}
holds.
\end{Lem}
\begin{proof}
This is obvious
by $ |I(\psi_t,t)| \le I(|\psi_t|,t) \le I(\tilde{\psi}_t,t) \to 0 $ as $ t \to \infty $.
\end{proof}
\begin{Lem} \label{lem: imp lem3}
Suppose that $ \psi $ is a non-negative integrable function and satisfies
\begin{align}
\lim_{t \to \infty } \kbra{ t \sup_{u>t} \psi(u) } = 0 .
\label{eq: assump sup psi}
\end{align}
Then $ \lim_{t \to \infty } I(\psi,t) = 0 $.
\end{Lem}
\begin{proof}
Let $ 0<\ensuremath{\varepsilon}<1 $ be fixed.
We split $ I(\psi,t) $ into a sum $ I(\psi_1,t) + I(\psi_2,t) $
where
$ \psi_1 = \psi 1_{(\ensuremath{\varepsilon} t,\infty )} $
and
$ \psi_2 = \psi 1_{(0,\ensuremath{\varepsilon} t)} $.
By the definition of $ I(\psi_1,t) $ and changing variables to $ v=ut $, we have
\begin{align}
I(\psi_1,t)
=&
\int_{\ensuremath{\varepsilon} t}^t
\kbra{ \cbra{ 1-\frac{u}{t} }^{\gamma -1} -1 } \psi(u) \d u
\\
\le&
t \sup_{u>\ensuremath{\varepsilon} t} \psi(u)
\int_{\ensuremath{\varepsilon} t}^t
\kbra{ \cbra{ 1-\frac{u}{t} }^{\gamma -1} -1 } \frac{\d u}{t}
\\
=&
\frac{1}{\ensuremath{\varepsilon}} \kbra{ \ensuremath{\varepsilon} t \sup_{u>\ensuremath{\varepsilon} t} \psi(u) }
\int_{\ensuremath{\varepsilon}}^1 \kbra{ (1-v)^{\gamma -1} - 1 } \d v .
\label{}
\end{align}
By the assumption \eqref{eq: assump sup psi}, we obtain
$ \lim_{t \to \infty } I(\psi_1,t) = 0 $
for any fixed $ \ensuremath{\varepsilon}>0 $.
By the definition of $ I(\psi_2,t) $, we have
\begin{align}
I(\psi_2,t)
\le
\kbra{ \frac{1}{(1-\ensuremath{\varepsilon})^{1-\gamma }} -1 }
\int_0^{\infty } \psi(u) \d u .
\label{}
\end{align}
Hence we have $ \limsup_{t \to \infty } I(\psi_2,t) $ vanishes as $ \ensuremath{\varepsilon} \to 0+ $.
Now the proof is completed.
\end{proof}
\begin{Lem} \label{lem: imp lem2}
Suppose that $ \psi_t(u)=\psi_1(u) \psi_2(t-u) $
where $ \psi_1 $ is integrable
and $ \psi_2 $ is bounded measurable
with $ \lim_{u \to \infty } \psi_2(u) = \psi_2(\infty ) > 0 $.
Suppose, in addition, that the function
$ t \mapsto \int_0^t (t-u)^{\gamma -1} \psi_t(u) \d u $
is ultimately non-increasing as $ t $ increases.
Then $ \lim_{t \to \infty } I(\psi_t,t) = 0 $.
\end{Lem}
\begin{proof}
Taking the Laplace transform, we have
\begin{align}
\int_0^{\infty } \d t {\rm e}^{-qt} \int_0^t (t-u)^{\gamma -1} \psi_t(u) \d u
=&
\int_0^{\infty } {\rm e}^{-qu} \psi_1(u) \d u
\int_0^{\infty } {\rm e}^{-qt} t^{\gamma -1} \psi_2(t) \d t
\\
\sim&
\Gamma (\gamma ) q^{-\gamma } \psi_2(\infty ) \int_0^{\infty } \psi_1(u) \d u
\qquad \text{as} \
q \to 0+ .
\label{}
\end{align}
Hence we may apply the tauberian theorem.
By the monotonicity assumption, we obtain
\begin{align}
\int_0^t (t-u)^{\gamma -1} \psi_t(u) \d u
\sim
t^{\gamma -1} \psi_2(\infty ) \int_0^{\infty } \psi_1(u) \d u
\qquad \text{as} \ t \to \infty .
\label{}
\end{align}
On the other hand, we have
\begin{align}
\int_0^t \psi_t(u) \d u
= \int_0^t \psi_1(u) \psi_2(t-u) \d u
\to
\psi_2(\infty ) \int_0^{\infty } \psi_1(u) \d u
\qquad \text{as} \ t \to \infty .
\label{}
\end{align}
Therefore we obtain $ \lim_{t \to \infty } I(\psi_t,t) = 0 $.
\end{proof}
\section{Penalisation with a function of the local time at the origin}
\label{sec: LT penal}
\subsection{Results}
\begin{Thm} \label{thm: LT mart op}
Let $ f $ be a non-negative function on $ [0,\infty ) $. Then
it holds that
\begin{align}
M_t(f(L_{\infty }))
= h(X_t) f(L_t) + \int_{L_t}^{\infty } f(l) \d l
, \qquad t \ge 0 .
\label{eq: LT mart op}
\end{align}
Consequently, it holds that
\begin{align}
\ensuremath{\mathscr{P}}[f(L_{\infty })]
= M_0(f(L_{\infty }))
= \int_0^{\infty } f(l) \d l .
\label{}
\end{align}
\end{Thm}
\begin{Rem}
As an outcome of \eqref{eq: LT mart op},
we have established that its right hand side is a $ (P_0,\ensuremath{\mathcal{F}}_t) $-martingale,
a well-known fact for $ \alpha =2 $ (see \cite[Prop. VI.4.5]{MR1725357})
\end{Rem}
\begin{Thm} \label{thm: LT Penal}
Let $ f $ be a non-negative function on $ [0,\infty ) $ such that
\begin{align}
I(f) := \int_0^{\infty } f(l) \d l \in (0,\infty ) .
\label{eq: LT Penal assump}
\end{align}
Then it holds that
\begin{align}
\frac{f(L_t) \cdot P_0}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ \stackrel{t \to \infty }{\longrightarrow} \
f(L_{\infty }) \cdot \ensuremath{\mathscr{P}}
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $.}
\label{eq: LT conv}
\end{align}
Consequently, the penalisation with the weight functional $ \Gamma_t=f(L_t) $
is given as
\begin{align}
\frac{f(L_t) \cdot P_0}{P_0[f(L_t)]}
\ \stackrel{t \to \infty }{\longrightarrow} \
\frac{f(L_{\infty }) \cdot \ensuremath{\mathscr{P}}}{I(f)}
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $.}
\label{}
\end{align}
\end{Thm}
\subsection{Proofs}
\begin{proof}[Proof of Theorem \ref{thm: LT mart op}]
Let $ t>0 $ be fixed
and $ Z_t $ a non-negative $ \ensuremath{\mathcal{F}}_t $-measurable functional.
On the one hand,
since $ L_{\infty } = L_t $ on $ \{ g \le t \} $, we have
\begin{align}
\ensuremath{\mathscr{P}}[Z_t f(L_{\infty }) 1_{\{ g \le t \}}]
=& \ensuremath{\mathscr{P}}[Z_t f(L_t) 1_{\{ g \le t \}}]
\\
=& \int_0^t P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 } [Z_t f(L_t)]
\\
=& \int_0^t P_0[\d L_u] \cbra{ Q^{(u)} \bullet \ensuremath{\mbox{{\boldmath $n$}}} } [Z_t f(L_t) h(X_t)]
\\
=& \int_0^t \ensuremath{\mbox{{\boldmath $n$}}}(R>t-u) P_0[\d L_u] \cbra{ Q^{(u)} \bullet M^{(t-u)} } [Z_t f(L_t) h(X_t)]
\\
=& P_0 [Z_t f(L_t) h(X_t)] .
\label{}
\end{align}
On the other hand, we have
\begin{align}
\ensuremath{\mathscr{P}}[Z_t f(L_{\infty }) 1_{\{ g>t \}}]
=& \int_t^{\infty } P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 } [Z_t f(L_u)]
\\
=& \int_t^{\infty } P_0[\d L_u] Q^{(u)} [Z_t f(L_u)]
\\
=& P_0 \ebra{ Z_t \int_t^{\infty } f(L_u) \d L_u }
\\
=& P_0 \ebra{ Z_t \int_{L_t}^{\infty } f(l) \d l } .
\label{}
\end{align}
Hence we obtain
\begin{align}
\ensuremath{\mathscr{P}}[Z_t f(L_{\infty })]
= P_0 \ebra{ Z_t \kbra{ f(L_t) h(X_t) + \int_{L_t}^{\infty } f(l) \d l } } .
\label{}
\end{align}
Therefore we have completed the proof.
\end{proof}
\begin{proof}[Proof of Theorem \ref{thm: LT Penal}]
We need only to prove the first assertion that
\begin{align}
\frac{f(L_t) \cdot P_0}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ \stackrel{t \to \infty }{\longrightarrow} \
f(L_{\infty }) \cdot \ensuremath{\mathscr{P}}
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $.}
\label{}
\end{align}
Set $ \psi(u) = p_u(0) Q^{(u)}[f(L_u)] $.
We will prove in Lemma \ref{lem: Penal P1 psi} below
that $ \psi $ satisfies the assumption of Lemma \ref{lem: imp lem3}.
Now we apply Lemma \ref{lem: imp lem3} for the function $ \psi $
and we obtain
\begin{align}
\int_0^t & \cbra{1-\frac{u}{t}}^{\frac{1}{\alpha }-1} P_0[\d L_u] Q^{(u)}[f(L_u)]
\to
\int_0^{\infty } P_0[\d L_u] Q^{(u)}[f(L_u)]
\label{eq: penal conv P1-0}
\end{align}
as $ t \to \infty $.
Let $ s>0 $ be fixed
and let $ Z_s $ be a bounded $ \ensuremath{\mathcal{F}}_s $-measurable functional.
Then
\begin{align}
\int_0^t P_0[\d L_u] \cbra{ Q^{(u)} \bullet M^{(t-u)} } [Z_s f(L_u)]
\to
\int_0^{\infty } P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 } [Z_s f(L_u)]
\label{}
\end{align}
as $ t \to \infty $
by Lebesgue's convergence theorem.
Hence we can apply Lemma \ref{lem: imp lem1}, and we obtain
\begin{align}
\frac{P_0[Z_s f(L_t)]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
=& \int_0^t \cbra{1-\frac{u}{t}}^{\frac{1}{\alpha }-1}
P_0[\d L_u] \cbra{ Q^{(u)} \bullet M^{(t-u)} } [Z_s f(L_u)]
\\
\to&
\int_0^{\infty } P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 } [Z_s f(L_u)]
\qquad \text{(as $ t \to \infty $)}
\label{eq: penal conv P1}
\\
=& \ensuremath{\mathscr{P}}[Z_s f(L_{\infty })]
\end{align}
as $ t \to \infty $.
This completes the proof.
\end{proof}
\begin{Lem} \label{lem: Penal P1 psi}
Set
\begin{align}
\psi(u) = p_u(0) Q^{(u)}[f(L_u)] .
\label{eq: Penal P1 psi}
\end{align}
Then the function $ \psi(u) $ is continuous and $ u \psi(u) \to 0 $ as $ u \to \infty $.
In particular, the function $ \psi $ satisfies
the assumption \eqref{eq: assump sup psi} of Lemma \ref{lem: imp lem3}.
\end{Lem}
\begin{proof}
For any non-negative Borel function $ \phi $, we have
\begin{align}
\int_0^{\infty } \phi(u) \psi(u) \d u
=& P_0 \ebra{ \int_0^{\infty } \phi(u) f(L_u) \d L_u }
\\
=& \int_0^{\infty } P_0[\phi(\tau_s)] f(s) \d s
\\
=& \int_0^{\infty } P_0[\phi(s^{1/\beta } \tau_1)] f(s) \d s
\intertext{
where $ \beta = 1 - 1/\alpha $.
If we denote $ \rho^{(\beta )}(v) = P_0(\tau_1 \in \d v) / \d v $, we have }
=& \int_0^{\infty } \d s f(s) \int_0^{\infty } \phi(s^{1/\beta } v) \rho^{(\beta )}(v) \d v
\\
=& \int_0^{\infty } \d u \phi(u) \int_0^{\infty } s^{-1/\beta }
\rho^{(\beta )}(s^{-1/\beta } u) f(s) \d s .
\label{}
\end{align}
Hence we obtain
\begin{align}
\psi(u) =
\int_0^{\infty } s^{-1/\beta } \rho^{(\beta )}(s^{-1/\beta } u) f(s) \d s .
\label{}
\end{align}
Since the function $ \rho^{(\beta )}(v) $ is unimodal (see, e.g., Sato \cite{MR1739520}),
we see that
$ v \rho^{(\beta )}(v) $ is bounded in $ v>0 $
and that $ v \rho^{(\beta )}(v) \to 0 $ as $ v \to \infty $.
Therefore, by the assumption that $ \int_0^{\infty } f(s) \d s < \infty $,
we obtain the desired result.
\end{proof}
\begin{Rem}
In the Brownian case $ \alpha =2 $,
the corresponding $ \beta $ equals $ 1/2 $ and
\begin{align}
\rho^{(1/2)}(v) = \frac{1}{2 \sqrt{\pi v^3}} {\rm e}^{-\frac{1}{4v}} .
\label{}
\end{align}
\end{Rem}
\section{Feynman--Kac penalisations}
\label{sec: FK penal}
\subsection{Results}
Recall that our Feynman--Kac penalisation is the penalisation with the weight functional
\begin{align}
\ensuremath{\mathcal{E}}^V_t = \exp \kbra{ - \int L(t,x) V(\d x) }
, \qquad t \ge 0
\label{}
\end{align}
for a non-negative measure $ V(\d x) $ on $ \ensuremath{\mathbb{R}} $.
\begin{Thm} \label{thm: FK Penal}
Let $ V $ be a non-negative measure on $ \ensuremath{\mathbb{R}} $ such that
\begin{align}
0 < \int (1+|y|^{\alpha -1}) V(\d y) < \infty .
\label{eq: FK Penal assump}
\end{align}
Let $ x \in \ensuremath{\mathbb{R}} $. Then it holds that
\begin{align}
0 < \ensuremath{\mathscr{P}}_x[\ensuremath{\mathcal{E}}^V_{\infty }] < \infty
\label{}
\end{align}
and that
\begin{align}
\frac{( \ensuremath{\mathcal{E}}^V_t 1_{\{ T_{\{ 0 \}} > t \}} ) \cdot P_x}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ \stackrel{t \to \infty }{\longrightarrow}& \
( \ensuremath{\mathcal{E}}^V_{\infty } 1_{\{ T_{\{ 0 \}} = \infty \}} ) \cdot \ensuremath{\mathscr{P}}_x
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $,}
\label{eq: FK conv 1}
\\
\frac{( \ensuremath{\mathcal{E}}^V_t 1_{\{ T_{\{ 0 \}} \le t \}} ) \cdot P_x}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ \stackrel{t \to \infty }{\longrightarrow}& \
( \ensuremath{\mathcal{E}}^V_{\infty } 1_{\{ T_{\{ 0 \}} < \infty \}} ) \cdot \ensuremath{\mathscr{P}}_x
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $}
\label{eq: FK conv 2}
\intertext{and}
\frac{\ensuremath{\mathcal{E}}^V_t \cdot P_x}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ \stackrel{t \to \infty }{\longrightarrow}& \
\ensuremath{\mathcal{E}}^V_{\infty } \cdot \ensuremath{\mathscr{P}}_x
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $.}
\label{eq: FK conv}
\end{align}
\end{Thm}
\begin{Cor}
Let $ V $ be a non-negative measure on $ \ensuremath{\mathbb{R}} $ such that \eqref{eq: FK Penal assump} holds.
Then the penalisation with the weight functional
$ \Gamma_t = \ensuremath{\mathcal{E}}^V_t $ is given as
\begin{align}
\frac{\ensuremath{\mathcal{E}}^V_t \cdot P_x}{P_x[\ensuremath{\mathcal{E}}^V_t]}
\ \stackrel{t \to \infty }{\longrightarrow} \
\frac{\ensuremath{\mathcal{E}}^V_{\infty } \cdot \ensuremath{\mathscr{P}}_x}{\ensuremath{\mathscr{P}}_x[\ensuremath{\mathcal{E}}^V_{\infty }]}
\qquad \text{along $ (\ensuremath{\mathcal{F}}_s) $.}
\label{eq: FK penal}
\end{align}
\end{Cor}
\begin{Thm} \label{thm: FK mart op}
Let $ V $ be a non-negative measure on $ \ensuremath{\mathbb{R}} $ such that \eqref{eq: FK Penal assump} holds.
Set
\begin{align}
C_V = \ensuremath{\mathscr{P}}[\ensuremath{\mathcal{E}}^V_{\infty }] .
\label{}
\end{align}
Let $ x \in \ensuremath{\mathbb{R}} $.
Then it holds that
\begin{align}
\varphi^1_V(x)
:=& \lim_{t \to \infty } \frac{P_x[\ensuremath{\mathcal{E}}^V_t;T_{\{ 0 \}}>t]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
= \ensuremath{\mathscr{P}}_x[\ensuremath{\mathcal{E}}^V_{\infty };T_{\{ 0 \}}=\infty ]
= h(x) P^{\dagger}_x[\ensuremath{\mathcal{E}}^V_{\infty }] ,
\label{eq: FK func 1}
\\
\varphi^2_V(x)
:=& \lim_{t \to \infty } \frac{P_x[\ensuremath{\mathcal{E}}^V_t;T_{\{ 0 \}} \le t]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
= \ensuremath{\mathscr{P}}_x[\ensuremath{\mathcal{E}}^V_{\infty };T_{\{ 0 \}}<\infty ]
= C_V P_x[\ensuremath{\mathcal{E}}^V_{T_{\{ 0 \}}}]
\label{eq: FK func 2}
\intertext{and}
\varphi_V(x)
:=& \lim_{t \to \infty } \frac{P_x[\ensuremath{\mathcal{E}}^V_t]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
= \ensuremath{\mathscr{P}}_x[\ensuremath{\mathcal{E}}^V_{\infty }]
= h(x) P^{\dagger}_x[\ensuremath{\mathcal{E}}^V_{\infty }] + C_V P_x[\ensuremath{\mathcal{E}}^V_{T_{\{ 0 \}}}]
\\
\equiv &
\varphi^1_V(x) + \varphi^2_V(x) .
\label{eq: FK func}
\end{align}
Moreover, for $ t \ge 0 $, it holds that
\begin{align}
M_{t,x}(\ensuremath{\mathcal{E}}^V_{\infty } 1_{\{ T_{\{ 0 \}} = \infty \}})
=& \varphi^1_V(X_t) 1_{\{ T_{\{ 0 \}} > t \}} \ensuremath{\mathcal{E}}^V_t ,
\label{eq: FK mart op 1}
\\
M_{t,x}(\ensuremath{\mathcal{E}}^V_{\infty } 1_{\{ T_{\{ 0 \}} < \infty \}})
=& \kbra{
\varphi^1_V(X_t) 1_{\{ T_{\{ 0 \}} \le t \}}
+ \varphi^2_V(X_t)
} \ensuremath{\mathcal{E}}^V_t ,
\label{eq: FK mart op 2}
\intertext{and}
M_{t,x}(\ensuremath{\mathcal{E}}^V_{\infty })
=& \varphi_V(X_t) \ensuremath{\mathcal{E}}^V_t .
\label{eq: FK mart op}
\end{align}
\end{Thm}
We divide the proofs of Theorems \ref{thm: FK Penal} and \ref{thm: FK mart op}
into several steps in the following subsections.
\begin{Rem}\footnote{
Tildes in this remark have nothing to do with our previous notation's
for independence.}
For the Feynman--Kac penalisations (Theorems \ref{thm: FK Penal} and \ref{thm: FK mart op})
in the brownian case,
Roynette--Vallois--Yor (\cite{MR2261065},\cite{MR2229621}, \cite{MR2253307})
have given more characterisations of the limit measure
than the contents of Theorem \ref{thm: FK mart op}.
For convenience, we consider the Wiener measures $ (W_x:x \in \ensuremath{\mathbb{R}}) $
normalized with the weight functional
\begin{align}
\tilde{\ensuremath{\mathcal{E}}}^V_t = \exp \kbra{ - \frac{1}{2} \int L(t,x) V(\d x) } .
\label{}
\end{align}
For each $ x \in \ensuremath{\mathbb{R}} $, let $ \ensuremath{\mathscr{W}}_x $ denote the law of $ (x+X_t:t \ge 0) $ under $ \ensuremath{\mathscr{W}} $.
Then the function
\begin{align}
\tilde{\varphi}_V(x)
= \lim_{t \to \infty } \sqrt{t} W_x[\tilde{\ensuremath{\mathcal{E}}}^V_t]
= \ensuremath{\mathscr{W}}_x[\tilde{\ensuremath{\mathcal{E}}}^V_{\infty }]
\label{}
\end{align}
is the unique solution of the Sturm--Liouville differential equation
\begin{align}
\d \tilde{\varphi}'_V(x) = \tilde{\varphi}_V(x) V(\d x)
\label{eq: Sturm--Liouville}
\end{align}
subject to the boundary conditions
\begin{align}
\lim_{x \to \infty } \tilde{\varphi}_V'(x) = \sqrt{\frac{2}{\pi}}
\qquad \text{and} \qquad
\lim_{x \to - \infty } \tilde{\varphi}_V'(x) = - \sqrt{\frac{2}{\pi}} .
\label{}
\end{align}
Moreover, the limit measure $ W^V_x $ (instead of $ P^V_x $)
is the law of the unique solution of the stochastic differential equation
\begin{align}
\d X_t = \d B_t + \frac{\tilde{\varphi}'_V(X_t)}{\tilde{\varphi}_V(X_t)} \d t
, \qquad X_0 = x .
\label{eq: SDE}
\end{align}
We do not know how to develop these arguments in the stable L\'evy case,
for which it would be interesting to obtain counterparts of
\eqref{eq: Sturm--Liouville} and \eqref{eq: SDE}.
\end{Rem}
\subsection{Penalisation weighed by a general multiplicative functional}
In this subsection, we make a general study.
Let $ \ensuremath{\mathcal{E}} = (\ensuremath{\mathcal{E}}_t:t \ge 0) $ be an $ (\ensuremath{\mathcal{F}}_t) $-adapted process
which satisfies $ 0 \le \ensuremath{\mathcal{E}}_t \le 1 $, $ t \ge 0 $
and
is a multiplicative functional:
\begin{align}
\ensuremath{\mathcal{E}}_{t+s} = \ensuremath{\mathcal{E}}_t \cdot (\ensuremath{\mathcal{E}}_s \circ \theta_t)
, \qquad t,s \ge 0 .
\label{}
\end{align}
Note that the process $ t \mapsto \ensuremath{\mathcal{E}}_t $ is necessarily non-increasing;
in fact, $ \ensuremath{\mathcal{E}}_{t+s} = \ensuremath{\mathcal{E}}_t \cdot (\ensuremath{\mathcal{E}}_s \circ \theta_t) \le \ensuremath{\mathcal{E}}_t $
for any $ t,s \ge 0 $.
\begin{Thm} \label{thm: general FK penal}
Let $ x \in \ensuremath{\mathbb{R}} $ be fixed.
Suppose that
\begin{align}
\int_0^{\infty } P_0[\d L_u] Q^{(u)} [\ensuremath{\mathcal{E}}_u] < \infty .
\label{eq: assump for general FK penal}
\end{align}
Then it holds that
\begin{align}
\frac{\ensuremath{\mathcal{E}}_t \cdot P_x}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\ \stackrel{t \to \infty }{\longrightarrow} \
\ensuremath{\mathcal{E}}_{\infty } \cdot \ensuremath{\mathscr{P}}_x
\qquad \text{along} \
(\ensuremath{\mathcal{F}}_s) .
\label{}
\end{align}
\end{Thm}
\begin{proof}
We may suppose that $ x=0 $ without loss of generality.
By the multiplicativity property, we have
\begin{align}
\ensuremath{\mathscr{P}}[\ensuremath{\mathcal{E}}_{\infty }]
=& \int_0^{\infty } P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 } [\ensuremath{\mathcal{E}}_{\infty }]
\\
=& \kbra{ \int_0^{\infty } P_0[\d L_u] Q^{(u)}[\ensuremath{\mathcal{E}}_u] } P^{\dagger}_0[\ensuremath{\mathcal{E}}_{\infty }] .
\label{}
\end{align}
Set
\begin{align}
\psi_t(u) = p_u(0) (Q^{(u)} \bullet M^{(t-u)}) [\ensuremath{\mathcal{E}}_t] .
\label{}
\end{align}
Then we have $ \psi_t(u) = \psi_1(u) \psi_2(t-u) $
where
$ \psi_1(u) = p_u(0) Q^{(u)}[\ensuremath{\mathcal{E}}_u] $
and $ \psi_2(t) = M^{(t)}[\ensuremath{\mathcal{E}}_t] $.
Let us check that
all the assumptions of Lemma \ref{lem: imp lem2} are satisfied for $ \psi_t(u) $.
Note that
\begin{align}
\int_0^{\infty } \psi_1(u) \d u
= \int_0^{\infty } P_0[\d L_u] Q^{(u)} [\ensuremath{\mathcal{E}}_u]
\label{}
\end{align}
and it is finite by the assumption \eqref{eq: assump for general FK penal}.
Note also that $ \psi_2 $ is bounded and that
$ \lim_{t \to \infty } \psi_2(t) = P^{\dagger}_0[\ensuremath{\mathcal{E}}_{\infty }] $
by Theorem \ref{thm: Penal P2 meander}.
Recall the following identity:
\begin{align}
P_0[\ensuremath{\mathcal{E}}_t]
= \int_0^t \ensuremath{\mbox{{\boldmath $n$}}}(R>t-u) P_0[\d L_u] \cbra{ Q^{(u)} \bullet M^{(t-u)} } [\ensuremath{\mathcal{E}}_t] .
\label{}
\end{align}
Since the left-hand side is non-increasing as $ t $ increases,
we see that the function $ t \mapsto \int_0^t (t-u)^{\frac{1}{\alpha }-1} \psi_t(u) \d u $
is non-increasing as $ t $ increases.
Hence we have verified all the assumptions of Lemma \ref{lem: imp lem2},
and we obtain $ \lim_{t \to \infty } I(\psi_t,t) = 0 $.
The remainder of the proof follows from Lemma \ref{lem: imp lem1}.
\end{proof}
\subsection{Non-degeneracy condition}
Now we return to the case where $ \ensuremath{\mathcal{E}}_t = \ensuremath{\mathcal{E}}^V_t $.
By the multiplicativity of $ (\ensuremath{\mathcal{E}}^V_t) $, we have
\begin{align}
C_V
= \ensuremath{\mathscr{P}}[\ensuremath{\mathcal{E}}^V_{\infty }]
=& \int_0^{\infty } P_0[\d L_u] \cbra{ Q^{(u)} \bullet P^{\dagger}_0 } [\ensuremath{\mathcal{E}}^V_{\infty }]
\\
=& \int_0^{\infty } P_0[\d L_u] Q^{(u)}[\ensuremath{\mathcal{E}}^V_u] P^{\dagger}_0[\ensuremath{\mathcal{E}}^V_{\infty }]
\\
=& \kbra{ \int_0^{\infty } P_0[\ensuremath{\mathcal{E}}^V_{\tau_s}] \d s } P^{\dagger}_0[\ensuremath{\mathcal{E}}^V_{\infty }] .
\label{}
\end{align}
\begin{Thm} \label{thm: conti case}
The following assertions hold:
\\ \quad
{\rm (i)}
If $ V \neq 0 $, then $ \displaystyle \int_0^{\infty } P_0[\ensuremath{\mathcal{E}}^V_{\tau_s}] \d s < \infty $;
\\ \quad
{\rm (ii)}
If $ V((-\ensuremath{\varepsilon},\ensuremath{\varepsilon})) < \infty $ for some $ \ensuremath{\varepsilon}>0 $,
then $ \displaystyle \int_0^{\infty } P_0[\ensuremath{\mathcal{E}}^V_{\tau_s}] \d s > 0 $;
\\ \quad
{\rm (iii)}
If $ \displaystyle \int h(x) V(\d x) < \infty $,
then $ P^{\dagger}_0[\ensuremath{\mathcal{E}}^V_{\infty }] > 0 $;
\\ \quad
{\rm (iv)}
If $ \displaystyle 0 < \int \{ 1+h(x) \} V(\d x) < \infty $,
then $ 0<C_V<\infty $.
\end{Thm}
For the proof of Theorem \ref{thm: conti case}, we need the following
\begin{Lem} \label{lem: expect of LT}
The following statements hold:
\\ \quad {\rm (i)}
$ \ensuremath{\mbox{{\boldmath $n$}}}[L(R,x)] = 1 $ for any $ x \in \ensuremath{\mathbb{R}} \setminus \{ 0 \} $;
\\ \quad {\rm (ii)}
$ P^{\dagger}_0[L(t,x)] = h(x) P_x(T_{\{ 0 \}}<t) $ for any $ t \ge 0 $ and any $ x \in \ensuremath{\mathbb{R}} $;
\\ \quad {\rm (iii)}
$ P^{\dagger}_0[L(\infty ,x)] = h(x) $ for any $ x \in \ensuremath{\mathbb{R}} $.
\end{Lem}
Remark that $ \ensuremath{\mbox{{\boldmath $n$}}}[L(R,0)]=0 $;
in fact, $ L(R,0)=0 $ $ \ensuremath{\mbox{{\boldmath $n$}}} $-almost everywhere.
\begin{proof}
(i)
For a non-negative Borel function $ f $, we have
\begin{align}
\int \d x f(x) \ensuremath{\mbox{{\boldmath $n$}}}[L(R,x)]
= \int_0^{\infty } \ensuremath{\mbox{{\boldmath $n$}}}[f(X_t)] \d t
= \int \d x f(x) \int_0^{\infty } \rho(t,x) \d t
= \int \d x f(x) .
\label{}
\end{align}
Hence we obtain $ \ensuremath{\mbox{{\boldmath $n$}}}[L(R,x)] = 1 $ for almost every $ x \in \ensuremath{\mathbb{R}} $.
By the scaling property, we obtain the desired conclusion.
(ii)
Let $ t \ge 0 $ be fixed.
For a non-negative Borel function $ f $, we have
\begin{align}
\int \d x f(x) P^{\dagger}_0[L(t,x)]
=& \int_0^t P^{\dagger}_0[f(X_s)] \d s
= \int_0^t \ensuremath{\mbox{{\boldmath $n$}}}[f(X_s)h(X_s)] \d s
\\
=& \int \d x f(x) h(x) \int_0^t \rho(s,x) \d s
\\
=& \int \d x f(x) h(x) P_x(T_{\{ 0 \}}<t) .
\label{}
\end{align}
Hence we see that
$ P^{\dagger}_0[L(t,x)] = h(x) P_x(T_{\{ 0 \}}<t) $ for almost every $ x \in \ensuremath{\mathbb{R}} $.
Since $ t \mapsto P^{\dagger}_0[L(t,1)] $ is continuous by the monotone convergence theorem,
we see, by the scaling property,
that $ \ensuremath{\mathbb{R}} \setminus \{ 0 \} \ni x \mapsto P^{\dagger}_0[L(t,x)] $
is continuous.
Noting that $ L(t,0)=0 $ $ P^{\dagger}_0 $-almost surely,
we complete the proof.
(iii)
Letting $ t \to \infty $ in (ii),
we obtain
$ P^{\dagger}_0[L(\infty ,x)] = h(x) $
by the monotone convergence theorem.
\end{proof}
Now we prove Theorem \ref{thm: conti case}.
\begin{proof}[Proof of Theorem \ref{thm: conti case}]
Note that
\begin{align}
\ensuremath{\mathcal{E}}^V_{\tau_s}
= \exp \kbra{ - s V(\{ 0 \})
- \sum_{l \in D, \ l \le s} \int_{\{ x \neq 0 \}} L(R,x)[e_l] V(\d x) }
\label{}
\end{align}
where $ L(R,x)[e_l] $ is the local time at $ x $ of the excursion $ e_l $ up to its lifetime.
Hence we have
$ P_0[ \ensuremath{\mathcal{E}}^V_{\tau_s} ] = \exp \kbra{ - s K_V } $
where
\begin{align}
K_V = V(\{ 0 \}) + \ensuremath{\mbox{{\boldmath $n$}}} \ebra{ 1-\exp \kbra{ - \int_{\{ x \neq 0 \}} L(R,x) V(\d x) } } .
\label{}
\end{align}
Consequently we have
\begin{align}
\int_0^{\infty } P_0 \ebra{ \ensuremath{\mathcal{E}}_{\tau_s} } \d s = \frac{1}{K_V} .
\label{}
\end{align}
{\rm (i)}
If $ \int_0^{\infty } P_0[\ensuremath{\mathcal{E}}^V_{\tau_s}] \d s = \infty $,
then we have $ K_V=0 $, which implies that $ V=0 $.
Hence the assertion is proved by contraposition.
{\rm (ii)}
Suppose that $ V((-\ensuremath{\varepsilon},\ensuremath{\varepsilon}))<\infty $ for $ \ensuremath{\varepsilon}>0 $.
Then, by Lemma \ref{lem: expect of LT}, we have
\begin{align}
\ensuremath{\mbox{{\boldmath $n$}}} \ebra{ \int_{(-\ensuremath{\varepsilon},\ensuremath{\varepsilon})} L(R,x) V(\d x) } = \int_{(-\ensuremath{\varepsilon},\ensuremath{\varepsilon})} \ensuremath{\mbox{{\boldmath $n$}}}[ L(R,x) ] V(\d x)
= V((-\ensuremath{\varepsilon},\ensuremath{\varepsilon}))<\infty .
\label{}
\end{align}
Now we obtain
\begin{align}
& \ensuremath{\mbox{{\boldmath $n$}}} \ebra{ 1-\exp \kbra{ - \int L(R,x) V(\d x) } ; \sup_{t \ge 0} |X(t)| <\ensuremath{\varepsilon} }
\\
=& \ensuremath{\mbox{{\boldmath $n$}}} \ebra{ 1-\exp \kbra{ - \int_{(-\ensuremath{\varepsilon},\ensuremath{\varepsilon})} L(R,x) V(\d x) } ; \sup_{t \ge 0} |X(t)| <\ensuremath{\varepsilon} }
\\
\le& \ensuremath{\mbox{{\boldmath $n$}}} \ebra{ \int_{(-\ensuremath{\varepsilon},\ensuremath{\varepsilon})} L(R,x) V(\d x) } < \infty .
\label{}
\end{align}
Since $ \ensuremath{\mbox{{\boldmath $n$}}}(\sup_{t \ge 0} |X(t)| \ge \ensuremath{\varepsilon}) < \infty $, we obtain $ K_V<\infty $.
Hence the assertion is proved.
{\rm (iii)}
By Lemma \ref{lem: expect of LT}, we obtain
\begin{align}
P^{\dagger}_0 \ebra{ \int L(\infty ,x) V(\d x) }
= \int h(x) V(\d x)
< \infty .
\label{}
\end{align}
This implies that
\begin{align}
P^{\dagger}_0 \cbra{ \int L(\infty ,x) V(\d x) < \infty }
= P^{\dagger}_0(\ensuremath{\mathcal{E}}^V_{\infty }>0)
= 1 ,
\label{}
\end{align}
which proves $ P^{\dagger}_0[\ensuremath{\mathcal{E}}^V_{\infty }]>0 $.
{\rm (iv)}
Suppose that $ 0 < \int \{ 1+h(x) \} V(\d x) < \infty $.
Then the assumptions of (i)-(iii) are all satisfied.
Noting that $ \ensuremath{\mathcal{E}}^V_{\infty } \le 1 $,
we obtain $ 0 < C_V < \infty $.
\end{proof}
\subsection{Proof of Theorems}
\begin{proof}[Proof of Theorem \ref{thm: FK Penal}]
Note that $ (\ensuremath{\mathcal{E}}^V_t) $ and $ (\ensuremath{\mathcal{E}}^V_t 1_{\{ T_{\{ 0 \}} > t \}}) $
are multiplicative functionals which take values in $ [0,1] $.
By Theorem \ref{thm: conti case},
we may apply Theorem \ref{thm: general FK penal}
to obtain \eqref{eq: FK conv} and \eqref{eq: FK conv 1}.
Subtracting both sides of \eqref{eq: FK conv 1} from \eqref{eq: FK conv},
we obtain \eqref{eq: FK conv 2}.
\end{proof}
\begin{proof}[Proof of Theorem \ref{thm: FK mart op}]
The second equalities of \eqref{eq: FK func 1}, \eqref{eq: FK func 2} and \eqref{eq: FK func}
are obvious by Theorem \ref{thm: FK Penal}.
The last equality of \eqref{eq: FK func 1} is obvious
by Lemma \ref{lem: Penal P2 meander}.
The last equality of \eqref{eq: FK func 2} is obtained as follows:
\begin{align}
\frac{P_x[\ensuremath{\mathcal{E}}^V_t 1_{\{ T_{\{ 0 \}} \le t \}}]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
=&
\int_0^t P_x[\ensuremath{\mathcal{E}}^V_s;T_{\{ 0 \}} \in \d s] \frac{P_0[\ensuremath{\mathcal{E}}^V_{t-s}]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\\
\stackrel{t \to \infty }{\longrightarrow}&
\int_0^{\infty } P_x[\ensuremath{\mathcal{E}}^V_s;T_{\{ 0 \}} \in \d s] \ensuremath{\mathscr{P}}[\ensuremath{\mathcal{E}}^V_{\infty }]
= C_V P_x[\ensuremath{\mathcal{E}}^V_{T_{\{ 0 \}}}] .
\label{}
\end{align}
Now we obtain the last equality of \eqref{eq: FK func}
by adding \eqref{eq: FK func 1} and \eqref{eq: FK func 2}.
Let $ 0 \le s < t $.
By the Markov property of $ (P_x) $ and by Theorem \ref{thm: FK Penal}, we have
\begin{align}
P_x \ebra{ \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} >s \}}
\frac{ P_{X_s}[\ensuremath{\mathcal{E}}^V_{t-s};T_{\{ 0 \}} >t-s ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} }
=&
\frac{P_x[\ensuremath{\mathcal{E}}^V_t;T_{\{ 0 \}} >t ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\\
\stackrel{t \to \infty }{\longrightarrow}&
\ensuremath{\mathscr{P}}_x[\ensuremath{\mathcal{E}}^V_{\infty } 1_{\{ T_{\{ 0 \}} = \infty \}} ]
\\
=&
h(x) P^{\dagger}_x[\ensuremath{\mathcal{E}}^V_{\infty }] .
\label{}
\end{align}
By the Markov property of $ (P^{\dagger}_x) $, we have
\begin{align}
h(x) P^{\dagger}_x[\ensuremath{\mathcal{E}}^V_{\infty }]
= h(x) P^{\dagger}_x[\ensuremath{\mathcal{E}}^V_s P^{\dagger}_{X_s}[\ensuremath{\mathcal{E}}^V_{\infty }] ]
= P_x[\ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} >s \}} \varphi^1_V(X_s)] .
\label{}
\end{align}
Hence, by Scheff\'e's lemma, we obtain
\begin{align}
\frac{ P_{X_s}[\ensuremath{\mathcal{E}}^V_{t-s};T_{\{ 0 \}} >t-s ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
\stackrel{t \to \infty }{\longrightarrow}&
\varphi^1_V(X_s)
\qquad \text{in $ L^1(\ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} >s \}} \cdot P_x) $.}
\label{}
\end{align}
Therefore, for any bounded $ \ensuremath{\mathcal{F}}_s $-measurable functional $ Z_s $, we have
\begin{align}
\frac{P_x[Z_s \ensuremath{\mathcal{E}}^V_t;T_{\{ 0 \}} >t ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
=&
P_x \ebra{ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} >s \}}
\frac{ P_{X_s}[\ensuremath{\mathcal{E}}^V_{t-s};T_{\{ 0 \}} >t-s ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} }
\label{eq: FK penal conv1}
\\
\stackrel{t \to \infty }{\longrightarrow}&
P_x [ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} >s \}} \varphi^1_V(X_s) ] .
\label{}
\end{align}
Combining this with \eqref{eq: FK conv 1}, we obtain
\begin{align}
\ensuremath{\mathscr{P}}_x[Z_s \ensuremath{\mathcal{E}}^V_{\infty } 1_{\{ T_{\{ 0 \}} = \infty \}} ]
=
P_x [ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} >s \}} \varphi^1_V(X_s) ] .
\label{}
\end{align}
This implies the identity \eqref{eq: FK mart op 1}.
By similar arguments, we have
\begin{align}
\frac{P_x[Z_s \ensuremath{\mathcal{E}}^V_t 1_{\{ T_{\{ 0 \}} \le s \}} ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
=& P_x \ebra{ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} \le s \}} \frac{P_{X_s}[\ensuremath{\mathcal{E}}^V_{t-s}]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} }
\\
\stackrel{t \to \infty }{\longrightarrow}&
P_x \ebra{ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} \le s \}} \varphi_V(X_s) }
\label{}
\end{align}
and
\begin{align}
\frac{P_x[Z_s \ensuremath{\mathcal{E}}^V_t 1_{\{ s < T_{\{ 0 \}} \le t \}} ]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)}
=& P_x \ebra{ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} > s \}}
\frac{P_{X_s}[\ensuremath{\mathcal{E}}^V_{t-s};T_{\{ 0 \}} \le t-s]}{\ensuremath{\mbox{{\boldmath $n$}}}(R>t)} }
\\
\stackrel{t \to \infty }{\longrightarrow}&
P_x \ebra{ Z_s \ensuremath{\mathcal{E}}^V_s 1_{\{ T_{\{ 0 \}} > s \}} \varphi^2_V(X_s) } .
\label{}
\end{align}
Combining these two limits together with \eqref{eq: FK conv 2},
we obtain \eqref{eq: FK mart op 2}.
The remainder of the proof is now obvious.
\end{proof}
\section{Characterisation of non-negative martingales} \label{sec: further}
For a non-negative $ \ensuremath{\mathscr{P}} $-integrable functional $ G $ such that $ \ensuremath{\mathscr{P}}[G]>0 $,
we define the probability measure $ P^G $ on $ \ensuremath{\mathcal{F}}_{\infty } $ as
\begin{align}
P^G = \frac{G \cdot \ensuremath{\mathscr{P}}}{\ensuremath{\mathscr{P}}[G]} .
\label{}
\end{align}
We say that a statement holds {\em $ \ensuremath{\mathscr{P}} $-almost surely}
if it holds $ P^G $-almost surely
for some $ \ensuremath{\mathscr{P}} $-integrable functional $ G $ such that
$ G>0 $ $ \ensuremath{\mathscr{P}} $-almost everywhere.
By the Radon--Nikodym theorem,
$ \ensuremath{\mathscr{P}} $-almost sure statement does not depend on the particular choice of such a functional $ G $.
The following theorem is the stable L\'evy version of \cite[Corollary 1.2.6]{NRY}.
\begin{Thm} \label{thm: char of pos mart}
Let $ (N_t) $ be a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-martingale.
Then $ (N_t) $ is represented as $ N_t = M_t(F) $
for some $ F \in L^1(\ensuremath{\mathscr{P}}) $
if and only if it holds that
\begin{align}
\frac{N_t}{1+h(X_t)}
\stackrel{t \to \infty }{\longrightarrow}
F
\ \text{$ \ensuremath{\mathscr{P}} $-almost surely}
\quad \text{and} \quad
\ensuremath{\mathscr{P}}[F] = N_0 .
\label{eq: conv sP a.s.}
\end{align}
\end{Thm}
Although it is completely parallel to that of \cite{NRY},
we give the proof for completeness of the paper.
\begin{Lem} \label{lem: ratio mart conv}
Let $ F $ and $ G $ be a non-negative $ \ensuremath{\mathscr{P}} $-integrable functional
and suppose that $ G>0 $ $ \ensuremath{\mathscr{P}} $-almost everywhere.
Then it holds that
\begin{align}
\frac{M_t(F)}{M_t(G)} = P^G \ebra{ \frac{F}{G} \biggm| \ensuremath{\mathcal{F}}_t } .
\label{}
\end{align}
Consequently, it holds that
\begin{align}
\frac{M_t(F)}{M_t(G)}
\stackrel{t \to \infty }{\longrightarrow}
\frac{F}{G}
\qquad \text{$ P^G $-almost surely.}
\label{}
\end{align}
\end{Lem}
\begin{proof}
Let $ Z_t $ be a non-negative $ \ensuremath{\mathcal{F}}_t $-measurable functional.
On the one hand, we have
\begin{align}
P^G[Z_t F/G] = \ensuremath{\mathscr{P}}[Z_t F] = P_0[Z_t M_t(F)] .
\label{}
\end{align}
On the other hand, we have
\begin{align}
P^G[Z_t M_t(F)/M_t(G)] = \ensuremath{\mathscr{P}}[Z_t (M_t(F)/M_t(G)) G]
= P_0[Z_t M_t(F)] .
\label{}
\end{align}
Hence we obtain $ P^G[Z_t F/G]=P^G[Z_t M_t(F)/M_t(G)] $, which completes the proof.
\end{proof}
\begin{Lem} \label{lem: MtF/1+hXt}
Let $ F $ be a non-negative $ \ensuremath{\mathscr{P}} $-integrable functional.
Then
\begin{align}
\frac{M_t(F)}{1+h(X_t)}
\stackrel{t \to \infty }{\longrightarrow}
F
\qquad \text{$ \ensuremath{\mathscr{P}} $-almost surely.}
\label{}
\end{align}
\end{Lem}
\begin{proof}
We apply Theorem \ref{thm: LT mart op} with $ f(l)={\rm e}^{-l} $ to see that
$ G={\rm e}^{-L_{\infty }} $ is a positive $ \ensuremath{\mathscr{P}} $-integrable functional such that
\begin{align}
M_t(G) = (1+h(X_t)) {\rm e}^{-L_t} .
\label{}
\end{align}
Hence we obtain
\begin{align}
\frac{M_t(G)}{1+h(X_t)}
\stackrel{t \to \infty }{\longrightarrow}
G
\qquad \text{$ P^G $-almost surely.}
\label{}
\end{align}
Hence, by Lemma \ref{lem: ratio mart conv}, we obtain
\begin{align}
\frac{M_t(F)}{1+h(X_t)}
=
\frac{M_t(G)}{1+h(X_t)}
\cdot
\frac{M_t(F)}{M_t(G)}
\stackrel{t \to \infty }{\longrightarrow}
G \cdot \frac{F}{G}
= F
\qquad \text{$ P^G $-almost surely.}
\label{}
\end{align}
This completes the proof.
\end{proof}
The following proposition states an interesting representation
of any non-negative $ (P_0,\ensuremath{\mathcal{F}}_t) $-supermartingale,
a component of which is a certain $ (M_t(F)) $ martingale.
\begin{Prop} \label{prop: supermart decomp}
Let $ (N_t) $ a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-supermartingale.
\\ \quad {\rm (i)}
There exists a non-negative $ \ensuremath{\mathscr{P}} $-integrable functional $ F $ such that
\begin{align}
\frac{N_t}{1+h(X_t)} \stackrel{t \to \infty }{\longrightarrow} F
\qquad \text{$ \ensuremath{\mathscr{P}} $-almost surely;}
\label{}
\end{align}
\quad {\rm (ii)}
Denote the $ P_0 $-almost sure limit of $ (N_t) $ as $ t \to \infty $ by $ N_{\infty } $.
Then $ (N_t) $ decomposes uniquely in the following form:
\begin{align}
N_t = M_t(F) + P_0[N_{\infty }|\ensuremath{\mathcal{F}}_t] + \xi_t
\label{eq: supermart decomp}
\end{align}
where:
\\ \quad {\rm (iia)}
$ (M_t(F)) $ is a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-martingale such that
\begin{align}
M_t(F) \stackrel{t \to \infty }{\longrightarrow} 0
\ \text{($ P_0 $-a.s.)}
\qquad \text{and} \qquad
\frac{M_t(F)}{1+h(X_t)} \stackrel{t \to \infty }{\longrightarrow} F
\ \text{($ \ensuremath{\mathscr{P}} $-a.s.);}
\label{}
\end{align}
\quad {\rm (iib)}
$ (P_0[N_{\infty }|\ensuremath{\mathcal{F}}_t]) $ is a non-negative uniformly-integrable $ (\ensuremath{\mathcal{F}}_t,P_0) $-martingale
with $ P_0 $-integrable terminal value $ N_{\infty } $ such that
\begin{align}
P_0[N_{\infty }|\ensuremath{\mathcal{F}}_t] \stackrel{t \to \infty }{\longrightarrow} N_{\infty }
\ \text{($ P_0 $-a.s.)}
\qquad \text{and} \qquad
\frac{P_0[N_{\infty }|\ensuremath{\mathcal{F}}_t]}{1+h(X_t)} \stackrel{t \to \infty }{\longrightarrow} 0
\ \text{($ \ensuremath{\mathscr{P}} $-a.s.);}
\label{}
\end{align}
\quad {\rm (iic)}
$ (\xi_t) $ is a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-supermartingale such that
\begin{align}
\xi_t \stackrel{t \to \infty }{\longrightarrow} 0
\ \text{($ P_0 $-a.s.)}
\qquad \text{and} \qquad
\frac{\xi_t}{1+h(X_t)} \stackrel{t \to \infty }{\longrightarrow} 0
\ \text{($ \ensuremath{\mathscr{P}} $-a.s.)}
\label{}
\end{align}
\end{Prop}
\begin{proof}
(i)
Let $ G={\rm e}^{-L_{\infty }} $.
For any non-negative $ \ensuremath{\mathcal{F}}_s $-measurable functional $ Z_s $,
we see that
\begin{align}
P^G[Z_s N_t/M_t(G)] = \ensuremath{\mathscr{P}}[Z_s N_t G/M_t(G)] = P_0[Z_s N_t] \le P_0[Z_s N_s] .
\label{eq: Nt/MtG is supermart}
\end{align}
Hence we conclude that
$ (N_t/M_t(G)) $ is a non-negative $ (\ensuremath{\mathcal{F}}_t,P^G) $-supermartingale.
Thus there exists a non-negative $ \ensuremath{\mathcal{F}}_{\infty } $-measurable functional $ \zeta $ such that
$ N_t/M_t(G) \to \zeta $ $ P^G $-almost surely.
By Lemma \ref{lem: MtF/1+hXt}, we see that
\begin{align}
\frac{N_t}{1+h(X_t)}
= \frac{N_t}{M_t(G)}
\cdot \frac{M_t(G)}{1+h(X_t)}
\stackrel{t \to \infty }{\longrightarrow}
\zeta G =: F
\qquad \text{$ P^G $-almost surely.}
\end{align}
(ii)
For any non-negative $ \ensuremath{\mathcal{F}}_t $-measurable functional $ Z_t $, we have
\begin{align}
P_0[Z_t M_t(F)] = \ensuremath{\mathscr{P}}[Z_t F] = P^G[Z_t \zeta] .
\end{align}
By Fatou's lemma, the last expectation is dominated by
\begin{align}
\liminf_{u \to \infty } P^G \ebra{ Z_t \cdot \frac{N_u}{M_u(G)} }
= \liminf_{u \to \infty } P_0[Z_t N_u] \le P_0[Z_t N_t] .
\end{align}
This proves that $ M_t(F) \le N_t $ $ P_0 $-almost surely.
Now we see that $ (\bar{N}_t := N_t - M_t(F)) $
is a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-supermartingale.
Since $ M_t(G) \to 0 $ $ P_0 $-almost surely as $ t \to \infty $,
we see that
\begin{align}
\lim_{t \to \infty } \bar{N}_t
=
\lim_{t \to \infty } N_t
=
N_{\infty }
\qquad \text{$ P_0 $-almost surely.}
\end{align}
For any non-negative $ \ensuremath{\mathcal{F}}_t $-measurable functional $ Z_t $, we have
\begin{align}
P_0[Z_t N_{\infty }]
\le
\liminf_{u \to \infty }
P_0[Z_t \bar{N}_u]
\le
P_0[Z_t \bar{N}_t] ,
\end{align}
we see that $ P_0[N_{\infty }]<\infty $ and that
$ (\xi_t:=\bar{N}_t-P_0[N_{\infty }|\ensuremath{\mathcal{F}}_t]) $
is still a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-supermartingale.
Now the proof is completed by (i) and by Lemma \ref{lem: MtF/1+hXt}.
\end{proof}
Finally, we proceed to prove Theorem \ref{thm: char of pos mart}.
\begin{proof}[Proof of Theorem \ref{thm: char of pos mart}]
The necessity is immediate from Lemma \ref{lem: MtF/1+hXt}.
Let us prove the sufficiency.
Let $ (N_t) $ be a non-negative $ (\ensuremath{\mathcal{F}}_t,P_0) $-martingale.
Then, by Proposition \ref{prop: supermart decomp},
we have the decomposition \eqref{eq: supermart decomp}.
Letting $ t=0 $, we have
\begin{align}
N_0 = M_0(F) + P_0[N_{\infty }] + \xi_0 .
\end{align}
Since $ N_0=M_0(F)=\ensuremath{\mathscr{P}}[F] $ by the assumption, we have
\begin{align}
P_0[N_{\infty }] = \xi_0 = 0 .
\end{align}
This proves that $ N_t=M_t(F) $ $ P_0 $-almost surely,
which completes the proof.
\end{proof}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 2,301 |
Q: NodeJS memory leak I'm making a game with socket.io and mysql, and at every client connect, I check if the user is banned by IP with my custom MySQL function:
MySQL.Query('SELECT * FROM bans WHERE user = ?', username, function(rows) {
// do something
})
but when I connect > 5 clients, I have this error on the server console:
(node) warning: possible EventEmitter memory leak detected. 11 error listeners added. Use emitter.setMaxListeners() to increase limit.
Trace
at PoolConnection.addListener (events.js:239:17)
at C:\Users\user\Documents\_server\mysql\mysql.js:26:21
at Ping.onOperationComplete [as _callback] (C:\Users\user\node_modules\mysql\lib\Pool.js:99:5)
at Ping.Sequence.end (C:\Users\user\node_modules\mysql\lib\protocol\sequences\Sequence.js:96:24)
at Ping.Sequence.OkPacket (C:\Users\user\node_modules\mysql\lib\protocol\sequences\Sequence.js:105:8)
at Protocol._parsePacket (C:\Users\user\node_modules\mysql\lib\protocol\Protocol.js:280:23)
at Parser.write (C:\Users\user\node_modules\mysql\lib\protocol\Parser.js:73:12)
at Protocol.write (C:\Users\user\node_modules\mysql\lib\protocol\Protocol.js:39:16)
at Socket.<anonymous> (C:\Users\user\node_modules\mysql\lib\Connection.js:96:28)
at emitOne (events.js:77:13)
Here's mysql.js file :
var config = require('../config.js');
var mysql = require('mysql');
var pool = mysql.createPool({
connectionLimit : 100,
host : config.sqlhost,
user : config.sqluser,
password : config.sqlpass,
database : config.sqlbase,
debug : false
});
var MySQL = {
Query: function(req, reqvars, callback) {
pool.getConnection(function(err,connection){
if (err) {
connection.release();
console.log('[MySQL] Error while connecting to the database');
}
connection.query(req, reqvars, function(err,rows){
connection.release();
if (typeof callback == 'function') callback(rows);
});
connection.on('error', function(err) {
console.log('[MySQL] Error while attempting query');
});
});
},
}
module.exports = MySQL;
Where's the problem?
A: Node.js warns you whenever you attach more than 10 (the default) listeners for the same event on a single event emitter.
In your particular case, you attach an error listener to the connection object each and every time you make an SQL query. This is probably not what you want, because if the first x queries executed fine, but then one fails, those error handlers for the previous queries will still get called.
Remedy
While I have no practical experience with the mysql library, the first thing hitting me is that you are completely ignoring the err parameter of your connection.query() callback. You must not ignore these, unless you would like to spend hours of debugging a seemingly valid piece of code.
It would seem that instead of attaching an error handler to the connection, you simply check the err argument.
PS. It is common practice to always pass along any errors that occur during execution as the first argument in callback-based APIs (specifying null in case of no error). Your callback function seems to not adhere to this practice.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 4,619 |
2019 MMEA Awards for Excellence
MMEA honors excellence in music education. Nominations will be accepted through January 22, 2019.
Access the nomination form of each award below and complete an online nomination. Feel free to contact Ashley Ashman through email at mmea.member.at.large@gmail.com with any concerns or questions. Alternatively nominations may be submitted through mail to: Robert Frost Middle School, Attention: Ashley Ashman, 9201 Scott Drive, Rockville, MD 20850. Contact NAfME at 1-800-828-0229 or Call: 703-860-4000 or Email: memberservices@nafme.org for any membership questions.
MMEA/NAfME EXEMPLARY MUSIC PROGRAM AWARD
This award is to recognize outstanding music programs. The award will be based on the extent to which the program meets the standards set forth in the NAfME publication, The School Music Program: Description and Standards. Any school that has not received the award within the previous five years is eligible.
MMEA Hall of Fame
This award is to recognize an individual for excellence in teaching and/or administration, contributions and improvements made in music education, betterment of the profession through exemplary service or acts, professional offices, publications, awards, recognitions, performances, and professional ideals and academic integrity.
OUTSTANDING MUSIC TEACHER AWARD
This award is to recognize the music educators whom you've seen in action either in the classroom, at concerts, or at in-services or conferences who share their experiences and knowledge and are mentors in their profession.
CORWIN TAYLOR MUSIC EDUCATION LEADERSHIP AWARD
This award is to recognize an individual who has made a significant contribution to the music education of Maryland's youth, whether as a teacher, administrator, parent, industry representative, performer, or in another capacity. The Corwin Taylor Music Education Leadership Award was established in 1993 in memory of Dr. Corwin H. Taylor (1905-1992), noted musician, composer, author, and educator. Dr. Taylor, an active member of the Music Educators National Conference (now National Association for Music Education, NAfME) and the Maryland Music Educators Association, was Supervisor of Instrumental Music for the Baltimore Public School system from 1945 to 1968 and a faculty member at the University of Maryland at College Park from 1968 to 1976. He will always be remembered for the great impact he had on the lives and careers of his students and colleagues. The award may be presented annually.
MMEA School Administrator Award
This award is to recognize the administrators who are music advocates and supporters, working with the music team to build incredible music programs for their schools and communities.
Rosemary and James Walters Service Award
This award is to recognize an individual who has made a significant contribution to the Maryland Music Educators Association. The Rosemary and James Walters Award for Service to the Maryland Music Educators Association was established in 2001 in memory of Rosemary and James Walters. Their teaching careers were in Maryland, spanning 30 years until a tragic accident in April 2001. Jim taught instrumental music, most recently at Francis Scott Key Middle School in Montgomery County. Rosemary taught choral music. Together, they characterized the essence of supporting the music education profession. They supported their colleagues, young and seasoned, and they supported every professional activity of the association and the school music community. Jim served as MBDA president, but his contribution was not limited to band activities. Rosemary eventually left the classroom to work full time in church music, but her involvement as a substitute music teacher and as an accompanist continued. At nearly any MMEA event, it was typical to see Jim and Rosemary offering to assist when they were not "on duty," doing whatever needed to be done. The award may be presented annually.
From local activities to national issues, MMEA/NAfME offer the combination of networking, professional development and resources you need as a music educator. Become a member today! | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 36 |
Q: Comparing a Bayesian model with a Classical model for linear regression I have fitted following Bayesian linear regression model using rjags package in R, with the help of cars data set.
I used some weakly informative priors for parameters.
require(rjags)
dim(cars)
N=length(cars$speed)
bayes_model="model {
for(i in 1:N){
dist[i] ~ dnorm(mu[i],tau)
mu[i] = beta[1] + beta[2]*speed[i]
}
for (l in 1:2) { beta[l] ~dnorm(0, 100) }
tau ~ dgamma(.001,.001)
sigma_tau = 1/tau
}"
model2 <- jags.model(textConnection(bayes_model),
data = list(dist=cars$dist,N=N,speed=cars$speed),
n.chains=2)
params <- c('beta','sigma_tau')
samps.1 <- coda.samples(model2, params, n.iter = 2000)
burn.in=1000
summary.model.1=summary(window(samps.1, start = burn.in))
Stat.model.1=as.data.frame(summary.model.1$statistics)
This is how the results looks like .
> Stat.model.1
Mean SD Naive SE Time-series SE
beta[1] 9.937366e-03 0.09806290 0.002191658 0.002238168
beta[2] 1.650041e-01 0.09903592 0.002213404 0.002330977
sigma_tau 2.341437e+03 522.81381343 11.684631408 11.700676273
When I fit a classical linear regression model , following results are obtained .
summary(lm(dist~speed ,data=cars))
Call:
lm(formula = dist ~ speed, data = cars)
Residuals:
Min 1Q Median 3Q Max
-29.069 -9.525 -2.272 9.215 43.201
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -17.5791 6.7584 -2.601 0.0123 *
speed 3.9324 0.4155 9.464 1.49e-12 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 15.38 on 48 degrees of freedom
Multiple R-squared: 0.6511, Adjusted R-squared: 0.6438
F-statistic: 89.57 on 1 and 48 DF, p-value: 1.49e-12
It can be observed that the results based on Bayesian and classical method is not similar. What may be the reason for this ? Is there any problem with my prior distributions ?
Are there any diagnostic plots that should look into ? Also please feel free to let me know if I have missed any important steps in this analysis.
I am relatively new to Bayesian and I am working with different sort of examples to learn how to assign correct prior distributions.
A: Try:
bayes_model="model {
for(i in 1:N){
dist[i] ~ dnorm(mu[i],tau)
mu[i] = beta[1] + beta[2]*speed[i]
}
for (l in 1:2) { beta[l] ~dnorm(0, 0.001) }
tau ~ dgamma(.01,.01)
sigma_tau = 1/tau
}"
You'll get something like:
> Stat.model.1
Mean SD Naive SE Time-series SE
beta[1] -16.88098 6.2008860 0.138586750 0.53168398
beta[2] 3.89631 0.3812761 0.008521332 0.03262728
sigma_tau 244.69856 53.4120568 1.193733180 1.25568735
Why? Because in JAGS the parameters of the normal distribution are mean $\mu$ and precision $\tau$, and $\tau=\sigma^{-2}$, so beta[l] ~ dnorm(0, 100) is a very strong, not a weakly informative prior, it forces beta[l] to be as close to zero as possible.
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 450 |
Lightweight and with shock-cord linked framework, Helinox have paved the way for innovative and sustainable outdoor camping chairs.
All of the chairs live up to the company's ultra lightweight reputation, weighing from 1485 g with case for the Beach Chair to 1335 g with case for the Camp Chair, in step with most competitors' offerings. Scott and I have significantly different body types — I refer to him lovingly as "my tank"— but each chair felt suitable for both of us. I didn't feel like Goldilocks swallowed up in Papa Bear's lounger, and Scott didn't tumble or strain the seams. The chairs are rated for 145 kg, and durable stitching plus reinforcing panels promise sustained usage over the long-term.
Both the Sunset and Beach models offer a pillow option, where the storage case can be stuffed with a jacket and attached to the top of the chairback. When the chairs are stored in their cases, they fit easily into a car kit with a packed length of 47 cm to 50 cm, a necessity for those of us who tend to overstuff with gear (guilty), or for longer sojourns where every centimetre is spoken for. The Beach version has a ground to seat base distance of only 17 cm, making it challenging to get in and out of, and far more likely to sink into sand or soft ground. Both the Sunset and Camp models lift 36 cm off the ground, an easier height for general use. The Camp chair's weight helps for gram counters, but if you're camping from the car, opt for the more luxurious Sunset option, with higher seat back and greater depth (70 cm versus 60 cm). | {
"redpajama_set_name": "RedPajamaC4"
} | 6,489 |
\section{Introduction}
Hillslopes evolve topographically through a variety of erosional mechanisms ranging from slow diffusive processes (e.g. soil creep), to fast, localized processes (e.g. landslides). Over short timescales ($10^0$ - $10^1$ yr), hillslope sediment transport determines the redistribution of sediment and its delivery to the slope base. Over long timescales ($10^2$ - $10^5$ yr) the balance between, and integral of, individual erosional events determines the topographic form of hillslopes. Where advective processes dominate, hillslopes tend to be concave up, and where diffusive processes are more pronounced hillslopes become convex (e.g. \citep{carson1972hillslope,kirkby1971}). It is well-acknowledged that the processes shaping landscapes are inherently dynamic and stochastic \citep{dietrich2003geomorphic,roering2004soil,tucker2010modelling}, yet landscape evolution model (LEM) characterization of hillslope processes is often based on geomorphic transport laws (GTLs), mathematical formulations expressing erosion as an averaged process operating over long timescales \citep{dietrich2003geomorphic}. This discrepancy gives rise to a mathematical disconnect between the stochastic processes operating at the grain scale over the short term, and the evolution of hillslope topography over the long term.
In this paper we demonstrate a principled probabilistic scaling argument by which a particle-based description of hillslope sediment transport can be scaled to a continuum one representing long-term hillslope evolution. In other words, we present a mathematical argument for deriving a continuum description of hillslope erosion while remaining faithful to the particle-scale dynamics that operate over short time and space scales.
GTLs are a compromise between a comprehensive physics-based description, which may be too complex to be parametrized through field observation, and rules-based modeling, which may lack a testable mechanistic footing \citep{dietrich2003geomorphic}. LEMs typically consist of a statement of mass conservation, GTLs for describing sediment transport in the form of differential equations, and numerical methods to approximate solutions to the GTLs \citep{tucker2010modelling}. Despite inherent simplifying assumptions associated with this approach, GTLs have been successful in simulating landform development in some environments, particularly associated with diffusive processes like creep and bioturbation (e.g. \citep{roering1999evidence}).
Particle-based models, which display a rich range of behavior despite their simplicity and ease of implementation, are an important alternative to this prescription of landscape evolution modeling \citep{tucker2010trouble,kessler2003self,davies2011discrete}. Traditionally, particle models have been criticized for using ``ad-hoc'' evolution rules and experimentally inaccessible parameters, and for neglecting the underlying transport physics \citep{dietrich2003geomorphic}. Accordingly, as continuum models have long been numerically implementable and experimentally verifiable, the use of GTLs has dominated studies of landscape evolution. However, the experimental validation of particle-based models is now possible using techniques for tracking grain motion \citep{mcnamara2004observations,habersack2001radio,roering2004soil,fathel2015experimental,roseberry2012probabilistic}. This, combined with their ability to incorporate particle mechanics and motion statistics, leads \citet{tucker2010trouble} to argue that particle-based models are no less fundamental than GTLs and should be used to complement continuum models.
While the case against the use of particle-based models has been undermined by experimental innovation, it is the theoretical development of nonlocal transport on hillslopes which best underscores the case for their use. Continuum models like those of \citet{culling1963soil} and \citet{andrews1987fitting} rely on locality assumptions, the assumption that sediment transport at position $x$ on a slope is a function of the hillslope conditions at $x$ (i.e. local land-surface slope) \citep{furbish2013sediment}. Locality assumptions are only valid when hillslope particles move short distances relative to the hillslope length \citep{tucker2010trouble}. Examples of local transport mechanisms are soil creep \citep{furbish2009statistical}, rainsplash \citep{dunne2010rain,furbish2009rain}, bioturbation and tree throw \citep{gabet2000gopher,gabet2003effects}. Nonlocal transport occurs when sediment transport at position $x$ depends on the hillslope characteristics a significant distance upslope or downslope of position $x$ \citep{furbish2013sediment} such that occurs in sheetwash sediment transport \citep{michaelides2012sediment,michaelides2014impact} and dry ravel \citep{gabet2012particle} on steep slopes. Accordingly, formulations of nonlocal transport must specify the relationship between flux and relative upslope or downslope, ultimately leading to assumptions on the distribution of particle travel distances \citep{furbish2010divots,furbish2013sediment} or the fitting of a fractional derivative operator \citep{foufoula2010nonlocal}. However, such relationships change as hillslopes evolve, and so particle-based approaches may be more appropriate \citep{gabet2012particle,dibiase2017slope}.
In order to effectively combine their strengths, the particle model must correspond, in some sense, to the continuum description. However, as \citet{tucker2010trouble} indicate, it is not clear how to identify such a pair.
Indeed, referring to the particle-based models of \citet{tucker2010trouble} and \citet{gabet2012particle}, \citet{ancey2015stochastic} observe, ``there is no technique for deriving continuum equations from the rules used to describe particle behavior in this environment.'' Here, we demonstrate a principled \textit{probabilistic scaling} argument by which a particle-based description can be scaled to a continuum one with the two descriptions corresponding to one another. The probabilistic scaling procedure, illustrated in Figure~\ref{fig:scaling_schematic}, consists of scaling space and time by a small parameter, ultimately converting the microscopic evolution rules into a partial differential equation governing the macroscopic observables \citep{kipnis1999scaling,olla1993hydrodynamical,bahadoran2010strong}.
In Sections~\ref{sec:model} and \ref{sec:hydro}, we introduce a simple particle-based model of hillslope evolution and provide a heuristic scaling argument, which identifies a corresponding continuum description in the form of an advection-diffusion equation. Critically, the particles of our model correspond to units of hillslope \textit{gradient}, not hillslope height. This element of indirection softens the distinction between local and nonlocal transport and, for this reason, our model can represent diverse geomorphic processes and the scaling argument applies uniformly across various transport regimes.
Finessing nonlocal transport through indirection comes at the expense of immediate access to information about particle hopping distances and fluxes. This contrasts the convolutional approaches of \citet{foufoula2010nonlocal} and \citet{furbish2010divots}, which express sediment flux arising from nonlocal transport as an integral over relative upslope. While such methods enable detailed calculation of fluxes, they require as input assumptions about the distribution of particle hopping distances and hillslope topography \citep{gabet2012particle,furbish2010divots,furbish2013sediment}. When these detailed outputs are unnecessary, the requisite inputs are unavailable, or corresponding simulations are computationally expensive, a particle-based approach may be preferable.
Section~\ref{sec:sim} describes simulations of the particle system for various choices of microscopic parameters, including both linear and nonlinear slope dependence, to exhibit the types of hillslope profiles which form and how fluxes arise in response to hillslope perturbations. Additionally, to translate simulation results into empirically testable predictions, we suggest a principled way of fitting model parameters from data and assigning dimensions to model outputs. Finally, in Section~\ref{sec:disc}, we discuss the relation of this paper to the hillslope evolution and nonlocal transport literature and suggest future work, which takes advantage of a dual, particle-based and continuum approach.
\begin{figure}[htbp]
\centering
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=0.9\textwidth]{figures/1_scaling_schematic_wide}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0.21,0.78) {\LARGE{\textbf{$i$}}};
\node at (0.25,0.02) {\LARGE{\textbf{$x$}}};
\node at (0.02,0.87) {\huge{\textbf{$h$}}};
\node at (0.06,0.15) {\huge{\textbf{$h$}}};
\node at (1.01,0.76) {\huge{\textbf{$\tau$}}};
\node at (0.537,0.87) {\huge{\textbf{$h$}}\LARGE{$(i)$}};
\node at (1.01,0.05) {\huge{\textbf{$t$}}};
\node at (0.537,0.15) {\huge{\textbf{$h$}}\LARGE{$(x)$}};
\draw[thick,->] (0.59,0.76) -- (0.98,0.76);
\draw[thick,->] (0.59,0.76) -- (0.59,.96);
\draw[thick,->] (0.59,0.05) -- (0.98,0.05);
\draw[thick,->] (0.59,0.05) -- (0.59,0.25);
\draw[thick] (0.095, 0.05) -- (0.095,0.25);
\draw[thick] (0.095,0.05) -- (0.42,0.05);
\draw[thick] (0.25,0.05) -- (0.25,0.04);
\draw[dashed] (0.25,0.05) -- (0.25,0.21);
\node[draw, align=center, text width=4cm] (A) at (0.205,1.02) {\large \textbf{Rescaling space ($i \mapsto x = \varepsilon i$)}};
\node[draw, align=center, text width=4cm] (B) at (0.745,1.02) {\large \textbf{Rescaling~time ($\tau \mapsto t = \varepsilon^2 \tau$)}};
\draw[thick] (A) -- (B);
\draw[thick,->] (0.47,1.02) -- (0.47,0.25) node[draw,midway, fill=white] {\Large{\textbf{$\varepsilon \rightarrow 0$}}};
\draw[thick] (0.6,0.9) -- (0.65,0.9) -- (0.65,0.85) -- (0.7,0.85) -- (0.7,0.87) -- (0.75,0.87) -- (0.75,0.82) -- (0.8,0.82) -- (0.8,0.78) -- (0.85,0.78) -- (0.85,0.85) -- (0.9,0.85) -- (0.9,0.92) -- (0.95,0.92);
\draw [thick] plot [smooth] coordinates { (0.6,0.17) (0.65,0.145) (0.7,0.135) (0.75,0.139) (0.775,0.12) (0.835,0.145)(0.885,0.18) (0.95,0.15) };
\draw[dashed] (0.6,0.9) -- (0.6,0.76);
\draw[dashed] (0.95,0.92) -- (0.95,0.76);
\draw[dashed] (0.6,0.745) -- (0.6,0.65);
\draw[dashed] (0.95,0.745) -- (0.95,0.65);
\draw[dashed] (0.95,0.65) -- (0.780,0.25);
\draw[dashed] (0.6,0.65) -- (0.77,0.25);
\draw[dashed] (0.78,0.25) -- (0.78,0.05);
\draw[dashed] (0.77,0.25) -- (0.77,0.05);
\draw[thick] (0.6,0.76) -- (0.6,0.75);
\draw[thick] (0.95,0.76) -- (0.95,0.75);
\draw[thick] (0.78,0.05) -- (0.78,0.04);
\draw[thick] (0.77,0.05) -- (0.77,0.04);
\node at (0,0.96) {\huge{\textbf{A}}};
\node at (0.045,0.25) {\huge{\textbf{B}}};
\node at (0.54,0.96) {\huge{\textbf{C}}};
\node at (0.54,0.25) {\huge{\textbf{D}}};
\end{scope}
\end{tikzpicture}
\caption[Schematic of space and time rescaling.]{Schematic of space and time rescaling. Discrete space in a particle model of a hillslope, indexed by $i$ (\textbf{A}), is rescaled by a small parameter $\varepsilon$. In the limit as $\varepsilon$ approaches $0$, discrete space becomes continuous; accordingly, we replace $i$ with a continuous quantity $x$ = $\varepsilon i$ (\textbf{B}). After the rescaling, particles originally spaced by unit distance are spaced by $\varepsilon$. Consider instead the hillslope height at a particular site $i$, which changes in response to particle movements occurring on a timescale $\tau$ (\textbf{C}). After the rescaling of space, changes in hillslope height on timescale $\tau$ are too small to be observed, so the dynamics must be quickened by rescaling $\tau$ to $t$ with $\varepsilon^2$. Rescaling both space and time results in a macroscopic height $h (x)$ evolving on timescale $t$ (\textbf{D}).}
\label{fig:scaling_schematic}
\end{figure}
\section{A particle-based model of hillslope evolution}\label{sec:model}
\subsection{Specifying state space and dynamics}\label{sub:dynamics}
As our goal is to model hillslope profiles, we begin by considering a 1D grid of $L+1$ labeled sites, which each contain some number of identical ``units'' of hillslope (Figure~\ref{fig:fixedheightzrp}A). We fix the number of units at site $1$ to be $H$, and the number at site $L+1$ to be $0$. The process of hillslope evolution could then occur via the rearrangement of the units across sites $2$ to $L$, according to some dynamics. However, our analysis becomes simpler if we instead consider a corresponding ``gradient'' particle system, where the particles represent differences in the number of hillslope hunks between adjacent grid sites (Figure~\ref{fig:fixedheightzrp}B). That is, if there are $h_\tau (i)$ units at site $i$ and time $\tau$ and $h_\tau (i+1)$ at site $i+1$, we place $\omega_\tau (i) = h_\tau (i) - h_\tau (i+1)$ particles at site $i$ of the gradient system. Note that, because we fixed site $L+1$ to have $0$ units, $\omega_\tau (L) = h_\tau (L)$. Additionally, because we fixed site $1$ to have $H$ units, $\sum_{i=1}^L \omega_\tau (i) = H$; we have conservation of gradient particles. In order to complete the specification of the gradient process, we need to describe the ways in which particles are allowed to move.
Figure~\ref{fig:fixedheightzrp} summarizes the rules governing the dynamics. Particles hop after exponentially-distributed waiting times, with rates given as follows. For sites $i \neq 1,\,L$, a particle will hop $i \rightarrow i+1$ with rate $pf(\omega_\tau (i))$ and $i \rightarrow i-1$ with rate $qf(\omega_\tau (i))$, where $p, q \in (0,1)$ and $p+q = 1$, and $f(\omega_\tau (i))$ is a nondecreasing function of $\omega_\tau (i)$ with $f(0) = 0$. The requirement that $f$ be nondecreasing in $\omega_\tau (i)$ formalizes the intuition that the dynamics on steep slopes happen at least as quickly as those on gradual slopes. At the left boundary $i=1$, a particle hops $1 \rightarrow 2$ with rate $pf(\omega_\tau (1))$ and, at the right boundary $i=L$, a particle hops $L \rightarrow L-1$ with rate $qf (\omega_\tau (L))$. As the number of gradient particles, $\omega_\tau (i)$, represents the steepness of the hillslope at site $i$, a gradient particle hopping to site $i$ corresponds to the hillslope becoming steeper at $i$. In terms of hillslope profile $h_\tau$, this could reflect deposition at site $i$ or removal at site $i+1$, both of which would cause the hillslope to become steeper at $i$.
Our model is a type of \textit{continuous-time Markov process}, known in the statistical physics and probability literature as a ``zero-range process'' because particles hop at rates which depend on the occupancy of their current site. In this sense, there is a zero-range interaction between particles occupying the same site. Note that particles in the gradient process only hop unit distances, unlike particles in the model of \citet{tucker2010trouble}. While gradient particles redistribute locally, the corresponding changes in the original height profile need not be.
\begin{figure}[htbp]
\centering
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=0.7\textwidth]{figures/2_ips}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0.15,0.97) {\huge{\textbf{$H$}}};
\node at (0.15,0.54) {\huge{\textbf{$1$}}};
\node at (0.15,0.115) {\huge{\textbf{$1$}}};
\node at (-0.02,0.35) {\LARGE{\textbf{$p f (\omega (1))$}}};
\node at (0.94,0.35) {\LARGE{\textbf{$q f (\omega (L))$}}};
\node at (0.33,0.40) {\LARGE{\textbf{$q f (\omega (i))$}}};
\node at (0.61,0.40) {\LARGE{\textbf{$p f (\omega (i))$}}};
\node at (0.46,0.54) {\huge{\textbf{$i$}}};
\node at (0.46,0.115) {\huge{\textbf{$i$}}};
\node at (0.77,0.115) {\huge{\textbf{$L$}}};
\node at (0.77,0.54) {\huge{\textbf{$L$}}};
\node at (0.875,0.61) {\huge{\textbf{$0$}}};
\node at (0.465,0.84) {\huge{\textbf{$h (i)$}}};
\node at (0.46,0.0) {\LARGE{\textbf{$\omega (i) = h (i) - h (i+1)$}}};
\node at (0.015,0.97) {\huge{\textbf{A}}};
\node at (0.015,0.445) {\huge{\textbf{B}}};
\end{scope}
\end{tikzpicture}
\caption[Mapping changes in a height profile to a gradient ZRP.]{Schematic of the mapping between the hillslope height (\textbf{A}) and corresponding hillslope gradients (\textbf{B}) of the particle model. The height of the hillslope's leftmost site ($i$ = $1$) is fixed at a height of $H$ and the rightmost site ($i=L+1$) is fixed at $0$ (\textbf{A}). In the gradient process (\textbf{B}), particles in the bulk ($1< i < L$) hop to the left and right with rates $qf(\omega (i))$ and $pf(\omega (i))$, respectively; particles at the left boundary move right at rate $pf(\omega (1))$ and those at the right boundary move left at rate $qf(\omega (L))$.}
\label{fig:fixedheightzrp}
\end{figure}
\subsection{Identifying the stationary distributions of the particle model}\label{sec:stationary}
The stationary distributions of the gradient process are those probability distributions over occupancies $\omega (i)$ which are unchanged by the dynamics specified in Section \ref{sub:dynamics}. To find the stationary distributions, it suffices to enforce a ``detailed balance'' condition between configurations
\begin{linenomath*}\[\omega = \{ \omega (1), \, \dots, \, \omega (i),\, \omega (i+1), \, \dots, \, \omega (L) \} \quad \text{and}\]\end{linenomath*} \begin{linenomath*}\[\omega^{i\rightarrow i+1} = \{ \omega (1), \, \dots, \, \omega (i) - 1,\, \omega (i+1) + 1, \, \dots, \, \omega (L) \},\]\end{linenomath*} which reads
\begin{linenomath*}\begin{equation}\label{eq:detailedbalance}
\mathds{P} (\omega) \cdot pf(\omega (i)) = \mathds{P} (\omega^{i\rightarrow i+1}) \cdot qf(\omega (i+1)+1).
\end{equation}\end{linenomath*} That is, in equilibrium, the frequency of moving from one configuration to a second is exactly balanced by the frequency of the reverse process.
Surprisingly, despite the many interactions between particles, the probability distribution $\mathds{P}(\omega)$ of observing the particle system in configuration $\omega$ in equilibrium can be expressed as a product of decoupled marginal distributions for each site $\mathds{P}(\omega) = \prod_{i=1}^L {\mathds{P}_i}^{\theta_i} (\omega (i))$. Informally, at equilibrium, the probability of seeing a certain number of gradient particles at a site is independent of all other sites. This property enables us to study the simpler marginal distributions $\mathds{P}_i^{\theta_i}$ instead of $\mathds{P}$, and would not be present if we had instead modeled the hillslope directly, with particles representing units of hillslope height. The marginal distributions have the form
\begin{linenomath*}\begin{equation}\label{eq:dist}
{\mathds{P}_i}^{\theta_i} (\omega (i)) = \frac{e^{\theta_i \omega (i)}}{f(\omega (i))! \, Z(\theta_i) } \quad \theta_i \in \mathds{R},
\end{equation}\end{linenomath*} with $f(z)! = \prod_{k=1}^z f(k)$ and $f(0)! = 1$. $Z(\theta_i) = \sum_{k=0}^{\infty} e^{\theta_i k} / f(k)!$ is a normalization constant, which is assumed to be finite. In Appendix \ref{app:detailedbalance}, we show that Equation \ref{eq:dist} indeed satisfies the detailed balance condition of Equation \ref{eq:detailedbalance}, so long as $\exp (\theta_{i+1} - \theta_i) = p/q$.
Using the stationary distributions ${\mathds{P}_i}^{\theta_i}$, we would like to calculate the stationary density, that is, the expected number of gradient particles occupying each site in equilibrium. Technically, this quantity depends on the choice of hillslope height $H$, and so we should calculate the \textit{conditional} expected number of gradient particles at each site. For an arbitrary choice of $f(\omega (i))$, parameter $\theta_i$, and fixed height $H$, the density at a site $i$ is
\begin{linenomath*}\begin{equation}\label{eq:occupancy}
{\rho}^{\theta_i | H} (i) = \mathds{E}^{\theta_i} \Bigg( \omega (i) \, \Bigg| \sum_{j=1}^L \omega (j) = H \Bigg) = \sum_{k=0}^H k \cdot \mathds{P}^{\theta_i} \Bigg( \omega (i) = k \, \Bigg| \sum_{j=1}^L \omega (j) = H \Bigg),
\end{equation}\end{linenomath*} where $\mathds{E}^{\theta_i}$ is the expectation with respect to the distribution $\mathds{P}^{\theta_i}$ and the notation $\Big| \sum \omega = H$ indicates conditioning on the sum of gradient particles being $H$. The sum over $k$ in \eqref{eq:occupancy} is an average over the numbers of gradient particles which could be at site $i$, with a weighting based on the probability of observing $k$ particles at site $i$, subject to the configuration having a total of $H$ gradient particles.
Note that this density describes the average number of particles at each site in equilibrium for the gradient process, not the original hillslope profile. In order to obtain a typical hillslope profile, the density must be inverted using $\omega (i) = h (i) - h (i+1)$, which leads to \begin{linenomath*}\begin{equation}\label{eq:inversion} h (i) = \sum_{j=i}^L \omega (j).\end{equation}\end{linenomath*}
\subsection{Hillslope profiles for linear rate}\label{sec:example}
We can calculate Equation \ref{eq:occupancy} explicitly for the choice of linear rate, $f(\omega (i)) = \omega (i)$, corresponding to the gradient particles hopping with rate proportional to local gradient. For this choice of $f(\omega (i))$, the stationary distributions are Poisson
\begin{linenomath*}\begin{equation}\label{eq:poisson}
{\mathds{P}_i}^{\theta_i} (\omega (i)) = \frac{e^{\theta_i\,\omega (i)}\cdot e^{-e^{\theta_i}}}{\omega (i)!}.
\end{equation}\end{linenomath*} In Appendix \ref{app:density}, we show that using (\ref{eq:poisson}) with (\ref{eq:occupancy}) gives
\begin{linenomath*}\begin{equation}\label{eq:density}
{\rho}^{\theta_i | H} (i) = H\frac{e^{\theta_i}}{\sum_{j=1}^L e^{\theta_j}} = H\cdot \Bigg(\frac{p}{q}\Bigg)^{i-1}\frac{\Big(\frac{p}{q}\Big) -1}{\Big(\frac{p}{q}\Big)^L - 1}.
\end{equation}\end{linenomath*} where the second equality follows from condition (i) on the $\theta_j$.
We can invert Equation~\ref{eq:density} with $h (i) = \sum_{j=i}^L \omega (j)$ to get the corresponding hillslope profile
\begin{linenomath*}\begin{equation}\label{eq:heights}
h (i) = H\frac{\Big(\frac{p}{q}\Big)^i - \Big(\frac{p}{q}\Big)^{L+1}}{\Big(\frac{p}{q}\Big) - \Big(\frac{p}{q}\Big)^{L+1}} \quad \text{for} \quad 1 \leq i \leq L,
\end{equation}\end{linenomath*} which describes the expected hillslope profile in equilibrium. Examples of such profiles are provided for a range of $p/q$ values in Figure~\ref{fig:exp_curves}.
\begin{sidewaysfigure
\centering
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=\textwidth]{figures/3_DRAFT_exp_curves_broad}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0.013,0.53) {\Huge{\textbf{$h (i)$}}};
\node at (0.52,0.035) {\Huge{\textbf{$i$}}};
\node[fill=white, opacity=0.35] at (0.615,0.72) {\Huge{\textbf{soil creep}}};
\node[fill=white, opacity=0.35] at (0.51,0.55) {\Huge{\textbf{sheet wash}}};
\node[fill=white, opacity=0.35] at (0.39,0.4) {\Huge{\textbf{sheet wash}}};
\node[fill=white, opacity=0.35] at (0.39,0.32) {\Huge{\textbf{with rills/gullies}}};
\end{scope}
\end{tikzpicture}
\caption[Example hillslopes produced by the gradient ZRP.]{The hillslope profiles produced by the particle model for $f(\omega (i)) = \omega (i)$, fixed $H = L = 100$, and values of $p$ shown in the legend, where we fix $p+q = 1$. We indicate soil creep, sheet wash, and sheet wash with rills/gullies as geomorphic processes which could be modeled by these curves, in analogy with the characteristic-form profiles of \citet{kirkby1971}.}
\label{fig:exp_curves}
\end{sidewaysfigure}
\subsection{Hillslope profiles for constant rate}\label{sec:stepexample}
We can also calculate $\rho^{\theta_i} (i) = \mathds{E}^{\theta_i} \omega (i)$, in absence of conditioning on $H$, for a choice of constant rate: $f(\omega (i)) = 1$ if $\omega (i) > 0 $ and $f(\omega (i)) = 0$ if $\omega (i) = 0 $. Whereas, in the case of linear rate, the dynamics depended on the local gradient, the constant rate case corresponds to a dynamics which evolves steep slopes at the same rate as gradual slopes. The fact that the conditioning matters little to the stationary hillslope profile follows from a large-deviations-type argument, which we omit here for brevity.
The occupancies $\omega (i)$ are distributed as geometric random variables, that is \begin{linenomath*}\[
\mathds{P_i}^\theta_i(\omega (i)) = \frac{e^{\theta_i \omega (i) }}{Z(\theta_i)f(\omega (i))!}=\frac{ e^{\theta_i \omega (i)}}{Z(\theta_i)} = e^{\theta_i \omega (i)} (1 - e^{\theta_i}),
\]\end{linenomath*}
thus the density of gradient particles has the following simple form
\begin{linenomath*}\begin{equation}\label{eq:steprho}
\rho^{\theta_i} (i)=\frac{e^{\theta_i}}{1-e^{\theta_i}}, \end{equation}\end{linenomath*}
valid for $(\theta_i <0)$. The stationary distribution requires $e^{\theta_{i+1} - \theta_i} = p/q$, or
\begin{linenomath*}\[
e^{\theta_i}=c\cdot\left(\frac {p}{q}\right)^i,\qquad c>0\,,\quad i<\frac{-\ln c}{\ln p-\ln q}
\]\end{linenomath*} which, combined with \eqref{eq:steprho}, gives the discrete gradient of the hillslope
\begin{linenomath*}\begin{equation}\label{eq:steprho2}
\rho^{\theta_i} (i) = \frac{c\cdot\left(\frac {p}{q}\right)^i}{1-c\cdot\left(\frac pq\right)^i}.
\end{equation}\end{linenomath*} To obtain the expected hillslope profile corresponding to \eqref{eq:steprho2}, we apply $h(i) = \sum_{j=i}^L \omega (j)$ and substitute \eqref{eq:steprho2}, resulting in
\begin{linenomath*}\[
h(i) = \sum_{j=i}^L \frac{c\cdot\left(\frac {p}{q}\right)^j}{1-c\cdot\left(\frac pq\right)^j}.
\]\end{linenomath*} We note that $c$ can be chosen to fit the left boundary condition for height $h(1) = H$.
\subsection{Particle model recap}\label{sec:particlemodelrecap}
We recall some key points from Section~\ref{sec:model} before proceeding to the scaling.\begin{enumerate}
\item The particles of the model represent units of slope, not units of height.
\item Particles move according to a rate function $f$ which is not necessarily linear.
\item To obtain a hillslope profile, the gradient particle profile must be summed according to \eqref{eq:inversion}.
\item Hillslope profiles can be calculated explicitly when $f$ is linear or constant; simulated otherwise.
\end{enumerate}
\section{Heuristic derivation of the continuum equation}\label{sec:hydro}
We now return to a general setting, where the form of $f(\omega_\tau (i))$ is unspecified, to identify the continuum equation corresponding to the particle-based model of Section~\ref{sec:stationary}. As in Section~\ref{sec:example}, the density $\rho_\tau (i) = \mathds{E}^{\theta_i} \omega_\tau (i)$ is the object of interest, the scaling of which wholly characterizes the gradient process in the limit of macroscopic time and space scales. We denote the particle model's time by $\tau$ and choose the scaling $t = \tau/dL^2$, $x = i/L$, with the interpretation that we ``zoom out'' by a factor of $L$ and speed up the process by a factor of $L^2$, in order to observe changes on the new spatial scale. This is the idea expressed in Figure~\ref{fig:scaling_schematic} with the small parameter $\varepsilon$ chosen in terms of the hillslope length as $\varepsilon = 1/L$, so $\varepsilon \rightarrow 0$ as $L \rightarrow \infty$. The time constant $d$ will become relevant in Section~\ref{sec:dimension}. We thus identify the rescaled density as \begin{linenomath*}\begin{equation*}\label{eq:rescaled_density} \varrho_t (x) := \mathds{E}^{\rho} \omega_{tdL^2} (xL), \end{equation*}\end{linenomath*} where the expectation with respect to $\rho$ is justified in Appendix~\ref{app:rho}.
We require that $p$ and $q$ become increasingly close in value when scaling $\rho_\tau (i)$. The intuition for this choice comes from the $f(\omega_\tau (i)) = \omega_\tau (i)$ curves of Figure~\ref{fig:exp_curves}, which indicate that increasing $p$ relative to $q$ results in a profile more closely resembling a step function. The scaling procedure will only serve to accentuate this resemblance and so, to avoid a degenerate rescaled density $\varrho_t (x)$, we choose the ``weakly asymmetric'' limit, where $p = \frac{1}{2} + \frac{E}{L}$ and $q = \frac{1}{2} - \frac{E}{L}$, and where $E$ is a positive parameter. Note that, while our choices force $p > q$, we could just as easily address $p < q$ by swapping them.
We proceed to examine the time evolution of the density for a site $i$, which results from adjacent particles hopping to $i$ and particles at $i$ hopping away
\begin{linenomath*}\begin{equation*}
\frac{d}{d\tau}\rho_\tau (i) = \frac{d}{d\tau}\mathds{E}^{\rho} \omega_\tau (i) = \mathds{E}^{\rho} p f(\omega_\tau (i-1)) + \mathds{E}^{\rho} q f(\omega_\tau (i+1)) - \mathds{E}^{\rho} p f(\omega_\tau (i)) - \mathds{E}^{\rho} q f(\omega_\tau (i)).
\end{equation*}\end{linenomath*} We now substitute the weak asymmetry condition in the following way
\begin{linenomath*}\begin{align*}
\frac{d}{d\tau} \mathds{E}^\rho \omega_\tau (i) &= -\mathds{E}^\rho f\,(\omega_\tau (i)) + \frac{1}{2} \mathds{E}^\rho f\,(\omega_\tau (i+1))\\ &- \frac{E}{L} \mathds{E}^\rho f\,(\omega_\tau (i+1)) + \frac{1}{2} \mathds{E}^\rho f\,(\omega_\tau (i-1)) + \frac{E}{L} \mathds{E}^\rho f\,(\omega_\tau (i-1))\\
&= \frac{1}{2} \Big[ \mathds{E}^\rho f\,(\omega_\tau (i+1)) - 2 \mathds{E}^\rho f\,(\omega_\tau (i)) + \mathds{E}^\rho f\,(\omega_\tau (i-1)) \Big]\\ &- \frac{E}{L} \Big[ \mathds{E}^\rho f\,(\omega_\tau (i+1)) - \mathds{E}^\rho f\,(\omega_\tau (i-1)) \Big].
\end{align*}\end{linenomath*} We continue by defining $G(\rho) := \mathds{E}^\rho f\,(\omega)$ and substitute the rescaled $t$ and $x$ variables
\begin{linenomath*}\begin{align*}
\frac{1}{dL^2} \frac{\partial}{\partial t} \mathds{E}^{\rho} \omega_{tdL^2} (xL) &= \frac{1}{2} \Big[ G\big(\rho_{tdL^2} (xL+1)\big) - 2 G\big(\rho_{tdL^2} (xL) \big) + G\big( \rho_{tdL^2} (xL -1) \big) \Big] \\
& \quad - \frac{E}{L} \Big[ G\big( \rho_{tdL^2} (xL+1) \big) - G\big( \rho_{tdL^2} (xL -1) \big) \Big].
\end{align*}\end{linenomath*} Rearranging and identifying $\varrho_t (x)$, we find
\begin{linenomath*}\begin{align*}
\frac{\partial}{\partial t} \varrho_t (x) &= \frac{dL^2}{2} \Big[ G\left(\varrho_t \left(x + L^{-1}\right)\right) - 2 G\left(\varrho_t \left(x\right) \right) + G\left( \varrho_{t} \left(x - L^{-1}\right) \right) \Big] \\
& \quad - d E L \Big[ G\left( \varrho_t \left(x+L^{-1}\right) \right) - G\left( \varrho_t \left(x - L^{-1}\right) \right) \Big].\\
&\simeq \frac{d}{2} \frac{\partial^2}{\partial x^2} G(\varrho_t(x)) - 2dE \frac{\partial}{\partial x} G (\varrho_t (x)).
\end{align*}\end{linenomath*}
We conclude
\begin{linenomath*}\begin{equation}\label{eq:hydro}
\frac{\partial}{\partial t} \varrho_t(x) \simeq \frac{d}{2} \frac{\partial^2}{\partial x^2} G\big(\varrho_t (x) \big) - 2dE \frac{\partial}{\partial x} G\big(\varrho_t (x)\big).
\end{equation} \end{linenomath*}
To find the proper boundary conditions, we repeat the argument for the leftmost site
\begin{linenomath*}\begin{align*}
\frac{\partial}{\partial \tau} \mathds{E}^\rho \omega_\tau (1) &= \frac{1}{2} \left[ \mathds{E}^\rho f\,(\omega_\tau (2)) - \mathds{E}^\rho f\,(\omega_\tau (1)) \right] - \frac{E}{L} \left[ \mathds{E}^\rho f\,(\omega_\tau (2)) + \mathds{E}^\rho f\,(\omega_\tau (1)) \right]\\
\implies \frac{1}{L}\frac{\partial}{\partial t} \varrho_t (L^{-1}) &= \frac{dL}{2} \left[ G \left( \varrho_t \left(2L^{-1}\right) \right) - G \left( \varrho_t \left(L^{-1}\right) \right) \right] - dE \left[ G \left( \varrho_t \left(2L^{-1}\right) \right) + G \left( \varrho_t \left(L^{-1}\right) \right) \right]\\
&\simeq \frac{d}{2} \frac{\partial}{\partial x} G\left( \varrho_t (0) \right) - 2dE G\left( \varrho_t(0)\right).
\end{align*}\end{linenomath*} In the limit as $L\rightarrow\infty$, the $\frac{\partial}{\partial t}$ term drops out and we have the Robin boundary \begin{linenomath*}\begin{equation}\label{eq:lbc}
\frac{\partial}{\partial x} G \left( \varrho_t (0) \right) = 4 E G \left( \varrho_t (0) \right).
\end{equation}\end{linenomath*} Similarly, we obtain the following boundary condition for the rightmost site
\begin{linenomath*}\begin{equation}\label{eq:rbc}
\frac{\partial}{\partial x} G(\varrho_t (1)) = 4 E G (\varrho_t (1)).
\end{equation}\end{linenomath*} The boundary conditions \eqref{eq:lbc} and \eqref{eq:rbc} are consistent with the time-stationary solution of \eqref{eq:hydro}, together implying \begin{linenomath*}\begin{equation}\label{eq:statsol} \frac{\partial}{\partial x} G(\varrho_t (x)) = 4EG(\varrho_t (x)) \quad 0 \leq x \leq 1, \end{equation}\end{linenomath*} the general solution of which is $G (\varrho_t (x) ) = C e^{4Ex}$.
Equation~\ref{eq:hydro}, along with \eqref{eq:lbc} and \eqref{eq:rbc}, is the continuum description of the particle-based hillslope model. Note that, as in Section~\ref{sec:stationary}, this equation describes the evolution of the gradient process, and so its solutions must be integrated to obtain the corresponding hillslope profiles. In the special case of $f(\omega (i)) = \omega (i)$, $G(\varrho_t(x)) = \varrho_t (x)$, so the continuum equation is an advection-diffusion equation \begin{linenomath*}\begin{equation}\label{eq:specialhydro}
\frac{\partial}{\partial t} \varrho_t(x) \simeq \frac{d}{2} \frac{\partial^2}{\partial x^2} \varrho_t(x) - 2dE\frac{\partial}{\partial x} \varrho_t(x)
\end{equation}\end{linenomath*} with Robin boundary conditions \begin{linenomath*}\[ \frac{\partial}{\partial x} \varrho_t (0) = 4 E \varrho_t (0) \quad\text{and}\quad \frac{\partial}{\partial x} \varrho_t (1) = 4 E \varrho_t (1).\]\end{linenomath*} In the special case of $f(\omega (i)) = 1$ for $\omega (i) > 0$, $G(\varrho_t (x) ) = \varrho_t (x) / (1+\varrho_t (x))$, so the continuum equation has the following form \begin{linenomath*}\[ \frac{\partial}{\partial t} \varrho_t(x) \simeq \frac{d}{2} \frac{\partial^2}{\partial x^2} \frac{\varrho_t(x)}{1+\varrho_t (x)} - 2dE\frac{\partial}{\partial x} \frac{\varrho_t(x)}{1+\varrho_t (x)} \]\end{linenomath*} with Robin boundary conditions \begin{linenomath*}\[ \frac{\partial}{\partial x} \frac{\varrho_t (0)}{1+\varrho_t (0)} = 4 E \frac{\varrho_t (0)}{1+\varrho_t (0)} \quad\text{and}\quad \frac{\partial}{\partial x} \frac{\varrho_t (1)}{1+\varrho_t (1)} = 4 E \frac{\varrho_t (1)}{1+\varrho_t (1)}.\]\end{linenomath*} Appendix~\ref{sec:fokker} describes the solution to \eqref{eq:hydro} subject to the boundary conditions \eqref{eq:lbc} and \eqref{eq:rbc}.
\subsection{Scaling recap}\label{sec:scalingrecap}
We recall some key points from Section~\ref{sec:hydro} before describing simulations and dimensionalization.\begin{enumerate}
\item The scaling procedure consists of three steps: balancing incoming and outgoing particles, substituting the weak asymmetry condition, and substituting the rescaled variables.
\item The resulting continuum equation describes the density of gradient particles and is of advection-diffusion type.
\item The continuum equation contains a function $G$ which has simple, explicit forms when the rate function is linear or constant.
\item The scaling argument confirms that, even if the continuum equation is complicated, its solutions can easily be approximated by simulating the corresponding particle model.
\end{enumerate}
\section{Simulation and dimensionalization}\label{sec:sim}
The analysis of Section~\ref{sec:hydro} tells us that if we want to study the evolution of hillslopes according to \eqref{eq:hydro}, we can simulate the particle model of Section~\ref{sec:model} instead. As choices of rate $f(\omega) \neq \omega$ generally lead to a nonlinear PDE \eqref{eq:hydro}, simulating the particle model may often be preferable to an analytic approach or a numerical scheme. In addition to simulating the equilibrium hillslope profiles under various choices of $p$ and rate function $f$, we would also like to simulate the response of hillslopes to perturbations, such as river erosion or climate change (usually implemented by a change in a diffusion coefficient \citep{fernandes1997hillslope,mudd2004influence,roering2001hillslope}). We begin with simulations of equilibrium hillslope profiles.
\subsection{Equilibrium hillslope profiles}\label{sec:simequil}
When the hopping rates of the gradient particle system are chosen to be $f(\omega (i)) = \omega (i)$, the hillslope gradients satisfy Equation~\ref{eq:specialhydro}, which is solved by a drifting diffusion. For other choices of rates, the gradients evolve according to Equation~\ref{eq:hydro}. \citet{balazs2007convexity} showed that convex (concave) $f(\omega (i))$ implies convexity (concavity) of $G(\rho)$. To demonstrate these two cases, we pick constant and quadratic rates given by \big($f(\omega (i)) = 1$ for $\omega (i) > 0$, $f(\omega (i)) = 0$ for $\omega (i) = 0$ \big) and $f(\omega (i)) = {\omega (i)}^2$, respectively. As a result of Section~\ref{sec:hydro}, the behavior of these solutions can be understood by simulating the corresponding particle model. Stationary hillslope and gradient profiles are compared in Figure~\ref{fig:rate_profile}. In particular, Figure~\ref{fig:rate_profile}A and \ref{fig:rate_profile}B highlight that, for different choices of $p$, the profiles arising from linear, quadratic, and constant rates can be made relatively similar, but their curvatures differ. Figure~\ref{fig:rate_profile}C and \ref{fig:rate_profile}D show that, when $p$ is fixed, the profile arising from a constant rate is far steeper than those from linear and quadratic rates. Note that the profiles in the linear rate case can be calculated from (\eqref{eq:heights}), while the constant and quadratic results can be produced with the following simulation procedure.
We begin by specifying $f(\omega (i))$, parameters $H$, $L$, $p$, and the number of simulation time steps, $N$. We choose an initial height profile, which satisfies the boundary conditions, and use $\omega (i) = h (i) - h (i+1)$ to get the corresponding gradient profile. For each time step, we (i) apply $f(\omega (i))$ to $\omega (i)$, (ii) sample hop latencies from independent exponential distributions with rates $f(\omega (i))$, and (iii) update $\omega (i)$ and $h (i)$ to reflect the first hop, contingent on satisfying boundary conditions. We implemented this procedure and conducted all simulations in MATLAB (R2016b, The MathWorks, Inc., Natick, Massachusetts, United States); the code can be obtained by emailing the corresponding author.
\begin{figure}[htbp]
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=\textwidth]{figures/4_DRAFT_rate_profile_summary}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0.02,0.77) {\LARGE{\textbf{$h (i)$}}};
\node at (0.02,0.3) {\LARGE{\textbf{$h (i)$}}};
\node at (0.515,0.77) {\LARGE{\textbf{$\omega (i)$}}};
\node at (0.515,0.3) {\LARGE{\textbf{$\omega (i)$}}};
\node at (0.255,0.55) {{\LARGE{\textbf{$i$}}}};
\node at (0.255,0.07) {{\LARGE{\textbf{$i$}}}};
\node at (0.75,0.55) {{\LARGE{\textbf{$i$}}}};
\node at (0.75,0.07) {{\LARGE{\textbf{$i$}}}};
\node at (0.02,0.95) {\huge{\textbf{A}}};
\node at (0.52,0.95) {\huge{\textbf{B}}};
\node at (0.02,0.48) {\huge{\textbf{C}}};
\node at (0.52,0.48) {\huge{\textbf{D}}};
\end{scope}
\end{tikzpicture}
\caption[Choice of rate changes profile]{Equilibrium hillslope (\textbf{A} and \textbf{C}) and gradient profiles (\textbf{B} and \textbf{D}) for quadratic \big($f(\omega) = {\omega}^2$\big), linear \big($f(\omega) = \omega $\big), and constant \big($f(\omega) = 1$ if $\omega > 0$ \big) rates. For \textbf{A} and \textbf{B}, parameters were $p=0.52$ (quadratic), $p=0.51$ (linear), $p=0.505$ (constant), $H=L=100$. For \textbf{C} and \textbf{D}, parameters were $p=0.51$ (all rates), $H=L=100$. All curves were obtained as the average over 10 identical trials.}
\label{fig:rate_profile}
\end{figure}
\subsection{Hillslope perturbations and empirical flux}\label{sec:simperturb}
We now turn our attention to hillslopes perturbed away from equilibrium, to study the timescales over which hillslopes relax and the influence the parameters have over this process. Consider the gradient process with $f(\omega (i)) = \omega (i)$, $L = 100$, $H=1\times 10^4$, and $p = 0.51$. We initialize the process with $\omega (i)$ corresponding to $\texttt{ceil}(h (i))$, where the $h (i)$ are given by Equation~\ref{eq:heights}. We introduce a river-erosion-inspired perturbation, which conserves total gradient particle count, by skimming $50$ gradient particles from each site with at least that many. All of the skimmed particles are added to a single site, and we track $h (i)$ and $\omega (i)$ as the hillslope relaxes back to equilibrium (Figure~\ref{fig:vert_panels}). Figure~\ref{fig:vert_panels}A depicts the hillslope and gradient profiles maintaining equilibrium after $1$ $\times$ $10^6$ timesteps. Immediately after this frame, the perturbation was applied. Figure~\ref{fig:vert_panels}B and \ref{fig:vert_panels}C show the profiles smoothing and refilling the base at timesteps $1.1$ $\times$ $10^6$ and $2.5$ $\times$ $10^6$, respectively. By timestep $5$ $\times$ $10^6$, the hillslope resembles the equilibrium hillslope.
It is natural to wonder about the affect $p$ has on the rate of hillslope relaxation in response to perturbations which do not change the underlying dynamics. Consider the same process, with $p = 0.51$, $p = 0.55$, or $p = 0.60$. Take \begin{linenomath*}\begin{equation}\label{eq:relaxdist}\Delta h_t (i) : = \big| h_t (i) - h_0 (i) \big| \qquad\text{and}\qquad \Delta h_t := \sum_{i=1}^{L} \Delta h_t (i) \end{equation}\end{linenomath*} as measures of distance from the $h_0$ equilibrium. The results for $t = 0$ to $t = 8 \times 10^7$ are shown in Figure~\ref{fig:decay}. It seems that the larger $p$ is, the greater the asymmetry in hopping rates, and the faster the hillslope returns to equilibrium. However, the perturbation depends on the gradient profile, and larger values of $p$ are associated with steeper hillslopes, meaning the local slope is not controlled in the experiment.
\begin{figure}[htbp]
\centering
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[height=0.8\textheight]{figures/5_DRAFT_vert_panels2}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (-0.01,0.985) {\Large{\textbf{$H$}}};
\node at (-0.01,0.885) {\Large{\textbf{$h(i)$}}};
\node at (-0.01,0.785) {\Large{\textbf{$0$}}};
\node at (1.048,0.985) {\Large{\textbf{$\rho (L)$}}};
\node at (1.048,0.885) {\Large{\textbf{$\omega (i)$}}};
\node at (1.035,0.785) {\Large{\textbf{$0$}}};
\node at (0.437,0) {\Large{\textbf{$L+1$}}};
\node at (0.05,0) {\Large{\textbf{$1$}}};
\node at (0.967,0) {\Large{\textbf{$L$}}};
\node at (0.55,0) {\Large{\textbf{$1$}}};
\node at (0.1,0.80) {\huge{\textbf{A}}};
\node at (0.1,0.55) {\huge{\textbf{B}}};
\node at (0.1,0.30) {\huge{\textbf{C}}};
\node at (0.1,0.05) {\huge{\textbf{D}}};
\node at (0.265,1.015) {\Large{\textbf{Height}}};
\node at (0.765,1.015) {\Large{\textbf{Gradient}}};
\end{scope}
\end{tikzpicture}
\caption[Response to perturbation.]{Simulated hillslope response to a river-erosion-like perturbation. A hillslope in equilibrium (\textbf{A}) with linear rate $f(\omega)$ = $\omega$ is perturbed (\textbf{B}) and relaxes (\textbf{C} and \textbf{D}). The rows depict time steps $1$ $\times$ $10^6$ (\textbf{A}), $1.1 \times 10^6$ (\textbf{B}), $2.5 \times 10^6$ (\textbf{C}), and $5 \times 10^6$ (\textbf{D}), for a perturbation applied near the righthand boundary immediately after time step $1\times 10^6$. The particle system was initialized at equilibrium (Equation~\ref{eq:heights}) with parameters $p=0.51$, $H=1\times 10^4$, and $L=100$. At equilibrium, $\rho (i)$ is given by Equation~\ref{eq:density}.}
\label{fig:vert_panels}
\end{figure}
\begin{figure}[htbp]
\centering
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=0.8\textwidth]{figures/6_DRAFT_decay_comparison3}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0,0.54) {\huge{\textbf{$\Delta h_t $}}};
\node at (0.55,0.02) {{\huge{\textbf{$t$}}} {\Large($\times \, 10^7$ steps)}};
\end{scope}
\end{tikzpicture}
\caption[Relaxation to equilibrium.]{Hillslope profile relaxation in response to a perturbation, for a particle system with $f(\omega (i)) = \omega (i)$, $p=0.51$, $p=0.55$, or $p=0.60$, $H=1\times 10^4$, $L=100$, and $t = 0$ to $t = 8 \times 10^7$. $\Delta h_t$ (defined by \eqref{eq:relaxdist}) was normalized by its largest value over the simulation. Each curve is the average over 25 trials.}
\label{fig:decay}
\end{figure}
To separately test the affects of $p$ and local slope on the rate of hillslope relaxation, we identified contiguous, 10-site regions of equilibrium hillslopes, for various choices of $p$, which had slope similar to that of an equilibrium profile for a different choice of $p$ (Figure~\ref{fig:independent_response_profiles}A and \ref{fig:independent_response_profiles}B). We then perturbed these regions of similar slope by adding one quarter of the total number of gradient particles in that region to a single drop site. For the linear rate model, the time series of $\Delta h_t (i)$ (where $i$ was the drop site) were well-fit by exponential decays ($R^2 > 0.995$ in all cases) with identical time constants (Figure~\ref{fig:independent_response_profiles}C). We then conducted the same perturbation, but for all possible contiguous 10-site windows. The resulting exponential decays for sites $i=10, 15, \dots, 90$ had time constants which agreed with that of Figure~\ref{fig:independent_response_profiles} and are summarized in (Figure~\ref{fig:independent_response_profiles}D). These simulation results suggest that, for linear rate, the timescale over which hillslopes relax does not depend on $p$ or the local slope; this conclusion is in agreement with the calculation of Section~\ref{sec:fokker} and \eqref{eq:relax} in particular. We emphasize that this is \textit{not} the case in general.
\begin{figure}[htbp]
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=\textwidth]{figures/7_perturb_profile_summary}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0.02,0.77) {\LARGE{\textbf{$h (i)$}}};
\node at (0.02,0.3) {\LARGE{\textbf{$\Delta h_t (i)$}}};
\node at (0.515,0.77) {\LARGE{\textbf{$\omega (i)$}}};
\node at (0.515,0.3) {\LARGE{\textbf{$\Delta h_t (i)$}}};
\node at (0.27,0.55) {{\LARGE{\textbf{$i$}}}};
\node at (0.27,0.07) {{\LARGE{\textbf{$t$}}} {\Large($\times \, 10^3$ steps)}};
\node at (0.76,0.55) {{\LARGE{\textbf{$i$}}}};
\node at (0.76,0.07) {{\LARGE{\textbf{$t$}}} {\Large($\times \, 10^3$ steps)}};
\node at (0.015,0.95) {\huge{\textbf{A}}};
\node at (0.515,0.95) {\huge{\textbf{B}}};
\node at (0.015,0.48) {\huge{\textbf{C}}};
\node at (0.515,0.48) {\huge{\textbf{D}}};
\end{scope}
\end{tikzpicture}
\caption[The role of $p$ and gradient on hillslope relaxation in the linear rate model]{The role of $p$ and gradient on hillslope relaxation in the linear rate model. Equilibrium hillslope profiles for a variety of choices of $p$ (\textbf{A}) and the corresponding gradient profiles (\textbf{B}). Parameters were $H$ = $1$ $\times $ $10^4$ and $L$ = $100$, with linear rate $f (\omega)$ = $\omega$. The gradient profiles overlap around $i$ = $70$, so we can control for the affect of local slope on the rate of hillslope relaxation by perturbing in the overlap region. For each choice of $p$, the perturbation (applied at the beginning of the simulation) consisted of taking one quarter of the gradient particles from each of 10 sites in an interval centered on $i$ = $70$, and adding them all to the leftmost site in the interval. The resulting time series of $\Delta h_t (i)$ (defined by \eqref{eq:relaxdist}) were well-fit by exponential decay with common time constant $1.47\times 10^{-4}$ (\textbf{C}). $R^2$ $>$ $0.995$ in all cases. In (\textbf{D}), we fixed $p$ = $0.51$ and performed the perturbation experiment using a sliding, 10-site window, in order to test various local gradients along the hillslope. The resulting, normalized $\Delta h_t $ decays for $i = 10, 15, \dots, 85, 90$ are shown (thin black curves) with the exponential fit superimposed (thick red curve). Each curve in \textbf{C} and \textbf{D} was the average of 25 identical trials.}
\label{fig:independent_response_profiles}
\end{figure}
Fluxes develop along the hillslope during the process of equilibration which, while not directly accessible via the methods of Section~\ref{sec:hydro}, can be approximated by an ``empirical'' flux inferred from height changes along the hillslope. For example, growth downslope of site $i$ suggests that a flux arose upslope of site $i$. As this indirect measurement of flux relies on height changes, it depends on two observations times $t$ and $t + \Delta t$. We calculate the empirical flux at site $i$, relative to time steps $t$ and $t + \Delta t$ as
\begin{linenomath*}\begin{equation}\label{eq:empiricalflux}
\phi_{t+\Delta t} (i) - \phi_t (i) = r \, \Delta t + \sum_{j > i} \big( h_{t+\Delta t} (j) - h_t (j) \big).
\end{equation}\end{linenomath*}
Here, $r$ is a constant flux coming from the right boundary and we adopt the convention that a positive value of flux at a site $i$ indicates a net, relative height change for sites $j > i$.
To demonstrate the use of the empirical flux, we consider a hillslope with $H=L=100$, initially at equilibrium with $p=0.51$. For convenience, we choose $\Delta t$ to be the length of one time step in the simulation. Immediately after $t=0$, we switch to $p=0.55$, producing a net positive flux toward the righthand boundary, as the hillslope tries to equilibrate. To isolate the flux contributions driven by equilibration from those of the constant flux $r$, we instead track the \textit{cumulative} flux through site $i$ as
\begin{linenomath*}\begin{equation}\label{eq:phibar}
\overline{\phi}_{t} (i) := \phi_{t} (i) - \phi_{0} (i) - r t.
\end{equation}\end{linenomath*} Figure~\ref{fig:flux}A shows the before-and-after hillslope profiles, corresponding to $p=0.51$ and $p=0.55$, and Figure~\ref{fig:flux}B shows the cumulative flux through sites $i=25$, $50$, and $75$ during equilibration.
\begin{figure}[htbp]
\centering
\begin{tikzpicture}
\node[anchor=south west,inner sep=0] (image) at (0,0) {\includegraphics[width=\textwidth]{figures/8_flux_decay_comparison}};
\begin{scope}[x={(image.south east)},y={(image.north west)}]
\node at (0.01,0.54) {\LARGE{\textbf{$h(i)$}}};
\node at (0.25,0.04) {\LARGE{\textbf{$i$}}};
\node at (0.55,0.54) {\LARGE{\textbf{$\overline{\phi}_t (i)$}}};
\node at (0.8,0.04) {\LARGE{\textbf{$t$}}};
\node at (0.01,0.92) {\huge{\textbf{A}}};
\node at (0.55,0.92) {\huge{\textbf{B}}};
\end{scope}
\end{tikzpicture}
\caption[Empirical fluxes during equilibration.]{A hillslope equilibrated for $p$ = $0.51$, $H$ = $100$, $L$ = $100$, and linear rate $f(\omega)$ = $\omega$, is perturbed by an abrupt change in the dynamics to $p$ = $0.55$. In \textbf{A}, the initial profile (solid line) evolves with updated $p$ to the final, near-equilibrium hillslope (dotted line). In \textbf{B}, cumulative fluxes $\overline{\phi}_t (i)$ (defined by \eqref{eq:phibar}) develop in response to the perturbation. Cumulative fluxes are shown for sites $i=25$, $i=50$, and $i=75$, averaged over 100 identical trials. By convention, a flux at site $i$ is positive if it indicates net hillslope height increase for sites $j > i$.}
\label{fig:flux}
\end{figure}
\subsection{Adding dimensions and fitting parameters}\label{sec:dimension}
In order to reliably translate simulation results into empirically testable predictions, we need a principled way of assigning dimensions to otherwise dimension-less model quantities (e.g. particle model length $L$ and the length $\ell$ of an observed hillslope, in meters). Additionally, we need to specify how hillslope data are used to fit model parameters. We suggest the following procedure, which is partly motivated by the calculations in Appendix~\ref{sec:fokker}.
Recall that sites in the particle model of Section~\ref{sub:dynamics} are indexed by $i = 1,2,\dots,L$. Let $i$ count the number of sediment grains in the length of the hillslope. If a typical grain has a diameter of $2$ millimeters and the hillslope length is measured to be $\ell = 200$ meters, then set $L$ = $200$ meters / $2$ millimeters = $100\,000$. Similarly, if the crest of the hillslope is $h = 100$ meters above the height at the end of the hillslope (at a distance $\ell$ meters from the crest), assign $H$ = $100$ meters / $2$ millimeters = $50\,000$. In this way, we relate dimensionless particle model quantities $L$ and $H$ to observable hillslope quantities with dimension, $\ell$ and $h$.
We now consider fitting $E$, which encapsulates the asymmetry in the underlying gradient process, and adding dimension to the simulation timesteps. For simplicity, we consider the case of the linear rate model, but the following procedure can be applied to nonlinear rate models using the contents of Appendix~\ref{app:nonlin}. We can estimate the parameter $E$ from measurements of the equilibrium or near-equilibrium shape of the hillslope, by fitting \eqref{eq:hx}. Next, we can add dimension to the simulation timesteps by fitting the time constant $d$, which was introduced in the scaling argument of Section~\ref{sec:hydro}. Fitting $d$ requires that a small perturbation $r_0$ be added to the hillslope, the relaxation of which obeys \eqref{eq:relax}. Ideally, the location of the perturbation and the timescale of relaxation should be such that the boundaries do not play a significant role. To summarize, we suggest the following, three-step approach. \begin{enumerate} \item Measure typical grain diameter to add units to $H$ and $L$. \item Fit $E$ to equilibrium hillslope shape. \item Fit $d$ to hillslope relaxation in response to a perturbation.\end{enumerate}
While the first step does not depend on the choice of rate function, the second and third steps do, as the form of the rate affects the the relationship between $E$ and the equilibrium hillslope shape, and relationship between $d$ and the relaxation of perturbations. We also note that this procedure makes use of both small-scale and large-scale measurements, as well as information about hillslope equilibrium and nonequilibrium.
\subsection{Simulation recap}\label{sec:simrecap}
We collect some key points from Section~\ref{sec:sim} before continuing on to the discussion.\begin{enumerate}
\item We simulated perturbations in two ways: rearranging the gradient particles (through $\omega$) and changing the dynamics (through $p$ or, equivalently, $E$).
\item Hillslope relaxation in response to perturbation can be tracked by comparing it with the corresponding stationary profile or by tracking the empirical fluxes.
\item In the linear rate case, hillslope relaxation timescale is independent of $E$, $H$, and $L$ .
\item The simulation results can be assigned dimensions to facilitate comparison with observations, according to the procedure of Section~\ref{sec:dimension}.
\end{enumerate}
\section{Discussion}\label{sec:disc}
The key ingredient of the particle model of Section~\ref{sec:model} is indirection: the decision for particles to represent units of hillslope gradient, instead of units of hillslope height. Consider again the scenario of Figure~\ref{fig:fixedheightzrp}. Had we specified similar dynamics on the hillslope profile directly, the resulting profiles could be unrealistic (e.g. large particle build-up next to sites with no particles) and the dynamics would require awkward constraints to prevent such profiles. Most importantly, this process would not have stationary profiles which are amenable to analysis, and a scaling argument like that of Section~\ref{sec:hydro} would not apply. In this sense, the gradient particle model is a natural choice, but one made at the expense of direct access to information about sediment flux and particle hopping distances. Indeed, although we can obtain the hillslope profile from the gradient particle profile (using \eqref{eq:inversion}), our model does not prescribe a dynamics on the hillslope profile and so is agnostic to fluxes of hillslope particles and the distances they typically travel. Critically, this circumvents the issue of specifying whether transport on the hillslope is local or nonlocal and, as a result, our model can represent a variety of geomorphic processes and the scaling argument holds across transport regimes.
We are free to accessorize our model with fluxes, defined in terms of hillslope gradient, which evolve according to the particle model of Section~\ref{sec:model} or, in the continuum, according to \eqref{eq:hydro}. In Section~\ref{sec:simperturb}, for example, we proposed a nonlocal flux \eqref{eq:empiricalflux} in terms of changes in the hillslope height (equivalently, changes in hillslope gradient via \eqref{eq:inversion}).
Alternatively, we could specify a local flux like those of \citet{culling1963soil} (linear dependence on slope), \citet{andrews1987fitting} (nonlinear dependence on slope), and \citet{furbish2009rain} (nonlinear, includes height and slope), or a nonlocal flux of the form favored by \citet{furbish2013sediment}. This freedom reflects the \textit{hillslope-first} nature of our particle model, for which we formulate the dynamics of the hilllslope gradients and infer the flux, as opposed to formulating the dynamics of the flux, from which we then infer the hillslope profile.
Such a hillslope-first approach may be more natural than a nonlocal, transport-first approach for conducting perturbation experiments like those described in Section~\ref{sec:simperturb}. For example, consider the experiment illustrated by Figures~\ref{fig:vert_panels} and \ref{fig:decay}, which simulates hillslope recovery from river erosion. Nonlocal formulations of transport require as input a distribution of particle travel distances \citep{furbish2010divots} or an assumption about the degree of nonlocality \citep{foufoula2010nonlocal}, but these features depend on the hillslope gradient, and so should vary throughout the experiment \citep{gabet2012particle}. In contrast, our model fixes the law governing the redistribution of hillslope gradient through the rate function $f$, which is an input of the modeler.
Given a choice of $f$, the parameter $p$ can be determined from an observation of hillslope shape, according to the procedure described in Section~\ref{sec:dimension}. Intuitively, for a given rate function, $p > \frac{1}{2}$ specifies a deposition-type process; $p < \frac{1}{2}$ specifies a washing-out-type process. For example, in Figure~\ref{fig:exp_curves}, $p = 0.49$ produces a stationary hillslope profile resembling one formed under sheet wash with gullies, while $p=0.51$ results in a profile which more closely resembles one formed under soil creep. The parameter $p$ can also be used to conduct perturbation experiments, as in Figure~\ref{fig:flux}, where the hillslope begins as the stationary profile under a process associated with $p=0.51$ and must equilibrate after an external driver (e.g. climate change) alters the dynamics to $p=0.55$. Unlike the case of a river-erosion-like perturbation, it may be possible to use a nonlocal, transport-first approach to conduct similar experiments, for example, by making a small change to a parameter in the distribution of particle travel distances.
The particle-based model of Section~\ref{sec:model} is purely probabilistic, unlike those of \citet{kirkby1975surface}, \citet{gabet2012particle}, which incorporate frictional forces associated with particle motion, and that of \citet{dibiase2017slope}, which also accounts for variations in grain size and is extended to motion in two spatial dimensions. These approaches benefit from directly incorporating hillslope microtopography, but are computationally-expensive in a way which may prohibit the simulation of hillslope evolution over long timescales \citep{dibiase2017slope}, and cannot be scaled to corresponding continuum equations \citep{ancey2015stochastic}. Our model is most similar to that of \citet{tucker2010trouble}, which is also purely probabilistic, rules-based, and computationally-inexpensive, but for which a corresponding continuum description is unavailable.
The scaling argument of Section~\ref{sec:hydro} claims that, under the appropriate scalings of time and space variables, and in the limit as $L\rightarrow\infty$, the model behaves according to an advection-diffusion equation. Note that this governs the scaled gradient process, not the hillslope itself -- we must integrate the solutions to obtain the corresponding hillslope. For the linear rate case, we can solve the continuum equation directly; in the nonlinear rate case, numerical methods may be required. Both the scaling argument and the resulting continuum equation are general; they hold for any non-negative, non-decreasing rate function $f$. Of course, if $f$ is complicated, so too will the continuum equation be (as in Appendix~\ref{app:nonlin}), and simulating the corresponding particle model will likely be preferable. As described in Section~\ref{sec:dimension}, we can use the continuum equation to fit the time constant $d$ with field data, which allows simulation timesteps to be translated into the timescale of the data. We emphasize that the scaling procedure both identifies a continuum model, as well as justifies the continuum model's approximation by simulations of the particle model, assuming $L$ is relatively large. The dimensionalization procedure of Section~\ref{sec:dimension} confirms that this condition will be satisfied in practice, as typical values for grain diameter and hillslope length give $L = 10^5$.
We anticipate that the modeling approach described here will be particularly useful for long-timescale simulations and simulations of landscape relaxation in response to perturbations. As simulations of the particle model are easy to implement and computationally inexpensive, they could be used to evaluate the long-term impact of external drivers or could be incorporated as one component of a larger landscape model (e.g. hillslope with runoff into a river) while respecting modest computational resources. In addition, the simplicity of the particle model makes it possible to simulate the interaction of sophisticated perturbations, such as intermittent weather patterns or avalanching, with baseline geomorphic processes. Equipped with the dimensionalization procedure, these simulations can be informed by observations of individual grains and entire hillslopes, as well as stationary and perturbed hillslopes, and ultimately translated into concrete predictions.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 6,689 |
The check can be used to monitor the temperature status of Fujitsu Siemens (FSC) Systems.
This check uses the data provided by the SNMP agents supporting the SNI-SERVER-CONTROL-MIB MIB. The check has been created using Fujitsu Siemens Primergy Servers but should work on other hardware too.
The check uses the thresholds provided by the SNMP agent to raise WARNING/CRITICAL states.
The check creates one service for each temperature sensor provided in the SNMP data. | {
"redpajama_set_name": "RedPajamaC4"
} | 3,353 |
On the roof of the swimming pool !
In Paris, we waited for the sun and the heat for a long time. For the past three weeks, the weather is wonderful. Unfortunately, I broke a little bone in my foot... So I have to stay quiet in my flat to be cured as soon as possible and to be able to walk properly in Manchester !
I live on the seventh floor. From my kitchen window, I can see the rooftop of the swimming pool just down among the buildings. After their bath, people go to sunbathe on the roof. But they can't imagine that I'm capturing them with my brushes and my colors!
I really like to observe them, to catch their behaviours, how they look when they think that nobody's watching.
The sun plays with the glass panes of the roof of the swimming pool ; it's very funny for me to create a dialogue between the cool ligths and shades of the roof and the warm hues of the bodies.
I have to paint very quickly because people move all the time and they can't stay perfectly still. I like this challenge.
I can't walk along the beach and swim in the sea but I have an incredible series of swimming-pool watercolors ! | {
"redpajama_set_name": "RedPajamaC4"
} | 6,009 |
\section{Introduction}
We start from the famous arithmetic-geometric mean inequality which is often called Young inequality:
\begin{equation} \label{ineq01_rev}
(1-\nu) a+ \nu b \geq a^{1-\nu} b^{\nu}
\end{equation}
for nonnegative real numbers $a$, $b$ and $\nu \in [0,1]$.
Recently, the inequality (\ref{ineq01_rev}) was refined by F.Kittaneh and Y.Manasrah in the following form,
for the purpose of the study on matrix norm inequalities.
\begin{Prop} {\bf (\cite{BK,KM,IN})} \label{prop02}
For $a, b\geq 0$ and $\nu \in [0,1]$, we have
\begin{equation} \label{ineq02_rev}
(1-\nu) a+ \nu b \geq a^{1-\nu} b^{\nu}+ r (\sqrt{a}-\sqrt{b})^2,
\end{equation}
where $r\equiv \min\{\nu,1-\nu\}$.
\end{Prop}
It is notable that the inequality (\ref{ineq02_rev}) was first proved in (6.32) on page 46 of the reference \cite{BK}.
In the section 2 of this paper, we give refined Young inequalities for two positive operators based on the scalar inequality (\ref{ineq02_rev}).
As for the reverse inequalities of the Young inequality, M.Tominaga gave the following interesting operator inequalities.
He called them {\it converse} inequalities, however we use the term {\it reverse} for such inequalities, throughout this paper.
\begin{Prop} {\bf (\cite{Tom1})}
Let $\nu \in [0,1]$, positive operators $A$ and $B$ such that $0<mI \leq A,B \leq MI $ with $h\equiv \frac{M}{m}>1$.
Then we have the following inequalities for every $\nu \in [0,1]$.
\begin{itemize}
\item[(i)] (Reverse ratio inequality)
$$
S(h) A \sharp_{\nu}B \geq (1-\nu) A + \nu B,
$$
where the constant $S(h)$ is called Specht's ratio \cite{Specht,JIFUJII} and defined by
$$
S(h) \equiv \frac{h^{\frac{1}{h-1}}}{e\log{h^{\frac{1}{h-1}}}},\quad (h \neq 1)
$$
for positive real number $h$.
\item[(ii)] (Reverse difference inequality)
\begin{equation}\label{prop_rev_diff_ineq}
h L(m,M)\log S(h) +A \sharp_{\nu}B \geq (1-\nu) A + \nu B,
\end{equation}
where the logarithmic mean $L$ is defined by
$$
L(x,y)\equiv \frac{y-x}{\log y-\log x},\,\,(x\neq y)\quad L(x,x)\equiv x
$$
for two positive real numbers $x$ and $y$.
\end{itemize}
\end{Prop}
In the section 3 of this paper, we give reverse ratio type inequalities of the refined Young inequality for positive operators.
In the section 4 of this paper, we also give reverse difference type inequalities of the refined Young inequality for positive operators.
\section{Refined Young inequalities for positive operators}
Let $\fH$ be a complex Hilbert space. We also represent the set of all bounded operators on $\fH$ by $B(\fH)$.
If $A\in B(\fH)$ satisfies $A^*=A$, then $A$ is called a self-adjoint operator.
A self-adjoint operator $A$ satisfies $\langle x \vert A \vert x\rangle \geq 0$ for any $\vert x \rangle \in \fH$, then $A$ is called a positive operator.
For two self-adjoint operators $A$ and $B$, $A\geq B$ means $A-B\geq 0$.
It is well-known that we have the following Young inequalities for invertible positive operators $A$ and $B$:
\begin{equation} \label{orig_Young_ineq}
(1-\nu) A+ \nu B \geq A\sharp_{\nu}B \geq \left\{(1-\nu) A^{-1}+ \nu B^{-1} \right\}^{-1},
\end{equation}
where $A\sharp_{\nu}B \equiv A^{1/2}(A^{-1/2}BA^{-1/2})^{\nu}A^{1/2}$ defined for $\nu\in[0,1]$.
The power mean was originally introduced in the paper \cite{KA}.
The simplified and elegant proof for the inequalities (\ref{orig_Young_ineq})
was given in \cite{FY}. See also \cite{Furuta} for the reader having interests in operator
inequalities.
As a refinement of the inequalities (\ref{orig_Young_ineq}), we have the following refined Young inequality for positive operators.
\begin{The} \label{the01}
For $\nu \in [0,1]$ and positive operators $A$ and $B$, we have
\begin{eqnarray}
(1-\nu )A+\nu B &\geq& A\sharp_{\nu} B + 2r \left(\frac{A+B}{2} -A\sharp_{1/2}B \right) \label{the01-ineq01} \\
&\geq & A\sharp_{\nu} B \label{the01-ineq02} \\
&\geq& \left\{ A^{-1}\sharp_{\nu} B^{-1}+2r \left(\frac{A^{-1}+B^{-1}}{2} -A^{-1}\sharp_{1/2}B^{-1} \right)\right\}^{-1} \label{the01-ineq03} \\
&\geq& \left\{ (1-\nu )A^{-1}+\nu B^{-1} \right\}^{-1} \label{the01-ineq04}
\end{eqnarray}
where $r\equiv \min\left\{\nu,1-\nu\right\}$ and $A\sharp_{\nu}B \equiv A^{1/2}(A^{-1/2}BA^{-1/2})^{\nu}A^{1/2}$ defined for $\nu\in[0,1]$.
\end{The}
To prove Theorem \ref{the01}, we use the following lemma.
\begin{Lem} \label{lem_hg}
For invertible positive operators $X$ and $Y$, we have
$$
(X+Y)^{-1} = X^{-1} -X^{-1}(X^{-1}+Y^{-1})^{-1}X^{-1}.
$$
\end{Lem}
{\it Proof}:
Since $(X+Y)(X+Y)^{-1}X=X$, we have $ X(X+Y)^{-1}X + Y(X+Y)^{-1}X =X$.
Thus we have
$X(X+Y)^{-1}X =X-Y(X+Y)^{-1}X =X-(X^{-1}(X+Y)Y^{-1})^{-1}=X-(X^{-1}+Y^{-1})^{-1}$.
Multiplying $X^{-1}$ from both sides, we obtain the lemma.
\hfill \qed
{\it Proof of Theorem \ref{the01}}:
The second inequality (\ref{the01-ineq02}) is clear, since we have $2r \left(\frac{A+B}{2} -A\sharp_{1/2}B \right) \geq 0$. We prove the first inequality.
From the inequality (\ref{ineq02_rev}), we have for $\nu \in [0,1]$ and $x \geq 0$
$$\nu x+1-\nu-x^{\nu} -r(\sqrt{x}-1)^2 \geq 0.$$
By the standard operational calculus, we have
\begin{eqnarray}
\nu T +1-\nu &\geq& T^{\nu} +r (T^{1/2}-1)^2 \nonumber \\
&=&T^{\nu} +r (T-2T^{1/2}+1) \label{ineq01}
\end{eqnarray}
for a positive operator $T$ and $\nu \in [0,1]$.
From here, we suppose that $A$ is an invertible.
(For a general case, we consider the invertible positive operator $A_{\epsilon}\equiv A+\epsilon I$
for positive real number $\epsilon$. If we take a limit $\epsilon \to 0$, the following result also holds.
Throughout this paper, we apply this continuity argument, however, from now on, we omit such descriptions for simplicity.)
Substituting $T=A^{-1/2}BA^{-1/2} $ into the inequality (\ref{ineq01}), we have
$$
\nu A^{-1/2}BA^{-1/2} +1-\nu \geq \left(A^{-1/2}BA^{-1/2}\right) ^{\nu} +r \left\{ A^{-1/2}BA^{-1/2}
-2\left(A^{-1/2}BA^{-1/2}\right) ^{1/2} +1 \right\}
$$
Multiplying $A^{1/2}$ to the above inequality from both sides, we have
$$
(1-\nu) A+ \nu B \geq A\sharp_{\nu}B +r \left( A+B -2 A \sharp_{1/2}B\right),
$$
which proves the inequality (\ref{the01-ineq01}).
Replacing $A$ and $B$ by $A^{-1}$ and $B^{-1}$ in the inequality (\ref{the01-ineq01}),
respectively and taking the inverse of both sides, then we have the last inequality (\ref{the01-ineq04}).
By Lemma \ref{lem_hg}, the right hand side of the inequality (\ref{the01-ineq03}) can be calculated as
$
R.H.S.= A\sharp_{\nu}B -\left(A\sharp_{\nu}B\right) \left[ A\sharp_{\nu}B + \left\{ 2r\left( \frac{A^{-1}+B^{-1}}{2}-A^{-1}\sharp_{1/2}B^{-1}\right) \right\}^{-1} \right]^{-1} \left(A\sharp_{\nu}B\right).
$
Since, $\left(A^{-1}\sharp_{\nu}B^{-1}\right)^{-1}=A\sharp_{\nu}B \geq 0$
and $2r\left( \frac{A^{-1}+B^{-1}}{2}-A^{-1}\sharp_{1/2}B^{-1}\right)\geq 0$, we have the third inequality (\ref{the01-ineq03}),
which completes the proof.
\hfill \qed
In the paper \cite{Furuta1}, the equivalent relation between the Young inequality and the H\"older-McCarthy inequality \cite{Mc}
was shown by a simplified elegent proof.
Here we show a kind of the refinement of the H\"older-McCarthy inequality applying Theorem \ref{the01}.
\begin{Cor}
For $\nu\in[0,1]$ and any positive operator $A$ on the Hilbert space $\fH$ and any unit vector $\vert x \rangle \in \fH$, if
$\langle x \vert A \vert x \rangle \neq 0$, then we have
\begin{equation} \label{ineq02}
1- \langle x \vert A\vert x\rangle ^{-\nu} \langle x \vert A^{\nu }\vert x\rangle
\geq r\left( 1- \langle x \vert A\vert x\rangle ^{-1/2} \langle x \vert A^{1/2 }\vert x\rangle \right)^2,
\end{equation}
where $r\equiv \min\left\{\nu,1-\nu\right\}$.
\end{Cor}
{\it Proof}:
If $\nu =0$, then the inequality (\ref{ineq02}) is trivial. It is sufficient that we prove it for the case of $\nu\in (0,1]$.
In the inequality (\ref{ineq01}) ,
we put $T=k^{\frac{1}{\nu}}A$, for any positive real number $k$ and by the unit vector $\vert x \rangle \in \fH$, we have
\begin{equation}\label{ineq03}
\nu k^{\frac{1}{\nu}} \langle x \vert A \vert x \rangle +1-\nu \geq k \langle x \vert A^{\nu} \vert x \rangle
+r\left( k^{\frac{1}{2\nu}} \langle x \vert A^{1/2} \vert x \rangle -1\right)^2
\end{equation}
In the inequality (\ref{ineq03}), if we put $k=\langle x \vert A \vert x \rangle ^{-\nu}$, then we obtain the inequality (\ref{ineq02}).
\hfill \qed
\begin{Rem}
From H\"older-McCarthy inequality \cite{Mc}:
\begin{equation} \label{holder01}
\langle x \vert A \vert x \rangle ^{\nu} \geq \langle x\vert A^{\nu} \vert x\rangle
\end{equation}
for any unit vector $\vert x \rangle \in \fH$,
if $\langle x \vert A \vert x \rangle \neq 0$, then we have
$$
1- \langle x \vert A\vert x\rangle ^{-\nu} \langle x \vert A^{\nu }\vert x\rangle \geq 0.
$$
The inequality (\ref{ineq02}) gives a refined one for the above inequality
which is equivalent to the inequality (\ref{holder01}) in the case of $\langle x \vert A \vert x \rangle \neq 0$.
\end{Rem}
\section{A reverse ratio inequality for a refined Young inequality}
For positive real numbers $a,b$ and $\nu\in[0,1]$, M.Tominaga showed the following inequality \cite{Tom1}:
\begin{equation}\label{ineq21}
S\left(\frac{a}{b} \right) a^{1-\nu} b^{\nu} \geq (1-\nu) a + \nu b,
\end{equation}
which is called the converse ratio inequality for the Young inequality in \cite{Tom1}.
In this section, we show the reverse ratio inequality of the refined Young inequality (\ref{ineq02_rev}).
\begin{Lem} \label{lem21}
For positive real numbers $a, b$ and $\nu \in [0,1]$, we have
\begin{equation}\label{ineq22}
S\left(\sqrt{\frac{a}{b}} \right)a^{1-\nu}b^{\nu} \geq (1-\nu) a + \nu b -r(\sqrt{a}-\sqrt{b})^2,
\end{equation}
where $r\equiv \min\left\{\nu,1-\nu \right\}$.
\end{Lem}
{\it Proof}:
\begin{itemize}
\item[(i)]For the case of $\nu \leq1/2$, $r=\nu$.
We consider the following function.
$$
g_b(\nu) \equiv \frac{\nu b+(1-\nu) -\nu(\sqrt{b} -1)^2}{b^{\nu}},\quad \left(0\leq \nu \leq \frac{1}{2}\right) .
$$
Then we have
$$
g'_b(\nu) = \frac{2(\sqrt{b}-1)-\left\{2(\sqrt{b}-1)\nu + 1\right\}\log b}{b^{\nu}}
$$
so that the equation $g'_b(\nu) = 0$ implies
$$\nu = \frac{1}{\log b}-\frac{1}{2(\sqrt{b}-1)} \equiv \nu_b.$$
From the Klein inequality:
$$
1-\frac{1}{\sqrt{b}} \leq \log \sqrt{b} \leq \sqrt{b} -1,\quad (b>0)
$$
we have $\nu_b \in [0,\frac{1}{2}]$.
We also find that $g'_b(\nu) >0$ for $\nu < \nu_b$ and $g'_b(\nu) <0$ for $\nu > \nu_b$.
Thus the function $g_b(\nu)$ takes a maximum value when $\nu=\nu_b,\, (b\neq 1)$ and it is calculated as follows.
\begin{eqnarray*}
\max_{0\leq\nu\leq \frac{1}{2}} g_b(\nu)= g_b(\nu_b)
&=& \frac{2(\sqrt{b}-1)\left(\frac{1}{\log b}-\frac{1}{2(\sqrt{b}-1)}\right)+1}{b^{\frac{1}{\log b}}b^{\frac{-1}{2(\sqrt{b}-1)}}}\\
&=&\frac{\frac{2(\sqrt{b}-1)}{\log b}}{eb^{\frac{-1}{2(\sqrt{b}-1)}}}
=\frac{\left(\sqrt{b}\right)^{\frac{1}{\sqrt{b}-1}}}{e\log{\left(\sqrt{b}\right)}^{\frac{1}{\sqrt{b}-1}}}=S(\sqrt{b}).
\end{eqnarray*}
Thus we have the following inequality.
\begin{equation}
S(\sqrt{b})b^{\nu} \geq \nu b+(1-\nu)-\nu(\sqrt{b}-1)^2.
\end{equation}
In the case of $b=1$, we have the equality in the above inequality, since we have $S(1)=1$.
Replacing $b$ by $\frac{b}{a}$ and then multiplying $a$ to both sides, we have
\begin{equation}
S\left(\sqrt{\frac{a}{b}}\right)a^{1-\nu}b^{\nu} \geq (1-\nu) a+ \nu b-\nu(\sqrt{a}-\sqrt{b})^2,
\end{equation}
since we have $S(x)=S(1/x)$ for $x>0$.
\item[(ii)] For the case of $\nu \geq 1/2$, $r=1-\nu$.
We consider the following function.
$$
h_a(\nu) \equiv \frac{\nu+(1-\nu)a-(1-\nu)(1-\sqrt{a})^2}{a^{1-\nu}},\quad \left(\frac{1}{2}\leq \nu \leq 1 \right).
$$
By the similar way to (i), we have
$$
h_a'(\nu)=0 \Leftrightarrow \nu = 1- \left(\frac{1}{\log a}-\frac{1}{2(\sqrt{a}-1)} \right) \equiv \nu_a.
$$
We also find $\nu_a \in [\frac{1}{2},1]$ and $h'_a(\nu) >0$ for $\nu < \nu_a$ and $h'_a(\nu) <0$ for $\nu > \nu_a$.
Thus the function $h_a(\nu)$ takes a maximum value when $\nu=\nu_a,\, (a\neq 1)$ and it is calculated by
$$\max_{\frac{1}{2}\leq \nu \leq 1}h_a(\nu)=h_a(\nu_a)=S(\sqrt{a}).$$
Therefore we have the following inequality.
\begin{equation}
S(\sqrt{a})a^{1-\nu} \geq \nu +(1-\nu)a-(1-\nu)(1-\sqrt{a})^2.
\end{equation}
In the case of $a=1$, we have the equality in the above inequality, since we have $S(1)=1$.
Replacing $a$ by $\frac{a}{b}$ and then multiplying $b$ to both sides, we have
\begin{equation}
S\left(\sqrt{\frac{a}{b}}\right)a^{1-\nu}b^{\nu} \geq (1-\nu) a+ \nu b -(1-\nu)(\sqrt{a}-\sqrt{b})^2.
\end{equation}
\end{itemize}
From (i) and (ii), the proof is completed.
\hfill \qed
\begin{Rem}
We easily find that both sides in the inequality (\ref{ineq22}) is less than or equal to thoes in the inequality (\ref{ineq21}) so that
neither the inequality (\ref{ineq22}) nor the inequality (\ref{ineq21}) is uniformly better than the other.
In addition, our next interest is the ordering between $S\left(\sqrt{\frac{a}{b}}\right) a^{1-\nu}b^{\nu}$ and $(1-\nu) a+ \nu b$.
However we have no ordering between them, because we have the following examples.
For example, let $a=1$ and $b=10$.
If $\nu=0.9$, then $(1-\nu) a+ \nu b - S\left(\sqrt{\frac{a}{b}}\right) a^{1-\nu}b^{\nu} \simeq -0.246929$.
And if $\nu=0.6$, then $(1- \nu) a+ \nu b - S\left(\sqrt{\frac{a}{b}}\right) a^{1-\nu}b^{\nu} \simeq 1.71544$.
\end{Rem}
Applying Lemma \ref{lem21}, we have the reverse ratio inequality of the refined Young inequality for positive operators.
\begin{The}
We suppose two invertible positive operators $A$ and $B$ satisfy $0<mI \leq A,B \leq MI$, where $I$ represents an identity operator and $m,M \in \mathbb{R}$.
For any $\nu\in[0,1]$, we then have
\begin{equation} \label{ineq_reverse_ratio_Young_op}
S(\sqrt{h})A\sharp_{\nu}B \geq (1- \nu) A + \nu B-2r \left(\frac{A+B}{2}-A\sharp_{1/2}B\right),
\end{equation}
where $h \equiv \frac{M}{m}>1$ and $r \equiv \min\left\{\nu,1-\nu\right\}$.
\end{The}
{\it Proof}:
In Lemma \ref{lem21}, we put $a=1$, then we have for all $b>0$,
$$
S(\sqrt{b})b^{\nu} \geq \nu b +(1-\nu)-r(\sqrt{b}-1)^2
$$
We consider the invertible positive operator $T$ such that $0<mI\leq T\leq MI$.
Then we have the following inequality
\begin{equation}
\max_{m\leq t\leq M} S(\sqrt{t}) T^{\nu} \geq \nu T + (1-\nu) -r (T-2T^{1/2}+1),
\end{equation}
for any $\nu \in [0,1]$.
We put $T= A^{-1/2}BA^{-1/2}.$ Since we then have $\frac{1}{h} =\frac{m}{M} \leq A^{-1/2}BA^{-1/2} \leq \frac{M}{m}=h$,
we have
\begin{eqnarray*}
&& \max_{\frac{1}{h}\leq t\leq h} S(\sqrt{t}) \left(A^{-1/2}BA^{-1/2} \right)^{\nu} \\
&&\hspace*{-5mm} \geq \nu A^{-1/2}BA^{-1/2} + (1-\nu)
-r \left\{A^{-1/2}BA^{-1/2} -2 \left(A^{-1/2}BA^{-1/2}\right)^{1/2}+1\right\}.
\end{eqnarray*}
Note that $h>1$ and $S(x)$ is monotone decreasing for $0<x<1$ and
monotone increasing for $x>1$ \cite{Tom1}.
Thus we have
\begin{eqnarray*}
&& S(\sqrt{h}) \left(A^{-1/2}BA^{-1/2}\right)^{\nu} \\
&& \hspace*{-5mm} \geq \nu A^{-1/2}BA^{-1/2} + (1-\nu) -r \left\{A^{-1/2}BA^{-1/2} -2 \left(A^{-1/2}BA^{-1/2}\right)^{1/2}+1\right\}.
\end{eqnarray*}
Multiplying $A^{1/2}$ to the above inequality from both sides, we have the present theorem.
\hfill \qed
\section{A reverse difference inequality for a refined Young inequality}
For the classical Young inequality, the following reverse inequality is known.
For positive real numbers $a,b$ and $\nu\in[0,1]$, M.Tominaga showed the following inequality \cite{Tom1}:
\begin{equation}\label{ineq100}
L(a,b)\log S\left(\frac{a}{b}\right) \geq (1- \nu) a + \nu b -a^{1-\nu} b^{\nu},
\end{equation}
which is called the converse difference inequality for the Young inequality in \cite{Tom1}
In this section, we show the reverse difference inequality of the refined Young inequality (\ref{ineq02_rev}).
\begin{Lem} \label{the100}
For positive real numbers $a, b$ and $\nu \in [0,1]$, we have
\begin{equation}\label{ineq101}
\omega L(\sqrt{a},\sqrt{b})\log S\left( \sqrt{\frac{a}{b}} \right) \geq (1-\nu) a + \nu b -a^{1-\nu} b^{\nu} -r\left(\sqrt{a}-\sqrt{b}\right)^2,
\end{equation}
where $\omega \equiv \max \left\{\sqrt{a},\sqrt{b}\right\}$.
\end{Lem}
{\it Proof}:
\begin{itemize}
\item[(i)] For the case of $\nu \leq1/2$, $r=\nu$.
We consider the following function.
$$
g_b(\nu) \equiv \nu b +(1-\nu)-b^{\nu}-\nu(\sqrt{b}-1)^2,\quad \left(0\leq \nu \leq \frac{1}{2}\right).
$$
From $g_b'(\nu)=2(\sqrt{b}-1)-b^{\nu}\log b$, we have
$$
g_b'(\nu)=0 \Leftrightarrow \nu = \frac{\log \frac{\sqrt{b}-1}{\log \sqrt{a}}}{\log b} \equiv \nu_b.
$$
We also find that $\nu_b \in [0,\frac{1}{2}]$ by elementaly calculations with the following inequalities:
$$
1-\frac{1}{\sqrt{b}}\leq \log \sqrt{b} \leq \sqrt{b} -1, \quad (b > 0).
$$
In addition, we have $g_b''(\nu) =-b^{\nu}\left(\log b\right)^2 <0$.
Therefore $g_b$ takes a maximum value when $\nu = \nu_b$, and it is
calculated as $g_b(\nu_b)=L(1,\sqrt{b}) \log S(\sqrt{b})$ by simple but slightly complicated calculations.
Thus we have
$$
L(1,\sqrt{b})\log S\left(\sqrt{b}\right) \geq \nu b+ (1-\nu) -b^{\nu}-\nu(\sqrt{b}-1)^2.
$$
We put $\frac{b}{a}$ instead of $b$ in the above inequality, and then multiplying $a$ to both sides, we have
\begin{equation} \label{ineq11}
\sqrt{a} L(\sqrt{a},\sqrt{b})\log S\left(\sqrt{\frac{a}{b}}\right) \geq (1-\nu) a + \nu b -a^{1-\nu}b^{\nu}-\nu(\sqrt{a}-\sqrt{b})^2,
\end{equation}
since $L(x,y)=L(y,x)$ and $S(x)=S(1/x)$ for $x>0$.
\item[(ii)] For the case of $\nu \geq1/2$, $r=1-\nu$.
We consider the following function.
$$
h_a(\nu) \equiv \nu +(1-\nu)a -a^{1-\nu}-(1-\nu)(1-\sqrt{a})^2,\quad \left(\frac{1}{2}\leq \nu \leq 1\right).
$$
By the similar way to (i), we have
$$
h_a'(\nu)=0 \Leftrightarrow \nu = 1- \frac{\log \frac{\sqrt{a}-1}{\log \sqrt{a}}}{\log a} \equiv \nu_a.
$$
By the similar way to (i), we have $\nu_a \in[\frac{1}{2},1]$ and $h_a''(\nu)=-a^{1-\nu}\left(\log a\right)^2 <0$ so that
$h_a$ takes a maximum value when $\nu=\nu_a$, and it is calculated as $h_a(\nu_a) =L(1,\sqrt{a})\log S(\sqrt{a})$.
Thus we have
$$
L(1,\sqrt{a})\log S(\sqrt{a}) \geq \nu +(1-\nu)a-a^{1-\nu}-(1-\nu)(1-\sqrt{a})^2,
$$
which implies
\begin{equation} \label{ineq12}
\sqrt{b} L(\sqrt{b},\sqrt{a})\log S\left(\sqrt{\frac{a}{b}}\right) \geq (1-\nu) a + \nu b -a^{1-\nu}b^{\nu}-(1-\nu)(\sqrt{a}-\sqrt{b})^2,
\end{equation}
by replacing $a$ by $\frac{a}{b}$ and then multiplying $b$ to both sides.
\end{itemize}
From the inequalities (\ref{ineq11}) and (\ref{ineq12}), we have the present theorem,
since $L(x,y)=L(y,x)$ and $S(x)=S(1/x)$ for $x>0$.
\hfill \qed
\begin{Rem} \label{remark_comparison}
We easily find that the right hand side of the inequality (\ref{ineq100}) is greater than that of the inequality (\ref{ineq101}).
Therefore, if the left hand side of the inequality (\ref{ineq101}) is greater than that of the inequality (\ref{ineq100}),
then Theorem \ref{the100} is trivial one. However, we have not yet found any counter-example such that
\begin{equation} \label{ineq102}
L(a,b)\log S\left(\frac{a}{b}\right) \geq \omega L(\sqrt{a},\sqrt{b} )\log S\left( \sqrt{\frac{a}{b}} \right),
\end{equation}
where $\omega = \max\left\{\sqrt{a},\sqrt{b}\right\}$
for any $a,b >0$ by the computer calculations. Here we give a remark that we have the following inequalities:
\begin{equation}
L(a,b) \leq \omega L(\sqrt{a},\sqrt{b}),\quad and \quad \log S\left(\frac{a}{b}\right) \geq \log S\left( \sqrt{\frac{a}{b}} \right)
\end{equation}
for any $a,b >0$.
At least, we actually have many examples satisfying the inequality (\ref{ineq102})
so that we claim that Theorem \ref{the100} is nontrivial as a refinement of the inequality (\ref{ineq100}).
In addition, it is remarkable that we have no ordering between
$$ L(a,b)\log S\left(\frac{a}{b}\right)$$
and
$$\omega L(\sqrt{a},\sqrt{b} )\log S\left( \sqrt{\frac{a}{b}} \right)+r\left( \sqrt{a}-\sqrt{b} \right)^2 $$
for any $a,b>0$ and $\nu \in [0,1]$.
Therefore we may claim that Theorem \ref{the100} is also nontrivial from the sense of finding a tighter upper bound of
$(1-\nu) a + \nu b -a^{1-\nu} b^{\nu} $.
\end{Rem}
Finally we prove the following theorem. It can be proven by the similar method in \cite{Tom1}.
\begin{The}\label{the_reverse_Young_op}
We suppose two invertible positive operators satisfy $0<mI \leq A,B \leq MI$, where $I$ represents an identity operator and $m,M \in \mathbb{R}$.
For any $\nu\in[0,1]$, we then have
\begin{equation} \label{ineq_reverse_Young_op}
h \sqrt{M} L(\sqrt{M},\sqrt{m})\log S(\sqrt{h}) \geq (1-\nu) A + \nu B-A\sharp_{\nu}B -2r\left(\frac{A+B}{2} -A\sharp_{1/2}B\right),
\end{equation}
where $h \equiv \frac{M}{m}>1$ and $r \equiv \min\left\{\nu,1-\nu\right\}$.
\end{The}
{\it Proof}:
From the inequality (\ref{ineq101}), we have
\begin{equation}
\omega L(\sqrt{b},1)\log S(\sqrt{b}) \geq \nu b + (1-\nu) -b^{\nu} -r (b-2\sqrt{b}+1),
\end{equation}
for all $\nu\in[0,1]$, putting $b=1$.
We consider the invertible positive operator $T$ such that $0<mI\leq T\leq MI$.
Then we have the following inequality
\begin{equation}
\max_{m\leq t\leq M} \max\left\{\sqrt{t},1\right\} L(\sqrt{t},1)\log S(\sqrt{t}) \geq \nu T + (1-\nu) -T^{\nu} -r (T-2T^{1/2}+1),
\end{equation}
for any $\nu \in [0,1]$.
We put $T= A^{-1/2}BA^{-1/2}.$ Since we then have $\frac{1}{h} =\frac{m}{M} \leq A^{-1/2}BA^{-1/2} \leq \frac{M}{m}=h$,
we have
\begin{eqnarray*}
&& \max_{\frac{1}{h}\leq t\leq h} \max\left\{\sqrt{t},1\right\} L(\sqrt{t},1)\log S(\sqrt{t}) \\
&&\hspace*{-5mm} \geq \nu A^{-1/2}BA^{-1/2} + (1-\nu)
-\left(A^{-1/2}BA^{-1/2}\right)^{\nu} -r \left\{A^{-1/2}BA^{-1/2} -2 \left(A^{-1/2}BA^{-1/2}\right)^{1/2}+1\right\}.
\end{eqnarray*}
Note that $h>1$ and $L(u,1)$ is monotone increasing function for $u>0$. In addition, we note that $S(x)$ is monotone decreasing for $0<x<1$ and
monotone increasing for $x>1$ \cite{Tom1}.
Thus we have
\begin{eqnarray*}
&& \sqrt{h}L(\sqrt{h},1)\log S(\sqrt{h}) \\
&& \hspace*{-5mm} \geq \nu A^{-1/2}BA^{-1/2} + (1-\nu) -\left(A^{-1/2}BA^{-1/2}\right)^{\nu} -r \left\{A^{-1/2}BA^{-1/2} -2 \left(A^{-1/2}BA^{-1/2}\right)^{1/2}+1\right\}.
\end{eqnarray*}
Multiplying $A^{1/2}$ to the above inequality from both sides, we have
$$
\sqrt{h} L(\sqrt{h},1)\log S(\sqrt{h})A \geq (1-\nu) A+ \nu B- A\sharp_{\nu}B -2r\left(\frac{A+B}{2}-A\sharp_{1/2}B\right).
$$
Since the left hand side in the above inequality is less than
$$\sqrt{h} L(\sqrt{h},1)\log S(\sqrt{h})M= h \sqrt{M}L(\sqrt{M},\sqrt{m})\log S(\sqrt{h}),$$
the proof is completed.
\hfill \qed
\begin{Rem}
As mentioned in Remark \ref{remark_comparison}, we have not yet found the ordering between
the right hand side of the inequality (\ref{ineq_reverse_Young_op}) and that of the inequality (\ref{prop_rev_diff_ineq}).
Therefore Theorem \ref{the_reverse_Young_op} is not a trivial result.
\end{Rem}
\section{Concluding remarks}
As we have seen, we gave refined Young inequalities for two positive operators.
In addition, we gave reverse ratio type inequalities and reverse difference type inequalities for the refined Young inequality for positive operators.
Closing this paper, we shall give a refinement of the weighted arithmetic-geometric mean inequality for $n$ real numbers by a simple proof.
\begin{Prop} \label{the_gen01}
Let $a_1,\cdots,a_n\geq 0$ and $p_1,\cdots,p_n > 0$ with $\sum_{j=1}^n p_j=1$ and $\lambda \equiv \min \left\{ p_1,\cdots ,p_n \right\}$.
If we assume that the multiplicity attaining $\lambda$ is $1$,
then we have
\begin{equation} \label{gen_ineq01}
\sum_{i=1}^n p_i a_i - \prod_{i=1}^n a_i^{p_i} \geq n \lambda \left(\frac{1}{n}\sum_{i=1}^n a_i- \prod_{i=1}^n a_i^{1/n} \right),
\end{equation}
with equality if and only if $a_1=\cdots =a_n$.
\end{Prop}
{\it Proof}:
We suppose $\lambda = p_j$. For any $j=1,\cdots,n$,
we then have
\begin{eqnarray*}
\sum_{i=1}^n p_i a_i - p_j \left(\sum_{i=1}^n a_i - n \prod_{i=1}^n a_i^{1/n} \right)
&=& np_j \left( \prod_{i=1}^n a_i^{1/n} \right) + \sum_{i=1,i\neq j}^n ( p_i-p_j) a_i \\
&\geq& \prod_{i=1,i\neq j}^n \left( a_1^{1/n} \cdots a_n^{1/n} \right)^{np_j} a_i^{p_i-p_j} \\
&=& a_1^{p_1} \cdots a_n^{p_n}.
\end{eqnarray*}
In the above process, the classical weighted arithmetic-geometric mean inequality \cite{HLP,Bu} for
$a_1,\cdots,a_n\geq 0$ and $p_1,\cdots,p_n > 0$ with $\sum_{j=1}^n p_j=1$;
\begin{equation} \label{classical_w_ineq}
\sum_{j=1}^n p_j a_j \geq \prod_{j=1}^n a_j^{p_j},
\end{equation}
with equality if and only if $a_1=\cdots =a_n$, was used.
We note that $p_i-p_j>0$ from the assumption of the proposition.
The equality in the inequality (\ref{gen_ineq01}) holds if and only if
$$(a_1a_2\cdots a_n)^{\frac{1}{n}}=a_1=a_2=\cdots=a_{j-1}=a_{j+1}=\cdots=a_n$$
by the equality condition of the classical weighted arithmetic-geometric mean inequality (\ref{classical_w_ineq}).
Therefore $a_1=a_2=\cdots=a_{j-1}=a_{j+1}=\cdots=a_n\equiv a$, then we have $a_j^{\frac{1}{n}}a^{\frac{n-1}{n}}=a$ from the first equality.
Thus we have $a_j=a$, which completes the proof.
\hfill \qed
The inequality (\ref{gen_ineq01}) gives a refinement of the classical weighted arithmetic-geometric mean inequality (\ref{classical_w_ineq}).
At the same time, it gives a natural generalization of the inequality (\ref{ineq02_rev}) proved in \cite{KM}.
It is also notable that Proposition \ref{the_gen01} can be proven by using the bounds for
the normalized Jensen functional, which were obtained by S.S.Dragomir in \cite{Dra}, except for the equality condtions.
Note that the inequality (\ref{gen_ineq01}) itself holds without the assumption that the multiplicity attaining $\lambda$ is $1$.
In addition, when we do not impose on this assumption, the equality in the inequality (\ref{gen_ineq01}) holds if $p_i=\frac{1}{n}$ for all $i=1,\cdots,n$.
\section*{Ackowledgement}
The author thanks Professor S. M. Sitnik for letting me know the valuable infromation on the reference \cite{BK} and
Dr.F.C.Mitroi for giving me the valuable comment on the equality condition of Propisition \ref{the_gen01}.
The author also thanks the anonymous referee for valuable comments to improve the manuscript.
The author was supported in part by the Japanese Ministry of Education, Science, Sports and Culture, Grant-in-Aid for
Encouragement of Young Scientists (B), 20740067.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 735 |
{"url":"https:\/\/plainmath.net\/31176\/financial-pressures-surgeries-performed-physicians-nationwide-increasing","text":"# Driven by tech advances and financial pressures,the number of surgeries performed in physicians' offices nationwide has been increasing over the years\n\nAyaana Buck 2021-09-26 Answered\n\nDriven by technological advances and financial pressures, the number of surgeries performed in physicians' offices nationwide has been increasing over the years. The function $$\\displaystyle{f{{\\left({t}\\right)}}}=-{0.00447}{t}^{{{3}}}+{0.09864}{t}^{{{2}}}+{0.05192}{t}+{0.8}{\\left({0}\\leq{t}\\leq{15}\\right)}$$ gives the number of surgeries (in millions) performed in physicians' offices in year t, with t=0 corresponding to the beginning of 1986.\na. Plot the graph of f in the viewing window [0,15]\u00a0$$\\times$$ [0,10].\nb. Prove that f is increasing on the interval [0, 15].\n\n### Expert Community at Your Service\n\n\u2022 Live experts 24\/7\n\u2022 Questions are typically answered in as fast as 30 minutes\n\u2022 Personalized clear answers\n\n### Plainmath recommends\n\n\u2022 Ask your own question for free.\n\u2022 Get a detailed answer even on the hardest topics.\n\u2022 Ask an expert for a step-by-step guidance to learn to do it yourself.\n\n## Expert Answer\n\nAubree Mcintyre\nAnswered 2021-09-27 Author has 8283 answers\n1. [Graph]\n2. We substitute t=1 (any number from (0,15)) into f'(t)'\n$$\\displaystyle{f}'{\\left({1}\\right)}=-{\\frac{{{1341}\\dot{{1}}^{{{2}}}-{19728}\\dot{{1}}-{5192}}}{{{100000}}}}={0.23578}$$\n\n### Expert Community at Your Service\n\n\u2022 Live experts 24\/7\n\u2022 Questions are typically answered in as fast as 30 minutes\n\u2022 Personalized clear answers\n\n### Plainmath recommends\n\n\u2022 Ask your own question for free.\n\u2022 Get a detailed answer even on the hardest topics.\n\u2022 Ask an expert for a step-by-step guidance to learn to do it yourself.\n...","date":"2021-11-28 12:21:13","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 1, \"mathjax_asciimath\": 0, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.2015378773212433, \"perplexity\": 5616.738566798846}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": false}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2021-49\/segments\/1637964358520.50\/warc\/CC-MAIN-20211128103924-20211128133924-00179.warc.gz\"}"} | null | null |
{"url":"https:\/\/stats.stackexchange.com\/questions\/325504\/bias-caused-by-optional-stopping","text":"# Bias caused by optional stopping\n\nThis question is about a kind of bias that is slightly more subtle than doing multiple tests without caution (related to multi-hypothesis testing).\n\nExperimenters are willing to reject an hypothesis $H_0$ that a coin is fair, expecting to find a higher probability for heads. When flipping 100 coins, significance of $\\alpha=5$% is obtained with 59 heads.\n\nAn obvious bias consists in forgetting negative results and only publishing positive ones. At worst: do 100 flips multiple times until you get more than 59\/100 heads and publish only the last experiment.\n\nInstead, the experimenter does multiple flips and stops when \"he feels like stopping\" and publish all the previous flips. If he is secretly willing to reject $H_0$ this might create a bias: stopping at a point the results are rather positive.\n\nAt worst, he might stop when the test for all previous flips is positive. Actually, I'm quite sure the stopping time $T:$\"first time the test is significant on all previous flips\" is almost surely finite. Do you confirm this?\n\nIs there a name for this kind of bias? Do you know of any studies or texts about it?\n\nSome relevant posts here and here.\n\nAccording to an answer from the second post, it seems that as the number of flips goes to infinity, at some point the significance test will be positive (almost surely), which is to say, there exists some finite number of samples after which it will almost surely happen.\n\nAccording to the first post, the number of flips it will actually take has a finite median but infinite expectation. The median grows very quickly as a function of the required $z$-score to pass the test, so it may be that this sort of bias can be effectively mitigated by demanding a lower $\\alpha$ threshold.\n\n\u2022 Thanks a lot. You say: \"there exists some finite number of samples after which it will almost surely happen.\". Would it be rather \" almost surely, there exists some finite number of samples after which it will happen\"? Jan 30, 2018 at 9:06\n\u2022 Formally, the statement is $\\lim \\sup_k \\frac{S_k}{\\sqrt{k}} = \\infty$, which is by definition $\\lim_{n \\rightarrow \\infty} \\sup_{k \\geq n} \\frac{S_k}{\\sqrt{k}} = \\infty$ hence the strange wording I used, but I am pretty sure this would imply $\\sup_{k \\in \\mathbf{N}} \\frac{S_k}{\\sqrt{k}} = \\infty$. So yes, I think it works out. Jan 30, 2018 at 9:11\n\nI did some simulations (under $H_0$: a fair coin $p=0.5$). I limited the number of flips to $n_\\max$ because the raw stopping time $T$ has such a huge tail that sometimes the computer wouldn't stop in a reasonable time. Anyway it's more realistic with a limit.\n\nThe experiment is:\n\n\u2022 do some first $100$ flips to initialize\n\u2022 do a z-test with $\\alpha=5$%. If it is significant or you flipped more than $n_\\max$, stop.\n\u2022 otherwise flip one more time and go back to the previous step\n\nThe false discovery rate (type I error) differs a lot from $\\alpha$:\n\n\u2022 For $n_\\max=1000$ : 26%\n\u2022 For $n_\\max=10000$ : 40%\n\nHowever something happens because of the optional stopping theorem: when \"meta-anlyszing\" several of these experiments (simply merging them into one big flipping session and do a z-test), the bias on false discovery rate tends to disappear:\n\nIt might sound a bit paradoxical: we have many experiments where 26% are falsely significant on average, but the global experiment still has the right type I error of 5%. And the global estimator $\\hat p$ is still (asymptotically) unbiased. It can be explained by the fact that the longest experiments, having more weight, are the least favourable to rejecting $H_0$.\n\nAs a conclusion, optional stopping can cause a strong bias for tests on each single experiment, but the bias tends to disappear when doing several experiments.","date":"2022-08-09 02:41:37","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 1, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 0, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.6671265959739685, \"perplexity\": 588.4472990358431}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2022-33\/segments\/1659882570879.37\/warc\/CC-MAIN-20220809003642-20220809033642-00291.warc.gz\"}"} | null | null |
lo-tech lame version
corz.org uses cookies to remember that you've seen this notice explaining that corz.org uses cookies, okay!
01. Help Me Gawd
02. Keep the Priest
03. Seeing Things
04. inside out
05. Both Eyes
06. tTalking
07. Burned Down
08. Get This Straight
09. Darknesses Everywhere
10. Marmalademore
11. Where She Lives
12. Believe it when I..
Click any track to download and play, keep free for personal use. Share with music lovers you yourself love.
intro..
Maybe you know the story, I put this together for a friend last year, and it escaped into the real world.
That's the short version.
Perhaps it's time to tell a longer one. These are sketches, of course. One-offs. Just what it says on the can; sketches. Unworked, unwashed, first-takes, fresh out of the womb of songs, and barely able to stand, let alone walk. Often I miss these myself, they have a peculiar quality in their raw form.
The friend is a heroin addict, hopelessly lost, and once heard track one. Could I make a copy? Of course.
I'm not AOL, and I couldn't just put one track on the disk, could I? I have a folder, an inbox for songs, and started scrolling through it, looking for, well, fillers, I guess. Can any sort of song be just that? I don't think so - the silliest, most insignificant works seem to ring true somewhere, who can tell such things - so I selected, the addiction is significant in the selection, I remember, and the more I listen to it now, as a selection, the more it seems to make more sense like that. Was I trying to say something? Who isn't?
So I'll break it down, track-by-track, I can't tell you *everything*, you'll just have to listen. But I can throw you some clues, perhaps you can see the patterns better from there. Connexions. You'll be glad to hear that old analogue echo-box has since bit the dust, a God-Send really. What isn't?
Yes, a noisy mess of a piece, but still, I want to play it over and over, loud, and so will you! This was how it escaped; her son made a copy, you see, and so it goes. Perhaps my third-favourite piece in the selection at this time, its simple riffs and childish hooky lyrics do their job well enough, teasing and toying with my ears. The childishness fits, describing this approach to God perfectly (she has troubles with God, you see, Big troubles). The vocal sonics progress from a vague, lost, restless anger, to something resembling powerful self-awareness, finally throwing themselves into the next Valley. The phonetic spelling of God is your starter for ten.
Note: there's a lot of electric guitar in sketches, much more than I generally do, so perhaps that thing I was trying to say, by picking these noisy tracks, I was trying to Say It Loud. Though more likely just a phase I was going through.
For me this an interesting track because it's a rare recording of how I sounded on this Dholak drum before I soaked it and tuned it (actually, sorry Dek, I ripped the bottom clean off it and used fat rope to tie the cords into! Sounds great! You didn't really want it back, though, did you, good man!) and actually sat down to practice with the thing. Nevertheless, I managed to capture the gist all right (tapping at your computer keyboard gets you to about this level of mastery!). That Dholak track is also the only overdub in sketches. Which means I went back and added it afterwards. Has anyone trained their feet to play the Tabla? That might be a worthwhile endeavour for my next drum.
The lyric is as clear as a bell, a vibrant reinforcement of the evolution of track one. It was an obvious choice for track two, finger-tapping good, and with a simple message; "There is no man between me and God". This was one of those songs that arrived with all the lyrics fully-formed and I strummed like a bastard hoping I wouldn't forget the ending.
It took a while for this one to grow on me, but I always liked the idea of it. Your idea of success and mine are different things. That influences almost everything about us, the way we say and do things, our seeing. It offers the idea that maybe if you fast forwarded your life, or I did, we'd see more things the same. And that that possibility exists right now, and we could even take it, and see the same thing, about everything. Also, Union regulations say I have to play a Major 7th chord every so many songs, so it was easy enough to slip that in here.
A ditty. I'm still not sure why I dropped this in here. The obvious lyric stuff makes sense, but it's not like me to work on exteriors, and the underlying thread would snap if that's all there was. I'm still thinking about what I might do with this ditty. The ending is a perfect intro for the next track, though. Maybe that's why it's here.
Still on seeing. Both Eyes, is about seeing things from both sides, in order to choose one path. I'm talking about decision, in its Latin sense - not that I know much Latin - we split reality with every decision we make. We chose this or that. We go left or right. How can we know the real consequences of any decision? Yet we let these things influence decisions. Both Eyes is about making decisions based on what we know to be right, rather than what we fear the least. It's about two people making a decision that splits their lives, they walk separate paths, and neither understands why the other did what they did.
Opening both eyes may be a cute wee metaphor for seeing both sides of a thing, like seeing 3D instead of 2D, but it's probably significant that there's a current running through lots of my stuff where the right side equates to the future, the left, the past. So both eyes being open is also about focusing on the moment, free of the imbalance of worry about the future, regrets from the past, it's about being here and feeling, rather than the seeing, knowing, rather than thinking, what is and isn't the right thing to do.
I've always been fascinated and overwhelmed by the power of words. But when they are Spoken! Oh fit! Perhaps that's why my written words often come out speaked, folk mention it, so that's two people saying the same thing to me, at least, and that's usually worth at least a listen. It's intentional.
But it's so much noise when it's just talking. I don't have a lot of time for small talk. It interferes with the music of living, and devalues the good stuff. This is what I'm tTalking about..
Heroin seems to shut down the higher parts of the mind. Imagine that you were addicted to drinking India Ink. Every morning you guzzle a whole bottle of the stuff down, can't help yourself. Then, a while later, when the convulsions have stopped, you head out to the shops. Although you know it will take a whole day to wash the India Ink from your mouth and face, you head out in the unquestioned belief that no one will notice. More than that, you will be shocked when they point to the large streaks of black on your face. You will say it is acid rain, or a felt-tip pen burst, or..
Small talk, before they get close enough for you to see those Gigantic pupils you'll be hit with it. If your lover won't stare you in the face, and talks continuously about the weather or TV or any old trivial shite instead of real stuff, well, maybe they aren't having an affair, maybe it's an addiction.
Okay, I'm illustrating. It's a curse! A scourge on society! It was designed for visiting the Post Office, and even then, only until you get to know the person who runs it. Small talk. Talktalktalktalktalk. If you're new to me, you'll discover I touch on this more than once.
Firstly, this is based on a true story. Three, in fact. The details don't matter, what matters is the lyric. It's about someone getting themselves into a state of complete powerlessness, a state where their deepest, darkest desires and longings have become manifest in their thoughts, become real in their life. And her house burned down.
It starts with her birth. The rose being her soul. The storm being life, but also the fire, the inner fire, the one that burns and wants expression, and ends with individual possessions burning. The spices, these were rarely used, for show. The faces, are photographs. Her family weren't so good to her. Those horses, they run no more. The final possession, the indestructible toy - I was thinking like a Tonka Toy from the 70's - obviously has defied it's nature and been destroyed (a real Tonka would survive) leaving nothing. This implies rebirth, the down being into the earth, like ashes do, and kundalini doesn't. Now think of the rose, how it is red, how we plucked it from the earth.
Quietly, right at the end, you hear a small prayer for rain, and so the cycle is re-initiated. Clever stuff, huh.
08. Get This Straight.
A ditty, but a complete one, and perfectly in its place. Fitting neatly as an evolution of the previous piece, made statement. "What the mind can conceive and believe ..." is preparation, I guess, for "Those things you want, especially the things you tell yourself you don't, WILL manifest". Get This Straight is simply the realisation that past desires have become manifest, an acceptance that had this not happened, those desires would still be hidden, causing trouble, instead of guiding.
Some might say they want recognition, when really they mean fame. They might say they want security, or peace, or comfort, but really they're talking about money, or vice versa. Someone might say "I love You", but really they just want attention, or they might love you, of course.
Get This Straight is about accepting that while I'm a long way off perfect, the events in my life are my own doing, no one is to blame, thanks, I'm fine with being me. I'd not intentionally hurt another, so it's crazy to beat myself up about passed mistakes; especially when they were so valuable, such essential evolution. We live and learn and live and learn and (repeat to fade - or just loop it) I was too close to the mic, or something.
Forgiving the uncropped intro, this is probably my favourite piece in the collection. From the getgo is sings to the muses, you know. And that's what it's all about, the struggle, between the animal and spiritual in every human. That activity of intellect prevents humans from truly being, makes them unreal.
I didn't mean the piece to go the way it did, there's a hesitant moment near the start where you can hear it pull away from me, I have little control after that. The noise in the middle (and again right at the end) is a take on the sound from Close Encounters of the Third Kind, where they raise their arm and scream to the other aliens.. a human has been located.
If you've ever lived near the junkie world, geographically or otherwise, you'll know that every morning there is a flurry of zombie activity. To the untrained eye their undead ways may be difficult to spot, but they are all over now. And what can you do about it? You could try something to pull them out, but really, the ticket to Junkie-City is usually one-way.
This piece is about eating marmalade. And then some more marmalade. And then we have the after-taste. It's all perfectly clear, I think.
On the surface, it's an interrogation. Did I kill her? Over and over and over. The tiredness at the start, the slurring, that's not drugs, you know. And over and over. I didn't kill her.
And then there is a breakthrough, as Morse probably wouldn't say, and it starts to get personal. That's all I have to say about the surface of things here
So I'm interrogating myself, and I'm asking, did I have an impact somewhere? How can we see all the effects of our actions, no matter how "good" the intentions, the effects may be "bad", that is, bad for someone. How can we tell, like that old lady we help across the busy road with her messages, she gets home three minutes earlier, and doesn't miss the burglars, now lies on her kitchen floor, head smashed in, robbed and dead, instead of coming home to find her holiday savings teapot in pieces. How can we tell? And that's all I'm saying about that, for now.
Here's how we can tell. Well, partly. Here in purgatory, we have action, right. We do shit. Thinking, speaking, we're clever bastards, yup. We can do it all. And this creates products, and some of those products can be bought and sold, and some can't, and you don't need to say "I believe in my knees", because they are right there in the middle of your legs, belief is wasted on them. Faith is another matter.
It's a joke, see? "I'll believe it when I see it". Did you ever think about that? At the point of seeing, belief is no longer required. Belief is for things you can't see, things you don't know, a framework to hang your other experiences on. Did I spoil the joke for you? Maybe. But when you're certain of a thing, what need is there for belief?
And it works both ways.
Drag & Drop Pickers!
Welcome to corz.org!
Since switching hosts (I hope you are alright, Ed! Wherever you are …) quite a few things seems to be wonky.
Juggling two energetic boys (of very different ages) on Coronavirus lockdown, I'm unlikely to have them all fixed any time soon. Mail me! to prioritise!
Please enable JavaScript for corz.org.
There are things I want to show you!
© 2021 « corz.org » 17.1.21
little timer computer say this page generated in 0.000 seconds
random word from corzblog: thursday
yadsruht :golbzroc morf drow modnar
Machine Translations
NOTE: This Will Mangle Technical Output (e.g. scripts).
Underfooter Tools
P2P Port Probe
Tag-Tools!
home • search • corzblog • devblog • test blog • downloads • contact • hire cor • give back •
• bbtags guide • fun stuff • that bot •
Save money online shopping • Free money! Get in NOW!!! • Think Slack! • DropBox=Essential | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 8,942 |
Rasmi Beekeeping Company ("Rasmi") is a thriving beekeeping operation in the Awdal region of Somaliland. The owner is a passionate beekeeper and resilient businessman who has been practicing his craft for over a decade. Rasmi honey is sold in Hargeisa, Djibouti, Amoud, and occasionally in the United Arab Emirates. Rasmi was founded as a small operation with a desire to grow, but had no capital with which to expand its business.
Financing facilitated by Shuraako has enabled Rasmi to increase production by 153% and acquire a larger market share. The business has grown its number of employees and increased its export capacity by expanding honey production and improving the quality of its products. | {
"redpajama_set_name": "RedPajamaC4"
} | 5,060 |
Q: Synchronous request using Alamofire I have checked other solutions like How to make a synchronous request using Alamofire?, I followed all instructions but I haven't been able to complete the requests and then use the information.
I need to retrieve distance and route points from google maps and then complete an array with that information.
This is the function:
func getGoogleMapsInfo(startLocation: CLLocationCoordinate2D, restaurant: Customer, completion: @escaping (_ distance: Int, _ routePoints: String) -> ()){
let endLocation = CLLocationCoordinate2D(latitude: restaurant.latitude, longitude: restaurant.longitude)
let origin = "\(startLocation.latitude),\(startLocation.longitude)"
let destination = "\(endLocation.latitude),\(endLocation.longitude)"
var distance = 0
var routePoints = ""
let url = "https://maps.googleapis.com/maps/api/directions/json?origin=\(origin)&destination=\(destination)&mode=walking"
Alamofire.request(url).responseJSON { response in
switch response.result {
case .success:
do {
self.json = try JSON(data: response.data!)
distance = Int(truncating: self.json["routes"][0]["legs"][0]["distance"]["value"].numberValue)
let route = self.json["routes"].arrayValue.first
let routeOverviewPolyline = route!["overview_polyline"].dictionary
routePoints = (routeOverviewPolyline?["points"]?.stringValue)!
completion(distance, routePoints)
break
} catch {
print("error JSON")
}
case .failure(let error):
print(error)
completion(distance, routePoints)
break
}
}
}
And this is how I call it:
for index in 0...nearbyRestaurants.count - 1 {
getGoogleMapsInfo(startLocation: currentLocation, restaurant: nearbyRestaurants[index]) { (distance, routePoints) in
nearbyRestaurants[index].distance = distance
nearbyRestaurants[index].routePoints = routePoints
}
}
I'd really appreciate if someone can help me.
A: Don't try to make an asynchronous request synchronous
Instead use DispatchGroup,notify is called when all network requests are completed.
let group = DispatchGroup()
for index in 0..<nearbyRestaurants.count {
group.enter()
getGoogleMapsInfo(startLocation: currentLocation, restaurant: nearbyRestaurants[index]) { (distance, routePoints) in
nearbyRestaurants[index].distance = distance
nearbyRestaurants[index].routePoints = routePoints
group.leave()
}
}
group.notify(queue: DispatchQueue.main) {
print("all info data has been received")
}
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 2,119 |
\part{}
\setlist{nolistsep}
\maketitle
\begin{abstract}
We consider the projected gradient algorithm for the nonconvex best subset selection
problem that minimizes a given empirical loss function under an $\ell_0$-norm
constraint.
Through decomposing the feasible set of the given sparsity constraint
as a finite union of linear subspaces, we present two acceleration
schemes with global convergence guarantees, one by same-space
extrapolation and the other by subspace identification.
The former fully utilizes the problem structure to greatly accelerate
the optimization speed with only negligible additional cost.
The latter leads to a two-stage meta-algorithm that first uses
classical projected gradient iterations to identify the correct
subspace containing an optimal solution, and then switches to
a highly-efficient smooth optimization method in the identified
subspace to attain superlinear convergence.
Experiments demonstrate that the proposed accelerated algorithms are
magnitudes faster than their non-accelerated counterparts as well as
the state of the art.
\end{abstract}
\section{Introduction}
We consider the sparsity-constrained optimization problem in $\Re^n$:
\begin{equation}
\label{eq:problem}
\min\nolimits _{w \in A_s} f(w),
\end{equation}
where $f$ is convex with $L$-Lipschitz continuous gradient,
$s \in \N$,
and $A_s$ is the sparsity set given by
\begin{equation}
A_s \coloneqq \{ w \in \Re^n : \|w\|_0 \leq s\},
\label{eq:As}
\end{equation}
where $\|w\|_0$ denotes the $\ell_0$-norm that indicates the number of
nonzero components in $w$.
We further assume that $f$ is lower-bounded on $A_s$.
A classical problem that fits in the framework of
\cref{eq:problem} is the best subset
selection problem in linear regression \citep{BeaKM67a,HocL67a}. Given a
response vector $y\in \Re^{m}$
and a design
matrix of explanatory variables $X\in \Re^{m\times n}$, traditional linear regression minimizes a
least squares (LS) loss
function
\begin{equation}
f(w) = \| y - Xw\|^2/2.
\label{eq:ls_loss}
\end{equation}
However, due to either high dimensionality in terms
of the number of features $n$ or having significantly fewer instances $m$ than
features $n$ (i.e., $m\ll n$), we often seek a linear model that selects only a
subset of the
explanatory variables that will best predict the outcome $y$. Towards
this goal, we can solve
\cref{eq:problem} with
$f$ given by \cref{eq:ls_loss} to fit the training data while simultaneously
selecting the best-$s$
features. Indeed, such a sparse linear regression problem is
fundamental in many
scientific applications, such as high-dimensional statistical learning and
signal processing \citep{JK17}.
The loss in \cref{eq:ls_loss} can be generalized to the following
linear empirical risk to cover various tasks in machine learning beyond regression
\begin{equation}
f(w) = g(Xw), \quad g(z) = \sum\nolimits_{i=1}^m g_i (z_i),
\label{eq:erm}
\end{equation}
where $g$ is convex.
Such a problem structure makes evaluations of
the objective and its derivatives highly efficient, and such efficient
computation is a key motivation for our algorithms for
\cref{eq:problem}.
\paragraph{Related Works.} The discontinuous cardinality constraint in
\cref{eq:problem} makes
the problem difficult to solve.
To make the optimization problem easier, a popular approach is to
slightly sacrifice the quality of the solution (either not strictly
satisfying the sparsity level constraint or the prediction performance
is deteriorated)
to use continuous surrogate functions for
the $\ell_0$-norm, which lead to a continuous nonlinear programming
problem, where abundant algorithms are at our disposal.
For instance, using a convex penalty surrogate such as the
$\ell_1$-norm in the case of LASSO \citep{Tibshirani96}, the problem
\cref{eq:problem} can be relaxed into a convex (unconstrained) one that can be
efficiently solved by many algorithms.
Other algorithms based on continuous nonconvex
relaxations such as the use of smoothly clipped absolute deviation
\citep{FanLi01} and the minimax concave penalty \citep{Zhang10}
regularizers are also popular in scenarios with a higher level of noise
and outliers in the data.
However, for applications in which enforcing the constraints or
getting the best prediction performance is of utmost importance, solving
the original problem \cref{eq:problem} is inevitable. \alert{(For a detailed
review, we refer the
interested reader to \cite[Section 1]{BKM16}.)}
Unfortunately, methods for \cref{eq:problem} are not as well-studied as
those for the surrogate problems.
Moreover, existing methods are indeed
still preliminary and too slow to be useful in large-scale problems
often faced in modern machine learning tasks.
In view of the present unsatisfactory status for scenarios that
simultaneously involve high-volume data and need to get the best prediction performance,
this work proposes efficient algorithms to directly solve \cref{eq:problem}
with large-scale data.
\alertt{To our knowledge, all the most popular algorithms that directly tackles
\cref{eq:problem} without the use of surrogates involve using the
well-known projected gradient (PG) algorithm, at least as a major component}
\alert{\citep{BT11,BKM16,Blu12,BluDav09,BahRB13a}}.\footnote{\cite{GotTT18a} proposed
an
algorithm for a similar
optimization problem that minimizes $f(w) + C\norm{w}_0$ for some
$C > 0$. But whether it is equivalent to \cref{eq:problem} is unclear
because both problems are nonconvex, and for any prespecified sparsity
level $s$, it is hard to find $C$ that leads to a solution $w^*$
with $\norm{w^*}_0 = s$.}
\cite{BT11} proved
linear convergence of the objective value with the LS loss
function \cref{eq:ls_loss} for the iterates generated by PG under a
scalable restricted isometry property, which also served as their tool
to accelerate PG. However, given any problem instance, it is hard, if not
computationally impossible, to verify whether the said property holds.
On the other hand, \cite{BKM16} established
global subsequential convergence to a stationary point for the iterates of PG
on \cref{eq:problem} without the need for such isometry conditions, and their
results
are valid for general loss functions $f$ beyond \cref{eq:ls_loss}.
While some theoretical guarantees are
known, the practicality of PG for solving
\cref{eq:problem} remains a big problem in real-world applications
as its empirical convergence speed tends to be slow.
\alertt{The PG approach is called iterative hard thresholding (IHT) in studies of
compressed sensing \citep{BluDav09} that mainly focuses on the
LS case.
To accelerate IHT, several approaches that alternates between a PG
step and a subspace optimization step are also proposed
\cite{Blu12,BahRB13a}, but such methods mainly focus on the
LS case and statistical properties, while their convergence speed is
less studied from an optimization perspective.}
Recently, ``acceleration'' approaches for PG on general nonconvex regularized problems have
been studied in \cite{NIPS2015_f7664060,WCP18}. While their proposed
algorithms are also applicable to \cref{eq:problem}, the obtained convergence
speed for nonconvex problems is
not faster than that of PG.
\Alertt{This work is inspired by our earlier work \cite{JHA22a}, which
considered a much broader class of problems without requiring convexity nor
differentiability assumptions
for $f$, and hence obtained only much weaker convergence results,
with barely any convergence rates, for such general problems.}
\paragraph{Contributions.} In this work, we revisit the PG
algorithm for solving the general problem
\cref{eq:problem} and propose two acceleration schemes by leveraging the
combinatorial nature of $\ell_0$-norm.
In particular, we decompose the feasible set $A_s$ as the finite union
of $s$-dimensional linear subspaces, each representing a subset of the
coordinates $\{1,\dotsc,n\}$, as
detailed in \cref{eq:As_decomp} of \cref{sec:pgm}. Such subspaces are utilized
in devising techniques to efficiently accelerate
PG.
Our first acceleration scheme is based on a same-space extrapolation
technique such that we
conduct extrapolation only when two consecutive iterates $w_{k-1}$ and
$w_k$ lie in the same
subspace, and the step size for this
extrapolation is determined by a spectral initialization combined with
backtracking
to ensure sufficient function decrease.
This
is motivated by the observation that for \cref{eq:erm},
objective and derivatives at the extrapolated point can be inferred
efficiently through a linear combination of $X w_{k-1}$ and $X w_{k}$.
The second acceleration technique starts with plain PG, and
when consecutive iterates stay in the same subspace, it
begins to alternate between a full PG step and a truncated Newton step
in the subspace to obtain superlinear convergence with extremely low
computational cost.
Our main contributions are as follows:
\begin{enumerate}[leftmargin=*]
\item We prove that PG for \cref{eq:problem} is globally
convergent to a local optimum with a local linear rate, improving upon the sublinear results of \citet{BKM16}. We emphasize
that our framework, like
\citep{BKM16}, is applicable to
general loss functions $f$ satisfying the convexity and smoothness requirements, and therefore
covers not only the classical sparse regression problem but also
many other ones encompassed by the empirical risk minimization
(ERM) framework.
\item By decomposing $A_s$ as the union of linear subspaces,
we further show that PG is provably capable of identifying a subspace containing a
local optimum of \cref{eq:problem}.
By exploiting this property,
we propose
two acceleration strategies with practical implementation and convergence
guarantees
for the general problem class \cref{eq:problem}.
Our acceleration provides both computational and theoretical
advantages for convergence, and can in particular obtain
superlinear convergence.
\item In comparison with existing acceleration methods for
nonconvex problems \citep{NIPS2015_f7664060,WCP18}, this work
provides new acceleration schemes with faster theoretical speeds
(see \cref{thm:rate2,thm:newton}), and beyond being applied to
the classical PG algorithm,
those schemes can
also easily be combined with existing accelerated PG approaches to
further make them converge even faster.
\item Numerical experiments exemplify the significant improvement
in both iterations and running time brought by our
acceleration methods, in particular over the projected gradient
algorithm by \cite{BKM16} as well as the accelerated proximal
gradient method for nonconvex problems proposed by
\cite{NIPS2015_f7664060}.
\end{enumerate}
This work is organized as follows. We review the projected gradient algorithm
and prove its local linear convergence and subspace identification
for arbitrary smooth loss functions in \cref{sec:pgm}. In \cref{sec:accelerate},
we propose the acceleration schemes devised through
decomposing the constraint set in \cref{eq:problem} into
subspaces of $\Re^n$.
Experiments in \cref{sec:exp} then illustrate the effectiveness of
the proposed acceleration techniques, and \cref{sec:conclusion} concludes this
work.
All proofs, details of the experiment settings, and additional
experiments are in the appendices.
\section{Projected Gradient Algorithm}\label{sec:pgm}
The \emph{projected gradient algorithm} for solving \cref{eq:problem} is given
by the iterations
\begin{equation}
w ^{k+1} \in \Tpg (w^k) \coloneqq P_{A_s}(w ^k - \lambda \nabla f(w
^k)), \label{eq:pgm}
\end{equation}
where $P_{A_s}(w )$ denotes the projection of $w $ onto $A_s$,
which is set-valued because of the nonconvexity of $A_s$.
When $f$ is given by \cref{eq:ls_loss},
global linear convergence of this algorithm under a restricted
isometry condition is established in \citep{BT11}.
For a general convex $f$ with $L$-Lipschitz continuous
gradients, that is,
\begin{equation}
\label{eq:L-lipschitz}
\|\nabla f(w) - \nabla f(w') \| \leq L \|w - w '\| \quad \forall
w, w' \in \Re^n,
\end{equation}
the global subsequential convergence of \cref{eq:pgm} is proved in
\cite{BKM16}, but neither global nor local rates of convergence is provided.
In this section, we present an alternative proof of global convergence and
more importantly establish its local linear convergence.
A useful observation that we will utilize in the proofs of
our coming convergence results is that the nonconvex set
$A_s$ given by \cref{eq:As} can be decomposed as
a finite union of subspaces in $\Re^n$:
\begin{equation}
\label{eq:As_decomp}
A_s = \bigcup\nolimits _{J\in \J _s} A_{J}, \quad A_{J} \coloneqq {\rm
span} \{ e_j: j\in J\}, \quad
\J_s \coloneqq \left\{ J\subseteq \{1,2,\dots,n\}: |J| = s\right\},
\end{equation}
where $e_j$ is the $j$th standard unit vector in $\Re^n$. \alert{Throughout
this paper, we assume that $\lambda\in (0,L^{-1})$.}
\begin{theorem}
\label{thm:pgm}
\alert{Let $\{w ^k\}$ be a sequence
generated by \cref{eq:pgm}.} Then:
\begin{enumerate}[leftmargin=*]
\item[(a)] (Subsequential convergence)
\alert{Either $\{f(w^k)\}$ is strictly decreasing, or there exists
$N>0$
such that $w^k = w^N$ for all $k\geq N$. In addition,} any
accumulation point $w^*$ of $\{w^k\}$ satisfies
$w^* \in P_{A_s} (w^* - \lambda \nabla f(w^*))$, and is
hence a stationary point of \cref{eq:problem}.
\item[(b)] (Subspace identification and full convergence) There
exists $N \in \N$ such that
\begin{equation}
\{w^k\}_{k=N}^{\infty}\subseteq \bigcup\nolimits
_{J\in\I_{w^*}} A_J,\quad
\quad \I_{w^*} \coloneqq \{ J\in \J_s: w^* \in
A_J\}.
\label{eq:identification}
\end{equation}
whenever $w^k\to w^*$. In particular, if $\Tpg (w ^*)$ is a singleton
for an accumulation point $w^*$ of $\{w^k\}$, then
$w^*$ is a local minimum for \cref{eq:problem}, $w ^k\to w ^*$, and
\cref{eq:identification} holds.
\item[(c)] ($Q$-linear convergence) If $\Tpg (w ^*)$ is a singleton
for an accumulation point $w^*$ and $w \mapsto w -
\lambda \nabla f(w)$ is a contraction over $A_J$ for all $J\in
\I_{w^*}$, then $\{w^k\}$ converges to $w^*$ at a $Q$-linear rate.
In other words, there is $N_2 \in \N$ and $\gamma \in [0,1)$ such that
\begin{equation}
\label{eq:Q-linear}
\norm{w^{k+1} - w^*} \leq \gamma \norm{w^{k} - w^*}, \quad \forall
k \geq N_2.
\end{equation}
\end{enumerate}
\end{theorem}
It is well-known that an optimal solution of \cref{eq:problem} is
also a
stationary point of it \cite[Theorem 2.2]{BE13}, and therefore (a) proves the
global subsequential convergence of PG to candidate solutions of
\cref{eq:problem}. Consider $z^* \coloneqq w^* - \lambda \nabla f(w^*)$,
and let $\tau$ be a permutation of $\{1,\dotsc,n\}$ such that
$z^*_{\tau(1)} \geq z^*_{\tau(2)} \geq \cdots \geq z^*_{\tau(n)}$.
The requirement of $\Tpg (w ^*)$ being a singleton in
\cref{thm:pgm} (b) then simply means the mild condition of
$z^*_{\tau(s)} > z^*_{\tau(s+1)}$, which is almost always true in
practice.
The requirement for (c) can be fulfilled when $f$ confined to $A_J$ is
strongly convex, even if $f$ itself is not.
This often holds true in practice when $f$ is of the form
\cref{eq:erm} and we restrict $s$ in \cref{eq:problem} to be
smaller than the number of data instances $m$, and is thus also mild.
The existence of a stationary point can be guaranteed when $\{w^k\}$
is a bounded sequence, often guaranteed when $f$ is coercive on
$A_J$ for each $J \in \J_s$.
In comparison to existing results in \cite{BKM16,HA13a,BolSTV18a},
parts (b) and (c) of \cref{thm:pgm} are new.
In particular, part (b) provides a full convergence result that usually
requires stronger regularity assumptions like the
Kurdyka-{\L}ojasiewicz (KL) condition \cite{HA13a,BolSTV18a} \alert{(see also
\cref{eq:KL2})} \alertt{that requries the objective value to decrease
proportionally with the minimum-norm subgradient in a neighborhood
of the accumulation point, but we just need the very mild
singleton condition right at the
accumulation point only}. Part (c) gives a local
linear convergence for the PG iterates even if the
problem is nonconvex, while the rates in \cite{BolSTV18a} requires a
KL condition and the rate is measured in the objective value.
The following result further provides rates of convergence of the
objective values even without the conventional KL assumption.
The first rate below follows from \cite{LeeW19a}.
\begin{theorem}
\label{thm:sublinear}
\alert{Let $\{w ^k\}$ be a sequence generated by \cref{eq:pgm}}.
If $w^k\to w^*$, such as when $\Tpg (w^*)$ is a singleton at an
accumulation
point $w^*$ of \cref{eq:pgm}, then
\begin{equation}
\label{eq:sublinear}
f(w^k) - f(w^*) = \alert{o}(k^{-1}).
\end{equation}
Moreover, under the hypothesis of \cref{thm:pgm} (c), the
objective converges to $f(w^*)$ R-linearly, i.e.,
\begin{equation}
f(w^k) - f(w^*) = O(\exp(-k)).
\label{eq:linear}
\end{equation}
\end{theorem}
By using \cref{thm:pgm}, we can also easily get
rates faster than \cref{eq:sublinear} under a version of the KL
condition that is easier to understand and verify than those assumed
in existing works.
In particular, existing analyses require the KL condition to hold in a
neighborhood in $\Re^n$ of an
accumulation point, but we just need it
to hold around $w^*$ within $A_J$ for the restriction $f|_{A_J}$ for each $J \in
\I_{w^*}$.
These results are postponed to \cref{thm:rate2} in the next section as
the PG method is a special case of our acceleration framework.
\section{Accelerated methods}
\label{sec:accelerate}
The main focus of this work is the proposal in this section of new
techniques with solid convergence guarantees to accelerate the PG
algorithm presented in the preceding section.
Our techniques fully exploit the subspace identification property
described by the inclusion \cref{eq:identification}, as well as the
problem structure of \cref{eq:erm} to devise efficient algorithms.
We emphasize that the two acceleration strategies described below can
be combined together, and they are also widely applicable such that
they can be employed to other existing algorithms for
\cref{eq:problem} as long as such algorithms have a property similar to
\cref{eq:identification}.
\subsection{Acceleration by extrapolation}
Traditional extrapolation techniques are found in the realm of convex
optimization to accelerate algorithms
\citep{BeckTeboulle09,Nesterov83} with guaranteed convergence
improvements, but were often only adopted as heuristics in the
nonconvex setting, until some recent works showed that theoretical
convergence can also be achieved \citep{NIPS2015_f7664060,WCP18}.
However, unlike the convex case, these
extrapolation strategies for nonconvex problems do not lead to faster
convergence speed nor an intuitive reason for doing so.
An extrapolation step proceeds by choosing a
\emph{positive} stepsize along the
direction determined by two consecutive iterates. That is, given two iterates
$w^{k-1}$ and $w^k$, an
intermediate point $z^k \coloneqq w^k + t_k (w^k-w^{k-1})$ for some stepsize
$t_k \geq 0$ is first calculated
before applying the original algorithmic map ($\Tpg$ in our
case).\footnote{it is clear that if $t_k \equiv 0$, we reduce back to
the original algorithm.}
Another popular acceleration scheme for gradient algorithms is the
spectral approach pioneered by \cite{JB88a}.
They take the differences of the gradients and of the iterates in two
consecutive iterations to estimate the curvature at the
current point, and use it to decide the step size for updating along
the reversed gradient direction.
It has been shown in \cite{WriNF09a} that equipping this step size
with a backtracking procedure leads to significantly faster
convergence for proximal gradient on regularized optimization
problems, which includes our PG for \cref{eq:problem} as a special case.
To describe our proposed double acceleration procedure that combines
extrapolation and spectral techniques, we first observe that all PG iterates
lie on $A_s$,
and that $A_s$ can be finitely decomposed as \cref{eq:As_decomp}. When two
consecutive iterates lie
on the same convex subspace $A_J$ for some $J\in \J_s$, within these two iterations, we
are actually conducting convex optimization.
In this case, an extrapolation step within $A_J$ is reasonable because it will
not
violate the constraint, and acceleration can be expected from the
improved rates of accelerated proximal gradient on convex problems in
\cite{BeckTeboulle09,Nes13a}.
Judging from \cref{thm:pgm} (b),
the corresponding $J$ is also a candidate index set that belongs to
$\I_{w^*}$, so extrapolation within $A_J$ makes further sense.
We set $t_k = 0$ to skip the extrapolation step if $d^k$ is not a
descent direction for $f$ at $w^k$.
Otherwise, we start from some $\hat t_k > 0$ decided by the curvature
information of $f$, and then execute a backtracking linesearch along
$d^k \coloneqq w^k - w^{k-1}$ to set
$t_k = \eta^i \hat t_k$ for the smallest integer $i\geq 0$ that provides
sufficient descent
\begin{equation}
f (w^k + t_k d^k) \leq f (w^k) - \sigma t_k^2 \|d^k\|^2,
\label{eq:suffdecrease}
\end{equation}
given parameters $\eta,\sigma \in (0,1)$.
We then apply \cref{eq:pgm} to $z^k = w^k +
t_k d^k$ to obtain $w^{k+1}$.
For the spectral initialization $\hat t_k$ for accelerating the
convergence,
instead of directly using approaches of \cite{JB88a,WriNF09a} that takes
the reversed gradient as the update direction,
we need to devise a different mechanism as
our direction $d^k$ is not directly related to the gradient.
We observe that for the stepsize
\begin{equation}
\label{eq:BB}
\alpha_k \coloneqq \inner{s^k}{s^k}/\inner{s^k}{r^k}, \quad
s^k \coloneqq w^{k} - w^{k-1}, \quad r^k \coloneqq \nabla f(w^k) -
\nabla f(w^{k-1})
\end{equation}
used in \cite{JB88a}, the final update $-\alpha_k\nabla f(w^k)$ is actually
the minimizer of the following subproblem
\begin{equation}
\label{eq:quad}
\min_{d\in\Re^n}\quad \inner{\nabla f(w^k)}{d} + \norm{d}^2/ (2
\alpha_k).
\end{equation}
By juxtaposing the above quadratic problem and the upper bound provided by
the descent lemma \cite[Lemma 5.7]{Beck17},
we can view $\alpha_k^{-1}$ as an estimate of the local Lipschitz
parameter that could be much smaller than $L$ but still guarantee
descent of the objective.
We thus follow this idea to decide $\hat t_k$ using such curvature
estimate and the descent lemma by
\begin{equation}
\label{eq:BB_extrapolation}
\hat t_k = \argmin_{t\geq 0}\; \inner{\nabla f(w^k)}{t d^k} +
\norm{t d^k}^2 / (2\alpha_k)\quad \Leftrightarrow \quad
\hat t_k = -\inner{\alpha_k \nabla f(w^k)}{
d^k}/ \norm{d^k}^2.
\end{equation}
Another interpretation of \cref{eq:BB} is that $\alpha_k^{-1} I$
also serves as an estimate of $\nabla^2 f(w^k)$,\footnote{As $\nabla f$ is Lipschitz
continuous, it is differentiable almost everywhere.
Here, we denote by $\nabla ^2 f(w^k)$ a generalized Hessian of $f$
at $w$, which is well-defined for $f$ with Lipschitz continuous
gradient \citep{HUJ84a}.}
and the objective in \cref{eq:quad} is a low-cost approximation of the
second-order Taylor expansion of $f$.
However, we notice that for problems in the form of \cref{eq:erm} and
with $d^k \in A_J$, the exact second-order Taylor expansion
\begin{equation}
\label{eq:taylor2}
f(w^k + t d^k) \approx f(w^k) + t\inner{\nabla f(w^k)}{d^k} +
t^2\inner{ \nabla ^2 f(w^k) d^k} { d^k }/2
\end{equation}
can be calculated efficiently.
In particular, for \cref{eq:erm}
and any $d^k \in A_J$, we get from $X d^k = X_{:,J} d^k_J$:
\begin{equation}
\label{eq:gradhess}
\begin{aligned}
\nabla f(w^k)^{\top} d^k &= \nabla g\left( (X w^k) \right)^{\top}
\left( X_{:,J} d^k_J \right),\\
\inner{\nabla^2 f(w^k) d^k}{d^k} &= \inner{(X_{:,J} d^k_J)}{\nabla^2
g\left( (Xw^k) \right) (X_{:,J} d^k_J)},
\end{aligned}
\end{equation}
which can be calculated in $O(ms)$ time by computing $X_{:,J} d^k_J$ first.
This $O(ms)$ cost is much cheaper than
the $O(mn)$ one for evaluating the full gradient of $f$ needed in the
PG step,
so our extrapolation plus spectral techniques has only negligible
cost.
Moreover, for our case of $d^k = w^k - w^{k-1}$,
we can further reduce the cost of calculate $X_{:,J} d^k_J$ and thus
\cref{eq:gradhess} to $O(m)$ by recycling intermediate computational
results needed in evaluating $f(w^k)$ through $X_{:,J} d^k_J = X w^k - X
w^{k-1}$.
With such tricks for efficient computation,
we therefore consider the more accurate approximation
to let $\hat t_k$ be the scalar that minimizes the quadratic function
on the right-hand side of \cref{eq:taylor2} for problems in the form
\cref{eq:erm}. That is, we use
\begin{equation}
\label{eq:tk}
\hat t_k \coloneqq - \inner{\nabla f(w^k)}{d^k}/ \inner{
\nabla ^2 f(w^k) d^k} { d^k }.
\end{equation}
Finally, for both \cref{eq:tk} and \cref{eq:BB_extrapolation},
we safeguard $\hat t_k$ by
\begin{equation}
\hat t_k \leftarrow P_{[c_k \alpha_{\min}
, c_k \alpha_{\max} ] } \left( \hat t_k \right)
\label{eq:hattk}
\end{equation}
for some fixed $\alpha_{\max} \geq \alpha_{\min} > 0$,
where
\begin{equation}
c_k \coloneqq \norm{(\nabla f(w^k))_{J} }/ (\zeta_k \norm{d^k}),
\quad
\zeta_k \coloneqq - \inner{d^k}{\nabla f(w^k)} /
(\norm{d^k} \norm{(\nabla f(w^k))_{J} }) \in (0,1].
\label{eq:cosine}
\end{equation}
We also note that the low cost of evaluating $X d^k$ is also the key
to making the backtracking in \cref{eq:suffdecrease} practical,
as each $f(w^k + \eta^i \hat t_k d^k)$ can be calculated in $O(m)$
time through linear combinations of $X w^k$ and $X d^k$.
The above procedure is summarized in
\cref{alg:extrapolation} \alert{with global convergence guaranteed
by \cref{thm:global_acce}. In \cref{thm:rate2}, we establish its
full convergence as well as its
convergence rates under a KL condition at $w^*$: there exists
neighborhood $U \subset \Re^n$ of
$w^*$, $\theta \in [0,1]$, and $\kappa > 0$ such that for every $J
\in \I_{w^*}$,
\begin{equation}
\left(f(w) - f(w^*)\right)^{{\theta}} \leq
{\kappa } \norm{(\nabla
f(w))_{J}},\quad \forall w\in A_J \cap U.
\label{eq:KL2}
\end{equation}
We denote
by $n_k$ the number of successful
extrapolation
steps in the
first $k$ iterations of \cref{alg:extrapolation}.}
The part of $\theta \in [0,1/2]$ with $f$ being convex in the last
item of \cref{thm:rate2} is directly from the result of \cite{CPL22b}.
\begin{theorem}
\label{thm:global_acce}
Under the hypotheses of \cref{thm:pgm}, any
accumulation point of a sequence generated by \cref{alg:extrapolation}
is a stationary point.
\end{theorem}
\begin{theorem}
\label{thm:rate2}
\alert{Consider either \cref{eq:pgm} or
\cref{alg:extrapolation} with }$\eta,
\sigma,{\epsilon}\in (0,1)$, and $\alpha_{\max} \geq \alpha_{\min} > 0$,
and suppose that there is an accumulation
point $w^*$ of the iterates at which
the KL condition holds. Then $w^k\to w^*$. Moreover, the following rates
hold:
\begin{enumerate}[leftmargin=*]
\item[(a)] If \alert{$\theta \in (1/2,1)$}: $f(w^k) - f(w^*) =
O((k + n_k)^{-1/\alert{(2\theta -1)}})$.
\item[(b)] If \alert{$\theta \in (0, 1/2]$}: $f(w^k) - f(w^*) = O(\exp(-(k
+ n_k)))$.
\item[\alert{(c)}] \alert{If \alert{$\theta = 0$, or $\theta \in [0,1/2]$ and $f$ is convex}:
there
is $k_0 \geq 0$ such that $f(w^k) = f(w^*)$ for all $k
\geq k_0$}.
\end{enumerate}
\end{theorem}
\alert{We stress that convexity of $f$ is not required in
\cref{thm:global_acce,thm:rate2} \Alertt{except the second half of the
last item of \cref{thm:rate2}}.}
There are several advantages of the proposed extrapolation strategy
over existing ones in \cite{NIPS2015_f7664060,WCP18}.
The most obvious one is the faster rates in \cref{thm:rate2} over
PG such that each successful extrapolation step in our method
contributes to the convergence speed, while existing methods only
provide the same convergence speed as PG.
Next, existing strategies only use prespecified step sizes without information
from the given problem nor the current progress,
and they only restrict such step sizes to be within $[0,1]$.
Our method, on the other hand, fully takes advantage of the function
curvature and can allow for arbitrarily large step sizes to better
decrease the objective.
In fact, we often observe $t_k \gg 1$ in our numerical
experiments.
Moreover, our acceleration techniques utilize the nature of
\cref{eq:As_decomp} and \cref{eq:erm} to obtain very efficient
implementation for ERM problems such that the per-iteration cost of
\cref{alg:extrapolation} is almost the same as that of PG, while the approach of
\cite{NIPS2015_f7664060} requires evaluating $f$ and $\nabla f$ at two
points per iteration, and thus has twice the per-iteration cost.
\begin{algorithm2e}[tb]
\DontPrintSemicolon
\caption{Accelerated projected gradient algorithm by extrapolation (APG)}
\label{alg:extrapolation}
Given an initial vector
$w^0\in\Re^n$ and parameters ${\epsilon},\eta,\sigma\in (0,1)$,
$\alpha_{\max} \geq \alpha_{\min} > 0$, $\lambda \in (0,1/L)$.
\For{$k=0,1,2,\dotsc$}{
\If{$k>0$; $w^{k-1}$ and $w^k$ activate the same $A_J$;
and ${\zeta_k \geq \epsilon}$}{
\label{line:if}
$d^k \leftarrow w^k - w^{k-1}$, and compute $\hat t_k$
from \cref{eq:hattk} with either \cref{eq:BB_extrapolation} or \cref{eq:tk}
\For{$i=0,1,\dotsc$}{
$t_k \leftarrow \eta^i \hat t_k$
\lIf{\cref{eq:suffdecrease} is satisfied}{$z^k
\leftarrow w^k + t_k d^k$, and break}
}
}
\lElse{
$z^k \leftarrow w^k$
}
$w^{k+1} \leftarrow\Tpg (z^k)$
}
\end{algorithm2e}
\subsection{Subspace Identification}
\label{sec:identify}
In line with the above discussion, we interpret
\cref{eq:identification}
as a theoretical property guaranteeing that the iterates of the projected gradient algorithm
\cref{eq:pgm}
will eventually identify the subspaces $A_J$ that contain a candidate solution
$w^*$ after a finite number of iterations. Consequently, the task of minimizing
$f$ over the nonconvex set $A_s$ can be reduced to a convex optimization
problem of minimizing $f$ over $A_J$. Motivated by this, we present a
two-stage algorithm described in \cref{alg:identification} that switches to
a high-order method for smooth convex optimization after a candidate
piece $A_J$ is identified to obtain even faster convergence.
Since $\nabla f$ is assumed to be
Lipschitz continuous, the generalized Hessian of $f$ exists everywhere
\citep{HUJ84a}, so we may employ a semismooth Newton
(SSN) method \citep{QiS93a} with backtracking linesearch to get a faster
convergence speed with low cost (details in \cref{subsec:ssn}). In particular, we
reduce the computation costs by considering the restriction of $f$ on the
subspace $A_J$ by
treating
the coordinates not in $J$ as non-variables so that the problem
considered is indeed smooth and convex.
As we cannot know a priori whether $I_{w^*}$ is indeed identified,
we adopt the approach implemented in
\cite{LeeW12a,YSL20a-nourl,LeeCP20-nourl} to consider it
identified when $w^k$ activates the same $A_J$ for long enough
consecutive iterations.
To further safeguard that we are not optimizing over a wrong subspace,
we also incorporate the idea of
\cite{Wri12a,BarIM20a,YSL20a-nourl,LeeCP20-nourl}
to periodically alternate to a PG step \cref{eq:pgm} after switching
to the SSN stage.
A detailed description of this two-stage algorithm is in
\cref{alg:identification}.
In the following theorem, we show that superlinear convergence can be
obtained for \cref{alg:identification} even if we take only one SSN
step every time between two steps of \cref{eq:pgm}, using a simplified
setting of twice-differentiability.
\alert{For our next theorem}, we need to introduce some additional notations.
Given any $w \in A_J$, we use $f_{J}(w_J) \coloneqq f(w)$ to
denote the function of considering only the coordinates of $w$ in
$J$ as variables and treating the remaining as constant zeros.
\alert{We assume that the conditions of \cref{thm:pgm} (b) hold with $w^* \in
A_s$, and that $f$ is
twice-differentiable around a neighborhood $U$ of $w^*$ with
$\nabla^2 f_J$ Lipschitz continuous in $U$ and $\nabla^2 f_J(w^*)$
positive definite for all $J \in \I_{w^*}$.}
\begin{theorem}
\alert{Suppose that starting after $k
\geq N$ and $P_{A_s}(w^k) \subset U$}, we conduct $t$ Newton steps
between every two steps of \cref{eq:pgm} for $t \geq 1$:
\begin{equation}
\label{eq:newton}
w^{k,0} \in P_{A_s}(w^k), \,
\begin{cases}
J &\in \I_{w^{k,0}},\\
w^{k,j+1}_{i} &= 0, \quad \forall i \notin J, \quad j=1,\dotsc,t-1,\\
w^{k,j+1}_J &= w^{k,j}_J -
\nabla^2 f_J(w^{k,j}_J)^{-1} \nabla f_J(w^{k,j}_J),\\
\end{cases}
\,
w^{k+1} \in \Tpg (w^{k,t}).
\end{equation}
Then $w^k \rightarrow w^*$ at a $Q$-quadratic rate.
\label{thm:newton}
\end{theorem}
In practice, the linear system for obtaining the SSN step is only
solved inexactly via a (preconditioned) conjugate gradient (PCG) method,
and with suitable stopping conditions for PCG and proper algorithmic
modifications such as those in \cite{MCY19a,MorYZZ20a},
superlinear convergence can still be obtained easily.
Interested readers are referred to \cref{subsec:ssn} for a more detailed
description of our implementation.
\begin{algorithm2e}[tb]
\DontPrintSemicolon
\caption{Accelerated projected gradient algorithm by subspace identification (PG+)}
\label{alg:identification}
Given an initial vector $w^0\in\Re^n$ and
$S,t\in \mathbb{N}$. Set Unchanged $\leftarrow 0$.
\For{$k=0,1,2,\dotsc$}{
\If{$k > 0$, and $w^{k-1}$ and $w^k$ activate the same component of $A_s$}{
Let $J\in \J_s$ correspond to the activated component
Unchanged $\leftarrow$ Unchanged $+1$
}
\lElse{
Unchanged $\leftarrow 0$
}
\If{Unchanged $\geq S$}
{
$y^k \leftarrow P_{A_J}(w^k)$ and
use $t$ steps of SSN described in \cref{subsec:ssn},
starting from $y^k$, to find $z^k$ that approximately
minimizes $f|_{A_J}$
\lIf{SSN fails}{
$z^k \leftarrow w^k$ and Unchanged $\leftarrow 0$.
}
}
\lElse{
$z^k \leftarrow w^k$
}
$w^{k+1} \leftarrow\Tpg (z^k)$
}
\end{algorithm2e}
\section{Experiments}
\label{sec:exp}
In this section, we conduct numerical experiments to demonstrate the accelerated techniques presented
in \cref{sec:accelerate}. We employ \cref{alg:extrapolation} (APG) with \cref{eq:tk} to
accelerate PG, and
further accelerate APG by incorporating subspace identification described in
\cref{alg:identification}, which we denote by APG+.\footnote{That is, if $Unchanged<S$ in
\cref{alg:identification}, we calculate $z^k$ as in \cref{alg:extrapolation}}
For $f$ in \cref{eq:problem}, we consider both LS
\cref{eq:ls_loss} and logistic regression (LR)
\begin{equation}
f(w) = \sum\nolimits_{i=1}^m \log\left( 1 + \exp\left( - y_i x_i^{\top}
w \right) \right) + \mu\norm{w}^2/2,
\label{eq:lr}
\end{equation}
where $(x_i,y_i) \in \Re^n \times \{-1,1\}$, $i=1,\dotsc, m$, are
the training instances, and $\mu>0$ is a small regularization parameter added to make the logistic loss
coercive.
The algorithms are implemented in MATLAB and tested
with public datasets in \Alert{\cref{tbl:data,tbl:data2} in
\cref{app:expdetails}. All algorithms compared start from $w^0=0$ and
terminate when
the first-order optimality condition
\begin{equation}
\text{Residual}(w) \coloneqq \norm{w - P_{A_s}\left( w - \lambda \nabla f\left(
w
\right) \right)}/ (1+\norm{w} + \lambda \norm{\nabla f \left( w
\right)}) < \hat \epsilon
\label{eq:opt}
\end{equation}
is met for some given $\hat \epsilon > 0$. More setting and parameter
details of our experiments are in \cref{app:expdetails}.
\paragraph{Comparisons of algorithms for large datasets.} To fit the practical
scenario of using \cref{eq:problem}, we specifically selected
high-dimensional datasets with $n$ larger than $m$.
We conduct experiments with various $s$ to widely test
the performance under different scenarios.
In particular, we consider $s \in \{ \ceil{0.01m}, \ceil{0.05m},
\ceil{0.1m}\}$ on all data except for the
largest dataset \webspam, for which we set $s \in \{ \ceil{0.001m},
\ceil{0.005m}, \ceil{0.01m}\}$. }
The results of the experiment with the smallest $s$ are summarized in
\cref{fig:main}, and results of the other two settings of $s$ are in
\cref{sec:moreexp}.
\begin{figure}[tbh]
\centering
Logistic regression
\begin{tabular}{ccc}
\begin{subfigure}[b]{0.29\textwidth}
\includegraphics[width=.86\linewidth]{fig_and_tabs/fig_news20_logistic_s160_time_res_pg}
\caption{\news, $s=\ceil {0.01m}$}
\end{subfigure}&
\begin{subfigure}[b]{0.29\textwidth}
\includegraphics[width=.86\linewidth]{fig_and_tabs/fig_rcv1_logistic_s203_time_res_pg}
\caption{\rcv, $s=\ceil {0.01m}$}
\end{subfigure}&
\begin{subfigure}[b]{0.29\textwidth}
\includegraphics[width=.86\linewidth]{fig_and_tabs/fig_webspam_logistic_s280_time_res_pg}
\caption{\webspam, $s=\ceil {0.001m}$}
\end{subfigure}
\end{tabular}
\smallskip
Least square
\begin{tabular}{cc}
\begin{subfigure}[b]{0.29\textwidth}
\includegraphics[width=.86\linewidth]{fig_and_tabs/fig_E2006-log1p_leastsquare_s161_time_res_pg}
\caption{\El, $s=\ceil {0.01m}$}
\end{subfigure}&
\begin{subfigure}[b]{0.29\textwidth}
\includegraphics[width=.86\linewidth]{fig_and_tabs/fig_E2006-tfidf_leastsquare_s161_time_res_pg}
\caption{\Et, $s=\ceil {0.01m}$}
\end{subfigure}
\end{tabular}
\caption{Experiment on sparse regularized LR and
LS. We present time v.s. residual in
\cref{eq:opt}.}
\label{fig:main}
\end{figure}
\begin{table}[h]
\caption{Comparison of algorithms for \cref{eq:problem} to meet
\cref{eq:opt} with $\hat \epsilon = 10^{-6}$, with
\cref{eq:lr} and \cref{eq:ls_loss} and with sparsity levels $s_1=\ceil{0.01m}$
and $s_2 = \ceil{0.05m}$ for all datasets
except \webspam where $s_1 = \ceil{0.001m}$ and $s_2 = \ceil{0.005m}$.
CPU: CPU time in seconds. GE: number of gradient
evaluations. In one iteration, PG, APG, and APG+ needs one
gradient evaluation , while PG-LL and PG-LL+ needs two.
CG: number of Hessian-vector products in the PCG
procedure for obtaining SSN steps.
PA: prediction accuracy (for \cref{eq:lr}).
MSE: mean-squared error (for \cref{eq:ls_loss}).
Time with $*$ indicates that the algorithm is terminated after
running $10000$ iterations without satisfying \cref{eq:opt}.}
\label{tbl:exp}
\begin{tabular}{@{}l@{}lrrrrrrrr@{}}
\toprule
\multirow{3}*{Dataset} & \multirow{3}*{Method} &
\multicolumn{4}{c}{$s_1$} &
\multicolumn{4}{c}{$s_2$}\\
\cmidrule(lr){3-6}
\cmidrule(lr){7-10}
& & \multicolumn{1}{c}{CPU} & \multicolumn{1}{c}{GE} &
\multicolumn{1}{c}{CG} &
\multicolumn{1}{c}{PA} & \multicolumn{1}{c}{CPU} &
\multicolumn{1}{c}{GE} &
\multicolumn{1}{c}{CG} &
\multicolumn{1}{c}{PA} \\
\midrule
\input{fig_and_tabs/summary_news20_s1s2}
\input{fig_and_tabs/summary_rcv1_s1s2}
\input{fig_and_tabs/summary_webspam_s1s2}
& &
\multicolumn{1}{c}{CPU} & \multicolumn{1}{c}{GE} &
\multicolumn{1}{c}{CG} &
\multicolumn{1}{c}{MSE} &
\multicolumn{1}{c}{CPU} & \multicolumn{1}{c}{GE} &
\multicolumn{1}{c}{CG} &
\multicolumn{1}{c}{MSE}\\
\midrule
\input{fig_and_tabs/summary_E2006-log1p_s1s2}
\input{fig_and_tabs/summary_E2006-tfidf_s1s2}
\end{tabular}
\end{table}
Evidently, the extrapolation procedure in APG provides a
significant improvement in the running time compared with the base algorithm
PG, and further
incorporating subspace identification as in APG+ results to a very fast algorithm
that outperforms PG and APG by magnitudes.
Since the per-iteration cost of PG and APG are almost the same as
argued in \cref{sec:accelerate}, we note that the convergence of APG in terms
of iterations is also superior to that of PG.
We next compare APG and APG+
with the extrapolated PG algorithm of
\citet{NIPS2015_f7664060} that we implement in MATLAB and denote by
PG-LL, which is a state-of-the-art approach for nonconvex regularized
optimization and thus suitable for \cref{eq:problem}.
We report the required time and number of gradient evaluations (which
is the main computation at each iteration) for the algorithms
to drive \cref{eq:opt} below $\hat \epsilon = 10^{-6}$.
For PG, APG, and APG+, one gradient evaluation is needed per
iteration, so the number of gradient evaluations is equivalent to the
iteration count.
For PG-LL, two gradient evaluations are needed per iteration, so its
cost is twice of other methods.
We also report the prediction performance on the test data,
and we in particular use the test
accuracy for \cref{eq:lr} and the mean-squared error for
\cref{eq:ls_loss}.
Results for the two smaller $s$ are in \cref{tbl:exp} while that for
the largest $s$ is in \cref{sec:moreexp}.
It is clear from the results in
\cref{tbl:exp} that APG
outperforms PG-LL for most of the test instances considered, while
APG+ is magnitudes faster than PG-LL.
When we equip PG-LL with our acceleration techniques
by replacing $\Tpg$ in \cref{alg:extrapolation,alg:identification}
with the algorithmic map defining PG-LL, we can further speed
up PG-LL greatly as shown under the name APG-LL+.
We do not observe a method that consistently possesses the best
prediction performance, as this is mainly affected by which local
optima is found, while no algorithm is able to find the best local
optima among all candidates.
With no prediction performance degradation, we see that APG+ and
APG-LL+ reduce the time needed to solve \cref{eq:problem} to a level
significantly lower than that of the state of the art.
\Alert{In \cref{sec:varyingresiduals}, we demonstrate the effect on
prediction performance when we vary the residual \cref{eq:opt} and illustrate
that tight residual level is \Alertt{indeed} required to obtain better prediction. Comparisons
with a greedy method is shown in \cref{sec:comparison_greedy}.}
\Alert{\paragraph{{Transition Plots.}} To demonstrate the behavior of the
algorithm for increasing values of $s$, we fit the smaller datasets in \cref{tbl:data2}
using logistic loss \cref{eq:lr} and least squares loss \cref{eq:ls_loss} for
varying
$s = \ceil{km}$, where $k=0.2,0.4,0.6,\dots,3$. \Alertt{The transition plots
are presented in
\cref{fig:transitionplot}.}} We note that the time is in log
scale.
We can see clearly that APG+ and APG-LL+ are consistently magnitudes faster
than the baseline PG method throughout all sparsity levels.
On the other hand, the same-subspace extrapolation scheme of APG is
consistently faster than PG and APG-LL and slower than the two Newton
acceleration schemes, although the performance is sometimes closer to
APG+/APG-LL+ while sometimes closer to PG.
APG-LL tends to outperform PG in most situations as well, but in
several cases when solving the least square problem, especially when
$s$ is small, it can sometimes be slower than PG.
Overall speaking, the results in the transition plots show that our
proposed acceleration schemes are indeed effective for all sparsity
levels tested.
\begin{figure}[tbh]
\centering
\begin{tabular}{@{}c@{}c@{}c@{}c@{}}
\multicolumn{4}{c}{Sparse regularized logistic regression}\\
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/colon-cancer_logistic_transition_plot}
\caption{\colon}
\end{subfigure}&
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/duke_logistic_transition_plot}
\caption{\duke}
\end{subfigure} &
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/gisette_scale_logistic_transition_plot}
\caption{\gisette}
\end{subfigure}&
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/leukemia_logistic_transition_plot}
\caption{\leu}
\end{subfigure}\\
\multicolumn{4}{c}{Sparse least squares regression}\\
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/colon-cancer_leastsquare_transition_plot}
\caption{\colon}
\end{subfigure}&
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/duke_leastsquare_transition_plot}
\caption{\duke}
\end{subfigure} &
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/gisette_scale_leastsquare_transition_plot}
\caption{\gisette}
\end{subfigure}&
\begin{subfigure}[b]{0.24\textwidth}
\includegraphics[width=\linewidth]{fig_and_tabs/leukemia_leastsquare_transition_plot}
\caption{\leu}
\end{subfigure}
\end{tabular}
\caption{\Alertt{Transition plots. We present sparsity levels versus
running time (in log scale). Top row: logistic loss. Bottom row:
least square loss.}}
\label{fig:transitionplot}
\end{figure}
\section{Conclusions}
\label{sec:conclusion}
In this work, we revisited the projected gradient algorithm for solving
$\ell_0$-norm constrained
optimization problems. Through a natural decomposition of the constraint set into subspaces
and the proven ability of the projected gradient method to identify a subspace that contains a
solution,
we further proposed effective acceleration schemes with provable
convergence speed improvements.
Experiments showed that our acceleration strategies improve
significantly both the convergence speed and the running time of the
original projected gradient algorithm,
and outperform the state of the art for $\ell_0$-norm constrained
problems by a huge margin.
We plan to extend our analysis and algorithm to the setting of a
nonconvex objective in the near future.
\Alert{\section*{Acknowledgments}
This work was supported in part by Academia Sinica Grand Challenge
Program Seed Grant No. AS-GCS-111-M05 and
NSTC of R.O.C. grants 109-2222-E-001-003 and 111-2628-E-001-003. }
\bibliographystyle{plainnat}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 450 |
The SC/ST & OBC cell has been reformed in the year 2013. The aim of the cell is to assist the students who belong to Scheduled Caste (SC), Scheduled Tribe (ST) and Other Backward Communities (OBC), for supporting academic performance and financial benefits from the government. The cell is also focused on monitoring the strict implementation of reservation policy as per UGC guidelines.
To integrate and promote the students who belong to SC/ST & OBC community at par with the main stream student body.
To observe the reservation policy for SCs/STs and OBCs in the institution.
Committee often meets the students and faculties belong to SC/ST & OBC communities, to understand their problems and to assist them in resolving the problems.
Creating awareness among the SC/ST & OBC students regarding the various Government and Non-Government scholarship schemes.
To disseminate and counsel SC/ST & OBC students of the college to utilize the benefits of the schemes offered by the Government and UGC. | {
"redpajama_set_name": "RedPajamaC4"
} | 7,981 |
<?php
/**
* Unit test Guest Toke
*
* I used that for some debug. It's incomplete and I guess
* It would be better to have a proper framework for unit
* test on PHP website. Anyway, it does not harm anyone for now
*
* @category Website
* @package Photoshow
* @license http://www.gnu.org/licenses/
*/
require_once(realpath(dirname(__FILE__)."/TestUnit.php"));
class CurrentUserTest extends TestUnit
{
/**
* Test login & logout
* @test
*/
public function test_login_logout(){
CurrentUser::login("testuser", "testpassword");
$this->assertEquals("testuser", $_SESSION['login']);
$this->assertNull($_SESSION['token']);
$this->assertNotNull(CurrentUser::$account);
$this->assertEquals("testuser", CurrentUser::$account->login);
$this->assertFalse(CurrentUser::$admin);
CurrentUser::logout();
//TODO: Failure because I do require_once. Autoload may solve the issue
$this->assertNull($_SESSION['login']);
$this->assertNull(CurrentUser::$account);
$this->assertNull($_SESSION['token']);
$this->assertFalse(CurrentUser::$admin);
}
}
?>
| {
"redpajama_set_name": "RedPajamaGithub"
} | 8,297 |
Q: Edit or create depending on boolean value I have an objects in my collection that might contain a boolean variable called verified, or not.
If it does have this variable and it is false then I want to set it to true.
If it does not have this variable then I want to create it as true.
query := bson.D{{Key: "email", Value: email}}
update := bson.D{
{Key: "$set",
Value: bson.D{
{Key: "verified", Value: true},
},
},
}
err := col.Collection("objects").FindOneAndUpdate(context.Background(), &query, &update).Err()
I have tried:
upsert := true
opts := options.FindOneAndUpdateOptions{
Upsert: &upsert,
}
filter := bson.D{{Key: "email", Value: email}}
query := bson.D{
{Key: "$set",
Value: bson.D{
{Key: "verified", Value: true},
},
},
}
err := col.Collection("objects").FindOneAndUpdate(context.Background(), &filter, &query, &opts).Err()
But it does not set the verified to true if the verified value does not exist. And even if it does exist and is false, it still isn't set to true.
A: how about using upsert: true?
upsert:
*
*Creates a new document if no documents match the filter Updates a
single document that matches the filter.
try passing upsert:true in findOneAndUpdate()
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 4,793 |
Montes Method
Meet Anthony
Learn The Montes Method. Call me (818) 400-2270
contact@themontesmethod.com
Tootsie & Me
Jun 16, 2017 | Posted by Anthony Montes | Uncategorized | 0 comments |
So, my best friend, who was also a struggling Actor and I, were sitting at the bar called the West Bank Cafe on 42nd St, in Manhattan in the early 80's, around 1982. We were talking about what we were always talking about… Acting. A gentleman at the bar overheard us, "You guys are Actors?," he says. "Yeah, " I said. " You too?," I asked him. "Nah, I'm a chauffeur for George Steinbrenner. "Really," I said, as we moved to the empty bar stools near him. I was thrilled cause I knew that Steinbrenner owned the Yankees, my favorite baseball team since I was a kid.
"You guys are actors, you should go down to that studio down the block, they're shooting a movie. You should get 'in it," he says. 'There's no way we can just go down there and, "Get in it," I said." We don't even have agents. "That don't matter," he says. "I was down there yesterday, and they had lots of actors just come in and say, 'Sylvia Fay' sent them, and they got in the movie." "Sylvia Fay," I asked? "Yeah, she's casting the movie. You should just go down there about 5am and say, "Sylvia Fay sent you." He looked at his watch, Downed his drink and mumbled to himself, "Shit," he said as pulled out a stick of gum and put it in his mouth. "I gotta go.," he said and as he was leaving he shouted back to us, "Good luck!"
"That's Great!," I said to JD. "We gotta go there tomorrow!" "I'm not going," JD said. "What do you mean you're not going? Don't you want to be in the movie?" "Sure I would if I knew for sure we could get in it, but how do know that this guy isn't just full of shit! How do we know he drives for Steinbrenner?," JD said. "Why would he lie? Why would he just tell us to go down there if he didn't know that we could get in this movie? Why would he do that?," I asked. "I don't know. Maybe just to fuck with us." "Well, I'm goin'!" I said. And I did.
The next morning I got up really early and got down to the set a little before 5am. As I walked in the front door, I saw a security guard walk away from his post, not seeing me, heading for the bathroom. Immediately a Women with a Clipboard saw me and walked over. "Can I help you?," she asked? "Yeah, Sylvia Fay sent me," I said. "Great, just follow me," she said. I could not believe how easy it was. She walked me over to where the makeup room was, and introduced me to the makeup people. There where 3 people waiting to go into makeup already. She showed me where the caterer was and invited me to help myself to some breakfast, and as we walked she told me that I would be the Cameraman in the scene and they might be giving me a line and off she went. Unbelievable, I thought. I was going to be the Cameraman and get a line, so that would mean that I would get Taft-Hartley'd and be able to get into the Screen Actors Guild. I was woken out of my daydream when Dabney Coleman walked over to me and introduced himself to me. "Hi, I'm Dabney Coleman and I'm playing Ron!" I looked at him in a state of shock, mouth opened and said nothing. "And you are?," he said. "Oh, me… I'm Tony Montes, I'm the Cameraman, I don't mean a real Cameraman, I mean, I'm the Cameraman in the scene, I'm playing the Cameraman." "Well nice to meet you," he said as he shook my hand. "You should get yourself something to eat, the foods great!" "I will," I said and he walked off.
I could not get to the pay phone quick enough to call JD and tell him what had happened and that he should come down right away, that Steinbrenner's Chauffeur was right, and that when he got down here, just say, "Sylvia Fay sent him. Jimmy said he would come down right away and I went over to to sit by the makeup trailer where there was now only 2 people waiting. I pulled out my copy of, A Hatful of Rain, and began to read. Jimmy and I were working on a scene from it for our class at H.B. Studios. After a few minutes, Dabney passed by us and said, "Tony, did you get yourself something to eat? The foods great!" I hadn't as my stomach was bit nervous and I was afraid to eat anything. Just then, a guy came walking down the hallway and the Woman with the Clipboard stopped him, right in front of us. "Can I help you?" she said. "Yeah, Sylvia Fay sent me." "That's strange…," the Woman with the Clipboard said. "Sylvia sending us 5 when we only needed 4… but, that's okay, we'll find something for you to do." "OH SHIT!" I thought…. Jimmy!
After some time the person to my left went into the makeup room, leaving me to be next with the person who was really supposed to be there sitting on my right. I had thought about getting the hell out of there, as soon as possible before I got myself in trouble, but the idea that I could maybe get my SAG card, would not allow me to leave. And then, JD came whistling down the hallway with his copy of, A Hatful of Rain in his hand. Hearing him whistling, the Woman with the Clipboard, hurried up to him. "Can I help you?," she demanded. "Yeah, Sylvia Fay told me to come here," he said. "Didn't Sylvia tell you to never where all black?," she demanded. "No, nobody told me anything," JD said. My heart was in my throat. This was not going to end well. "Who did you speak to at Sylvia's office?," she said growing more pissed off by the second. "Mark….," he said, making it up. "Wait here!" The Woman with Clipboard, stormed off into her office, next to the makeup room. JD came and sat on the other side of the guy who was really supposed to be there. "What's Shakin' Daddy'O?," JD said to me. "Don't talk to me!," I snapped, staying buried in my play.
After a couple of minutes the Security Guard came walking down the hallway and straight into the office with, the Woman with the Clipboard. I got up immediately and whispered to JD, "Let's go!," "What? Why? I just got here.," JD said. "I'll tell you when we get out of here.," I said. Reluctantly, JD got up and started to follow me. We got about 20 feet when I heard the Woman with the Clipboard shout, "Tony!" JD and I stopped dead in our tracks. The gig was up, I was busted. I'm going to jail, I thought. Slowly I turned to face her. "Sylvia Fay, didn't send you, did she Tony?," "No.," I said weakly. "I'm sorry then, I'm afraid we're not going to be able to use you after all.," she said. "I know… Sorry," I said as I turned and started to walk off. "Daddy'O… What just happened?," JD asked. Just then Dabney passed us and said, "Tony, did you get something to eat?" "No, I'm good…" I said. After he was out of sight, JD said, "Hey Daddy'O… Isn't that…." "I don't want to talk about it I said and we left. As I left I found out the movie was going to be called, TOOTSIE. It would star one of my favorite all time actors, Dustin Hoffman. It would take me many more years before I would finally get my SAG card.
Warning: A non-numeric value encountered in /homepages/40/d679859834/htdocs/clickandbuilds/TheMontesMethod/wp-content/themes/HighendWP/functions/theme-likes.php on line 55
About Anthony Montes
I am an Actor/Writer/Producer & Acting Teacher/Coach
Why It's Important To Have A Place To Work At Your Craft By Anthony Montes on May 18, 2017 9
My Porn…. No! By Anthony Montes on June 20, 2017 5
EVERYTHING HAPPENS THE WAY IT'S SUPPOSED TO HAPPEN By Anthony Montes on May 10, 2017 3
How BILLY JACK, saved my life. By Anthony Montes on July 23, 2017 3
Send me an email and I'll get back to you, as soon as possible.
Get familiar with how I work. Download my book. Download eBook
Please feel free to reach out to Anthony with any questions. We look forward to watching you progress in your acting career! See you soon.
Anthony Montes
10943 Camarillo St, North Hollywood, CA 91602
www.themontesmethod.com
anthony.montes56
2nd Annual International Acting Intensive May 25 – 29, 2020
Dominican Republic Workshop March 23 – 27th
How BILLY JACK, saved my life.
My Porn…. No!
© 2017 · The Montes Method.
Warning: count(): Parameter must be an array or an object that implements Countable in /homepages/40/d679859834/htdocs/clickandbuilds/TheMontesMethod/wp-content/plugins/the-events-calendar/common/src/Tribe/Customizer/Section.php on line 204 | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 9,426 |
{"url":"https:\/\/mechanicaldesign.asmedigitalcollection.asme.org\/energyresources\/article\/144\/2\/022305\/1109214\/Modeling-Nanosecond-Pulsed-Spark-Discharge-and","text":"## Abstract\n\nDilute combustion, either using exhaust gas recirculation or with excess air, is considered a promising strategy to improve the thermal efficiency of internal combustion engines. However, the dilute air-fuel mixture, especially under intensified turbulence and high-pressure conditions, poses significant challenges for ignitability and combustion stability, which may limit the attainable efficiency benefits. In-depth knowledge of the flame kernel evolution to stabilize ignition and combustion in a challenging environment is crucial for effective engine development and optimization. To date, a comprehensive understanding of ignition processes that result in the development of fully predictive ignition models usable by the automotive industry does not yet exist. Spark-ignition consists of a wide range of physics that includes electrical discharge, plasma evolution, joule-heating of gas, and flame kernel initiation and growth into a self-sustainable flame. In this study, an advanced approach is proposed to model spark-ignition energy deposition and flame kernel growth. To decouple the flame kernel growth from the electrical discharge, a nanosecond-pulsed high-voltage discharge is used to trigger spark-ignition in an optically accessible small ignition test vessel with a quiescent mixture of air and methane. Initial conditions for the flame kernel, including its thermodynamic state and species composition, are derived from a plasma-chemical equilibrium calculation. The geometric shape and dimension of the kernel are characterized using a multi-dimensional thermal plasma solver. The proposed modeling approach is evaluated using a high-fidelity computational fluid dynamics procedure to compare the simulated flame kernel evolution against flame boundaries from companion Schlieren images.\n\n## 1 Introduction\n\nDilute combustion is considered one of the promising technologies to improve the brake thermal efficiency of an internal combustion engine [1]. The dilute combustion strategy achieved by either high excess-air ratio or high exhaust gas recirculation can raise the charge specific heat and lower the combustion temperature, both leading to a thermal efficiency gain compared to conventional spark-ignition operation. Nevertheless, under dilute operation, the ignition system faces significant challenges in terms of ignition and combustion stability. Stable ignition requires the delivery of the proper amount of electrical energy into the mixture to generate a flame kernel large enough to initiate combustion. Mixture ignitability is closely related to the cycle-to-cycle variability; therefore, a reliable ignition system with optimized operating strategies is crucial to achieving dilute combustion and support engine development effectively.\n\nComputational fluid dynamics (CFD) can be an effective development tool that leverages high-performance computing resources. Within the last few decades, several studies have been devoted to developing spark-ignition models for engine simulation, and many of these models have been adopted in the engine multi-dimensional modeling community [210]. Duclos and Colin [3] have proposed an electrical circuit model to calculate the spark energy transfer rate from the secondary circuit to the gas. Spark channel elongation by the flow motion is mostly modeled with Lagrangian particles [35,810], in that Dahms et al. [4] have taken the effect of turbulent stretch and wrinkle on the arc motion into account. For a localized ignition under crossflow around the spark plug, some studies [4,9] have monitored ignitability along the spark channel based on mixture and flow conditions. The spherical flame kernel is usually initialized either in between the electrodes or at where local conditions permit, and its growth rate and temperature are computed by a set of one-dimensional (1D) governing equations. Tan and Reitz [2] have derived a plasma expansion velocity for the rate of change of kernel radius. In addition, Lucchini et al. [5] have applied a thermal expansion term for the kernel growth after the flame kernel temperature decreases below a certain threshold. Some researchers [8,10] have coupled the Eulerian-type energy deposition model with a Lagrangian-type evolution of the spark channel, instead of using 1D equations for describing the kernel growth. However, these models have not been consistently tested and evaluated at challenging operating conditions. In addition to dilute mixtures, the predictive performance of ignition models may further decrease due to the prevalence of severe in-cylinder conditions for advanced engine concepts, namely less reactive mixtures, high-speed cross flows, and intense turbulence levels around the spark plug. Moreover, complex electrical discharge features, including blow-out and re-strikes [11], pose numerical modeling challenges. To date, a comprehensive and fully predictive ignition model has not yet been developed that can be readily used by the automotive industry. To further improve model accuracy and capability, fundamental understandings of the entire ignition process are required that inform systematic model development.\n\nSpark-ignition consists of a large number of physical processes that include electrical discharge, plasma evolution, Joule-heating of gas, flame kernel initiation, and kernel growth. Timescales of these events cover a wide spectrum from nanoseconds to milliseconds [12]. In a conventional spark system, the breakdown event ionizes the gas between the anode and cathode to open an electrically conductive channel and results in a plasma that reaches extremely high pressure and temperature values within tens of nanoseconds. Next, a shock wave propagates away from the channel within a few microseconds. The breakdown event is followed by the arc phase, where the thin plasma channel expands by the heat conduction and diffusion and initiates the combustion reaction. Finally, the glow discharge phase takes place, where the remainder circuit energy is deposited to the gas at a relatively lower temperature. As the timescale of glow discharge is of the order of milliseconds, continuous electrical energy deposition overlaps heat release by combustion and both contribute to intermediate species production as well. Thus, for a conventional ignition system, it is impossible to decouple the kernel growth from the electrical discharge.\n\nIn this study, an advanced approach is proposed to model spark-ignition energy deposition and flame kernel growth. A nanosecond-pulsed discharge (NPD) experiment is designed to lead to spark-ignition in an optically accessible spark calorimeter at Sandia National Laboratories. The experiment is purposely designed to shorten the duration of the discharge, thereby decoupling the flame kernel growth from the energy source. The flame kernel initialization model used in this study has an assumption that a plasma-chemical equilibrium state is reached in the kernel after the NPD without losses of kinetic and thermal energy by shock or radiation. The electrical energy input to the equilibrium computation uses time-resolved discharge voltage and current data along with a developed circuit model. The shape and size of the plasma channel are derived from an equilibrium plasma simulation. The proposed modeling approach is evaluated using a high-fidelity CFD procedure to compare the simulated flame kernel evolution against flame boundaries from the companion Schlieren images. Finally, a parametric study is performed for both input energy and plasma channel volume. In the following sections, a detailed description of the experimental and numerical setup is first provided. This is followed by a description of the CFD modeling methodology and relevant assumptions for the flame kernel initialization. Lastly, model assessment and a parametric study of the model inputs are presented.\n\n## 2 Experimental and Numerical Setup\n\n### 2.1 Experimental Setup.\n\nAll tests are performed in a custom-made (stainless steel 316), optically accessible, spark calorimeter, which is shown in Fig. 1. The experimental apparatus has been discussed in detail previously [13,14], with a summary presented here. The total internal volume of the calorimeter is 29 cm3. Three optical UV grade quartz (Corning 7980) windows are installed in the calorimeter body to visualize the ignition and flame propagation. A 20 mm diameter front viewing window enables direct imaging. Two orthogonally installed side windows with 16 mm diameter clear apertures of 12.7 mm allow for a pulsed light for Schlieren imaging of the streamer volume and ignition kernel to pass through the chamber. A pin-to-pin (P2P) opposed electrode configuration is used; the high-voltage anode is a modified non-resistor spark plug with the J-hook removed, and the anode machined to a conical shape with a rounded tip (\u223c70 \u03bcm radius of curvature) to maximize local electric field strengths while maintaining relatively repeatable discharge characteristics. The anode is centrally positioned at the top of the calorimeter. A 1\/8 in. (3.18 mm) diameter stainless steel rod is used as the cathode. The cathode tip has a flat tip with a 200 \u03bcm plateau, as observed by a microscopic image. The cathode is installed from the calorimeter base and secured in place by a Swagelok fitting. A constant inter-electrode distance of 1.01 mm is maintained throughout the present study.\n\nFig. 1\nFig. 1\nClose modal\n\nThe calorimeter is filled to the desired pressure (5 bar absolute) using methane (CH4) and desiccated air (O2: 20.95%, N2: 79.05% by volume). The mixture is heated to 343 K using resistive heat tape connected to temperature controllers with the temperature monitored using embedded K-type thermocouples. Although the temperature and pressure are lower than a typical in-cylinder condition at the time of ignition, resulting initial densities of 4.88 and 4.92 kg\/m3 are comparable to engine-like conditions. For plasma discharges, density is the most relevant thermodynamic parameter. The equivalence ratios of the fuel-air mixture are varied from \u03c6 = 1 to \u03c6 = 0.55 at the given vessel condition. The lean flammability limit of the methane\u2013air mixture at 5 bar pressure and 343 K temperature is found to be \u03c6lean = 0.58.\n\nTo generate the NPD within the calorimeter, a Transient Plasma Systems Inc. SSPG-101-HF high-voltage pulse generator with a \u223c10 ns full width at half maximum (FWHM) pulse width and 5 ns rise time is used. A low-impedance inline attenuator probe is used to monitor pulse voltage and current for each discharge event at the upstream of the spark plug. Schlieren imaging is performed for each discharge using a pulsed green light-emitting diode, a Photron SA-Z high-speed camera, and a vertical knife-edge; a schematic of the setup is shown in Fig. 2. The camera is operated at 12 \u03bcs exposure time at 60,000 frames per second. Schlieren images are post-processed to remove noise and increase contrast.\n\nFig. 2\nFig. 2\nClose modal\n\n### 2.2 Numerical Setup.\n\nFlame kernel growth simulations are performed using converge [15], a CFD tool that automates the mesh generation process and permits using simple orthogonal grids, locally forced grid embedding, and the adaptive mesh refinement algorithm (AMR). The turbulent flow is modeled with the dynamic structure model [16] with the large eddy simulation (LES) framework. A multi-zone well-stirred reactor model is used for the combustion, in that the multi-zone method [17] is utilized to reduce the computational cost. The GRI-Mech 3.0 chemical mechanism [18], consisting of 53 species and 325 reactions, is employed. The walls enclosing the spark calorimeter are set as non-slip wall boundaries with a constant temperature of 343 K. The Werner and Wengle wall model [19] is applied. A conjugate heat transfer (CHT) simulation is not considered due to the lack of significant contact between the burned gas and the walls.\n\nThree equivalence ratios, one stoichiometric condition (\u03c6 = 1), and two lean conditions (\u03c6 = 0.7 and \u03c6 = 0.6) are numerically investigated. The three-dimensional computational domain covers the entire inner volume of the spark calorimeter, and the grid configuration near the electrode gap region is shown in Fig. 3. The base grid size is 1.6 mm, and the fixed grid embedding is introduced for two spherical volumes covering the electrode gap region. A 1D laminar premixed flame simulation reveals that the laminar flame thicknesses are ranging between 146 \u03bcm and 422 \u03bcm for three equivalence ratios. The AMR approach is defined to have at least five grid points across the flame thickness. So, a sub-grid scale-based AMR is activated for the mass fraction of HCO, temperature, and velocity. The minimum grid size is 25 \u03bcm for all cases by Y_HCO and temperature sub-grid scales-based AMR.\n\nFig. 3\nFig. 3\nClose modal\n\nFor the NPD-induced plasma, the characterization of the channel shape and size is carried out using vizspark, a multi-dimensional thermal plasma solver that computes magneto-hydrodynamic equations assuming single-temperature and equilibrium chemistry [20,21]. Figure 4 shows the axis-symmetry of the two-dimensional computational domain used in the vizspark simulation. To accurately capture the arc root on the electrode tips, the solid regions of electrodes are also included and solved via an immersed boundary method. The grid consists of the mixed triangle and quadrilateral cells, where the minimum grid size is 5 \u03bcm in the gap region, and the grid size is gradually increased up to 40 \u03bcm. To solve the current continuity equation, a potential boundary condition is imposed on the top of the anode, while a current density boundary condition is assigned to the bottom of the cathode.\n\nFig. 4\nFig. 4\nClose modal\n\n## 3 Modeling Flame Kernel Initialization\n\n### 3.1 Plasma-Chemical Equilibrium Approach.\n\nFlame kernel initialization is an outcome of the electrical discharge and plasma dynamics. The majority of ignition models used for engine simulations simply neglect to solve the breakdown phase and assume an initial flame kernel with arbitrary shape and size. Also, ionized species are not taken into account. Comprehensive modeling that couples plasma physics with a flow solver has been conducted [22]. However, such an approach is still prohibitively expensive.\n\nPlasma evolution by electrical discharge for the NPD used in this study is highly transient with multiple temporal and spatial scales, and complex physics. Relevant dynamics include streamers propagation, ionization and recombination, and joule-heating. Due to the great disparity in timescales, the result of streamer dynamics can be fed as an initial condition to the CFD solver. The output is expected to be extremely high pressure and temperature kernel with ionized species in equilibrium, which initiates the flame kernel and its early growth.\n\nIn this study, a plasma-chemical equilibrium approach is proposed for modeling the flame kernel initialization. Several assumptions are made. First, the thermodynamic properties and species composition are assumed to be in equilibrium because of the high-temperature plasma. From the experimental data, the high-temperature plasma is evident because of an enormously high current footprint. Second, ionized species from the plasma equilibrium are not introduced to the CFD domain. Adding ionized species to the CFD solver would require updated chemical kinetics that also capture multiple ionization and recombination pathways, which will significantly increase the cost of the simulation. Accordingly, ion-recombination is not taken into account. Third, the kinetic energy loss by shock is neglected in the equilibrium calculation. Heat losses due to radiative heat transfer are neglected as well.\n\nThe proposed approach calculates two equilibrium states to obtain initial kernel properties, in terms of temperature, pressure, and species composition. Only pure air is considered since the thermodynamic database for ionized hydrocarbon species is limited. Because the mole fraction of methane at the stoichiometric ratio is less than 10%, the pure air assumption is considered acceptable here. For plasma equilibrium, the air is equilibrated isochorically by maximizing the entropy with total energy deposited from the electrical discharge. A total of 14 species are considered: e\u2212, N, N+, N\u2212, NO, NO+, N2, N2+, N2\u2212, O, O+, O\u2212, O2, and O2+. Next, the methane\u2013air mixture is equilibrated isobarically by minimizing the Gibbs free energy at the resulting temperature and pressure from the plasma equilibrium. Fourteen key species are included in the equilibrium: N, N2, O, O2, NO, C, CO, CO2, H, OH, HO2, H2O, H2, and CH4. The thermodynamic database of all ionized and ground species is obtained from the NASA website [23], and an open-source chemical kinetic software, cantera,1 is used to calculate equilibrium. Thus, the initial kernel temperature and pressure are obtained from the plasma equilibrium, and its species compositions are taken from the second chemical equilibrium step.\n\nTo compute equilibrium, two key inputs are needed: (1) energy deposited from the electrical discharge and (2) plasma channel volume. The experimental data is used to obtain the former input, while the plasma solver is employed to compute the latter.\n\n### 3.2 Energy Calculation From Experiments.\n\nVoltage and current are coupled to the discharge event that takes place between the electrodes. It is difficult to measure these quantities without perturbing the electrode configuration. These quantities are typically measured somewhere upstream of the electrodes, leaving the voltage and current that pass through the discharge to be inferred. In an ideal simulation, the circuit and discharge phenomena would be coupled and manifested in the boundary conditions of the electric field and charge transport equations. The complexity of the routines that take transmission line effects into account, and their accompanying stiffness make such a detailed coupled approach unattractive. In this study, a simpler semi-empirical method to derive boundary voltage and current is proposed.\n\nProfiles of raw voltage and current measured upstream of the electrodes in the NPD experiment are provided in Fig. 5(a). Incident and reflected currents can be qualitatively demarcated from their sign. If the cable between the electrodes and the current probe was longer, the incident and reflected pulses would appear more separated in time; here, they appear to be almost merged due to the shortness of the cable. The reflected voltage is almost all positive, consistent with the transmission line theory. Since deconvoluting these waveforms require circuit parameters that are subject to uncertainty, the voltage and current across the discharge cannot be directly derived from these waveforms accurately. But the product of voltage and current, i.e., power, allows one to calculate the transmitted and reflected energies:\n$Etransmitted=Eincident\u2212Ereflected=\u222b(VI)incident\u2212\u222b(VI)reflected=\u222b(VI)$\n(1)\nwhere the rightmost term is the integral of power, and the leftmost is the transmitted energy. Figure 5(b) shows the energy and power corresponding to Fig. 5(a). It is seen that the long-time integral gives the net energy transmitted to the gas. The net resistance in such circuits is mostly in the discharge, where all the energy is dissipated. The minimum energy peak corresponds to the reflected pulse, i.e., negative power, where the energy undergoes a negative rate of change. This component does not enter the gas since its negative value would imply unphysical gas cooling. The power delivered to the gas can be derived by eliminating the non-monotonic part of the energy and then taking its derivative. The resulting power is positive everywhere, thus guaranteeing that all the delivered energy is dissipated in the gas. By taking the measured voltage as being applied across the gap, a current consistent with the transmitted power can be derived via I = P\u2044V, as shown in Fig. 5(a). This transmitted current serves as a suitable boundary condition for the charge transport equation. Although the voltage used here is not what is actually applied across the electrodes, the approximation used is justified as it conserves energy, does not distort the time of application of voltage too much and is straightforwardly applicable without resorting to more difficult numerical techniques which if implemented could detract from the accuracy of the solution in other ways.\nFig. 5\nFig. 5\nClose modal\n\n## 4 Results and Discussions\n\n### 4.1 Characterization of Plasma Channel.\n\nRetrieving the geometric shape and size of the plasma channel from the Schlieren images is impossible due to image saturation by the highly luminous discharge. In this study, vizspark is used to simulate the evolution of the plasma channel. Figure 6 shows the conduction current density, pressure, and temperature fields in the gap between the two electrodes. As shown earlier, the peak current density occurs at \u223c90 ns, and the NPD ends at \u223c110 ns. In such a short discharge duration, the electrical energy of 4 mJ is deposited into the gas. Accordingly, a very thin conductive channel is established along the electrode axis and allows the current flowing across the gap. The electrical power dissipation results in significant gas heating, and the gas temperature and pressure are extremely elevated. The mean temperature (\u223c25,000 K) is much greater than the adiabatic flame temperature, so it is fair to assume the equilibrium state for the gas inside the kernel. The high pressure induces a strong shock to propagate outward, and a clearly detached shock front is observed. The plasma channel is observed to be a cylindrical shape, despite some corrugations.\n\nFig. 6\nFig. 6\nClose modal\n\nTo obtain the diameter of the cylindrical plasma channel quantitatively, the temporal evolution of radial distributions of temperature and pressure is analyzed. Gas temperature and pressure are averaged in a number of bins in the radial direction, which is shown in Fig. 7. The sharp decrement of temperature can be seen as the interface between the high-temperature plasma channel and the fresh mixture. At the end of discharge, the diameter of the plasma channel is observed to be approximately 100 \u03bcm considering the axisymmetric domain.\n\nFig. 7\nFig. 7\nClose modal\n\n### 4.2 Flame Kernel Initialization.\n\nBy computing the plasma-chemical equilibrium with the inputs derived so far, Table 1 lists the thermodynamic conditions and species composition in the kernel. As the NPD is not varied with mixture equivalence ratio, the temperature and pressure that result from the plasma equilibrium are kept constant for all cases. Due to the high temperature, most species are dissociated to the atomic level, and only molecular N2 exists in small fractions.\n\nTable 1\n\nThermodynamic condition and species composition for kernel initialization (only mass fractions exceed 0.001 kg\/kg are listed)\n\nEquivalence ratio\u03c6 = 1\u03c6 = 0.7\u03c6 = 0.6\nTemperature [K]25685.4\nPressure [MPa]99.1\nMass fraction\n[kg\/kg \u00d7 102]\nN72.373.573.9\nN20.190.200.21\nO21.922.322.4\nC4.122.932.53\nH1.380.980.85\nEquivalence ratio\u03c6 = 1\u03c6 = 0.7\u03c6 = 0.6\nTemperature [K]25685.4\nPressure [MPa]99.1\nMass fraction\n[kg\/kg \u00d7 102]\nN72.373.573.9\nN20.190.200.21\nO21.922.322.4\nC4.122.932.53\nH1.380.980.85\n\nA standard energy deposition approach that sources power into a prescribed shape is also tested in the CFD simulations. The power profile from Fig. 5(b) is used, and a cylindrical source shape with a diameter of 100 \u03bcm is specified. Simulations that use the standard energy deposition are not able to handle the high-power deposition within the thin cylindrical volume and result in temperature beyond the code maximum temperature limit (\u223c95,000 K), which forcedly terminates the simulation during the power sourcing. This is because all the energy is deposited into the ground species without consideration of the ionization. Hence, the ionization process should be considered to predict an accurate gas temperature for the flame kernel initialization.\n\n### 4.3 Flame Kernel Evolution.\n\nThe modeling approach for flame kernel initialization is evaluated by comparing the flame kernel evolution against the experimental data. Figure 8 shows the comparison of kernel growth between simulation results and acquired Schlieren images at \u03c6 = 1 condition. The first few frames of Schlieren images are saturated by bright light emitted from the arc and electrodes, so it is not possible to discern any meaningful comparison from those initial frames. The pseudo-Schlieren image is built by taking the gradient of the density field and plotted on the cut plane through the electrode axis. Note that it is not post-processed with a line integral to obtaining a line-of-sight picture. As the Schlieren images indicate the gradient of the density field, its volume is much larger than the burned region enclosed by the reaction layer. Based on the 1D premixed flame simulation result, the inner reaction layer temperature of 1800 K is used to show combustion activities inside the kernel.\n\nFig. 8\nFig. 8\nClose modal\n\nThe optical measurement reveals that the flame kernel vigorously expands in the horizontal direction. It can be inferred that the flame kernel forms a toroidal shape and surrounds the electrode without a large contact area. To investigate the effect of heat loss on the walls on the kernel evolution, a simulation with adiabatic wall boundary conditions is performed. The difference of kernel evolution between an adiabatic wall and an isothermal wall turns out to be insignificant, which implies that a CHT simulation is not needed for this type of problem. This is mostly due to the extremely short duration of discharge and the initial kernel being blown away from the electrodes and consequently not losing significant heat to them.\n\nSimulation results are well-aligned with the experimental observations, in particular, the kernel size by initial expansion, and subsequent growth calculated by simulations match observations from the Schlieren images. As mentioned earlier, the reaction zone is embedded in the heated volume. It is interesting to note that the reaction zone is shredded due to vigorous expansion, as can be seen at 66.7 \u03bcs. However, the reactivity of the stoichiometric mixture is enough to compete with the blow-off mechanism and finally sustains the flame to propagate outward.\n\nAs the simulated initial kernel expansion matches Schlieren images, simulation results are further leveraged to analyze the dynamics that occur in the very initial period of the spark-ignition event. Figure 9 shows the temporal evolution of velocity and temperature fields. Two velocity components are used to interpret the shock and expansion dynamics. One microsecond after discharge, a shock departs away from the high-temperature kernel, and the kernel size vigorously expands from the initial diameter of 100 \u03bcm to \u223c700 \u03bcm. Then, a pressure equilibrium process starts as the pressure behind the shock front becomes lower than its initial value. Balancing pressure in a free volume would assist the flame kernel in expanding spherically. However, the electrode geometry triggers the non-trivial expansion dynamic of the flame kernel.\n\nFig. 9\nFig. 9\nClose modal\n\nThere is no obstacle against the horizontal expansion, so the kernel consistently expands from the electrode axis after the shock departure. In contrast, vertical expansion is much more complex than horizontal one. The plateau on the cathode tip fully reflects the initial shock wave compared to the rounded tip of the anode, so a steeper pressure gradient is formed that is normal to the plateau surface. On the other hand, the bottom of the anode insulator surface plays a similar role as the cathode tip, which reflects the shock and generates a downward momentum. Two parcels of gas with oppositely directed longitudinal momenta produce a clockwise rotating vortex and assist the horizontal expansion. As a result, the flame kernel shrinks vertically but expands horizontally, which can be seen by comparing the temperature field between 1 \u03bcs and 10 \u03bcs. The upward momentum by the shock reflection at the cathode tip lasts tens of microseconds and sustains the vortex motion. Finally, the combination of all expansion dynamics results in a toroidal flame kernel shape, and then the flame propagation continues. These observations align well with the previous experiment reported in Refs. [24,25].\n\nThe flame kernel initialization model is also evaluated for a lean condition (\u03c6 = 0.7) for which experiments show successful ignition. Figure 10 shows the comparison of flame kernel evolution between simulation and experiments. Note that the NPD characteristics do not change according to the mixture equivalence ratio. The adiabatic flame temperature decreases from 2280 K (\u03c6 = 1) to 1869 K (\u03c6 = 0.7), while the temperature of the inner reaction layer is 1638 K. Accordingly, the smaller kernel volume shown in the first Schlieren image (t = 66.7 \u03bcs) is due to the lowered mixture reactivity. The simulation result indicated by pseudo-Schlieren and iso-temperature contour shows that the horizontal kernel expansion at 66.7 \u03bcs is comparable to that of the \u03c6 = 1 condition. However, as the laminar flame speed decreases from 24.1 cm\/s (\u03c6 = 1) to 11.1 cm\/s (\u03c6 = 0.7), the kernel growth becomes slower. Despite the lower combustion rate, the kernel grows properly and transitions into a propagating flame, which agrees with the experimental observation.\n\nFig. 10\nFig. 10\nClose modal\n\nThe model is further evaluated at an ultra-lean condition (\u03c6 = 0.6) which is close to the experimental lean limit, and the comparison against the experimental data is shown in Fig. 11. The laminar flame speed of \u03c6 = 0.6 mixture at the given condition is 5.9 cm\/s, in which the temperature of the inner reaction layer decreases down to 1537 K. Similar to the previous two conditions, the simulation is able to capture the initial kernel expansion, but it fails to predict the later stage flame propagation. At 133.7 \u03bcs, the kernel volume represented by the pseudo-Schlieren has a comparable width, but a thinner height than that observed in the experiment. Consequently, the embedded reaction zone is packed into a small volume and broken into several pieces. The reaction is eventually quenched by blow-off, while the heated kernel volume starts to shrink.\n\nFig. 11\nFig. 11\nClose modal\n\nTo further analyze the combustion dynamics, Damk\u00f6hler number (Da) is computed for the early stage of kernel growth, as shown in Fig. 12. Da is defined as the ratio of flow residence timescale to chemical timescale, in that the former one is the reciprocal of strain rate (1\/(du\/dx)) and the latter one is the rate of change of methane molar concentration ([CH4]\/(d[CH4]\/dt)). After the initial kernel expansion up to 66.7 \u03bcs, the Da ahead of the reacting front for \u03c6 = 0.6 condition turns out to be much smaller than that results in \u03c6 = 1 condition. At 133.7 \u03bcs, the Da of the stoichiometric case is greater than 10, while that of the lean case becomes comparable to or smaller than unity. Hence, the former case can have vigorous flame kernel growth under the reactive mixture, but the reacting front of the latter case starts to shrink because of perturbations of a chemical reaction by the flow. The kernel expansion momentum, as well as the vortex motion, reduces the residence time for the flame to be self-sustained. As a result, \u03c6 = 0.6 represents a condition beyond the lean limit for CFD simulations. The CFD lean limit lies between \u03c6 = 0.7 and \u03c6 = 0.6, and it is very close to the lean limit observed in the experiments. The remaining discrepancy between CFD and experiments is likely due to the assumptions made for this study. The heat production by ion-recombination processes, the sensitivity of near limit propagation to the chemical kinetic mechanism, and the effect of heat loss by radiation are all worth considering in a future study. Despite a fine grid resolution across the flame thickness used in this study, a higher-order direct numerical simulation may be required to unveil the flame kernel evolution in such challenging mixture conditions.\n\nFig. 12\nFig. 12\nClose modal\n\nTo gain more insight into the impact of boundary conditions on the CFD results, the parametric study on the two inputs (electrical energy and plasma channel size) for the plasma-chemical equilibrium calculation is conducted as follows. The aim of this study is to investigate how different flame kernel evolution results from the input variations (i.e., uncertainties in measurement or plasma simulation) in the flame kernel initialization model. For the \u03c6 = 0.6 conditions, the very first observable Schlieren image (t = 66.7 \u03bcs) is set as the reference, and the inputs are varied for energies of 1, 2, 4 mJ, and diameters of 100, 200, 400 \u03bcm, respectively.\n\nFigure 13 presents the comparison between the acquired Schlieren image and the simulation results. Except for the case of the energy of 1 mJ and the diameter of 400 \u03bcm, a bigger diameter for kernel initialization results in a larger kernel expansion. The effect of energy input is more obvious than that of diameter; kernel expansion is largely under-predicted with half or quarter of the measured energy value. Thus, it can be concluded that energy played a key role in the proposed model. Also, it implies that accurate measurement of voltage and current signals is essential to the simulation. It is worth noting that, for the \u03c6 = 0.6 condition, neither the baseline case (E: 4 mJ, d: 100 \u03bcm) nor the other cases from the parametric study are able to sustain the flame after the expansion and lead to flame propagation. Further studies are required on the flame kernel evolution under the ultra-lean mixture condition close to the lean flammability limit.\n\nFig. 13\nFig. 13\nClose modal\n\n## 5 Conclusions\n\nIn this study, a high-fidelity CFD procedure to simulate the flame kernel initiation and growth from a nanosecond-pulsed spark is developed, and a plasma-chemical equilibrium approach for initializing the plasma energy deposition is proposed and evaluated.\n\nThe model assumes the equilibrium state in the initial flame kernel after the NPD without losses of kinetic and thermal energy by shock and radiation. By computing the plasma equilibrium, the model takes ionization into account to properly calculate the initial temperature and pressure inside the flame kernel. In order to provide accurate energy input, a semi-empirical method is developed to post-process the measured voltage\/current traces and extract the energy, power, and current being transmitted to the electrodes. Equilibrium plasma simulation is performed using vizspark to obtain the shape and size of the kernel. Combining all these inputs to the plasma-chemical equilibrium computation, the initial flame kernel is characterized by very high temperature and pressure, mostly filled with dissociated atoms of O, N, C, and H.\n\nThe proposed model is evaluated against Schlieren data from experiments performed in an optically accessible ignition test vessel. The computed flame kernel expansion matches well with the experimental measurements. The toroidal shape of the flame kernel observed from the experiment is properly reproduced by the CFD simulation. Simulations also provide insight into the mechanisms leading to the kernel evolution after the spark. Ultimately, the calculated flame kernel evolution is in good agreement with experiments for the stoichiometric (\u03c6 = 1) and one of lean conditions (\u03c6 = 0.7). Simulations fail to describe ignition success for the ultra-lean (\u03c6 = 0.6) case close to the lean flammability limit, which does not agree with experiments. Future investigations will focus on the role of chemical kinetics, ion-recombination, radiation, and grid size to bridge the gap between modeling and experiments in the flame kernel growth for the lean case.\n\n## Acknowledgment\n\nThe submitted paper has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (\u201cArgonne\u201d). Argonne, a US Department of Energy Office of Science laboratory, is operated under Contract No. DE-AC02-06CH11357. This research is funded by DOE\u2019s Vehicle Technologies Program, Office of Energy Efficiency and Renewable Energy. The authors would like to express their gratitude to Gurpreet Singh and Michael Weismiller, program managers at DOE, for their support. In addition, the authors gratefully acknowledge the computing resources provided on Bebop, a high-performance computing cluster operated by the Laboratory Computing Resource Center at Argonne National Laboratory. Finally, the authors would like to thank Anand Karpatne and Doug Breden at Esgee Technologies Inc. for their technical supports and valuable discussions on the use of the vizspark software.\n\n## Conflict of Interest\n\nThere are no conflicts of interest.\n\n## References\n\n1.\nHeywood\n,\nJ.\n, and\nMacKenzie\n,\nD.\n(Eds.)\n2015\n, \u201c\nOn the Road Toward 2050: Potential for Substantial Reductions in Light-Duty Vehicle Energy Use and Greenhouse Gas Emissions\n,\u201d MIT Energy Initiative Report.\n2.\nTan\n,\nZ.\n, and\nReitz\n,\nR. D.\n,\n2006\n, \u201c\nAn Ignition and Combustion Model Based on the Level-Set Method for Spark Ignition Engine Multidimensional Modeling\n,\u201d\nCombust. Flame\n,\n145\n(\n1\u20132\n), pp.\n1\n15\n.\n3.\nDuclos\n,\nJ.-M.\n, and\nColin\n,\nO.\n,\n2001\n, \u201c\nArc and Kernel Tracking Ignition Model for 3D Spark-Ignition Engine Calculations\n,\u201d\nProceedings of the Fifth International Symposium Diagnostics & Modeling of Combustion in IC Engines (COMODIA 2001)\n, pp.\n343\n350\n.\n4.\nDahms\n,\nR. N.\n,\nDrake\n,\nM. C.\n,\nFansler\n,\nT. D.\n,\nKuo\n,\nT.-W.\n, and\nPeters\n,\nN.\n,\n2011\n, \u201c\nUnderstanding Ignition Processes in Spray-Guided Gasoline Engines Using High-Speed Imaging and the Extended Spark-Ignition Model SparkCIMM. Part A: Spark Channel Processes and the Turbulent Flame Front Propagation\n,\u201d\nCombust. Flame\n,\n158\n(\n11\n), pp.\n2229\n2244\n.\n5.\nLucchini\n,\nT.\n,\nCornolti\n,\nL.\n,\nMontenegro\n,\nG.\n,\nD\u2019Errico\n,\nG.\n,\nFiocco\n,\nM.\n,\nTeraji\n,\nA.\n, and\nShiraishi\n,\nT.\n,\n2013\n, \u201c\nA Comprehensive Model to Predict the Initial Stage of Combustion in SI Engines\n,\u201d\nSAE Technical Paper No. 2013-01-1087, Detroit, MI, April 16\u201318, 2013\n.\n6.\nColin\n,\nO.\n, and\nTruffin\n,\nK.\n,\n2011\n, \u201c\nA Spark Ignition Model for Large Eddy Simulation Based on an FSD Transport Equation (ISSIM-LES)\n,\u201d\nProc. Combust. Inst.\n,\n33\n(\n2\n), pp.\n3097\n3104\n.\n7.\nZhu\n,\nG.\n,\nPattabiraman\n,\nK.\n,\nPerini\n,\nF.\n, and\nRutland\n,\nC.\n,\n2018\n, \u201c\nModeling Ignition and Combustion in Spark-Ignition Engines Based on Swept-Volume Method\n,\u201d\nSAE Technical Paper No. 2018-01-0188, Detroit, MI, April 10\u201312\n.\n8.\nGe\n,\nH.\n, and\nZhao\n,\nP.\n,\n2018\n, \u201c\nA Comprehensive Ignition System Model for Spark Ignition Engines\n,\u201d\nProceedings of the 0ASME 2018 Internal Combustion Engine Division Fall Technical Conference\n,\nSan Diego, CA\n,\nNov. 4\u20137\n.\n9.\nPyszczek\n,\nR.\n,\nHahn\n,\nJ.\n,\nPriesching\n,\nP.\n, and\nTeodorczyk\n,\nA.\n,\n2019\n, \u201c\nNumerical Modeling of Spark Ignition in Internal Combustion Engines\n,\u201d\nASME J. Energy Resour. Technol.\n,\n142\n(\n2\n), p.\n022202\n.\n10.\nScarcelli\n,\nR.\n,\nZhang\n,\nA.\n,\nWallner\n,\nT.\n,\nSom\n,\nS.\n,\nHuang\n,\nJ.\n,\nWijeyakulasuriya\n,\nS.\n,\nMao\n,\nY.\n,\nZhu\n,\nX.\n, and\nLee\n,\nS.\n,\n2019\n, \u201c\nDevelopment of a Hybrid Lagrangian-Eulerian Model to Describe Spark-Ignition Processes at Engine-Like Turbulent Flow Conditions\n,\u201d\nASME J. Eng. Gas Turbines Power\n,\n141\n(\n9\n), p.\n091009\n.\n11.\nSayama\n,\nS.\n,\nKinoshita\n,\nM.\n,\nMandokoro\n,\nY.\n, and\nFuyuto\n,\nT.\n,\n2019\n, \u201c\nSpark Ignition and Early Flame Development of Lean Mixtures Under High-Velocity Flow Conditions: An Experimental Study\n,\u201d\nInt. J. Engine Res.\n,\n20\n(\n2\n), pp.\n236\n246\n.\n12.\nMaly\n,\nR.\n, and\nVoge\n,\nM.\n,\n1979\n, \u201c\nInitiation and Propagation of Flame Fronts in Lean CH4-Air Mixtures by the Three Modes of the Ignition Spark\n,\u201d\nProc. Combust. Inst.\n,\n17\n(\n1\n), pp.\n821\n831\n.\n13.\nWolk\n,\nB.\n, and\nEkoto\n,\nI.\n,\n2017\n, \u201c\nCalorimetry and Imaging of Plasma Produced by a Pulsed Nanosecond Discharge Igniter in EGR Gases at Engine-Relevant Densities\n,\u201d\nSAE Int. J. Engine\n,\n10\n(\n3\n), pp.\n970\n983\n.\n14.\nBiswas\n,\nS.\n,\nEkoto\n,\nI.\n, and\nScarcelli\n,\nR.\n,\n2018\n, \u201c\nTransient Plasma Ignition (TPI) for Automotive Applications\n,\u201d\n4th International Conference on Ignition Systems for Gasoline Engines\n,\nBerlin, Germany\n,\nDec. 6\u20137\n.\n15.\nConvergent Science Inc.\n,\n2019\n,\nCONVERGE v3.0 Manual\n,\nConvergent Science Inc.\n,\nMadison, WI\n.\n16.\nPomraning\n,\nE.\n,\n2000\n, \u201c\nDevelopment of Large Eddy Simulation Turbulence Models\n,\u201d\nPh.D. dissertation\n,\nUniversity of Wisconsin-Madison\n,\nMadison, WI\n.\n17.\nRaju\n,\nM.\n,\nWang\n,\nM.\n,\nDai\n,\nM.\n,\nPiggott\n,\nW.\n, and\nFlowers\n,\nD.\n,\n2012\n, \u201c\nAcceleration of Detailed Chemical Kinetics Using Multi-Zone Modeling for CFD in Internal Combustion Engine Simulations\n,\u201d\nSAE Technical Paper, Detroit, MI, Apr. 24\u201326, 2012, SAE Paper No. 2012-01-0135\n.\n18.\nSmith\n,\nG. P.\n,\nGolden\n,\nD. M.\n,\nFrenklach\n,\nM.\n,\nMoriarty\n,\nN. W.\n,\nEiteneer\n,\nB.\n,\nGoldenberge\n,\nM.\n,\nBowman\n,\nC. T.\n,\nHanson\n,\nR. K.\n,\nSong\n,\nS.\n,\nGardiner\n, Jr.,\nW. C.\n,\nLissianski\n,\nV. V.\n, and\nQin\n,\nZ.\n,\n1999\n, http:\/\/combustion.berkeley.edu\/gri-mech\/\n19.\nWerner\n,\nH.\n, and\nWengle\n,\nH.\n,\n1991\n, \u201c\nLarge Eddy Simulation of Turbulent Flow Over and Around a Cube in a Plane Channel\n,\u201d\nProceedings of the Eighth Symposium on Turbulent Shear Flows\n,\nMunich, Germany\n,\nSept. 9\u201311\n.\n20.\nEsgee Technologies Inc.\n,\n2019\n,\nVizSpark v2.3.0 User Manual: A Multi-Dimensional Simulator for Thermal Plasma Systems\n,\nEsgee Technologies Inc.\n,\nAustin, TX\n.\n21.\nKarpatne\n,\nA.\n,\nBreden\n,\nD.\n, and\nRaja\n,\nL.\n,\n2018\n, \u201c\nSimulation of Arc Quenching in Hermetically Sealed Electric Vehicle Relays\n,\u201d\nSAE Int. J. Passeng. Cars\u2014Electron. Electr. Syst.\n,\n11\n(\n3\n), pp.\n149\n157\n.\n22.\nCastela\n,\nM.\n,\n2016\n, \u201c\nDirect Numerical Simulations of Plasma-Assisted Ignition in Quiescent and Turbulent Flow Conditions\n,\u201d\nPh. D. dissertation\n,\nUniversite Paris-Saclay\n,\nFrance\n.\n23.\nThermoBuild\n,\nNASA Glenn Research Center\n, https:\/\/cearun.grc.nasa.gov\/ThermoBuild\/, Accessed May 17, 2021.\n24.\nHaley\n,\nR. F.\n, and\nSmy\n,\nP. R.\n,\n1989\n, \u201c\nElectronically Induced Turbulence\u2014The Short Duration Spark\n,\u201d\nJ. Phys. D: Appl. Phys.\n,\n22\n(\n2\n), pp.\n258\n265\n.\n25.\nPitt\n,\nP. L.\n,\nClements\n,\nR. M.\n, and\nTopham\n,\nD. R.\n,\n1991\n, \u201c\nThe Early Phase of Spark Ignition\n,\u201d\nCombust. Sci. Technol.\n,\n78\n(\n4\u20136\n), pp.\n289\n314\n.","date":"2023-03-27 07:48:06","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 1, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 0, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.4617302417755127, \"perplexity\": 1740.7448811982892}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": false}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2023-14\/segments\/1679296948609.41\/warc\/CC-MAIN-20230327060940-20230327090940-00083.warc.gz\"}"} | null | null |
HISTORY.com works with a wide range of writers and editors to create accurate and informative content. All articles are regularly reviewed and updated by the HISTORY.com team. Articles with the "HISTORY.com Editors" byline have been written or edited by the HISTORY.com editors, including Amanda Onion, Missy Sullivan and Matt Mullen.
Kamala Harris becomes first female vice president
Kamala Harris makes history when she is sworn in as the 49th U.S. vice president on January 20, 2021, becoming the first woman, the first Black American and the first Asian American to occupy the office. When Harris was chosen as Joe Biden's running mate in August 2020, the ...read more
Daniel Webster (1782-1852) emerged as one of the greatest orators and most influential statesmen in the United States in the early 19th century. As an attorney, he argued several landmark cases before the Supreme Court that expanded the power of the federal government. A ...read more
William Jennings Bryan
William Jennings Bryan (1860-1925), the U.S. congressman from Nebraska, three-time presidential nominee and secretary of state, emerged near the end of the 19th century as a leading voice in the Democratic Party and the nation. A devout Protestant, his populist rhetoric and ...read more
Omar N. Bradley
Omar N. Bradley (1893-1981), was a senior U.S. Army officer who served as field commander of American soldiers during the Allied invasion of Normandy on D-Day and led Allied troops as they drove into Germany near the end of World War II. Known as the "G.I.'s General" for the ...read more
Erwin Rommel (1891-1944) was a German army officer who rose to the rank of field marshal and earned fame at home and abroad for his leadership of Germany's Afrika Korps in North Africa during World War II. Nicknamed "the Desert Fox," Rommel also commanded German defenses against ...read more
Quotes from 7 of Martin Luther King Jr.'s Most Notable Speeches
Martin Luther King Jr. was one of the most influential figures of the American civil rights movement—and a gifted orator. His stirring speeches touched on everything from social and racial justice, to nonviolence, poverty, the Vietnam War and dismantling white supremacy. And ...read more
U.S. Capitol riot
On the afternoon of January 6, 2021, a mob of President Donald Trump's supporters descend on the U.S. Capitol, attempting to interfere with the certification of electoral votes from the 2020 presidential election. The rioters assaulted the Capitol police force and ransacked the ...read more
Cleveland becomes first MLB team with numbers on back of jerseys
On April 16, 1929, the Cleveland Indians open the season with numbers on the back of each player's jersey, the first Major League Baseball team to do so. The numbers make it easier for scorekeepers, broadcasters and fans to identify players. Cleveland wins the game against the ...read more
First US cycling club formed
On February 11, 1878, the Boston Bicycle Club, the first organization for recreational cyclists, is formed. The following year, a club is formed in Buffalo, followed by a club in New York in 1880. In the ensuing decades, as middle-class participation in cycling grows, hundreds of ...read more
NHL game televised in US for first time
On February 25, 1940, the first telecast of a National Hockey League is transmitted over New York's W2XBS—the National Broadcasting Company's experimental station used to test TV technology. A viewing audience estimated at 300 subscribers watches the New York Rangers defeat the ...read more
In 2021, the United States—and the world—continued to confront the consequences of the momentous events of 2020, particularly the COVID-19 pandemic and the 2020 U.S. presidential election, in which former Vice President Joe Biden defeated the incumbent President Donald Trump. ...read more
US women win the first Winter Olympics hockey gold medal
On February 17, 1998, in Nagano, Japan, the United States defeats Canada, 3-1, to win the gold medal in the first women's hockey tournament held at the Winter Olympics. "After these Olympics, I hope the sport grows times a million," American forward Katie King says. "Anyone who ...read more
Albright becomes first female US figure skater to win world title
On February 15, 1953, Tenley Albright, a 17-year-old from Boston, becomes the first American female to win the world figure skating championship. All seven judges at the event at an outdoor rink in Davos, Switzerland give her a first-place vote. Albright, who was stricken as a ...read more
Lee Elder becomes first Black golfer to play in Masters
On April 10, 1975, 41-year-old Lee Elder becomes the first Black golfer to play in the Masters, considered the most prestigious event in the sport. Elder shoots 37 on the front and back nine for a 74 at the Augusta (Georgia) National Golf Club and trails leader Bobby Nichols by ...read more
The Assassination of Malcolm X
Civil rights leader Malcolm X took the stage at the Audubon Ballroom in the Washington Heights neighborhood of Manhattan on February 21, 1965. Just minutes later, shortly after 3 p.m., the former prominent Nation of Islam figure was gunned down by three men as his wife, Betty ...read more
Effa Manley becomes first woman elected to Baseball Hall of Fame
On February 27, 2006, baseball pioneer Effa Manley becomes the first woman elected to the Baseball Hall of Fame. Manley, who died in 1981, was co-owner of the Newark (New Jersey) Eagles, a Negro League powerhouse, and a huge advocate for Black ballplayers and civil rights causes. ...read more
NCAA adopts controversial Proposition 48
On January 13, 1986, NCAA schools vote to adopt Proposition 48, a controversial regulation that mandates minimum high school grades and scores on standardized college entrance exames for student-athletes to participate in sports as freshmen. The proposition, which passes by a ...read more
All-female team competes in America's Cup sailing for first time
On January 13, 1995, America3, an all-female sailing team, wins the first race of the America's Cup defender trials, easily beating Team Dennis Conner by a little more than a minute. The team is the sport's first all-women team to compete in the 144-year history of the America's ...read more
MLB commissioner suspends players in drug scandal
On February 28, 1986, Major League Baseball commissioner Peter Ueberroth suspends 11 players. including some of the sport's biggest names, for their involvement with illegal drugs. The suspensions are the most severe in the baseball since the infamous "Black Sox Scandal" in 1919. ...read more
PGA approves participation of Black golfers
On January 19, 1952, Professional Golfers Association president Horton Smith announces that a seven-man committee "almost unanimously" votes to allow Black golfers to compete in PGA co-sponsored events. With the announcement, Smith hopes that Black golfers participate in the next ...read more
Jason Collins, first openly gay athlete to play in NBA, makes U.S. sports history
On February 23, 2014, Brooklyn Nets center Jason Collins becomes the first openly gay athlete to play in a game in the United States' four major professional leagues. The 35-year-old journeyman plays 10 scoreless minutes, recording two rebounds and five fouls in the Nets' 108-102 ...read more
Multi-sport star Jim Thorpe signs MLB contract with Giants
On February 1, 1913, 25-year-old multi-sport star Jim Thorpe—who won two gold medals at the 1912 Olympics—signs a Major League Baseball contract with the New York Giants. The signing comes on the same day Thorpe returns his Olympic medals to Sweden for a violation of amateur ...read more
NCAA suspends SMU football program for 1987 season
On February 25, 1987, the NCAA suspends the Southern Methodist University football program for 1987 season for repeated rules violations but stops short of imposing the so-called "death penalty." Still, the sanctions are the most severe levied by the NCAA against a major college ...read more
In epic Super Bowl upset, Jets make good on Namath guarantee
On January 12, 1969, at the Orange Bowl in Miami, the New York Jets of the American Football League defeat the NFL's Baltimore Colts, 16-7, in Super Bowl III—a result considered one of the biggest upsets in sports history. Days earlier, Jets quarterback Joe Namath guaranteed a ...read more
IOC finds fraud, awards second gold in Winter Olympics skating event
On February 15, 2002, the International Olympic Committee announces it has sufficient evidence of fraud by a French judge and awards a second gold medal in pairs figure skating at the Winter Olympics in Salt Lake City, Utah. The decision comes after days of rumors and ...read more
Governing body for U.S. golf is founded
On December 22, 1894, the Amateur Golf Association of the United States—later renamed the United States Golf Association—is formed in New York to govern the sport. Five charter golf clubs join to form the association—Newport (Rhode Island) Golf Club; St. Andrews Golf Club in ...read more
Early 20th Century US
William Randolph Hearst (1863-1951) launched his career by taking charge of his father's struggling newspaper the San Francisco Examiner in 1887. By the 1930s, he had built the nation's largest media empire, including more than two dozen newspapers in major cities nationwide, ...read more
American League is founded
On January 28, 1901, professional baseball's American League is founded in Chicago. The league plans for a 140-game schedule, 14-man rosters and a players' union. Franchises are in Baltimore (Orioles), Boston (Americans), Chicago (White Stockings), Cleveland (Blues), Detroit ...read more
American becomes first non-Japanese to achieve highest rank in sumo wrestling
On January 25, 1993, American Chad Rowan becomes the first non-Japanese sumo wrestler to become a "yokozuna," the sport's highest rank. Rowan, a 23-year-old Hawaii native who stands 6-foot-8 and weighs 455 pounds, is the 64th person to hold the top rank in sumo, Japan's national ...read more
Colts win NFL title in 'Greatest Game Ever Played'
On December 28, 1958, the Baltimore Colts defeat the New York Giants, 23-17, in overtime in the NFL Championship Game—a back-and-forth thriller that later is billed as "The Greatest Game Ever Played." The nationally televised championship—the league's first overtime contest—is ...read more
MLB owners approve interleague play
On January 18, 1996, Major League Baseball owners unanimously approve interleague play for the 1997 season. The owners' vote, which called for each team to play 15 or 16 interleague games, breaks a 126-year tradition of teams only playing games within their league during the ...read more
NFL's Rams announce move to St. Louis
On January 17, 1995, the Los Angeles Rams announce they are leaving Southern California after 49 years and moving to St. Louis. The team, which reportedly lost $6 million in 1994, is lured to Missouri with a package that includes a new $260 million stadium and a $15 million ...read more
Dallas Cowboys win playoff game on 'Hail Mary' pass
On December 28, 1975, Dallas Cowboys quarterback Roger Staubach throws a 50-yard touchdown pass to Drew Pearson in the waning seconds to beat the Minnesota Vikings in a playoff game, 17-14. Afterward, Staubach calls the miraculous touchdown a "Hail Mary," thus cementing the term ...read more
First NFL playoff game is played indoors
On December 18, 1932, the Chicago Bears defeat the Portsmouth (Ohio) Spartans, 9-0, in the NFL's first playoff game—and first game played indoors. The victory gives the Bears the championship and leads to a playoff system for the first time. Because of frigid weather and ...read more
A. Philip Randolph was a labor leader and civil rights activist who founded the nation's first major Black labor union, the Brotherhood of Sleeping Car Porters (BSCP) in 1925. In the 1930s, his organizing efforts helped end both racial discrimination in defense industries and ...read more
John Marshall was the fourth chief justice of the U.S. Supreme Court (1801-35). In Marbury v. Madison (1803) and other landmark cases, Marshall asserted the Supreme Court's authority to determine the constitutionality of the nation's laws—a principle known as judicial review—and ...read more
Madison Square Garden formally opens with NHL game
On December 15, 1925, the New York Americans lose to the Montreal Canadiens, 3-1, in the formal opening of New York's Madison Square Garden, which becomes one of the world's most famous sporting venues. The game, played before 17,000 fans, is also the first NHL game played at the ...read more
First Thanksgiving college football game played
On November 30, 1876, Yale defeats Princeton, 2-0, in Hoboken, New Jersey in the first collegiate football game played on Thanksgiving. Nearly 1,000 fans attend the game, played in cold, rainy weather. "The friends of both colleges mustered in good force," the New York Times ...read more
First basketball game played
On December 21, 1891, 30-year-old James Naismith introduces the first game of basketball. Based on 13 rules created by Naismith, the game is tested by 18 students at the International Young Men's Christian Association Training School in Springfield, Massachusetts. Two teams of ...read more
Goaltender Manon Rheaume becomes first woman to play in pro hockey game
In Atlanta on December 13, 1992, Manon Rheaume becomes the first woman to play in a regular-season professional hockey game. In the Atlanta Knights' 4-1 loss to Salt Lake City, Rheaume enters at the start of the second period with the score tied at 1 in the International Hockey ...read more
Ernie Davis becomes first Black player to win Heisman Trophy
On December 6, 1961, Syracuse running back Ernie Davis becomes the first Black player to win the Heisman Trophy—college football's top individual award—beating Ohio State fullback Bob Ferguson. Earlier in day, Davis meets with President John Kennedy at the Waldorf-Astoria Hotel ...read more
First NBA game played
On November 1, 1946, the New York Knickerbockers beat the Toronto Huskies in the first NBA game, 68-66. The Knickerbockers are led by guard Leo Gottlieb, who scored 14 points in the game played before 7,090 fans at Maple Leaf Gardens in Toronto. Although they lost to the ...read more
Violet Palmer becomes first woman to officiate an NBA game
On October 30, 1997, 33-year-old Violet Palmer becomes the first woman to officiate an NBA game. Despite the watershed moment, there is little reaction from the crowd when she is announced before the tip-off of the Dallas Mavericks-Vancouver Grizzlies game. "This is a dream come ...read more
Rookie LeBron James' NBA debut impressive
On October 30, 2003, 18-year-old basketball prodigy LeBron James scores 25 points, grabs six rebounds and dishes out nine assists, but his Cleveland Cavaliers lose to the more experienced Sacramento Kings, 106-92. His debut is one of the most impressive in league history—only ...read more
Winter Olympics History
The Winter Olympics are an international sports competition held every four years. The event, also called the Winter Games, includes cold-weather events on snow (skiing, snowboarding, biathlon) and ice (figure skating, hockey, speed skating, curling, bobsled, luge, skeleton). ...read more
Chicago Bears beat Philadelphia Eagles in freaky Fog Bowl
On December 31, 1988, the Chicago Bears defeat the Philadelphia Eagles, 20-12, in a playoff game plagued by a thick fog starting late in the first half. Playing conditions at Soldier Field in Chicago become problematic, and fans in attendance and television viewers struggle to ...read more
Miami Dolphins beat Kansas City Chiefs in NFL's longest game
On December 25, 1971, Garo Yepremian boots a 37-yard field goal in the second overtime of an AFC playoff game to give the Miami Dolphins a 27-24 win over the Kansas City Chiefs in the longest game in NFL history. Game time elapsed for the Christmas contest: 82 minutes and 40 ...read more
Montreal's Jacques Plante becomes first NHL goaltender to wear facemask
On November 1, 1959, the day after Halloween, Jacques Plante of the Montreal Canadiens revolutionizes hockey by donning a facemask, the first NHL goaltender to do so in a regular-season game. Plante wears the custom-made fiberglas mask after suffering a badly cut nose and lip on ...read more
Minnesota Vikings' Jim Marshall runs wrong way with recovered fumble
On October 25, 1964, after recovering a fumble against the 49ers in San Francisco, Minnesota Vikings star defensive end Jim Marshall runs 66 yards the wrong way into his own end zone. The four-year veteran believes he has scored a touchdown, so he throws the ball out of bounds in ...read more
Chicago Cubs win first World Series title since 1908, snap 'curse'
On November 2, 2016, the Chicago Cubs win their first World Series championship since 1908, beating the Cleveland Indians, 8-7, in a thrilling Game 7 delayed by rain. "Let It Reign," reads the headline in the next day's Chicago Tribune sports section. The win snaps the "Billy ...read more | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 6,377 |
\section{Introduction}
Narrow-band surveys for Lyman-$\alpha$ (Ly$\alpha$) emitting galaxies at high
redshift have recently revealed a number of luminous (up to $5 \cdot 10^{43}$
erg s$^{-1}$), very extended (from a few times ten kpc to more than 150 kpc)
Ly$\alpha$-emitting objects, so-called Ly$\alpha$ ``blobs'' (Fynbo et al. 1999;
Keel et al. 1999; Steidel et al. 2000; Francis et al. 2001;
Matsuda et al. 2004; Palunas et al.
2004; Dey et al. 2005; Villar-Martin et al. 2005).
At least three mechanisms have been suggested as energy sources for
Ly$\alpha$ blobs. These are: \emph{i)} hidden QSOs (Haiman \& Rees 2001;
Weidinger et al. 2004, 2005), \emph{ii)} star formation and superwinds from
(possibly obscured) starburst galaxies (Taniguchi et al. 2001; Ohyama et al.
2003; Mori et al. 2004; Wilman et al. 2005), and \emph{iii)} so-called cold
accretion (Haiman, Spaans \& Quataert 2000; Fardal et al. 2001; Keres et al.
2004; Maller \& Bullock 2004; Birnboim \& Dekel 2003; Sommer-Larsen 2005;
Dijkstra et al. 2006(a,b); Dekel \& Birnboim 2006).
Cooling flows are phenomena observed in galaxy clusters for more than a decade
(Fabian 1994). These are explained by gas which is cooling
much faster than the
Hubble time through X-ray emission in the centres of the clusters.
However, cooling
emission from a galaxy, or a group sized halo can be dominated by
Ly$\alpha$ emission
(e.g. Haiman, Spaans \& Quataert 2000; Dijkstra et al. 2006(a,b)).
In this \emph{Letter} we present the discovery of a
Ly$\alpha$ blob at redshift $z \approx 3.16$ located in the GOODS South
field, which we argue is the first piece of evidence for cold gas
accretion onto a dark matter halo.
Throughout this paper, we assume a cosmology with $H_0=72$ km s$^{-1}$
Mpc$^{-1}$, $\Omega _{\rm m}=0.3$ and $\Omega _\Lambda=0.7$. All magnitudes are
in the AB system.
\section{Observations and Data reduction}
\label{obs}
A 400$\times$400 arcsec$^2$ section, centred on R.A.~$= 03^h 32^m 21.8^s$,
Dec~$= -27^{\circ} 45' 52''$ (J2000), of the GOODS South field
was observed
with FORS1 on the VLT 8.2 m telescope Antu
during two visitor mode nights on December 1--3, 2002.
A total of 16 dithered exposures were obtained over the two nights
for a combined exposure time of 30 ksec, all with the narrow band
filter OIII/3000+51 and using the standard resolution collimator
(0.2$\times$0.2
arcsec$^2$ pixels). For this setup the central wavelength of the
filter is 5055 {\AA} with a FWHM of 59 {\AA}, corresponding to the
redshift range $z = 3.126$~--~$3.174$ for Ly$\alpha$.
The observing conditions were unstable during the two nights with the
seeing FWHM varying between 0\farcs66 and 1\farcs25 on the first night
and 1\farcs4 and 3\farcs3 on the second night.
The images were reduced (de-biased, and corrected for CCD pixel-to-pixel
variations using twilight flats) using standard techniques.
The individual reduced images were combined using a modified version
of our code that optimizes the Signal-to-Noise (S/N) ratio for faint,
sky-dominated sources (see M{\o}ller \& Warren 1993, for details on this
code). The modification of the code was necessitated by the highly
variable seeing. The sky background was assumed to be constant. The
FWHM of the PSF of the final combined narrow-band image is 0\farcs8.
For object detection, we used the software package SExtractor
(Bertin \& Arnouts 1996).
A full description of our selection of Ly$\alpha$ emitters in the GOODS field
will be given in a subsequent paper. In this {\it Letter} we
discuss the nature of an extended, low surface brightness
blob with a centroid (of the Ly$\alpha$ emission) of
R.A.~$ = 03^h 32^m 14.6^s$ and Dec~$ = -27^{\circ} 43' 02$\farcs$4$ (J2000) detected in the combined narrow-band image.
Follow-up MOS spectroscopy was obtained in service mode using
FORS1/VLT UT2 over the time period December 2004 -- February 2005. The
total observing time was 6 hours.
We used a 1\farcs4 slitlet and grism 600V resulting in a wavelength range of
4650 {\AA} to 7100 {\AA} and a spectral resolution FWHM of approximately 700.
The seeing varied between 0\farcs77 and 1\farcs2 during the spectroscopic
observations.
The GOODS archival data used here and their detection limits are listed in
Table~\ref{mwtab}.
\begin{table}
\begin{center}
\caption{Specifications of deep, multi-wavelength data available in the
GOODS South field and the narrow-band image.
The last column gives the 3$\sigma$ limit as detected in a $2''$ radius
aperture and the narrow-band value gives
the blob flux in this aperture. }
\vspace{-0.5cm}
\begin{tabular}{@{}lccccccc}
\hline
Filter/Channel & $\lambda_c$ & Filter & 3$\sigma$ limit ($2''$ aperture)\\
& & FWHM & ($\mathrm{erg}$~$ \mathrm{cm}^{-2}$$\mathrm{s}^{-1}$$\mathrm{Hz}^{-1}$)\\
\hline
X-rays (\emph{Chandra}) & 4.15 keV & 3.85 keV & $9.90 \cdot 10^{-34}$ \\
U (\emph{ESO 2.2-m}) & 3630 \AA & 760 \AA & $8.62 \cdot 10^{-31}$ \\
B (\emph{HST}) & 4297 \AA & 1038 \AA & $9.25 \cdot 10^{-30}$ \\
Narrow (\emph{VLT}) & 5055 \AA & 60 \AA & $6.68 \cdot 10^{-30}$ \\
V (\emph{HST}) & 5907 \AA & 2342 \AA & $4.66 \cdot 10^{-30}$ \\
i (\emph{HST}) & 7764 \AA & 1528 \AA & $1.50 \cdot 10^{-29}$ \\
z (\emph{HST}) & 9445 \AA & 1230 \AA & $3.00 \cdot 10^{-29}$ \\
J (\emph{VLT}) & 1.25 $\mu$m & 0.6 $\mu$m & $5.31 \cdot 10^{-30}$ \\
H (\emph{VLT}) & 1.65 $\mu$m & 0.6 $\mu$m & $1.86 \cdot 10^{-29}$ \\
Ks (\emph{VLT}) & 2.16 $\mu$m & 0.6 $\mu$m & $1.56 \cdot 10^{-29}$ \\
Ch1 (\emph{Spitzer/IRAC}) & 3.58 $\mu$m & 0.75 $\mu$m & $2.51 \cdot 10^{-31}$ \\
Ch2 (\emph{Spitzer/IRAC}) & 4.50 $\mu$m & 1.02 $\mu$m & $6.43 \cdot 10^{-32}$ \\
Ch3 (\emph{Spitzer/IRAC}) & 5.80 $\mu$m & 1.43 $\mu$m & $5.01 \cdot 10^{-29}$ \\
Ch4 (\emph{Spitzer/IRAC}) & 8.00 $\mu$m & 2.91 $\mu$m & $4.65 \cdot 10^{-30}$ \\
\hline
\label{mwtab}
\end{tabular}
\end{center}
\vspace{-1.2cm}
\end{table}
\section{Results}\label{results}
The spectrum of the part of the Ly$\alpha$ blob covered by the
slitlet can be seen in the left-most panel of Fig.~\ref{spectrum}.
The line has the asymmetric profile expected for a high redshift
Ly$\alpha$ emitter.
We detect no other emission lines in the spectrum.
The most likely interloper is [OII] at redshift 0.36, but no emission
is observed in the spectrum where e.g. H$\beta$ or [OIII] are expected at this
redshift, see Fig.~\ref{spectrum}. This leads us to the
conclusion that we are observing a Ly$\alpha$-emitting object at $z = 3.157$.
The observed FWHM velocity width of the emission line is $505$~km~s$^{-1}$.
The instrument FWHM of the set-up is $290$~km~s$^{-1}$, hence
the Ly$\alpha$ intrinsic velocity width is marginally resolved. The
intrinsic width is less than $500$~km~s$^{-1}$. This is of the order or
smaller than for other published blobs,
with velocity widths of $500 - 2000$~km~s$^{-1}$ (Keel et al. 1999;
Steidel et al. 2000; Francis et al. 2001; Ohyama et al. 2003; Bower et al 2004;
Dey et al. 2005).
\begin{figure*}[!ht] \begin{center}
\epsfig{file=fig1.ps,width=18cm,clip=}
\caption{\emph{Left}~\emph{a)} Flux calibrated spectrum of the blob emission
line. The
line has the characteristic blue side absorption, indicating high redshift.
\emph{b)} The part of the spectrum (binned with a binsize equal to half
the resolution ($1.1$~{\AA})) where H$\beta$ and [OIII] should have been observed if the emission line was [OII] at a redshift of $z \approx 0.36$.
These lines are not observed and
therefore we conclude the observed line is due to Ly$\alpha$ at $z=3.16$.
\emph{Middle} Contour-plot of narrow-band emission from the Ly$\alpha$ blob
overlaid the
HST V-band image. The narrow-band image has been continuum
subtracted by subtracting the re-binned, smoothed and scaled HST/V-band image.
Contour levels are $2 \cdot 10^{-4}$, $4 \cdot 10^{-4}$ and $6
\cdot 10^{-4}$~erg~s$^{-1}$~cm$^{-2}$ in restframe flux (corresponding
to $1.2 \cdot 10^{-18}$, $2.5 \cdot 10^{-18}$ and $3.7 \cdot 10^{-18}$ in
observed flux). The image is $18'' \times 18''$ ($18''$ corresponds to a physical size of $\sim 133$~kpc).
Numbers refer to those used in section~\ref{results}. The dotted lines indicate
the slitlet position for our follow-up spectroscopy.
\emph{Right} Plot of surface brightness as function of radius.
The flux is the sky subtracted narrow-band flux.
The PSF of the image is illustrated by the solid line, and the
dotted line is the best fit model of Dijkstra et al. 2006.
The deficit at $\sim 45$~kpc is due to the asymmetric appearance of
the blob.}
\label{surfbright}
\label{spectrum}
\label{contour}
\end{center}
\vspace{-0.5cm}
\end{figure*}
A contour-plot of the blob superimposed on the HST/ACS
V-band image is shown in the middle panel of Fig.~\ref{contour} and a plot of
the surface brightness of the blob is seen in the right panel of the same
figure. The full set of thumb-nail images of the blob
in all 14 bands can be found in Fig.~\ref{thumbs}. No obvious continuum
counterpart is detected in any band. The radial
size is at least 30 kpc (60 kpc diameter) with fainter emission
extending out to 40 kpc radius. This can be seen as the extension to the SW
in the contour-plot in Fig.~\ref{contour}. The significance of the
lowest contour levels is of the order of $2\sigma$ per pixel.
The total Ly$\alpha$ luminosity, in a 30 kpc radius aperture, is
L$_{\mathrm{Ly}\alpha} = (1.09 \pm 0.07) \cdot 10^{43}$~erg~s$^{-1}$.
This coincides, after correction for the smaller area sampled in the
spectrum, to the Ly$\alpha$ flux detected in the spectrum within errors. A
conservative lower limit to
the restframe equivalent width (EW) of the emission line can be calculated from
upper limits on the broad-band fluxes in the HST B and V filters in the same
aperture.
This limit is EW~$\gtrsim 220$~{\AA} in the restframe.
This is in the range of previously published Ly$\alpha$
blobs, that have a Ly$\alpha$
flux to B-band flux density range between 50~--~1500 {\AA} in the restframe
(but typically these values are derived measuring the continuum flux in a
smaller aperture than the emission line flux).
\begin{figure*}[!ht] \begin{center}
\epsfig{file=x.ps,width=2.5cm,clip=}\epsfig{file=u.ps,width=2.5cm,clip=}\epsfig{file=b.ps,width=2.5cm,clip=}\epsfig{file=v.ps,width=2.5cm,clip=}\epsfig{file=i.ps,width=2.5cm,clip=}\epsfig{file=z.ps,width=2.5cm,clip=}\epsfig{file=n.ps,width=2.5cm,clip=}
\epsfig{file=j.ps,width=2.5cm,clip=}\epsfig{file=h.ps,width=2.5cm,clip=}\epsfig{file=k.ps,width=2.5cm,clip=}\epsfig{file=ch1.ps,width=2.5cm,clip=}\epsfig{file=ch2.ps,width=2.5cm,clip=}\epsfig{file=ch3.ps,width=2.5cm,clip=}\epsfig{file=ch4.ps,width=2.5cm,clip=}
\caption{Thumbnail images of all available multi-wavelength data in the GOODS South field, centred on the Ly$\alpha$ blob. All images are $18'' \times 18''$.}
\label{thumbs}
\end{center}
\vspace{-0.7cm}
\end{figure*}
There are seven objects detected in a wide range ($\ge 8$) of energy bands,
within a $10''$~radius
surrounding the blob. 8 other objects are detected within
the V-band and one further detected in the \emph{Spitzer/IRAC}
bands. The photometric redshift of these objects was calculated using
the public \emph{Hyperz}\footnote{http://webast.ast.obs-mip.fr/hyperz/}
code by Bolzonella et al. (2000). The resulting photometric redshifts for the
eight objects with most data points ($\ge 8$)
can be found in Table~\ref{photoz}. The other eight objects detected in the
V-band have only a few detections across the spectrum and hence their
photometric redshifts are unreliable.
The redshift of object \#~3 is similar to the blob redshift and indicates
that this galaxy may be near to the blob. Object \#~6 is
an intriguing object, undetected in the deep optical and near-IR imaging
but bright in the \emph{Spitzer/IRAC} bands. Its photometric redshift is
consistent with the redshift of the blob, but with a large uncertainty.
The object is also detected in the \emph{Spitzer/MIPS} 24~$\mu$m band.
Based on the Spitzer magnitudes, and on the diagnostic colour-colour
diagram of Ivison et al. (2004), object \#~6 is best fit by a star-burst
at high redshift ($z \sim 5.5$, consistent with the photometric redshift
estimate), hence unrelated to the blob.
\begin{table}
\begin{center}
\caption{Photometric redshifts of objects surrounding the blob. Numbering
refers to those given in Fig.~\ref{contour}. Errors given are 1~$\sigma$.
}
\vspace{-0.5cm}
\begin{tabular}{@{}lcccccc}
\hline
Obj \# & Dist. from blob & $z_{phot}$ & $\chi^2/d.o.f$ & Type & A$_V$ rest \\
& (arcsec) & & & & \\
\hline
1 & 4.6 & 1.1$^{+0.40}_{-0.30}$ & 1.3 & Burst & 0.20 \\
2 & 4.6 & 1.1$^{+0.34}_{-0.41}$ & 8.6 & Burst & 1.20 \\
3 & 6.8 & 2.9$^{+1.41}_{-0.59}$ & 4.8 & Spiral & 1.20 \\
4 & 8.4 & 0.6$^{+1.97}_{-0.63}$ & 8.1 & Burst & 0.40 \\
5 & 8.7 & 0.9$^{+2.46}_{-0.89}$ & 2.0 & Burst & 0.20 \\
6 & 3.0 & 4.5$^{+4.29}_{-1.54}$ & 0.9 & Spiral & 1.20 \\
7 & 6.3 & 1.1$^{+1.24}_{-0.81}$ & 1.9 & Burst & 1.20 \\
8 & 4.5 & 3.5$^{+1.27}_{-3.48}$ & 0.6 & Spiral & 0.00 \\
\hline
\label{photoz}
\end{tabular}
\end{center}
\vspace{-1.2cm}
\end{table}
\section{Discussion}
We first consider that the Ly$\alpha$ emission of the blob may be due to
recombination of gas, photo-ionized by an AGN or a starburst galaxy.
If the blob is a ``passive'' gas cloud
illuminated and photo-ionized by a nearby AGN, then, following Cantalupo
et al. (2005), one can show that for an AGN with luminosity L$_\nu=
\mathrm{L}_{\rm LL}(\nu/\nu_{\rm LL})^{-\alpha}$, and in order to result
in a
peak blob Ly$\alpha$ surface brightness of $\Sigma_{Ly\alpha}$, the AGN has to
be located at a distance of no more than
\[
270\, \mathrm{kpc} \left(\frac{\Sigma_{Ly\alpha}}{10^{-3}\, \mathrm{erg}\,
\mathrm{s}^{-1}
\mathrm{cm}^{-2}}\right)^{-1}
\sqrt{\frac{L_{\rm LL}}{10^{30}\, \mathrm{erg}\, \mathrm{s}^{-1}
\mathrm{Hz}^{-1}}}
\sqrt{\frac{0.7}{\alpha}},
\]
where equality applies to the case where the blob gas is optically
thick at the Lyman limit. No AGN has been detected in the deep \emph{Chandra}
image available
within this distance. Worsley et al. (2005) argue that a significant
fraction
of the unresolved X-ray background in hard bands consists of highly obscured
AGN. However, Worsley et al. (2005) also predict
that the AGN responsible for this background are situated at $z \sim 0.5-1.0$.
Furthermore, Silverman et al. (2005) present a luminosity function for AGN
at higher redshift. To the detection limit of the CDFS (L$_X (z = 3.15) \approx
1.9 \cdot 10^{43}$~erg~s$^{-1}$) and with our search volume
($3 \times 3$~Mpc~$\times \Delta z = 0.05$) we expect to detect only 0.06 AGN
in our entire search volume.
We also consider the possibility that galaxy
\#~3 can photo-ionise the blob. However, if we assume a power law for the
spectrum and extrapolating from the HST/B and HST/V detections we find that
the UV luminosity of galaxy \#~3 is not sufficient to photo-ionise the blob,
unless highly collimated towards the blob. We have no reason to believe that
this is the case.
The second possibility is that the blob Ly$\alpha$ emission
is somehow related to starburst driven, superwind outflows. A starburst would
be expected to be located within the blob to create such a
Ly$\alpha$
halo and no central continuum source has been detected. Even though a
very massive starburst can be made invisible in the
UV/optical range by dust obscuration, it should be visible in the IR, i.e. the
\emph{Spitzer/IRAC} bands.
The third option is that the Ly$\alpha$ emission is due to cold accretion of
predominantly neutral, filamentary gas onto a massive dark matter halo.
For cold accretion, the bulk of the Ly$\alpha$ emission is produced
by collisional
excitation, rather than recombination. Recently, Dijkstra et al.
2006(a,b) presented a theoretical model for Ly$\alpha$ cooling flows, along
with predictions of the emission line profile and the shape of the surface
brightness function. The S/N of our spectrum is not high enough to allow a
comparison of emission line profiles. However, the surface
brightness profile matches well the predictions for a centrally illuminated
, collapsing
cloud of Dijkstra et al. 2006(a), see Fig.~1. Further tests are needed
to determine how well their model fits.
To test whether this blob can be filamentary gas accreting
``cold'' onto a companion galaxy, we also conducted the following
experiment:
we calculated the Ly$\alpha$ surface brightness
in a 100$\times$100 kpc (projected) region for a proto-galaxy
of ``cooling'' radiation only (so all contributions from regions with young
stars were removed, as well as all emission, in general, from gas closer
than 10 kpc to any star-forming region). The calculation was based
on a cosmological simulation of the formation and evolution of an
M31-like disk galaxy (Sommer-Larsen 2005; Portinari \& Sommer-Larsen
2005).
The results at $z\sim3$ are presented in Sommer-Larsen (2005),
and get to
a surface brightness about an order of magnitude lower than the observed level.
This is interesting, and may point to a cold accretion origin
of the blob Ly$\alpha$ emission on a larger scale, such as
filamentary gas accretion onto a galaxy-group sized halo.
Another possibility is that the periods with high surface brightness
are shorter than 2.5 Myr (the resolution of the simulation). Given that in a
search volume of about
40000 co-moving Mpc$^3$, only one such blob has been detected, it is
actually comforting, that we could not reproduce the blob
characteristics, by cold accretion onto this, randomly selected, M31-like
galaxy. This has to be a rare phenomenon.
A test for the cold accretion model would be to observe the Balmer lines.
For collisionally excited hydrogen, neglecting extinction
effects, the flux in H$\alpha$ should only be about 3.5 percent of the
Ly$\alpha$ flux, whereas for recombining, photo-ionized gas this ratio is $\sim
11.5$~\% (Brocklehurst 1971). Hence, the relative H$\alpha$ luminosity is
expected to be significantly larger in the latter case. The situation is
similar for H$\beta$, and whereas the H$\alpha$ line will be very difficult to
detect from the ground, H$\beta$ should be observable.
\section{Conclusion}
We have here reported the results of an extensive multi-wavelength
investigation of a redshift $z = 3.16$ Ly$\alpha$~emitting blob discovered in
the GOODS South field. The blob has a diameter larger than 60~kpc
diameter and a total luminosity of $\mathrm{L}_{\mathrm{Ly}\alpha} \sim
10^{43}$~erg~s$^{-1}$. Deep HST imaging show no obvious optical counterpart,
and the lack of X-ray or IR emission suggest there are no AGN or dusty
starburst components associated with at least the centroid of the blob.
Two galaxies within a $10''$ radius have photometric redshifts consistent
with the redshift of the blob, but follow-up spectroscopy is needed to
establish if there is a connection.
We have run simulations of Ly$\alpha$ surface brightness arising from
cold accretion and found that such extended Ly$\alpha$ emission may be
explained by accretion
along a filament onto a galaxy group sized dark matter halo. Another
possibility is that such emission in very short lived, i.e. significantly
shorter than the 2.5 Myr resolution of our simulation. We argue that other
previously suggested origins of Ly$\alpha$ blobs (hidden AGN and
``super-winds'')
can be ruled out in this case due to the lack of detected continuum
counter-parts. Hence,
though our cold accretion simulation cannot perfectly match our data, it is the
only explanation that is plausible. Our results combined
with
the fact that previously studied blobs appear to be caused by superwinds
and/or AGN in turn implies that the energy sources for blob Ly$\alpha$
emission are diverse.
\vspace{-0.1cm}
\begin{acknowledgements}
KN gratefully acknowledges support from IDA~--~Instrumentcentre for Danish
Astrophysics. The numerical simulations used in this paper were performed on
the facilities provided by Danish Center for Scientific Computing (DCSC). The
Dark Cosmology Centre is funded by the DNRF. The authors thank the DDT panel
and the ESO Director General for granting time for follow-up spectroscopy.
\end{acknowledgements}
\vspace{-0.7cm}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 1,897 |
\section{Introduction} \label{sec:intro}
Coronal mass ejections (CMEs) are the most violent eruptive magnetic activities in the solar atmosphere. They often appear as stellar-sized magnetized plasma bubbles in white-light coronagraph observations, and can rapidly release a vast amount of mass and energy into the inner heliosphere making our near-earth space a hazardous place \citep[e.g.,][]{2011LRSP....8....1C,2013AdSpR..51.1967S}. To date, almost all CME theories suggested that the core of CMEs correspond to rapidly erupting magnetic flux ropes (MFRs) \citep[e.g.,][]{2000JGR...10523153F,2000JGR...105.2375L}. As the desirable progenitor of CME flux ropes, an MFR is defined as a set of highly coherent helical magnetic field lines winding around one common axis. In such a non-potential topology, a mature MFR often appears with high storage of magnetic free energy and helicity, and is prone to suffer ideal MHD instabilities. Hence, the present of a mature MFR prior to or even during CME initiations is always given great attention.
\par
In the past two decades, a lot of effort has been made in numerical simulations and observations on the formation and destabilization of active-region MFRs \citep[see reviews i.e.][]{2017ScChD..60.1383C,2019RSPTA.37780094G,2020SSRv..216..131P}.
One important reason is that active-region MFR proxies often repeatedly form in newly-emerged bipolar active regions (ARs) and their eruptions can result in recurrent CMEs and flares in rapid succession \citep[e.g.,][]{2002ApJ...566L.117Z,2012A&A...539A...7L,2017ApJ...844..141L}.
In accord with filament channel formations, it is widely believed that MFRs can be created along PILs in two ways: (1) the directly emergence of a mature MFR from the below photosphere \citep[e.g.,][]{2004ApJ...609.1123F,2004ApJ...610..588M,2008ApJ...673L.215O} or (2) successive reconnection of coronal fields induced by the continuously photospheric shearing or converging flows \citep[e.g.,][]{2000ApJ...529L..49A,2010ApJ...708..314A,2013ApJ...779..157G}. Due to differences in the spatial height of reconnection, the second one may respectively manifest as tether-cutting reconnection just at the onset of the eruption \citep[e.g.,][]{2010ApJ...725L..84L,2018ApJ...869...78C,2019ApJ...887..118C} or
slow flux cancellation in the photosphere long prior to the eruption \citep[e.g.,][]{2011A&A...526A...2G,2012ApJ...759..105S,2016ApJ...816...41Y}. In particular, apart from appearing in the formation phase of MFRs and filaments, flux cancelation is also thought as a popular triggering mechanism for CMEs and coronal jets \citep[e.g.,][]{2001ApJ...548L..99Z,2001ApJ...554..474Z,2011A&A...526A...2G,2018ApJ...869...78C,2018ApJ...864...68S}.
\par
In a bipolar configuration, the ideal MHD models suggest that the sudden onset of MFR eruptions is directly triggered by the ideal torus instability \citep{2004A&A...413L..27T,2010ApJ...718..433O}, ideal kink instability \citep{1979SoPh...64..303H,2006PhRvL..96y5002K}, and also catastrophe mechanisms \citep{1993ApJ...417..368I,2000JGR...105.2375L}. Instead, the non-MHD models emphasize that MFR eruptions tend to take place if pre-flare reconnection below MFRs reducing their overlying magnetic constraints \citep[e.g.;][]{2000ApJ...545..524C,2001ApJ...552..833M,2013ApJ...771L..30J,2014ApJ...797L..15C,2018ApJ...869...78C}.
However, real source regions of many large-scale CMEs often cover more broad spatial ranges, connecting two inter-coupled ARs, or even a cluster of ARs \citep[e.g.,][]{2006SoPh..239..257W,2006A&A...445.1133Z,2007SoPh..244...75W,2015SCPMA..58.5682W,2007SoPh..241..329Z}. Accordingly, their CME progenitors are more likely associated with larger-scale pre-eruption structures, such as transequatorial magnetic loops, inter-connecting filaments among ARs, and extended bipolar PILs \citep[e.g.,][]{2006A&A...445.1133Z,2019ApJ...873...23Z,2007SoPh..244...75W,2013ApJ...764...91S}.
Under these multipolar complex magnetic systems, CME initiations and eruptive dynamics are no an alone behavior of a bipolar AR anymore, instead, it may tied to magnetic coupling and instability in the whole magnetic flux system \citep{1999ApJ...510..485A,2007SoPh..244...75W,2015SCPMA..58.5682W,2008AnGeo..26.3077V,2013ApJ...773...93S,2020ApJ...905..150Z}.
\par
In addition, multipolar complex magnetic systems often breed frequent sympathetic eruptive events \citep{2003ApJ...588.1176M,2008ApJ...677..699J,2011ApJ...738..179J,2011ApJ...739L..63T,2011JGRA..116.4108S}. Under such a common dome of magnetic flux system, a preceding filament eruption can easily weaken the magnetic constraint overlying other filaments and thus trigger sympathetic CMEs \citep[e.g.,][]{2012ApJ...745....9Y,2015ApJ...803...68Y,2012ApJ...750...12S,2016ApJ...820..126J,2020A&A...640A.101H}. In particular, several observations reported that in helmet-streamer configurations, coronal jets can also drive streamer-puff \citep{2005ApJ...635L.189B,2007ApJ...661..543M,2016ApJ...822L..23P} or narrow CMEs \citep[e.g.,][]{2002ApJ...575..542W,2009SoPh..259...87N,2011ApJ...738L..20H,2020ApJ...901...94J}, supporting a physically link exist between these two types of eruptive phenomena. Recently, \citet{2016ApJ...822L..23P} found that a series of coronal jets occurring at the edge of the AR 12192 resulted in homogenous bubble-shaped CMEs. Based on the the release of magnetic twist from jet-base field into large-scale coronal loops and related coronal dimming during the jet eruptions, they infer that the jet-guiding coronal loops eventually blowed out as low-speed CME bubbles due to increasing twist.
\par
In this paper, we study the initiation of an Earth-directed CME from a multipolar complex magnetic system, in which a large-scale CME flux rope event is found to arise from an unwinding coronal jet via magnetic coupling. With multi-wavelength and stereoscopic observations from Solar Dynamics Observatory \citep[SDO,][]{2012SoPh..275....3P} and \textit{Solar Terrestrial Relations Observatories} (STEREO), this work presents a direct observation for the creation of a large-scale MFR by rapid twist transform in unwinding jet spire, and sheds some light on the complex magnetic coupling in large-scale CME initiations. The layout of the remaining paper is as follows: Section 2 gives the data and methods; Section 3 presents the observational analysis and results; Section 4 presents the conclusion and discussion.
\par
\section{Data and Methods} \label{sec:Data}
Multi-wavelength EUV imaging data obtained from Atmospheric Imaging Assembly (AIA) \citep{2012SoPh..275...17L} on board SDO and H$\alpha$ center images from the Global Oscillation Network Group (GONG) are used to study CME source activities. Line-of-sight (LOS) magnetograms from Helioseismic and Magnetic Imager (HMI) \citep{2012SoPh..275..207S} are applied to study the photospheric magnetic evolution in source region. Meanwhile, observations from the \textit{Reuven Ramaty High Energy Solar Spectroscopic Imager} \citep[RHESSI;][]{2002SoPh..210....3L} and the Nancay radio heliograph \citep[NRH;][]{1997LNP...483..192K} are also used. The solar rotation of all these solar disk imaging data was removed through registering to a proper reference moment (09:30 UT).
In addition, the CME and associated phenomena are detected from the inner to the outer corona with the combination observations from the Large Angle and Spectrometric Coronagraph \citep[LASCO;][]{1995SoPh..162..357B} on board the \textit{Solar and Heliospheric Observatory} and Extreme Ultraviolet Imager \citep[EUVI;][]{2004SPIE.5171..111W} on board STEREO-A and B.
\par
Based on the Active Region Patches \citep[SHARPs;][]{2014SoPh..289.3549B} vector field products of AR 11515, the photospheric vertical electric current ($J_{z}$) in source region is computed from the direct observed horizontal field according to the Ampere's law: $\vec{J}_{z}=\frac{1}{\mu_{0}}(\bigtriangledown \times \vec{B})_{z}$. To reveal the pre-eruption magnetic field configuration, we ``pre-processed" CEA vector magnetograms to best suit the force-free condition, and then used them as the photospheric boundary to conduct a series of NLFFF magnetic field modelling with weighted optimization method \citep{2000ApJ...540.1150W,2004SoPh..219...87W}.
\par
With these NLFFF and PF magnetic extrapolation modelings, coronal magnetic free energy ($E_f$) can be computed within a certain volume $V$ as follow: $E_{f}=\int_{V} \frac{B_{N}^2}{8\pi} dV - \int_{V} \frac{B_{P}^2}{8\pi} dV$,
where the subscripts $N$($P$) denotes NLFFF(PF) extrapolated magnetic field \citep{2012ApJ...748...77S}.
In addition, newly-injected magnetic energy from the below photosphere is also estimated from the energy (Poynting) flux that cross the photospheric surface as follow \citep{2012ApJ...761..105L}:
$\frac{d E_{in}}{dt}=\frac{1}{4\pi}\int_{S} B^{2}_{t}{V_{\bot n}} dS-\frac{1}{4\pi}\int_{S} (\textbf{\emph{B}}_{t}\cdot{\textbf{\emph{V}}_{\bot t}})B_{n} dS,$
where the emerge term (first term) comes from the emergence of magnetic tubes from the solar interior, and the shear term (second term) is generated by shear motions on the solar surface.
\section{Observational Results} \label{sec:Obs}
\begin{figure}[htb!]
\centerline{\includegraphics[width=0.8\textwidth,clip=]{01.eps}
}
\caption{(a) The temporal change of GOES flux in 1-8 \AA \ and 0.5-4 \AA , and the associated AIA flux that computed in panel (d). The numbers orderly mark four preceding emission enhancements before the major flare, which correspond to four recurrent jets before the main eruption. (b) The overview of the major solar eruption (SOL2012-07-02T10:48). (c) Two related CMEs, in which CME2 is closely related to our study. (d) and (e): The overall magnetic environment of this event. Their FOVs are identical and denoted by the white box in panel (b). The related field lines, including ``L1", ``L2", and ``L3", are well traced by PFSS technique, and the former two can be observed from AIA 211 \AA \ image; the green elliptical dashed line denotes ``L3", corresponding to a latter mentioned fan-spine configuration. The yellow box in panel (d) denotes the field of view (FOV) of \nfig{fig3}. Other features of interest are marked in each panel; see the text for details.}
\label{fig1}
\end{figure}
\par
The multipolar flux magnetic system that bred the solar eruptive event of our interest consists of two inter-coupled ARs (NOAA 11514 and 11515) (see \nfig{fig1}(b)). As shown in \nfig{fig1}(a), the major flare (SOL2012-07-02T10:48) belongs to a short-duration one, which started, peaked, and rapidly ended at around 10:43, 10:48, and 10:57 UT, respectively. Previously, \citet{2014A&A...562A.110L} studied the sunspot splitting of AR 11515 and emphasized its role in triggering the major flare. Different from their study, here we mainly investigate CME initiations and related magnetic coupling processes. Note that this solar eruption simultaneously resulted in two successive Earth-directed CMEs (see \nfig{fig1}(c1-c2)), but the second is the one of our interest since it is closely related to the formation and eruption of a large-scale CME flux rope.
\par
From 211 \AA \ observation and PF extrapolations in \nfig{fig1} (d) and (e), it can be seen that this multipolar flux system at least consisted of three groups of opposite-polarity magnetic fluxes: $P1$-$N1$, $P2$-$N2$, and $P3$-$N3$. As two bundles of high-lying coronal loops, $L1$ connected $N2$ and $P2$, while $L2$ connected $P1/P2$ and remote negative background magnetic fluxes; Instead, $L3$ connected $P1$ to $N3$ and disperse background fields. At the adjacent footpoints of $L1$ and $L2$, a small solar filament (F1) a length of $\sim$ 45 Mm resided in a newly emerged magnetic system ($P1$-$N1$).
\par
Via NLFFF modeling, the pre-eruption magnetic configuration above the source region are better presented in \nfig{fig2}(a) and (b). In 3D perspectives, one can notice that there exist two privileged positions for the occurrence of magnetic reconnection: an opposite-polarity interface in POS1 and a null point in POS2, according to 3D reconnection theory \citep{1994A&A...285.1023D}. POS1 corresponds to the interface between $L1$ and the emerging magnetic system $P1$-$N1$. Within this system, the filament structure of F1 is mainly constrained by a bundles of arch loops (yellow lines) that connects $P1$ and $N1$. Accordingly, the maximum twist number of F1 is derived as $-0.88$ turns from the NLFFF extrapolation by the equation: $T_{w} = \frac{1}{4\pi} \int_{L} \alpha dl $ \citep{2006JPhA...39.8321B}. Meanwhile, near the north section of the filament structure, there exists a null-point-type configuration (blue lines, $L3$), which arched its outer spine to the remote negative-polarity flux $N3$ and partially restrained the filament structure with its inner spine lines.
\par
\begin{figure}[htb!]
\centerline{\includegraphics[width=0.6\textwidth,clip=]{02.eps}
}
\caption{(a) and (b): Selected field lines from the NLFFF extrapolation seen from a close-up 3D view, and its background is the B$_z$ map at Z = 0. (c): AIA 193 \AA \ running-difference image presents the jet-induced coronal loop expansion. The red contour traces the recurrent jets that occurred at the footpoint of $L1$. (d) The high temperature composite image of coronal jet, which blends 94, 193 and 335 \AA \ images. (e) AIA 304 \AA \ image observed during recurrent jets. The FOV of (d) and (e) are denoted by the white dash box in panel (c). (f) AIA 304 \AA \ space-time plot along the green S1 in panel (e), in which the GOES 1.0-8.0 \AA \ flux is also plotted for comparison. Features of interest are marked in each panel; see the text for details. An animation of panels (c), (d), (e), and (f) is available. The animation covers 06:30 UT to 12:30 UT with 60 s cadence. The video duration is 14 s.}
\label{fig2}
\end{figure}
Before the occurrence of the major eruption, a so-called coronal geyser site \citep{2020ApJ...891..149P} formed the interface at the POS1. As shown in \nfig{fig2} (c-e) and its animation, four recurrent coronal jets successively took place there with similar narrow jet spires and evident plasma heating signals (especially the HXR emission and chromospheric flaring patch in \nfig{fig2}(d) and (e)). Such analogous dynamics features suggest that these recurrent jets are homologous ones \citep[e.g.;][]{2015ApJ...815...71C,2018ApJ...852...10L,2019ApJ...887..154L,2020ApJ...891..149P}.
\par
On the photosphere, as shown in \nfig{fig3}, the ongoing emerged $N1$ kelp converging toward $P2$ forming an emergence-driven canceling interface between $N1$ and $P2$, which is spatially coincide with POS1. During 00:00 to 18:00 UT, $N1$ successively increased its flux from $1.9\times 10^{22}$ to $3.2\times 10^{22}$ Mx, while the positive flux demonstrated a weak decrease from 06:00 to 12:00 UT (see \nfig{fig3}(d1).
Moreover, along this canceling interface, enhanced vertical current flux density $J_z$ concurrently built up with an elongated pattern (see \nfig{fig3}(a3)). The integrated curves of its unsigned vertical current, $I_{out}$ and $I_{in}$, both demonstrate an increasing trend during this 18 hrs (see \nfig{fig3}(c2)). As responses of recurrent jets, episodes of analogous flaring patches took place along the localized $J_z$ enhancement. These results support that recurrent jets were triggered by repetitive reconnection at POS1 \citep[e.g.,][]{2010A&A...512L...2A,2013A&A...555A..19G}.
\par
\begin{figure}[htb!]
\centerline{\includegraphics[width=0.6\textwidth,clip=]{03.eps}
}
\caption{Close-up view snapshots of the evolving source region (in the yellow dashed box in \nfig{fig1} (d)). (a1-a2): HMI $B_z$. (a3): Vertical current computed at the photosphere. (b1-b3) AIA 335\AA \ , GONG H$_{\alpha}$, and UV 1600 \AA \ observations. (c1): The unsigned magnetic fluxes computed in the source region. (c2): The profile of integral unsigned current $I_{z out}$ and $I_{z in}$ in the cyan dotted box of panel (a3). (d1): Computed magnetic free energy (\textit{E$_f$}) in the source region with time. (d2): Poynting flux cross the photospheric surface in the source region with time. Features of interest are marked in each panel; see the text for details.}
\label{fig3}
\end{figure}
\par
With the occurrence of recurrent jets, the highly-lying $L1$ underwent sudden expansions at the same time (see \nfig{fig2}(a) and animation), implying a sudden reconstruction of magnetic topology. More importantly, soon after the appearance of the 4th jet, F1 suddenly erupt from coronal geyser site at around 10:45 UT. To better describe this process, a space-time plot is made along the green slice 1 in \nfig{fig2}(e), which both goes through the southeast end of F1 and the ejecting trajectory of coronal jets. The result is presented in \nfig{fig2}(f) with the comparison of \textit{GOES} flux curve. From which, one can intuitively see that before its final eruption: (1) four preceding recurrent jets correspond to the four peaks in the GOES flux curves and source-region AIA fluxes (also see \nfig{fig1}(a)); (2) with the occurrence of recurrent jets, F1 displayed a quasi-static ascent. This indicates that such preceding flux peaks and recurrent jets might be considered as precursors for the imminent main eruption.
\par
As shown in in \nfig{fig3}(d1), during the occurrence of recurrent jets and the main eruption, the time profile of magnetic free energy in source region underwent three obvious drops. Thereinto, the former two respectively released $1.5 \times 10^{31}$ and $1.3 \times 10^{31}$ erg of free energy, while the last drop that includes the main eruption released $2.2 \times 10^{31}$ erg of free energy. Meanwhile, mainly due to the emergence term of poynting flux, around $4.5 \times 10^{31}$ erg of magnetic energy was injected upward from the below photospheric surface and refilled such free energy decrease (see \nfig{fig3}(d2)).
\par
\begin{figure}[htb!]
\centerline{\includegraphics[width=0.7\textwidth,clip=]{04.eps}
}
\caption{(a1-a3): The ascent and disintegration of the filament observed in AIA 1600 \AA \ images. (b1-b3) and (c1-c3): AIA 211 \AA \ images and running-difference 304 \AA \ images. (a4) and (b4): Selected AIA 94 \AA \ images illustrate post-flare loops that observed after the filament eruption. The first(second) post-flare loops formed above(outside) the source region. (c4): A newborn flare ribbon observed in base-difference 304 \AA \ image. (d1) and (d2): space-time plots made along S2 and S3 in panel (c3). The white dash box in panel (b1) denotes the FOV of panels in the top row. Features of interest are marked in each panel; see the text for details. An animation of panels (b3), (b4), and (c1) is available. The animation covers 10:30 UT to 12:29 UT with 60 s cadence. The video duration is 5 s.}
\label{fig4}
\end{figure}
\par
Via interacting with its overlying NP configuration (POS2, see \nfig{fig2}(b)), the erupting F1 soon underwent a rapid disintegration and triggered the M5.6-class flare (see \nfig{fig4}(a1)-(a3)). As a result, typical signals of null-point reconnection, including a quasi-circular flare ribbon and a remote brightenings, were detected around the main flare ribbon \citep[e.g.;][]{2009ApJ...700..559M,2016ApJ...827...27Z,2018Ap&SS.363...26L,2019ApJ...886L..34L}. As the filament disintegration proceeded, the north (also positive-polarity) feet of F1 remained line-tied at its original position, but its south feet appeared as an ``open'' jet spire. Accordingly, the filament plasma rapidly ejected towards the southeast possibly along $L1$ and also displayed a conspicuous anti-clockwise unwinding motion (see \nfig{fig4}(b1)-(b3) and (c1)-(c3)). Compared with those preceding narrow and collimated jets, this unwinding jet can be characterized as a so-called blowout jet due to its broader jet spire \citep{2010ApJ...720..757M}.
\par
Assuming the main axis of the unwinding jet spire can be regarded as a circular cylinder and its fine threads rotates rigidly, a rough twist calculation of such unwinding jet spire can be given \citep[e.g.;][]{2011ApJ...735L..43S,2012RAA....12..573C,2013RAA....13..253H,2015ApJ...814L..13L,2018ApJ...852...10L,2019FrASS...6...44L}.
In \nfig{fig4} (d1) and (d2), two space-time plots are made perpendicular to its jet spire (along the red slices S2 and S2 in \nfig{fig4}(c3)). one can see that rolling motions nearly perpendicular to the jet spire display as dark/bright strips, which last for 42 and 50 mins in (d1) and (d2), respectively. By tracing and conducting linear fittings along these inclined dark/bright strips, the average rotational speed ($\nu_r$) of this unwinding jet can be estimated as \speed{48.5} and \speed{38.2}, respectively. The average width of the jet spire (23.09 Mm and 21.40 Mm) is equal to its diameter ($d$). Accordingly, their angular speeds ($\omega =2\nu_r /d$) along S2 and S3 are computed as $3.57\times10^{-3}$ rad s$^{-1}$ (period 1758 s) and $4.20\times10^{-3}$ rad s$^{-1}$ (period 1494 s), respectively. So, the total amount of twist released in this jet spire is roughly at 1.43$-$2.01 turns, or 2.86$-$4.02$\pi$. This result reaches an agreement with other previous twist estimation (1.2$-$4.7 turns) in comparable coronal jets \citep[e.g.,][]{2011ApJ...735L..43S,2012RAA....12..573C,2013ApJ...769..134M,2015ApJ...806...11M,2015ApJ...814L..13L,2019FrASS...6...44L}. Note that the median twist number of pre-eruption F1 is derived as 0.88 turns from the NLFFF extrapolation. A similar disagreement is also found in \citep{2018ApJ...852...10L}. We think the twist estimation by the feature tracing method is more reliable, because: (1) the feature tracing method involves less assumptions; (2) the low-lying F1 might require a sophisticated non-force-free modeling due to its small scale.
\par
\begin{figure}[htb!]
\centerline{\includegraphics[width=0.8\textwidth,clip=]{07.eps}
}
\caption{Snapshots of the global eruption observed in (a-c): H$_{\alpha}$ and AIA 304\AA \ images, (d-f): running difference AIA 211 \AA \ images, and (g-i): NRH radio imaging observation at 298 MHz. Features of interest are marked in each panel; see the text for details.}
\label{fig5}
\end{figure}
\par
Due to the unwinding dynamics, a new filament feet gradually appeared within the ``open" jet spire and demonstrated an apparent slipping motion towards the south (see \nfig{fig4}(c3)). By the time of 11:28 UT, the jet-like plasma ejecta traced out a newborn close magnetic structure. With the inspection of AIA (see animation of \nfig{fig4}) and STEREO-A/EUVI observations, we found that the newborn close magnetic structure actually corresponds to a newborn large-scale MFR. Especially near its north feet, prominent magnetic twist fields and anti-clockwise unwinding (rolling) motion indeed exist (see \nfig{fig6}(c1) and its animation). It must be pointed out that the creation of this newborn large-scale MFR accomplished at a relatively high coronal height, thus it soon erupt upwards.
\par
\nfig{fig5} further presents this eruption process with a larger FOV. In H$_\alpha$ observations, this erupting large-scale MFR manifested as an obvious erupting filament against the solar disk, and rapidly disappeared by the time of 11:25 UT (see \nfig{fig5}(a) and (b)). In running-difference 211 \AA \ images, it is found that as the major flare triggered, an EUV wave propagated toward the south, which might be the coronal imprint of the first CME (CME1) (also see \nfig{fig1}(c1)). After the major flare, the large-scale MFR arose from the coronal geysers and slowly erupt toward southwest, which eventually led to the second CME (CME2). This jet-CME coupling eruption needs to differentiate from the simple extension of coronal jets in white-light observations \citep{2002ApJ...575..542W,2011ApJ...738L..20H,2020ApJ...901...94J} and other so-called twin CME events \citep{2012ApJ...745..164S,2019ApJ...881..132D,2019ApJ...877...61M}, because a newborn large-scale MFR was indeed created during the solar eruption by the rapid twist transport from jet base to background fields.
\par
In particular, two unique features are found in the eruption of the large-scale MFR.
First, as mentioned before, the newborn south feet of the large-scale MFR showed an apparent ``drift" toward the south in 304 and 211 \AA \ imaging observations. This evolution behavior is also evidenced by simultaneous radio imaging observations at 298 MHz from NRH, which corresponds to a computed height of around 170 Mm above the photosphere based on coronal density model of \citet{1999ApJ...523..812S} and the plasma-density relationship ($f = 8.98 \times \sqrt{n_e} $).
As energetic electrons are injected into and filled its erupting volume, the feet of the newborn large-scale MFR were clearly imaged as a pair of distinct radio sources in \nfig{fig5}(c). Similar feet signatures of eruptive MFRs were also reported by other previous radio imaging observations \citep[e.g.,][]{2020FrASS...7...79C,2020ApJ...895L..50C}, but the feet radio source usually remain stationary.
In this current observation, the north one remained stationary at its spatial position, while the south one first appeared at around 10:55 UT, and demonstrated a south-orientated ``drift" during 11:00 to 11:29 UT (see \nfig{fig5}(g-i)).
\par
Second, signatures of an extra flare reconnection was detected below the erupting large-scale MFR. As shown in \nfig{fig4}(b4) and (c4), by the time of 12:10 UT, a set of new post-flare loops (2th PFLs) formed behind the erupting large-scale MFR, bridging a newly-formed chromospheric ribbon and the main flare region. Accordingly, $RHESSI$ sources was also detected at the loop-top of new post-flare loops (see \nfig{fig5}(b4)), indicating the occurrence of particle acceleration. On the whole, the creation of large-scale MFR and the appearance of the 2th PFLs provide solid evidence for this jet-CME coupling eruption. In the GOES flux, as shown in \nfig{fig1}(a), the associated X-ray emission enhancement in 2th PFLs is also clearly detected as another an C-class flare. Together with the jet-driven M5.6-class main flare, this result supports that this coupling erption event involves two-stage of flare magnetic reconnection.
\par
\begin{figure}[htb!]
\centerline{\includegraphics[width=0.8\textwidth,clip=]{05.eps}
}
\caption{STEREO observations on the newborn CME flux rope. The left part: The erupting filament in running-difference EUVI 304 \AA \ image. Close-up view snapshots of fine twisted structures at the north feet of the CME flux rope are presented in (c1-c2). The right part: The corresponding CME captured by white coronagraph COR1 and COR2. Features of interest are marked in each panel; see the text for details.}
\label{fig6}
\end{figure}
\par
With the aid of stereoscopy observations from STEREO, it becomes more evident in \nfig{fig6} that the blowout jet first appeared along the high-lying loop $L1$, and then the newborn large-scale MFR displayed as a giant prominence eruption with two feet rooted at the source region \textit{``C"} (marked by black dashed box) and a remote region \textit{``D"}, respectively. The distance between \textit{``C"} and \textit{``D"} spans more than 270 Mm.
With a close-up inspection, one can even notice that an anti-clockwise rolling motion appeared in the north leg of the erupting large-scale MFR, which indicating an obvious transfer of magnetic twist took place from the feet \textit{``C"} to the another feet \textit{``D"} (also see \nfig{fig6}(c1-c2)).
In the right part of \nfig{fig6}, the white light coronagraphs from STEREO cor1 and cor2 well recorded its related CME. CME2 first appeared at 13:22 UT and successfully propagated to more than 18 $R_s$ with a relative low velocity of around \speed{300}.
\section{Conclusion and Discussion}
CMEs and coronal jets are two types of common solar eruptive phenomena, which often independently happen at different spatial scales. The former are well-known for stellar-sized violent magnetized plasma explosions and their potential capacity in causing hazardous space weather \citep[e.g.,][]{2011LRSP....8....1C,2013AdSpR..51.1967S}, while the latter, as ubiquitous smaller-scale eruptive phenomena, are best known for their important contributions to the heating and mass supply for the upper solar atmosphere \citep[e.g.,][]{2014Sci...346A.315T,2016SSRv..201....1R,2019Sci...366..890S}. In this paper, we study the origin of a large-scale CME flux rope event arising from an unwinding coronal jet. Based on the stereoscopic and multi-bands observational analysis, we find that this whole eruptive event started with a small-scale filament within a multipolar complex magnetic system whose eruption first triggered an unwinding blowout jet and an M5.6 short-duration flare. Due to the subsequent release of magnetic twist from the jet base, a newborn larger-scale MFR was then created in the unwinding jet spire, with its new south feet exchanged to a remote site (around 270 Mm far from the jet base). Finally, this newborn large-scale MFR successfully erupt into the outer coronae driving a stellar-sized Earth-directed CME, leaving another an C1.8 flare in its source region. On the whole, this event highlights the pathway of a real magnetic coupling process in the initiation of the Earth-directed CME, supporting the view that some large-scale coronal eruptive phenomena can originate from the magnetic coupling of different magnetic activities at various spatial scales \citep{2007SoPh..244...75W,2015SCPMA..58.5682W,2013ApJ...773...93S}.
\par
Despite that the release of magnetic twist phenomena are frequently detected in coronal jets \citep{2011A&A...532L...9C,2011ApJ...735L..43S,2012RAA....12..573C,2013RAA....13..253H,2013ApJ...769..134M,2015ApJ...806...11M,2015ApJ...814L..13L,2019ApJ...887..239Y,2019FrASS...6...44L} and filament eruption events \citep[e.g.;][]{2014ApJ...797...52Y,2020ApJ...904...15Y,2018MNRAS.476.1286J,2018ApJ...866...96J}, but their resultant consequence are poorly studied yet. This work provides a direct stereoscopic imaging observation on the creation of a large-scale erupting CME flux rope in an unwinding coronal jet. Different from previous mentioned flux-emergence formation mechanism \citep[e.g.,][]{2004ApJ...609.1123F,2004ApJ...610..588M,2008ApJ...673L.215O,2018SoPh..293...93C,2019ApJ...874...96Y} and reconnection-cancellation formation mechanism \citep[e.g.,][]{2012ApJ...759..105S,2017ApJ...839..128W,2018ApJ...859..148W,2018ApJ...869...78C,2019ApJ...887..118C}, this observation supports an alternative scenario: the rapid magnetic twist transport from the jet base to background fields directly result in the formation of newborn large-scale MFR. Moreover, a rough twist estimation of the blowout jet indicates that around 1.43$-$2.01 turns of magnetic twist was released in its jet spire. The twist characteristics of the newborn large-scale flux rope and its anti-clockwise rolling motion can also be clearly recognized in STEREO/EUVI observations. In addition, the location of its ``drift" remote feet is also confirmed as a moving radio source in simultaneous NRH radio observations.
\par
Apart from our observation, applying a state-of-the-art of data-constrained 3D MHD simulations and observations, \citet{2018ApJ...866...96J} also noticed that in NOAA 11283, an erupting sigmoidal flux rope breaks one of its legs, and quickly gives birth to a new tornado-like flux rope that is highly twisted and has multiple connections to the Sun (see also a similar 3D simulation from \citet{2020ApJ...903..129P}, which focus on the same event but mainly studies its related coronal dimmings). \citet{2016ApJ...822L..23P} also reported a series of coronal jets occurring at the edge of the AR 12192, which resulted in homogenous bubble-shaped CMEs. Based on the release of magnetic twist from the jet-base field into large-scale coronal loops and related coronal dimming during the jet eruptions, \citet{2016ApJ...822L..23P} inferred that the jet-guiding coronal loops eventually blew out as low-speed CME bubbles due to increasing twist. Focusing on one of these CME-producing jets, \citet{2015ApJ...814L..13L} also provided a detailed twist estimation in the jet spire with high-resolution chromospheric observations, and they reported that the total rotation angle is high up to 8$\pi$. In the current work, the estimated rotation angle in the unwinding jet ranges from around 2.8$-$4.0$\pi$, which should be enough for the creation of the newborn CME flux rope.
\par
Before the occurrence of the blowout jet, repetitive reconnection triggered four obvious recurrent jets above an elongate enhanced $J_z$ enhancement between opposite-polarity converging flux ($P2$ and $N1$) (namely at POS1 ). These recurrent jets demonstrate narrow jet spire and short lifetime, thus may termed as ``standard" jets, while the blowout jet of our interest has a broader unwinding spire and triggered by a microfilament eruption scenario \citep{2010ApJ...720..757M,2015Natur.523..437S}. In their source region, flux emergence provided a continual injection of poynting flux and magnetic free energy for recurrent eruptions (see \nfig{fig3}(d1-d2)). On the other hand, flux emergence also introduced an emergence-driven canceling interface. As a result, reconnection between F1-contained newly emerged flux and pre-existing field happened, which weaken its constraint above the F1 at some degree \citep{2007ApJ...669.1359S,2019ApJ...874...96Y}. Consistent with the result of \citet{2014A&A...562A.110L} and \citet{2016ApJ...821..100S}, these suggest that both flux emergence and its driven flux cancellation should play a role in the onset of this whole magnetic coupling eruption.
\par
At last remark, two issues worthy of mention here. First, in order to fully understand such jet-CME coupling eruption events, the energetics and interplanetary effects of jet-driven CMEs should be linked back to their jet base parameters in the future, especially the twisting effect of pre-jet filaments. Second, it must be pointed out that this newborn large-scale CME flux rope in the present observation immediately erupt due to its higher formation height, despite that it actually created along an extended PIL\footnote{See the online synoptic magnetogram: \url{http://jsoc.stanford.edu/data/hmi/synoptic/hmi.Synoptic_Ml.2125.png}.}. Therefore, whether this twist-transport formation scenario can apply to explain the appearance of those long-term exited large-scale MFRs among inter-coupled ARs \citep[e.g.,][]{2019ApJ...873...23Z} remains unclear.
\acknowledgments
The authors sincerely thank the referee for constructive suggestions and comments. H.C.C. thanks Dr. Guiping Zhou for helpful discussions after the on-line seminar, ``Frontiers in solar physics", held by the Key Laboratory of Solar Activity of NAO. H.C.C. is supported by the National Postdoctoral Program for Innovative Talents (BX20200013) and China Postdoctoral Science Foundation (2020M680201); This work is also supported by the National Key R$\&$D Program of China (2019YFA0405000), and the National Natural Science Foundation of China under grants 11633008, 11873088, 11933009, and 11703084.
\vspace{5mm}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 2,128 |
Q: requests.exceptions.ConnectionError: Failed to establish a new connection: [Errno 111] Connection refused I am trying to use py2cytoscape from python2.
from py2cytoscape import cyrest
cytoscape=cyrest.cyclient()
cytoscape.status()
On which I am getting :
Could not get status from CyREST:
HTTPConnectionPool(host='localhost', port=1234): Max retries exceeded with url: /v1/ (Caused by NewConnectionError(': Failed to establish a new connection: [Errno 111] Connection refused',))
What could be the problem?
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 7,781 |
{"url":"https:\/\/www.gamedev.net\/forums\/topic\/678449-i-doubt-i-understand-raii-just-yet\/","text":"# I doubt I understand RAII just yet\n\nThis topic is 786 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.\n\n## Recommended Posts\n\nHello dear forum! I have a question concerning RAII.\n\nFirst let me try to word it myself: RAII is the concept of wrapping up resources that shall be deleted once they go out of scope.\n\nThis \"wrapper\" then deletes the resource when it goes out of scope.\n\nEspecially because I cannot guarantee that I delete it myself, the risk of screwing this seems quite big.\n\nThough I'm a bit sceptical about this. I should know what object exists and should know how to clean it up - even RAII seems to have flaws, this gives me quite a headache.\n\nIs the resource management this hard? For now, I never ran into severe issues nor can I prove that there never were any dead memory entries.\n\nHowever the way I clean up is probably not the idomatic way and less reliable than the RAII-concept.\n\nNow taking a look at the implementation, which gets quite a bit difficult for me. Mainly looked at the Wikipedia article:\n\nhttps:\/\/en.wikibooks.org\/wiki\/More_C%2B%2B_Idioms\/Resource_Acquisition_Is_Initialization\n\nI would need one object that auto deletes my other resource. Let's say I have a class which contains multiple objects and quite some pointers.\n\nI would avoid everything but smart pointers, is this already a part of the RAII concept?\n\nWould I need to wrap an autodelete class around every object that the object owns (including itself)? Or should I simply put the object that is owning the other objects into the autodelete?\n\nAs the owner might live way longer than their owned objects, I probably want to free the memory earlier (when the destructor of the owned objects is being called). Which would let me tend to my first idea.\n\nThese all might be a wrong perception of mine. Is that even the right way of handling RAII?\n\nWhen the first example in the wiki writes:\n\nlock.acquire();\nlock.release();\n\nI totally have no idea what this does in that place. What does it mean? It probably has to do with freeing the memory but I hardly can grasp around its functionality.\n\nThough I know about the release() of pointers, I see no pointer being used here which stands in a relationship to this.\n\nAnother question that jumps into my head is the scopedLock part. Is this another idiom?\n\nMy last question is about when to use RAII and when not. Whenever I browse through other C++ code, I doubt I ever saw it being explictly used as in the Wikipedia article.\n\nDoes this mean there are better or different implementations?\n\nI would be really happy if someone could explain this to me, because I want to know how I can implement RAII myself.\n\nIt gets quite frustrating on how many idioms there are, it feels like bad practise to program C++ in the first place unless I know all idioms and design patterns..\n\nEdited by Angelic Ice\n\n##### Share on other sites\nLike always, I wouldn't implement it for the idea. If you have good resource management and no issues, it's good to know it's possible but not necessary to implement.\n\n?\n\n##### Share on other sites\nWhile I did not downvote you nor do I know the mind of the person who did, I kinda agree with the vote.\n\nYour advise feels like bad advise. RAII is so fundamental to working in C++, if you do not have a solid grasp on it you should strive to remedy that situation as soon as possible. The reason for not using RAII should be <well-reasoned rationale> not 'nah, never had to use it before'.\n\nI agree, thanks.\n\n1. 1\nRutin\n25\n2. 2\n3. 3\n4. 4\nJoeJ\n17\n5. 5\n\n\u2022 14\n\u2022 11\n\u2022 9\n\u2022 9\n\u2022 10\n\u2022 ### Forum Statistics\n\n\u2022 Total Topics\n631756\n\u2022 Total Posts\n3002111\n\u00d7\n\n## Important Information\n\nWe are the game development community.\n\nWhether you are an\u00a0indie, hobbyist, AAA developer, or just trying to learn,\u00a0GameDev.net is the place for you to learn, share, and connect with\u00a0the games industry. Learn more About Us or sign up!\n\nSign me up!","date":"2018-07-19 19:40:13","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 0, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 1, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.2448824942111969, \"perplexity\": 1396.2123845671954}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 5, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2018-30\/segments\/1531676591216.51\/warc\/CC-MAIN-20180719183926-20180719203926-00046.warc.gz\"}"} | null | null |
With Easter the biggest chocolate opportunity in the spring season, Asian Trader discovers how independent retailers can make the most of the occasion.
Easter is the greatest peak for confectionery sales across spring, with total confectionery growing by +3.9% in 2018 (Nielsen).
A key driver of that growth in 2018 came from the Easter Eggs sector, which achieved value sales worth £370m according to Nielsen.
Easter is all about creating memories with your loved ones through treats and traditions.
Consumers are seeking more experiences during the season, both in experiential and consumption.
According to a 2017 One Poll of 2000 parents with children aged 0-16 years old, the top activities for consumers at Easter are: eating chocolate, seeing friends and family, gifting chocolate and going on egg hunts.
As the market leader at Easter (IRI), Mondelēz holds the leading share of shell eggs and the number one position in miniature eggs, and single serve.
Mondelēz was the number one manufacturer during Easter 2018 with 44% share of the market. 82% of the UK population ate a Mondelēz Easter product during the season with Cadbury Creme Egg being the number one brand.
Four Cadbury Creme Eggs were sold every second in 2018, bringing with it £57m in brand sales – up by 22%3 thanks to the hunt for Cadbury's White Cadbury Creme Egg (IRI). In 2018, 157 units of Cadbury Mini Eggs were sold per minute.
Mondelēz drove the absolute largest value into the shell eggs category year-on-year and grew the subcategory by £10m year-on-year. It also remains number one in the single serve category with a 64% share.
For Easter 2019, Mondelēz's ambition is to remain a leader of the season and help bring families together, through its cracking portfolio of brands and products, the company said in a statement.
The company is bringing new innovation to all segments and investing £10m during the season.
Marketing support began on 1 January with the launch of a national Cadbury Creme Egg promotion 'Hunt the White Creme Egg' – which involves a convenience channel exclusive promotion, where the retailer could win £1,000 if a White Creme Egg is found in their store. Plus, Cadbury Creme Egg hunting season will return to TV, supported by a far reaching digital media campaign.
Two power brands will unite this Easter to bring the consumer a new flavour – Cadbury Oreo Egg. This new product aims to drive sales in the category incrementally, after it proved its success in Canada being 62% incremental to Cadbury Creme Egg.
Throughout February and March, shoppers are looking to buy gifts and sharing products for friends and family, especially their children, to enjoy during casual together time.
They are also looking to capture the spirit and tradition of Easter, while creating memorable moments through the egg hunt ritual, using a range of different products from brands their family knows and loves.
Sharing products, like Cadbury Mini Eggs and Cadbury Egg 'n' Spoon, are ideal during these times to help create special sharing moments at home.
At this time Cadbury will be organising egg hunts across the country, with a £6m campaign and Cadbury will be involved in public egg hunts, with the continuation of its partnership with the National Trust.
Cadbury Heroes favourite Cadbury Creme Egg Twisted, will be launching in a convenient bag format which will help to target shoppers who are looking for new products – £1.49 RRP, 94g, 10 packs per case.
Mondelēz's number one Easter brand will introduce the Cadbury Creme Egg Mega Egg – £4.99 RRP, 209g, 6 packs per case.
Also new for Easter, is the Cadbury Heroes Easter Pouch – £5.69 RRP, 384g, 8 packs per case.
In the three weeks prior to Easter, consumers are looking to create and extend the unique spirit of Easter through gifts, while continuing to inspire memorable moments through the egg hunt ritual.
To meet this consumer need, it is crucial to stock family sharing and gifting products during this time, Mondelez says.
Gifting accounts for more than 60% of chocolate occasions, which demonstrates the importance of stocking shell eggs (Engage Research), as consumers are on the hunt for surprises and gifts for their loved-ones.
New to Mondelēz shell egg portfolio, and aiming to build on the success of the inclusion range, is the Cadbury Crunchie Inclusion Egg.
This NPD is part of the £40m Crunchie brand and offers a gift worthy, standout product to the shopper – £12.00 RRP, 570g, 4 packs per case. Cadbury Picnic shell egg will also be launching, which is the sub-brand's first ever shell egg. £6.00 RRP, 274g, 6 packs per case.
Ferrero is introducing a number of egg products to appeal to a wide range of shoppers.
The Thorntons range will be bolstered by the introduction of three new luxury eggs, featuring strong colours and a premium look and feel to appeal to younger, affluent shoppers.
The premium eggs will be available in: White Chocolate & Raspberry, Milk Chocolate Almond & Hazelnut and Dark Chocolate & Orange.
The popular 212.5g and 275g luxury Ferrero Rocher eggs saw sales increase by +70% YoY in 2018, with the larger egg ranking 3rd overall in the Luxury Eggs category (Nielsen).
Both products will receive a makeover for 2019 to provide greater stand out on shelf, whilst also highlighting the quality of the products inside.
The Ferrero Rocher range will be supported across multiple channels at each occasion throughout spring.
For Kinder, two new 220g Surprise eggs are being launched to follow on from the success of the 100g in 2018 – where value sales grew by +21% and it was the number 1 performing SKU in the kids confectionery category (Nielsen).
The Kinder Surprise eggs will drive further growth in the category, with bigger toys and relevant licences (Teenage Mutant Ninja Turtles and Powerpuff Girls) having a wide appeal.
The Kinder range is being refreshed with a host of new products and line extensions. The five-strong range of Kinder Surprise 36g hollow figures will sport seasonally relevant characters and designs, likely to support early season sales for self-treating.
The entire Kinder portfolio will be supported by a media campaign across all key channels and haloing the full seasonal range.
Focussing on the key differentiator of Kinder Surprise products – the toy – the campaign will drive relevance with parents, showing the variety of the products and toys available while demonstrating the excitement they can bring.
For shoppers seeking more luxury, the new Ferrero Rocher mini eggs drive sales early in the season when seasonal sharing is at its peak.
The two Ferrero egg products (Hazelnut or Cocoa, 100g) are set to capitalise on the growing popularity of mini eggs, with the launch being supported by sampling activity at high-footfall areas to drive awareness across the UK.
Bolstering the range of novelties further is the introduction of the new Ferrero Rocher Squirrel.
The 90g product scored highly for likeability in development, with 63% of shoppers saying they would purchase as an additional item.
Completing Ferrero's range of new products for Easter 2019 are new bunny figures from Thorntons.
Two new 200g bunny products (one milk chocolate and one white) are being launched this spring, having scored highly in consumer testing with 82% of shoppers saying they would be happy to give one as a gift.
Ferrero will also be supporting retailers across Valentine's Day and Mother's Day through range extensions, limited edition product formats and seasonally relevant packaging, all underpinned with multi-channel media support.
The value of Thorntons boxed chocolate grew year on year at both occasions in 2018, at +12%6 (£3.1m) and +15%7 (£4.2m) respectively. The limited edition 'With Love' boxes make a return for Thorntons in 2019, with updated packaging to drive more interest and underline the premium product inside.
For Ferrero Rocher, Ferrero Collection and Raffaello the core range leads sales as shoppers trust the product and recognise it as a high-quality option when looking for a gift for their loved ones. Ferrero Collection will have two seasonally relevant packs for Mother's Day, one ribbon pack and one "Just for you" sleeve pack.
Raffaello saw retail sales value increase by +22% (Nielsen) in 2018. A New T8 80G Matchbox pack is being launched in 2019, featuring versatile designs to cater to multiple occasions all year round.
The special edition T14 heart also returns for spring 2019, after a +134% volume increase in 2018 (Nielsen). Ferrero will be supporting the line up with a digital campaign coupled with one of the UK's biggest sampling activities ever, set to reach five million potential consumers and provide three million samples.
"Confectionery is one of the few categories where shoppers are willing to spend money on products they love, especially at key seasonal trading spikes such as those throughout spring," adds Boorer.
"We always strive to offer retailers and shoppers the right balance of seasonal specials, which drive awareness and interest to the category, with our established core range, which helps to drive sales all year round. We're committed to investing in media campaigns alongside our significant NPD to help retailers capitalise on the peaks in consumer demand by carrying the right mix of products," Boorer concludes. | {
"redpajama_set_name": "RedPajamaC4"
} | 8,417 |
Q: Find duplicates in array without extra memory and can use variables for storing values How can we find the duplicate elements in an array with the following limitations:
*
*Without using extra memory
*Can use variables for storing data and not objects like HashMap/HashSet and
all.
*Time complexity can be O(n) and should not be O(n^2)
Note :
Here array is dynamic integer array.
A: If all elements of the array are in between [0, n), then we can do it in the following way
*
*Traverse the array from i = 0 to n
*for every element a[i], suppose
x = a[i] if a[i] >= 0
x = a[i] + n if a[i] < 0
check if a[x] >= 0. If yes then add -n to a[x]
*if for another element a[i]. if a[x] < 0 then that means that a similar value changed the index hence it is a duplicate.
A: Below is my solution for the above question.
public static boolean checkDuplicates(int[] arr, int length) {
int value = 0;
boolean isDupFound = false;
for (int i = 0; i < length; i++) {
int currChar = arr[i];
int bit_Position = currChar - 'a';
if ((value & (1 << bit_Position)) > 0) {
isDupFound = true;
break;
} else {
value = value | (1 << bit_Position);
}
}
return isDupFound;
}
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 9,596 |
Q: Highlighting a part of a text using html
Hello everyone.
The question which I have mentioned earlier is related to this. The letter which is shown in image was written by using unicodes.
పా
now I want to highlight the white part of that image. I have tried this by replacing the required part by using
ా but it is not suitable as it is having the dotted lines. And I even tried this by using the png image of that letter for which it is harder for me to change the colour of the image for highlighting. Please suggest me some methods.
A: you can do it with a font, create a custom font from https://icomoon.io/. For that you will need SVG file of both the symbol పా and ా and create a font. that way you will be able to change all propeties of font using CSS.
Here is a tutorial on How to Create a SVG from Illustrator
| {
"redpajama_set_name": "RedPajamaStackExchange"
} | 447 |
PHOENIX RISING
William W. Johnstone with J. A. Johnstone
PINNACLE BOOKS
Kensington Publishing Corp.
www.kensingtonbooks.com
All copyrighted material within is Attributor Protected.
Table of Contents
Title Page
Epigraph
CHAPTER ONE
CHAPTER TWO
CHAPTER THREE
CHAPTER FOUR
CHAPTER FIVE
CHAPTER SIX
CHAPTER SEVEN
CHAPTER EIGHT
CHAPTER NINE
CHAPTER TEN
CHAPTER ELEVEN
CHAPTER TWELVE
CHAPTER THIRTEEN
CHAPTER FOURTEEN
CHAPTER FIFTEEN
CHAPTER SIXTEEN
CHAPTER SEVENTEEN
CHAPTER EIGHTEEN
CHAPTER NINETEEN
CHAPTER TWENTY
CHAPTER TWENTY-ONE
CHAPTER TWENTY-TWO
CHAPTER TWENTY-THREE
CHAPTER TWENTY-FOUR
CHAPTER TWENTY-FIVE
CHAPTER TWENTY-SIX
Copyright Page
I may be changed by war,
but I will not be defeated by it.
—AUDIE MURPHY
CHAPTER ONE
Fort Rucker, Alabama—Tuesday, January 10
Major Jake Lantz was thirty-two years old. A helicopter pilot and flight instructor in the Army Aviation School at Fort Rucker, Alabama, he was in the peak of physical condition, recently scoring a perfect three hundred on his latest PT test, maxing out on the three required events: push-ups, sit-ups, and the two-mile run. A not-too-prominent scar on his right cheek, the result of shrapnel wound in Afghanistan, ran like a bolt of lightning from just below his eye to the corner of his mouth. He had blue eyes, and he wore his light brown hair closely cropped, in the way of a soldier.
Jake, who was a bachelor, lived alone in a three-bedroom ranch-style house on Baldwin Court in Ozark, Alabama, the town that proudly bills itself as the "Home of Fort Rucker." He had kept the heat down during the day to save on his gas bill. Now he shivered as he turned it up.
Stripping out of his flight suit, Jake pulled on a pair of sweatpants and a red sweatshirt, emblazoned with the word ALABAMA across the front. He had not gone to school at Alabama, but had become a big fan of University of Alabama football.
Checking the digital clock on his dresser, he saw that he had but one minute left until the program he wanted to watch came on, so he hurried into the living room, settled down on the couch, picked up the remote, and clicked it toward the TV.
The initials GG appeared on the screen, then the voice-over introduced the show.
From New York! It's the George Gregoire show!
And now, here is your host, George Gregoire!
The GG monogram moved into the background, and George Gregoire, with his signature crew-cut blond hair, slightly chubby face, and toothy smile, greeted his television audience.
Hello, America!
You are not going to want to miss the show today. I have information that, if I had been able to verify it before the election last November, might have saved our country the anguish, turmoil, and trouble we are going to go through over the next four years under President-elect Mehdi Ohmshidi.
In fact, I will say it here and now, this could be grounds for impeachment. Can a president be impeached even before he assumes office? I don't know, but if the men and women in the House and Senate would put our country ahead of party, they might just want to think about this.
Here is a video, recently surfaced, of President-elect Mehdi Ohmshidi giving an address to the OWG. The OWG stands for One World Government. Ohmshidi is—well, let's just let the video speak for itself.
The video was somewhat grainy, obviously taken not by a camera for broadcast, but by a small, personal camera. Nevertheless, it was quite clearly President-elect Mehdi Ohmshidi standing at a podium addressing a rather sizeable crowd. Many in the crowd were holding signs, saying such things as:
U.S. Is An Obsolete Concept
One People, One World, One Government
No More Flags, No More Wars
Patriotism Is Jingoistic
Ohmshidi began to speak and because the sound wasn't of the best quality, his words were superimposed in bright yellow, over the picture.
I see a world united! A world at peace! A world where there are no rich and there are no poor, a world of universal equality and brotherhood.
Such a world will surely come, my friends, but it will never be as long as we are divided by such things as religion, patriotism, the greed of capitalism, and the evil of so-called honorable military service. There is nothing honorable about fighting a war to advance one nation's principles over another. One world, one people, one government!
Ohmshidi's closing shout was met by thunderous applause and cheers from the audience.
The picture returned to George Gregoire on his New York set.
The question of Ohmshidi's membership in the OWG was raised during the election, but spokesmen for Ohmshidi said that it was merely a flirtation he had entered into when he was in college.
Really ?
Ohmshidi graduated from UC Berkeley twenty-one years ago. I'm going to bring the video up again, in freeze-frame. I want you to look at the sign on the curtain behind him.
In freeze-frame, on the curtain behind the speaker's stand were the words:
WELCOME TO THE ONE WORLD GOVERNMENT CONVENTION
Notice, that beneath the sign are the dates of the convention, June 6 to June 10—TWO YEARS AGO!
"Jake, are you in here?" a woman's voice called from the front door.
Jake picked up the remote and muted the TV. "In here, Karin," he called back.
Karin Dawes was a captain, an Army nurse, who was still wearing her uniform. She had short black hair, brown eyes, an olive complexion, and the same body she had when she was a college cheerleader. She was also a world-class marathoner who had just missed qualifying to represent the U.S. in the last Olympics. Seeing George Gregoire on the silent TV screen, Karin chuckled.
"You're watching Gregoire. Of course, it's six o'clock. What else would you be watching?"
"You should watch him," Jake said. "Maybe you would learn something."
"I do watch him," Karin said. "As much time as I spend over here with you, how can I help but watch him?"
"Ha! Now I know why you spend so much time over here. Here, I thought it was my charm. Now I find out it's just so you can watch George Gregoire."
"I confess, you are right," she said. She leaned over to kiss him, the kiss quickly deepening.
"Damn," Jake said, when they separated. "That's what I call a greeting. Do I sense a possibility that this could go further?"
"How can it go any further?" Karin said. "It's at least half an hour before Gregoire is over, isn't it?"
Jake picked up the remote again, and turned the TV off.
"You're sure I'm not taking you away from George Gregoire?" Karin teased. "I certainly would not want to be accused of alienation of affection."
"Woman, you talk too damn much," Jake said, kissing her again. "Besides," he said, "I've got a TV in the bedroom, I can always watch him while . . ."
"You try that, Major, and you'll have George Gregoire in bed with you before I split the sheets with you again," Karin said, hitting him playfully on the shoulder. Jake laughed out loud, then put his arm around her as they went into the bedroom.
There was an ease in their coupling, the assurance of being comfortable lovers who knew each other well, and yet their relationship was not so stale that it couldn't still be fresh with new discovery. Outside, the wind was blowing hard, and Jake could hear the dry rattle of the leafless limbs of an ancient oak.
Afterward they lay together under the covers, her head on his shoulder, his arm around her, his hand resting on her naked thigh. It was, as always, a feeling of total contentment.
"Jake?"
"Yes, my love?"
"Will we always have this? I don't mean are we going to get married, or anything like that. I just mean, will we always have this sense of joie de vivre?"
"Is there any reason why we shouldn't?"
"I don't know," Karin admitted. "I know I tease you about watching George Gregoire all the time, and about listening to all the right-wing radio shows. But, what if they are right? What if the country has made a big mistake in electing Ohmshidi?"
"There is no what-if," Jake said. "We did make a big mistake. Well, we didn't. I'm not a part of the we, because I didn't vote for him."
"I didn't either."
Jake raised his head and looked down at her. "What? You, Miss Liberal Incarnate? You didn't vote for him?"
"I couldn't bring myself to vote for him," she said. "Not knowing the way you felt about it."
Jake kissed her on the forehead. "Maybe there is some hope for you yet," he said.
"But you didn't answer my question. Will we always have this?"
A sudden gust of wind caused the shutters to moan.
When there was an uncomfortable gap in the conversation that stretched so long that Karin knew Jake wasn't going to answer, she changed the subject.
"I wonder if it is going to snow."
"Don't be silly," he said. "It never snows in Ozark, Alabama."
There were three inches of snow on the ground the next morning as Jake drove the ten miles into Fort Rucker. Because snow was so rare here—it had been fifteen years since the last snow—neither Ozark nor Dale County had the equipment to clean the roads. As a result, Jake drove slowly through the ruts that had been cut in the snow by earlier cars. He returned the salute of the MP at the Ozark gate, then drove down Anderson Road, which, like the streets in Ozark, was still covered with snow.
As chief of Environmental Flight Tactics, Jake had his own marked parking slot, though the sign was covered with snow. He exchanged salutes with a couple of warrant officer pilots as he covered the distance between his car and the front door of the building which held not only the offices of the faculty, but also classrooms for the ground school.
"Major, I thought you told me that it never snowed in Southern Alabama," Clay Matthews said. Sergeant Major Matthews was Jake's right-hand man, the non-commissioned officer in charge of EFT
"It doesn't," Jake said. "Disabuse yourself of any idea that this white stuff you see on the ground is snow. It's just a little global warming, that's all."
"Right," Clay said with a little chuckle. "Oh, Lieutenant Patterson called from General von Cairns's office. The general wants you to drop by sometime this morning."
"What's my schedule?"
"You don't have anything until thirteen hundred."
"All right, maybe I'll drop by his office now. I'm not surprised he wants to see me. I told him, he wouldn't be able to run this post without my help."
"Yes, sir, that's what I tell everyone about Environmental, too," Clay said. "You couldn't run the place without me."
Jake chuckled. "Yeah, well, the difference is, I'm just shooting off my mouth when I say that about the general. But when you say that about me, you are right."
Like Ozark, Fort Rucker had no snowplow equipment. But it did have a ready supply of man power and there were several enlisted men, under the direction of a sergeant, clearing off the parking lot and shoveling the sidewalks at the post headquarters. Because of that, Jake was able to walk from his car to the building without getting his boots wet.
Lieutenant Phil Patterson was on the phone when Jake stepped into the outer office, but he hung up quickly, and stood.
"Good morning, Major," he said. "Just a moment and I'll tell the general you are here."
"Thanks."
First Lieutenant Phil Patterson was a West Point graduate who had recently completed flight school. Jake remembered him when he was a student going through the Environmental Flight Tactics phase of his training. He was a bright, eager, and well-coordinated young man. Patterson had wanted an overseas assignment out of flight school, and was disappointed when he was chosen to stay at Fort Rucker as the general's aide-de-camp. But, he was a first lieutenant in a captain's slot, so the assignment wasn't hurting his career any.
Patterson stepped back out of the general's office a moment later. "The general will see you, sir."
Jake nodded his thanks, and stepped into the general's office. Major General Clifton von Cairns was pouring two cups of coffee.
"Have a seat there on the sofa, Jake," the general said. Jake had served in Iraq with von Cairns when he had been a captain and von Cairns had been a colonel. That was von Cairns's second time in Iraq—he had also been there during Operation Desert Storm.
"As I recall, you like a little bit of coffee with your cream and sugar," von Cairns said as he prepared the coffee.
"Yes, sir, thank you."
Carrying the two cups with him, von Cairns handed the one that was liberally dosed with cream to Jake. "I'm sorry I don't have any root beer," von Cairns said. "That is your drink, isn't it?"
"I like a root beer now and then," Jake said.
"Yes, I remember your 'beer' run when we were in Iraq," von Cairns said.
Jake's preference for root beer was well known by everyone who had ever worked with him. What the general was referring to was the time Jake had made a run to Joint Base Balad for beer and soft drinks. Beer wasn't actually authorized due to cultural concerns and was officially banned by the military; however the civilian contractors were not constrained by such rules and they were a ready source of supply for the Army. But Jake had come back with only one case of beer and nineteen cases of root beer in the helicopter. He was never asked to make a beer run again.
"How many students do you have in your cycle right now?" the general asked.
"I have twelve."
"Can you expedite them through? Double up on the flight hours?"
"Yes, sir, I suppose I could. It would mean rescheduling some of the ground schooling."
"I want you to do that," von Cairns said. He took a swallow of his coffee before he spoke again.
"Jake, I'm not much for politics—I've always thought that as a professional soldier I should leave the politics to others. But I don't mind telling you, this new man we're about to swear in scares the hell out of me. I've heard some disturbing talk from some of my friends at DA. They are afraid he is going to start cutting our budget with a hatchet. If we don't get this cycle through quickly, we may not get them through at all."
"Surely he wouldn't halt flight training, would he?" Jake asked. "So much of the Army is now oriented around aviation."
"Did you watch George Gregoire last night?" von Cairns asked.
"I rarely miss it."
"You might remember when Gregoire showed Ohmshidi speaking to the OWG group, he said, and I quote, 'the evil of so-called honorable military service .' This man doesn't just distrust the military, he hates the military. And he is about to become our commander in chief."
"I understand, General," Jake said. "I'll get the schedules revamped as quickly as I can."
"You are a good officer, Jake. Would that I had a hundred just like you. It is a pleasure to have you in my command."
"And I am honored to serve under you, General."
General von Cairns stood up then, a signal that the meeting was over. Jake stood as well, and started to leave.
"Jake, are you still seeing that nurse? What is her name?"
"Karin Dawes, sir. Captain Karin Dawes."
"Yes, she is the one I pinned the Bronze Star on last month, isn't she? She's a good woman. You could do worse."
CHAPTER TWO
Wednesday, January 18
Hello, America.
With just two days before we swear in our new president, I would like for us to take inventory of just where we are in this country.
Four decades of social engineering have begun to accrue in such a way as to presage disaster for the U.S.
Gregoire held his hands over his head and waved them as he rolled his eyes.
This is not just the ravings of—mad—George Gregoire. No, sir, and no, ma'am. Events over the last several years have borne me out.
Consider this. Stringent environmental laws have inhibited drilling in new fields for domestic oil. Those same laws have also limited refining capacity and dictated exotic cocktail blends of fuel for certain parts of the country. Even during times of critical fuel shortages, these blends cannot be transshipped from region to region.
Automobile companies are mandated CAFE standards and unnecessary safety features that add thousands of dollars to the base prices of cars.
Do you remember when we were young, how eagerly we looked for the new cars each year?
Gregoire changed the tone of his voice, mimicking the excitement.
Have you seen the new Ford? Yes, but wait until you see the new Chevy!
He was silent for a moment, masterfully playing his audience.
Tell me, America, when is the last time you greeted the new models with anything more than a yawn?
And have you noticed that fewer and fewer models are being introduced now? Proud names such as Plymouth, Oldsmobile, and Pontiac Trans Am—cars which we once lusted after, cars with style and performance—are no more.
He began to sing,
What a thrill to take the wheel, of my brand-new Oldsmobile.
America, we have had a century-old deep and abiding love affair with cars, but now we find them boring. We look back on the cars of the fifties and sixties with a reverent nostalgia, and like most nostalgia, this is an unrequited love—we will never return to those days. Do you remember those yesterdays when we were young? Do you remember the sweetness of life then, as rain upon my tongue?
He began singing Roy Clark's "Yesterday When I Was Young."
Oh, and how is this for intelligence? In California, federal courts, in order to preserve a two-inch inedible fish, have restricted the flow of water into some of the most productive agricultural areas in the country. And since California produces nearly fifty percent of the nation's fruits, nuts, and vegetables, this water restriction is already having a drastic impact on the market price.
Government interference with bank lending has caused the housing market to go bust, resulting in the loss of billions of dollars in personal equities across the country.
Gregoire, who was standing now, stuck his hands in his pockets and looked at the floor, silent for a long moment before he spoke again. The camera came in tight on his face so he could give the audience his most sincere look.
My friends, this is the country that elected Mehdi Ohmshidi, a naturalized American born forty-seven years ago in Islamabad, Pakistan. I can only pray that we survive this monumental mistake.
Thursday, January 19
"All right, Candidate Lewis," Jake told his flight student. "We've just received word from previous flights that the LZ is bracketed by small-arms fire from your nine o'clock, and shoulder-launched ground-to-air missiles from your three o'clock. How are you going to avoid the ground fire?"
"Make the approach below their angle of fire, sir," the warrant officer candidate replied.
"Make it so," Jake said, mimicking Captain Picard of Star Trek.
As WOC Lewis started his descent, Jake saw a flock of geese approaching from the right.
"Watch the geese on your three o'clock," Jake said.
"I see them," Lewis answered. Lewis pulled the collective to try and go over them, but the geese were making the same maneuver.
"Damn!" Lewis shouted as several of the geese collided with the helicopter. Blood and feathers from those that hit the main rotor suddenly appeared on the windshield. There was also a sudden and severe vibration at the same time they could hear the high-pitched whine of the tail rotor drive shaft spinning without any resistance.
"I've got it!" Jake shouted, taking the controls.
There was a loud bang as the tail rotor and a part of the tail fin separated from the aircraft. The center of gravity pitched forward and, without the anti-torque action of the tail rotor, the helicopter began to spin to the right. Instinctively, Jake depressed the left anti-torque pedal to halt the spin, even as he knew that without the tail rotor, it would be ineffective.
The spin was much faster than anything Jake had ever experienced, and earth and sky blended into a whirling pattern that made it impossible to separate one from the other.
Out of the corner of his eye, Jake saw Candidate Lewis start to grab the cyclic.
"Hands off !" Jake screamed.
They were about seventy-five feet above ground and had already spun around at least fifteen times. Jake knew he needed to kill the engines in order to lessen the torque, but the engine controls were on the cockpit roof and he had to fight the centrifugal force in order to get his arm up. Finally he managed to kill both engines. The whirling main rotor blades continued to generate torque but, mercifully, without the engines, the spinning slowed.
Then, just before impact, Jake jerked back on the cyclic and the nose of the helicopter came up. Now, with the spin rate down to half what it had been, and with the helicopter level, the Blackhawk made a hard but somewhat controlled landing.
Jake sat in his seat with dust streaming up around the helicopter and the rotor blades still spinning. He waited until the spinning was slow enough that he knew they would not generate lift, then pulled the collective up, putting enough pitch in the blades to slow them until they finally stopped.
"Are you okay?" Jake asked.
"What the hell happened?" Candidate Lewis asked.
"You got hit by an RPG," Jake said.
"What?"
"A goose, or some geese, took out the tail rotor," Jake said. "It was the same effect as being hit by an RPG."
"Damn. I'm glad I wasn't flying solo," Lewis said.
"Funny you would say that," Jake said. "I was just thinking I wish the hell you had been flying solo."
Although neither pilot was hurt, they were required by SOP to report to the hospital for a physical evaluation. Jake was in the examining room, just zipping his flight suit closed when Karin came in with a worried look on her face.
"I heard you were in a crash!" she said, the tone of her voice reflecting her worry.
"I resent that. I made a controlled landing," Jake replied. "A hard landing, yes, but it was controlled."
Karin threw her arms around him. "Oh," she said. "When I heard you had been brought in I was scared to death."
"It's nice to be worried about," Jake said. "But really, it was no big thing."
"Hah, no big thing, my foot. I heard some of the other aviators talking about it. You lost your tail rotor but were able to land. Everyone is calling you a hero."
"A hero?" Jake said. He smiled. "Yeah, I'll accept that."
"Well, now, don't let it go to your head," Karin teased. "You are hard enough to be around as it is."
"Really? How do you manage to be around me so much?"
"Because I'm a saint. Didn't you know that?" Karin asked. She kissed him.
"Careful. What if one of the other nurses came in now and caught you cavorting with a patient?"
"I'd tell them to get their own patient," Karin replied with a broad smile.
"I'm off tomorrow," Jake said. "Because of the aircraft incident, I'm supposed to take a forty-eight-hour stand-down. What about you?"
"I'm not off until next Tuesday, but I can trade with one of the other nurses."
"Come over to the house, we'll watch our new president be sworn in."
Friday, January 20
The pictures on the TV screen, taken from cameras stationed all through the nation's capitol, showed throngs of people ecstatically cheering as the car bearing President-elect Mehdi Ohmshidi drove by, headed for the capitol steps.
It is estimated that the crowd gathered in Washington for the inauguration of our nation's first ever foreign-born president numbers well over two million people.
The television reporter was speaking in breathless excitement.
The excitement is contagious and the atmosphere electric—enough to send a tingle running up this reporter's leg. History is being made here today. President-elect Ohmshidi is the first person ever to take advantage of the twenty-eighth amendment to the Constitution repealing Article Two, Section One and making any naturalized citizen eligible to be president of the United States. Think of it. America is now the world and the world is now America.
Jake was in his living room, eating popcorn and drinking a root beer as he watched the inaugural proceedings.
Jake had not voted for Ohmshidi, but then he had not really been enthusiastic about the other candidate either. His vote, as he had explained it to Karin, had been more against Ohmshidi, than it had been for Admiral Benjamin Boutwell, the former chairman of the Joint Chiefs of Staff. Jake had often declared that if he had omnipotent power he would replace everyone in government, regardless of their party, with someone new.
Ohmshidi had risen to national prominence as the federal prosecutor who tried the case against Masud Izz Udeen. Izz Udeen was an Islamic terrorist who released sarin gas into the ventilation system of Madison Square Garden, killing over seven hundred Americans.
As Izz Udeen received his sentence of death, he pronounced a fatwa against Ohmshidi and implored Muslims of the world to martyr themselves if need be in order to kill Ohmshidi. The fatwa against him, along with his successful prosecution of Izz Udeen, propelled Ohmshidi to national prominence, resulting in his election to President of the United States.
Jake watched as Ohmshidi stood on the steps of the nation's capitol building with his right hand raised, and his left hand very pointedly not on the Bible, but hanging by his side. The chief justice of the United States administered the oath of office, then concluded with, So help me, God.
Ohmshidi responded with, And this I, Mehdi Ohmshidi, affirm.
"Damn," Jake said aloud, speaking to himself. "What was that about?"
Ohmshidi moved to the microphone to present his inaugural address.
My fellow Americans. As your new president I make you this promise. It is not a campaign promise; it is not a mere statement of ambition. It is a promise that will be fulfilled. On this day we are embarking upon a world-altering journey that will bring about a new paradigm in American culture. This fundamental change will enable the poorest among us to share in the bounty of this, the wealthiest nation in the world. I will accomplish this goal by requiring more from those who have greatly profited by the opportunities offered them.
That means that the wealthiest among us will have to do their fair share in order to make all our citizens truly equal. But from their sacrifice will emerge a new order. Think of it. No more will there be people with no place to lay their heads, with no food upon their tables, without adequate health care, and with none of the finer things that make life worthwhile.
Such a thing has long been the goal of compassionate people, and in the past we have introduced welfare programs, food stamps, aid to dependent children, Medicaid, Medicare, and yes, even Social Security, to move in that direction. But any economist will tell you that all those programs have failed. I will not fail. We will have, before I complete my first four years, a universal program of shared wealth.
There was a light tap on the door, and when it was pushed open Captain Karin Dawes stuck her head in.
"It's me," she called.
"Come on in, Karin," Jake invited. "You're late. I've eaten almost all the popcorn already."
Karin walked over to the refrigerator and opened the door. "Don't you ever buy any kind of soft drink except root beer?"
"There is no soft drink except root beer."
"What a deprived life you have lived," Karin said as she grabbed one. "What have I missed?" She settled on the sofa beside him, pulling her legs up and leaning against him.
"Not much. Ohmshidi just admitted that he was a communist."
Karin popped the tab, and the root beer can spewed a fine mist. "You're kidding me!"
"Well, he as much as did. He's talking about sharing the wealth."
"Oh, that's all. Now, tell me the truth, Jake. Wouldn't you like to have some of Bill Gates's money?" Karin asked as she took a swallow of her drink.
"Not unless I did something to earn it. I believe in a fair wage for honest work, but I certainly don't believe in taking money from the successful to give to the losers who voted for this bozo."
"Come on, give him a chance. He hasn't been president for more than an hour, and you are already picking on him."
"It took him less than fifteen minutes to show his true colors," Jake said. "And forget the people who were calling him a pinko during the election. He isn't pink; he is red through and through."
Karin laughed. "Jake, I can't believe you are such a troglodyte. Just calling you a right-wing wacko doesn't quite get it. You are to the right of Attila the Hun. Are all Amish that way?"
"If you mean do they want to do for themselves, the answer is yes. And I agree with them. I didn't abandon everything the Amish believe in when I left the Life," Jake said. "I'm still a strong believer in the idea of individual self-reliance, rather than depending on the government for everything."
"He can't be all that bad," Karin said as she turned her attention to the TV screen. Ohmshidi was still talking.
"I thought you told me you didn't vote for him."
"I didn't vote for him, but millions of Americans did."
"I know. That's what frightens me."
"You will share the wealth tonight though, won't you?"
"What do you mean?"
Karin laughed. "Pass the popcorn. Unless part of your self-reliance means I have to pop my own."
Jake passed the popcorn bowl over to her. "Just listen to him," he said. "Every time he opens his mouth, he sticks his foot in it."
During my campaign, I promised you a transparent presidency, and, in adhering to that promise, I will keep you informed of my every action. So I am issuing now, in this inaugural address, my first executive order.
"Damn," Jake said. "An executive order in the inaugural address? I don't think that's ever been done before."
"You can't say he isn't up and running," Karin replied.
For too long the United States has been perceived by the rest of the world as a nation with an intrusive military presence. Since World War Two we have maintained a significant and, for much of the world, intimidating force in other countries. Therefore on this day, as my first official action as president, I am ordering all American troops, wherever they may be, to return to the United States. From this date forward, we will have no deployed forces anywhere in the world.
"Whoa!" Jake said, leaning forward. "What did he just say?"
"He said he is bringing all the troops back home."
"And do what with them? Where are they going to go?" Jake asked.
"I guess that means I won't ever make it to Germany," Karin said.
"That is pure insanity," Jake said. "If this is the first thing he does, where do we go from here?"
Karin picked up the remote and turned off the TV. "I am getting concerned about you. I'm afraid that you are going to get so mad watching this guy that you may well have an intracerebral hemorrhage."
"A what?"
"A stroke. It is my medical recommendation that we forget about him and go out for dinner."
"You're right. Even if I don't have a stroke, watching this commie bastard is going to make my head explode," Jake said. "We'll go out, but I choose."
"What do you mean, you choose? I'm the one who suggested we go out."
"That's just it. You suggested it," Jake replied. "That's like being challenged to a duel; the one challenged gets to choose the weapons. In this case the one invited to go out gets to choose the restaurant."
"I've never heard that. Is that some Amish rule?"
"Don't be silly, I never ate in a restaurant in my life until I was an adult. It's just the rule of common sense. You made the suggestion we go out, I get to choose where we go."
"Don't tell me," Karin said. "You are going to choose Bubba's All-You-Can-Eat Catfish Heaven, aren't you?"
"It is a great place, isn't it?"
"Oh, yes, it's just wonderful," Karin said, rolling her eyes.
"I'm glad you like it too," Jake said, purposely disregarding her sarcasm as he reached for his car keys.
"Jake, have you ever thought of maybe going to a quieter, more traditional restaurant where they have real silverware, elegant crystal, fine china, good wine, and maybe a strolling musician? You know. Something romantic?"
"You know what is romantic?" Jake asked.
"What?"
"Catfish fried golden brown, steaming-hot hush-puppies, a plate full of French fries liberally doused with hot sauce, a side of sliced onion, a dill pickle spear, and an ice-cold root beer."
"How can you possibly say something like that is romantic?"
"Because it is beautiful," Jake said. "And isn't romance supposed to be beautiful?"
"You are incorrigible."
"Not really. I'm just hungry," Jake replied as held the door open for her.
She laughed. "All right, Bubba's All-You-Can-Eat Catfish Heaven it is, then."
The restaurant was noisy and filled with customers, many of whom were soldiers from Fort Rucker. Half a dozen waiters scurried among the tables carrying trays upon which there were platters piled high with fried fish. Over in one corner a group of soldiers were doing their rendition of "All Out of Love," the singing discordant and loud.
"You wanted music," Jake said, nodding toward the table of singing soldiers. "You've got music."
"Oh, that's lovely. You think of everything," Karin said.
"I try."
"Any aftereffects from your incident yesterday?" Karin asked.
Jake took a swallow of root beer before he answered. "He tried to kill me, you know."
"What? Who tried to kill you?"
"The flight student I had yesterday. Oh, he might pretend that those geese hit us, but I know better. He went out looking for them. Every student I have ever had has tried to kill me. Oh, yeah, they all say they are just making honest errors, but I know better. I sincerely believe that it is a conspiracy."
"I'm sure it is. A left-wing conspiracy, no doubt," Karin said. "Every flight student you have ever had has been a part of the left-wing conspiracy."
"That's true. But it isn't just the flight students. I mean, think about this. Ever since I got my wings, people have been trying to kill me. Did you know that in Iraq and Afghanistan, they were actually shooting at my helicopter?"
Karin laughed. "As I recall, you were flying an Apache while you were in Afghanistan, and doing quite a bit of shooting of your own. You didn't get your Distinguished Flying Cross for making sightseeing trips."
"Still, you would think they would have more respect for a disparate collection of oscillating parts that, somehow, manage to levitate." Jake held up his right hand to call one of the harried waiters over, even as he was using his left to push another piece of fried catfish into his mouth.
"We'll need another platterful," he said when the waiter came over.
"We don't need another whole platter unless you are going to eat them all yourself. I'm absolutely stuffed," Karin said.
"No sweat, I'll eat them."
"Do you ever fill up?"
"Eventually," Jake answered.
CHAPTER THREE
Monday, January 23
Major General Clifton von Cairns, the commandant of Fort Rucker, used one of the larger classrooms to have an officers' call for all department, division, and section chiefs to talk about the troops that would be returning to the States. He admitted, during the meeting, that he had no idea what this would portend. The problem would be in finding billets for all of them.
"We don't have space for them, not in our CONUS TO and E units, and not in our training commands. Department of Army has asked every post commander to inventory their facilities with an eye toward absorbing the influx."
"General, will we be able to handle such an increase ?" a colonel asked.
"Yes, of course. We had much larger numbers of troops in garrison during World War Two. Of course, we also had a lot more military posts then. The problem now is that since BRAC, so many posts have been closed down in the last several years that it is going to make it difficult."
"How long has DA known about bringing all the troops back to CONUS?" another colonel asked. "What I mean is, why didn't they give us prior warning?"
General von Cairns looked at the colonel with an expression that mirrored his frustration. "Colonel Haney, from what I was told by the Army chief of staff this morning, Department of Defense learned about this at the same time we did: when the president announced it during his inaugural address."
Monday, February 27
Hello, America. George Gregoire here.
In his inaugural address, Mehdi Ohmshidi stated his intention to bring back to the United States every uniformed American stationed overseas. Well, he has done that. So, let's take a look at what has happened.
All military training in America has come to a complete halt. The bases are overcrowded. There is no place to put the returning military, not in any training capacity, nor in any operational unit. Morale has sunk to an all-time low as officers and men report to work daily, but with no real work to do.
And what has happened overseas ? With America's withdrawal from NATO, all NATO operations have come to a halt. Terrorism has increased in Europe, and in the Middle East.
Since 1945, U.S. troops in Korea have helped support the South Korean military, providing the security needed to lift that country from the third world to one of the economic giants of the world. But with the withdrawal of American troops, North Korea has become much more adventurous, last week sinking two South Korean fishing vessels and, just yesterday, penetrating the buffer zone that separates the two countries and killing three South Korean border guards.
But it isn't just Ohmshidi's foreign policy that is failing. Let me ask you this. How is this universal program of shared wealth working out for you ?
Let's do give Ohmshidi credit for establishing a degree of equality in the nation. He has not been able to improve the plight of the poor, but he has been quite successful in bringing down the living standards of the rest of us. And I'm not just talking about the wealthy; I'm talking about working Americans. In barely over one month, the value of the dollar has fallen by eighteen percent.
I don't mind telling you, friends, I don't see this situation getting any better. In fact, I see it getting worse, much worse. I will do my part. I will ask the bold questions and I will always tell you the truth.
Just over one month after Ohmshidi ordered all overseas military to return to the continental United States, Fort Rucker was filled to capacity with returning soldiers, and in order to accommodate the influx, all training activities were suspended. It would have been difficult to continue training activities anyway, because, in a cost-cutting measure, the Department of the Army was now regulating the total number of hours that could be flown in any week. Once Fort Rucker got their allocation, it would hold the hours in a pool and pilots who needed flight time for pay purposes would have to apply for that time. The problem was there were not enough hours allocated to the fort to enable all the pilots to make their minimums, and as additional rated officers arrived, the situation grew even more critical. General von Cairns had been correct in anticipating difficulty in completing the flight program and his recommendation to Jake to expedite the last cycle barely enabled the twelve students to complete their courses.
At the post hospital, Karin was having her own problems. Reductions in Medicare and Tricare denied civilian health care to military retirees and their families, so they were remanded to military base hospitals. As a result the case load at the hospital was greatly increased. VA facilities all across the country were closed, and those who were eligible for VA benefits were instructed to go to active-duty military hospitals on a space-available basis only.
This increased patient load meant that Karin was working longer, harder hours each day, and was often too tired to visit Jake. But tonight she came by, bringing hamburgers and French fries from a local drive-in.
"Do you have any idea how much two hamburgers and two orders of French fries cost?" she asked. Then, without waiting for a response, she answered her own question. "Eighteen dollars! Can you believe that?"
"Everything has gone up," Jake said. "In order to have enough to meet all the new projects, the government has begun printing money hand over fist. Gregoire says that the presses are running nonstop. Obviously the more money you have in circulation, the less value it has, so that things that have real value, like food, are seeing drastic increases in price. I bought two twelve-packs of root beer this afternoon on the way home from work. Forty-two dollars."
"You might want to think about giving up your root beer," Karin said.
"I'll give up my root beer when they pry the last can from my cold, dead fingers," Jake teased.
"Jake," Karin said, the smile on her face replaced by an expression of concern and even a hint of fear. "Where is all this going? What is going to happen to us?"
"I don't know, Karin. God help us, I don't know."
"That's not what I wanted to hear."
"What did you want to hear?"
"I wanted to hear you say that everything will be all right."
Jake was quiet for a moment; then he sighed. "Karin, for us—for you and me—everything will be alright. But there is no way the country is going to get through this without serious, serious consequences."
"How can you say then, that it will be alright for you and me?"
"Because I will make it alright for you and me," Jake said. "That is a promise."
"Get the drinks, let's eat," Karin said, wanting to change the subject.
"How about root beer?" Jake suggested as he started toward the refrigerator.
"Root beer? I don't know, let me think about it. Ummm, yes, I think I would like a root beer."
Jake brought the drinks into the living room and put the cold cans on coasters on the coffee table in front of the sofa.
"Did you fly today?" Karin asked as she handed a hamburger and fries to Jake.
"Nobody flew today," Jake answered as he unwrapped his burger. "I haven't flown in two weeks. We are limited to one thousand hours per month, and as of now there are only six hours remaining in this month's flight-hour pool. You know how many rated aviators we have on this post?"
"A lot," Karin replied.
"We had almost a thousand before the influx of troops from overseas, and that added at least two hundred more. That means there are twelve hundred pilots who are now in queue for six hours. That breaks down to eighteen seconds of flight time apiece."
Karin laughed, spewing root beer as she did so. She wiped her mouth with a napkin.
"It's not funny," Jake said. "If aviators can't keep up their minimums, aircraft are going to start falling out of the sky because the pilots aren't going to be safe."
"I know it's not really funny, Jake," Karin said. She laughed again. "But I'm just picturing someone getting into a helicopter for eighteen seconds." She hopped up from the couch. "I'm flying," she said. She plopped back down on the couch. "Oops, time is up."
"It's not funny, damn it," Jake said, but despite himself, he laughed as well.
"I'm going to turn on TV," Karin said. "Kentucky is playing LSU tonight."
"What do you care? You're not running cross-country for Kentucky anymore," Jake said. "And as far as I know, you are no longer a cheerleader."
"You think I'm not?" Karin replied. Getting up from the couch and standing flat-footed on the floor, she did a backflip, tucking in her legs at the top of the flip because she went so high that her feet would have hit the ceiling. Landing on her feet, she thrust her pelvis forward and held her arms over her head.
"Now, imagine me in my cheerleader outfit," she said.
"You're making me blush. Be nice now," Jake said.
"Are you sure you want me to be nice?" Karin asked, seductively.
"Maybe not that nice," Jake answered, pulling her to him for an open-mouthed kiss.
Gulf Shores, Alabama—Thursday, March 1
Bob Varney, chief warrant officer–4, United States Army retired, got a cookie and a cup of coffee from the welcome counter at the bank, then had a seat until he could speak to one of the bank officers.
"Bob?" Joel Dempster called, sticking his head out of his office.
Having finished both his cookie and the coffee, Bob dropped the paper cup into a trash can, then went into Joel's office.
"I read Summer Kill and Death Town. They were great," Joel said. "When is your next book coming out?"
"Within a month. It's Murder in Milwaukee. I'll be signing at the Page and Canvas in Fairhope when it comes out."
"Don't know if I'll be able to get there, but I'll for sure buy it."
"Thanks."
"I think Hollywood should make a movie of one of your books."
"From your lips to God's ear," Bob said.
"Now, what can I do for you?" Joel asked.
"I was just wondering. I went online to check my account; I didn't see the deposit for my Army retirement."
"Yes, I thought that might be why you were here. If it is any consolation to you, it isn't just you, Bob. There was no deposit for anyone. We got a notice from DFAS that all transactions are being halted while they undergo reorganization."
"Wow. Really? Everyone?"
"Everyone. You are lucky. With your writing you have another income, a good income, I might add. But as you know there are a lot of military retirees here. And many, if not most of them, depend entirely upon their military retirement and Social Security."
"I didn't even check for Social Security."
"Don't bother, there was no deposit for it, either."
"That's not good," Bob said.
"I'll tell you something else that isn't good. We have been ordered to submit a report to the federal government providing information on the amount of money every depositor has in all accounts."
"Are you going to do that? Do they have the authority to make you do it?"
"As long as we participate in the FDIC program, we have no choice but to comply."
"Maybe I should take out what I've got in there," Bob said.
"You can't."
"What do you mean, I can't?"
"You just deposited a royalty check earlier this week, didn't you? A rather substantial check?"
"Yes, it was for signing four contracts, and delivery and acceptance of a completed book. A little over forty thousand dollars."
"At this point any withdrawal, or check, in excess of ten thousand dollars, must be approved by the federal government."
"Why?"
"I'm sure you have noticed that the economy is a little shaky now, and is getting worse almost by the day. I think this is to prevent a run on the banks."
"Is the money safe?"
"It is as safe as money is safe," Joel said. "The problem is, how secure is the American dollar? I've been hearing things through the grapevine that make me wonder."
"Now you are getting me scared," Bob said. "First you say there are no retirement or Social Security payments, then you say I can't get the money I do have out of the bank. Joel, what the hell is going on?"
"I wish I could tell you, Bob, I really do. I've talked to all the other bankers; we are very worried about this. Banks are only as good as the service they are able to provide to their depositors. When you start breaking that trust, then you are putting into jeopardy a bank's ability to function. If I were you, and I'm cutting my own throat by telling you this, but if I were you, I would withdraw nine thousand, nine hundred and ninety-nine dollars and ninety-nine cents. As long as you don't go to ten thousand dollars on any one transaction, you are safe."
"Thanks, Joel, I guess that's the route I'll take."
"Then come back tomorrow and do it again, keep doing it until your account has just a few cents in it."
"I appreciate you telling me that, Joel," Bob said. "I'll do that too."
"Just write the check here, I'll cash it. There's no sense in causing anyone to get curious. And, if you would, be careful about who you tell this to."
"I will," Bob promised. "And again, thanks."
Bob wrote the check and handed it to Joel. Joel left the office, and returned a moment later with the cash in a bank envelope.
"Are you going to rent your house this summer?" Joel asked as he handed the cash to Bob.
"I don't know if we are, or not," Bob replied. "By this time last year, we had eight weeks rented already. So far this year, we don't have so much as a single nibble."
"I guess folks are a little frightened of what's ahead," Joel said.
"Yeah, it sure looks that way."
Bob got up and stuck his hand out toward Joel. "I appreciate what you are doing for me, Joel."
"You're a good customer and an interesting guy," Joel said. "And, I wouldn't worry too much about things. I'm sure it's all going to work out."
"If not, we'll just whistle past the graveyard, eh, Joel?"
Joel laughed out loud. "Sounds like a good plan," he said.
Fort Rucker—Thursday, March 1
From the office of the Commanding General, Fort Rucker, Alabama
Subject: Flight Time
Suspense Date: With immediate effect
**1.** All facility aircraft are herewith grounded. No flight will be authorized unless it is an emergency flight.
**a.** Emergency refers to national emergency only.
**b.** Authorization for emergency must come from Department of Defense.
**c.** Said authorization will require authenticator.
**2.** All aviators are hereby ordered to submit flight logbooks showing most recent flying time for analysis of flying patterns.
**3.** All aircraft maintenance logbooks will be surrendered to flight operations, and all aircraft will be rendered non-flyable by removing lines from fuel tank to fuel controls.
**4.** Flight school students who have less than 200 flying hours will be dismissed from the course and reassigned to non-flying billets.
**5.** Flight school students with more than 200 flying hours will be subject to flight instructor's evaluation for further disposition. If recommended by flight instructor, they will be awarded the wings of a rated aviator.
Distribution:
By Electronic Transfer
For the Commander MG Clifton von Cairns
Joseph A. Wrench
LTC
Avn
Adjutant
"What is this nonsense, Major?" Captain Greenly asked. Greenly was one of the instructor pilots in Environmental Flight Tactics. "All aircraft are grounded?"
"That's the way I read it," Jake replied.
"For how long?"
"That I can't tell you, Len," Jake replied. "But I can tell you, it doesn't look good."
"I've only got one student who meets the two hundred-hour requirement, but hell, that's because he's so uncoordinated he had to fly extra hours just to keep up. My two best students are still fifteen hours short. What am I going to tell them?"
"Tell them the truth," Jake said. "At this point it doesn't matter whether they have their wings or not. Nobody is flying."
"Have you thought about where that leaves us?" Greenly asked.
"What do you mean?
"I mean if nobody is flying, there isn't much need for flight instructors, is there?"
"I see your point."
Greenly sighed, and ran his hand through his hair. He was a veteran of two tours in Iraq and one in Afghanistan.
"Jake, I've got forty-five days of leave time accrued. If you have no objections to it, I think this might be a good time to take leave."
"Why burn your leave time?" Jake asked.
"What do you mean?"
"Why don't you just take off? Check in with me every week or so and let me know how to get ahold of you if anything comes up."
"You think that would be all right? I mean, to just leave without leave papers?"
"I don't know where all this is going, Len, but I seriously doubt that the Army will even know you are gone. I'm certainly not going to tell them."
Greenly smiled. "All right," he said. "Maybe I'll just do that. Drop in on the folks back in Kansas, unannounced." Greenly started toward the door; then he stopped and looked back toward Jake. "Jake, what would you say if I told you I've been thinking about resigning my commission?"
"You're a good man, Len. You've been a fine officer and an asset to Environmental Flight. I hate to say it, but, with the way things are, I would say that I can understand why you might."
Greenly stepped back into the room, then reached out to shake Jake's hand. "Good-bye, Jake," he said.
"Good-bye, Len."
Greenly came to attention and snapped a sharp salute. Jake returned it, then Greenly did a crisp about-face and left the room.
Jake stared at the empty doorframe for a moment; then he sat down and looked up at the wall. It was filled with photographs of Army helicopters from the H-13s and H-19s of Korea, to the Hueys, Chinooks, and Cobras of Vietnam, to the Blackhawks and Apaches of Iraq and Afghanistan. There was also a picture of a CH-47 Chinook in Afghanistan, and under it was a caption:
Yes, the Chinook is still here.
Nothing can ground this bird.
"Nothing except a dumbass president," Jake said aloud.
CHAPTER FOUR
Raised in the Amish community, Jake had no reason to believe that he would not be a farmer like his father, grandfather, and great-grandfather. Like all Amish boys, he had learned the skills necessary to live in a world that shunned modern conveniences. He was a good carpenter, he knew farming, he understood nature and knew what wild plants could be eaten and what plants would have medicinal value.
But even as a child he used to watch airplanes pass overhead and wonder about them. One day an Army helicopter landed in a field nearby. The occupants got out, opened the engine cowl, and made few adjustments, then closed the cowl, got back in, and took off. Jake knew, on that day, that he wanted to fly a helicopter. He also knew that such an ambition was not for an Amish boy.
When he was eighteen years old, Jake, like all other eighteen-year-old Amish, went through rumspringa, a period of time in which they were exposed to the modern world. Once this coming-of-age experience was over, the Amish youth would face a stark dilemma: commit to the Amish church—or choose to leave, which meant severing all ties with their community and family forever. Jake made the gut-wrenching decision to sever those ties.
Because of that, he was excommunicated from the church. Being expelled meant being shunned by everyone, including his own family. When he went back home, in uniform, after graduating from Officer Candidate School, his mother and father turned their backs and refused to speak to him. His sister shunned him also, but he saw tears streaming down her face and he knew it was not something she wanted to do.
After OCS, Jake went to college on the Servicemen's Opportunity College program, getting his BA degree from the College of William and Mary in two and a half years. After that, Jake attended flight school, fulfilling his ambition to be a pilot. His love for flying was not diminished even though he had three combat tours: one to Iraq and two in Afghanistan. There he flew the Apache armed helicopter and was awarded the Distinguished Flying Cross as well as the Air Medal with "V" device for heroic action against the enemy. He also received a Purple Heart when a shoulder-launched missile burst just in front of the helicopter, killing his gunner/copilot and opening up gaping wounds in Jake's face, side, and leg. He managed to return to his base, but had lost so much blood that when he landed he passed out in the helicopter, not regaining consciousness until he was in the hospital. That was where he met the nurse, Karin Dawes, who was then a first lieutenant.
Jake had never married, partly because before he met Karin, he had never met anyone he wanted to marry. He had been giving a lot of thought to asking Karin to marry him, but with the nation in turmoil, he wasn't sure it was the right thing to do.
Lancaster, Pennsylvania—Wednesday, March 14
It was the first leave Jake had taken in almost two years and because of that he had well over fifty days of leave time accrued. He took fifteen days, convinced Karin to take leave with him, but told her that the first thing he wanted to do was look in on his family if they would receive him. They flew to Harrisburg, Pennsylvania, and there, Jake rented a car for the drive to Lancaster.
"How long since you have been home?" Karin asked as they drove east on I-78.
"Twelve years."
"Twelve years? That's a very long time," Karin said.
"Yes, it is. Even longer when you realize that the last time I was home my parents wouldn't even speak to me."
"Oh, that's awful," Karin said. "What makes you think they will have anything to do with you now, if they didn't before?"
"Over the last couple of years, my sister and I have exchanged a few letters," Jake said. "She said that my dad has mellowed some."
"What about your mother?"
"I'm my mother's son," Jake said. "I think she was as hurt by my father's shunning of me as I was. But she must do what he says, so she had no other choice. Otherwise she would have accepted me back the very first day."
"That brings up another question. Why are you going back now?"
Jake was silent for a long moment.
"Jake?" Karin repeated, not certain that he heard her.
"I don't know that I can answer that question," Jake said. "It is just something that I feel I must do. Especially now, with our nation on the brink of disaster."
"I can understand that," Karin said. "I hope things go well for you."
"Karin, you understand why I can't take you with me to meet them, don't you? It would be . . ."
"Very awkward, I know. You don't have to explain, Jake. I understand." Karin reached her hand toward him and Jake moved his left hand to the steering wheel, then took her hand in his and lifted it to his lips.
They checked into a motel in Lancaster, and Jake brought their suitcases into the room. From his suitcase he removed a pair of gray, drop-front trousers, work boots, a blue shirt, a pair of suspenders, and a flat-brim hat. He stepped into the bathroom, and when he came out a moment later, the difference in his appearance was startling.
"What are you wearing?"
"Plain clothes," Jake said.
"You look—very handsome," Karin said.
"This is going to be hard enough without you teasing me," Jake said.
"I'm not teasing," Karin insisted. "You look very masculine and, I don't know how to explain it, but, very sexy, in an earthy way." She stepped up to him, then put her arms around his neck and pulled his lips down to hers for a deep kiss.
"You've got me all hot and bothered," she said.
"Can you hold on to that feeling until I get back?"
"It'll be hard," Karin said.
"I promise you, it will be," Jake replied with a bawdy grin.
Karin laughed out loud. "That was a statement, not a question. But go on, do what you must do. I'll find something to watch on television while you are gone."
Jake held up his cell phone. "I've got my phone," he said. "I'll call you when I can."
"I'll be fine," Karin said. "Don't worry about me."
Leaving the motel, Jake drove east on Old Philadelphia Pike; then, crossing the railroad, he turned left on Beachdale Road. Just before he reached the farm of his parents, he saw a large gathering of buggies and wagons parked at the Yoder farm. At first he wondered what was going on; then he saw that they were building a barn.
Jake stopped his car, and walked over toward the barn. They had just finished assembling the frame for one end and several were in position to lift it up.
"Wir brauchen jemanden, hier heraufzukommen und eine Hand zu verleihen," someone called from the bare eaves of the barn.
It was the language of Jake's youth, a request that someone come up to lend a hand.
"Ich komme," Jake answered, and he scurried up one of the ladders, then got into position. Half a dozen ropes were thrown to the men on top of the barn frame, and Jake grabbed one of them, and pulled with the others as the end frame was raised into position.
Soon the framework of the barn was all in place and now the only thing that remained was to fill in the siding and the roof. There were at least twenty men working, so the barn was erected with amazing speed. After all the siding and roofing was completed, everyone grabbed a paintbrush and bucket of paint and, within three hours after Jake arrived, the barn had been erected and painted, and all the scrap lumber around it picked up and thrown into the back of the wagons.
During the time the barn was being erected, the women were preparing a meal, and now one of the women began ringing a bell. It was still too cold to eat outside so everyone tromped into the house, where tables had been set up in the dining room, the living room, and in the big, central hall.
It had been more than fifteen years since Jake was last in this house, but it could have been yesterday. Every piece of furniture, every wall hanging, was exactly as it had been the last time Jake was here. He remembered that he had been here for the funeral of the elder Yoder.
Moses Yoder gave the blessing.
" Unser himmlischer Vater, I ask that you bless these wonderful people today for their generous hearts, helping hands, and loving souls. And we thank you for the women who prepared the meal that will sustain us through this day of toil. Segnen Sie dieses Essen zu unserer Verwendung, und wir zu Ihrem Dienst. Amen."
There had been very little conversation among the men during the work, except for that which was required to do the work. Nobody had called Jake by name, which meant they either didn't recognize him, or were continuing the shunning. As Jake passed the mashed potatoes to the man on his left, he saw his father sitting across the table at the far end. He had seen his father earlier while they were working, and he was sure that his father had seen him, but his father had made no sign of recognition. Now his father was looking directly at him. Jake nodded at his father, but his father didn't acknowledge the greeting.
After the meal, Moses Yoder stood at the front door and shook the hand of every man as he left, thanking each of them for helping him replace the barn that had been destroyed by fire.
"You have come from another order?" Yoder asked as Jake was leaving. "I thank you for your help." Yoder did not recognize him.
"It is the way of the Christian," Jake said, implying, though not saying, that he was from another order.
"Indeed it is, brother, indeed it is," Yoder said, putting his other hand on top of Jake's to add emphasis to his greeting.
When Jake stepped outside he saw his father standing under a tree, obviously waiting for him.
"Jacob," Solomon Lantz said. "Have you returned to the Life?"
"I have not, Father," Jake said. "Too many things have happened; I am too far removed to return to the Life now."
"But you are wearing plain clothes, and you helped with the labor."
"These are things you taught me, Father. I may have abandoned the Life, but I have not abandoned your teachings."
"Das ist gut," Solomon said.
"Father, I have been writing to Martha, and she has written me back," Jake said, speaking of his sister. "Did she tell you that I wanted to come and visit?"
"Ja."
"I do not wish to make you uncomfortable. If you want me to leave, tell me, and I will go away now."
"You come," Solomon said. "Your mother will want to see you. And your sister too."
Jake drove behind his father's carriage, following him very slowly for the two miles that separated the Yoder farm from his ancestral home. This was where he was born and where he grew up. As a child, the big white house, the barn, and the workshop had seemed quite normal to him. Now he viewed them with the eyes of an outsider, and recognized, for the first time, how beautiful they really were.
He parked under a big oak tree and got out of the car, waiting until his father went into the barn to unhitch the horse from the closed buggy. Solomon walked quickly from the barn to the house; then, after a couple of moments, his father stepped back out onto the wide front porch.
"Come, Jacob. Your mother waits," Solomon called.
As soon as Jake stepped in through the front door his mother greeted him with a great hug, pulling him as close to her as she could. Jake could smell cinnamon and flour during the embrace—aromas of his childhood—and it was almost as if he had never left.
"Jacob, Jacob, mein Lieblingskind, willkommene Heimat, willkommene Heimat," she said, weeping with joy as she welcomed him home.
Looking up, Jake saw his sister, Martha, and when he finished his embrace with his mother, Martha came into his arms. There was a small boy standing in the doorway that led from the entry hall into the parlor.
"And who is this?" Jake asked.
Shyly, the boy stepped behind the wall, and peeked around the corner of the door.
"This is my son, Jacob," Martha said.
"I did not know that you were married and had a child," Jake said. "How wonderful for you."
"Jacob, this is your Uncle Jacob," Martha said.
"His name is just like mine," young Jacob said.
Martha laughed. "That is because you are named after him."
"Where is his father?" Jake asked. "I would like to meet him."
Tears came to Martha's eyes. "He took sick and died, last year," she said.
"Oh, Martha, I am so sorry. Who was the father?"
"It was Emile Zook."
"Emile!" Jake said. "Emile is dead?"
Emile had been Jake's best friend as he was growing up.
"Yes."
"But, Martha, in your letters to me you said nothing of this. You did not tell me you were married, you did not tell me that it was to Emile, or that Emile had died."
"I did not want to trouble you with tales of your old life," Martha said. "If you decided to come to see us, I wanted it to be because you wanted to, not because you felt that you should."
Jake put his arms around Martha again, and pulled her to him. "I am sorry for your loss, my sister," he said. "Emile was a very good man. I have missed him these many years, and I will miss him the more, now that I know he is gone."
Although Jake had eaten an enormous meal at the Yoder farm, his mother insisted that he have a piece of apple pie and a cup of coffee. She served the pie hot, with a slab of melted cheese on top.
"Father, you have spoken little," Jake said.
"There is a reason God gave us two ears and one mouth," Solomon said. "It is because we should listen twice and speak once."
"Yes," Jake said. "I remember this is one of the things you taught me."
"Why have you come home?" Solomon asked.
"To see you, and mother, and my sister," Jake said. Then, looking over toward young Jacob he added, "And my nephew."
"You come home, but you do not stay. Sie sehen nicht, dass diese Mittel mehr Schmerz für Ihre Mutter?"
"It was not my intention to bring more pain to my mother," Jake said, answering his father's charge.
"Are you still in the Army?" Solomon asked.
"Yes," Jake answered. "Such as it is. The way things are going in the country now, this new president we have is doing everything he can to destroy the Army."
"Who is the new president?" Solomon asked.
"Father, I can't believe . . ." Jake started, but he stopped in midsentence. There was no television in this house, no radio, and no electricity. There was no news beyond Lancaster that could possibly be of any importance to Solomon Lantz.
"His name is Ohmshidi," Jake said.
"Ohmshidi? What sort of name is that for an English?"
Jake knew that by the term English, Solomon was referring only to those people who are not Amish. It really had no bearing on nationality.
"He is Pakistani," Jake said. "He has an American mother, but he was born in Pakistan and is a naturalized American."
For the next half hour Jake explained the condition of the country to his parents, sharing his fears that the president was only making matters much worse.
"It is not our concern," Solomon said.
"I hope you are right, Father. Believe me; I have never envied the Life more strongly than I do now. How I wish I could live here like everyone else in total ignorance to the world outside."
"Have you been to war, Jacob?" Solomon asked.
"Yes."
"Have you killed?"
"Father, that's not a fair question. Wars are fought as a matter of executive decisions. The men who fight them, men such as I, have absolutely no input into the decisions."
"Have you killed?" Solomon asked again.
Jake was quiet for a long moment. "Yes," he finally said. "I have killed. It was not something I wanted to do, but it was something I had to do."
Solomon got up and walked over to a long table that sat under the window and stared outside, as if trying to come to terms with the fact that his son had killed. Jake looked at the table and remembered how, as a child, when it would rain outside, his mother would sometimes drape a quilt over the table and make a tent. That way Jake could camp in the rain without ever getting wet.
"You say that you are afraid the country will collapse," Solomon said. "What does that mean?"
"I think it will mean no law and order. It will also mean runaway inflation and if that happens, money will become worthless. We may see wide-spread electrical outages, fuel and food shortages," Jake said. "If all that comes to pass, there will be riots in the streets."
"Will you survive?" Solomon asked his son.
"I—yes, I think I will."
"You only think you will?"
"I will survive," Jake said. "I am worried about you."
"Worry not about me," Solomon said. "All the troubles of the English will not trouble us. For many generations we have lived our lives and the English have lived theirs."
"Father, I fear that things may be different now. This new president . . ."
"Is not of our concern," Solomon repeated.
"Hide your food, Father," Jake said.
"Hide the food? Why do you say such a thing?"
"If, as I fear, there is a total breakdown of civilization, the English know that Amish keep a lot of food stored. They may come try to take it."
"I will not turn away a starving man," Solomon said.
"They will be more than starving. They will be desperate, and they could bring much harm to you and to the others. Please, Father, heed my warning. Hide your food and tell your neighbors that they must do so as well. For if you don't, I fear what may happen to you."
"I will heed your advice," Solomon said.
"Thank you."
"Will you stay with us now?" Jake's mother asked.
"No, Mother, I wish I could," Jake said. "But I cannot."
There was no real reason why he couldn't stay, but it wasn't entirely a lie either, because he did wish that he could. But he had Karin back in Lancaster, and he did not want to leave her alone. Also he felt a very strong and totally unexpected attraction for the Life that he had abandoned so long ago. It was an attraction that he could not succumb to. He needed to get away now, while his resolve was still strong.
Jake stood then, and walked over to retrieve his hat from a hat rack that was on the wall just inside the front door. The hat rack was a thirty-inch-long, highly polished piece of walnut. Carved into the hat rack were the words:
Die Lantz-Familie
Jake ran his hand over the smooth wood.
"It was a Christmas present you made for me," Jake's mother said.
"Yes, when I was twelve years old."
"If go you must, do it now," Solomon said.
Jake's mother embraced him again, and he could feel her tears on his cheek. Martha embraced him as well.
Jake waited until he was back on Old Philadelphia Pike before he dialed Karin's cell phone.
"How did it go?" Karin asked.
"All things considered, it went well," Jake said. "They accepted me without shunning."
"I'm glad."
"Do you need me to pick up something to eat?"
"No, I walked across the street to a place that serves Amish food. Or so the sign says."
Jake chuckled. "That's for the tourists," he said. "It's pretty close, but it isn't real."
"Real or not, it was very good," Karin said. "How much longer before you get here?"
"Fifteen minutes, more or less."
"Are you still wearing your plain clothes?" Karin's voice took on a deep seductive tone.
"Of course. What else would I be wearing?"
"You'll need to get out of them."
"Yes."
"I'll help," she said.
CHAPTER FIVE
Fort Rucker—Thursday, March 15
General Clifton von Cairns swiveled around in his chair and looked through the windows of his office out onto the parade ground. He was the commanding general of an Army base whose sole reason for existence was to train aviators and aircraft maintenance personnel—but, by order of the Department of the Army, all training had been suspended until further notice. In the meantime he had over twenty thousand soldiers wandering around on the base with no specific jobs.
Worse, he had aviators who weren't able to fly, not even to maintain their minimums. He drummed his fingers on the arm of his chair for a moment; then he reached for the telephone, and dialed the direct line to the deputy chief of staff, U.S. Army, G1, at the Pentagon.
"This is General von Cairns. I would like to speak to General Roxbury," he said when the phone was answered.
"Yes, General, what can I do for you?" General Roxbury said when he came on the line.
"Tell me, Bill, just when in the hell do you think we will be able to resume training?" von Cairns asked.
"We've been through all that, Clifton. Training will be resumed as soon as we can get reorganized. We have brought three hundred thousand troops back from overseas, the largest part of that number being Army personnel. That has put quite a strain on our military infrastructure as I'm sure you can understand. And right now, our first priority is reorganization."
"Alright, I can see that, but why restrict our flying time? As you know, I am not only CG of Fort Rucker and the Army Aviation School; I am also chief of the Army Aviation Branch. These flight-time restrictions are Army-wide, and they are having a serious impact in allowing our aviators to maintain their minimums. And that, Bill, could have dire, and I mean dire, consequences."
"I wish I could help you with that, Clifton, I really do. But that is out of my hands. The restriction of flight time isn't just for the Army. It is for all branches of the service, and it comes direct from the secretary of defense."
"Yes, someone who has never served one day in the military, who has never held a private-sector job, and who has never been in charge of anything larger than an office staff. Can't you talk to him, Bill? Can the chief of staff talk to him? Hell, how about the chairman of the Joint Chiefs? He's an Air Force man, a pilot; he ought to understand better than anyone what this is doing to training, to operational readiness, to say nothing of morale."
"Believe me, he does understand. And he has talked with the secretary of defense as well as the president. But the flight limitations remain in effect."
General von Cairns was quiet for a long moment.
"You still there, Clifton?"
"Yeah," von Cairns said. "I'm here."
"Look, I know what you are going through," General Roxbury said. "And I'm doing—we are doing—all we can to get this situation resolved as quickly as we can. All I can say now is for you to just hang loose and keep your personnel ready to resume training at a moment's notice. This can't last forever."
"No, it sure as hell can't," von Cairns said. "Good-bye, Bill, and give Connie my best."
"Will do," General Roxbury said. "I'm going to stay on this, Clifton, I promise."
General von Cairns hung up the phone. Nearly one hundred soldiers were moving across the parade ground performing police call. But there was nothing left for them to pick up, because another group of one hundred had performed a police call earlier this morning. This useless waste of manpower was the result of junior officers and NCOs "making work," for the men and women soldiers on the base. The lack of mission was having a serious impact on troop morale.
Von Cairns looked over at a shadow box on the wall. Inside the box was a Distinguished Service Cross, a medal of valor second only to the Medal of Honor.
The framed citation was right beside it.
Citation: Distinguished Service Cross
CLIFTON VON CAIRNS
The President of the United States takes great pride in presenting the Distinguished Service Cross to Clifton von Cairns, Major, U.S. Army, Avn, for his extraordinary heroism in military operations against an opposing armed force while serving as an Apache pilot with the Third Combat Aviation Brigade during Operation DESERT STORM on January 21, 1991. On that date, Major von Cairns was flying a search-and-destroy mission at a forward-operating location when he received tasking to look for another Apache crew that had been shot down the night before. For three hours of intensive searching deep inside enemy territory, he risked his life as he had to fly at absolute minimum altitude to pinpoint the survivors' location. When an enemy truck appeared to be heading toward the downed crew, Major von Cairns engaged and destroyed it, thus enabling a Blackhawk helicopter to secure the rescue. Once the crew recovery was effected, Major von Cairns flew cover for the rescue helicopter, taking out two more enemy gun positions on the return flight. It was his superior airmanship and his masterful techniques at orchestration that made this rescue happen. Through his extraordinary heroism, superb airmanship, and aggressiveness in the face of the enemy, Major von Cairns reflects the highest credit upon himself and the United States Army.
The door to his office being open, Lieutenant Phil Patterson stepped through it, then called out, "General von Cairns?"
Von Cairns swiveled back around. "Yes, Lieutenant ?"
"You wanted the 1352 forms? The Matériel Readiness Reports?"
"Yes, what do we have?"
"Forty-two percent of our aircraft could be made flyable by reconnecting the fuel control lines."
"What? Only forty-two percent? What's the problem ? You would think with as little flying as we are doing that we could at least keep our aircraft operational."
"Yes, sir, well, it isn't the fault of maintenance, General. Fifty-one percent of the red-X aircraft are grounded for parts."
"Fifty-one percent? What's the holdup? Are the parts on order?"
"Yes, sir, and they are on AOG, which as you know is the highest priority," Lieutenant Patterson said. "Evidently there is a hold on all resupply."
"How much authority can AOG have now, anyway ?" General von Cairns asked. "Hell, all aircraft are on ground."
"That's true, General. But I suppose that is the best they can do."
"Yes, I suppose it is," von Cairns agreed. "Thank you, Lieutenant."
Patterson returned to his desk. He had welcomed the assignment to run down the 1352s. That had given him something to do other than sit at his desk and read paperback novels.
Opening the middle drawer on his desk, he picked up Death Town by Robert Varney. The other officers sometimes teased him about his "high literary tastes," but he didn't care. He enjoyed the thrillers, and the way things were going right now, he needed all the escape he could get.
JFK Airport, New York—Friday, March 16
Pan World America Flight 103, out of Frankfurt, Germany, was just entering New York airspace for landing at JFK. Rena Woodward, the chief flight attendant, took the mic from its holder.
"Ladies and gentlemen, the captain has just turned on the seat-belt sign. Please make certain that all trays are stowed, your seats are in the upright position, and your seat belts are fastened. We thank you for flying PWA."
Suddenly, from the aisle seat in row twenty-three, Abdullah Ibrahim Yamaninan stood up and, using a cigarette lighter, lit the hem of his shirt.
"Death to all infidels! Allah hu Akbar!"
The shirt flamed as suddenly and as brightly as a magnesium flare.
"He's got a bomb!" someone shouted.
Mike Stewart, a former linebacker for Penn State, was a passenger one row behind. He grabbed the blanket off the woman who was seated next to him, then leaped up and wrapped it around Yamaninan, knocking him down as he did so.
"Get a fire extinguisher!" Stewart shouted as he tried to smother the flames with the blanket.
Yamaninan was screaming in pain from the severe burns all over his body. Reena arrived then with a foam fire extinguisher and she emptied it on Yamaninan and Stewart, whose clothes were, by now, also burning.
One of the other flight attendants called the flight deck to inform the pilot of what had just happened, and the pilot called JFK to declare an emergency.
"Pan World, what is the nature of your emergency ?"
"It appears we have a bomber on board."
"Say again, Pan World. You have a bomb on board?"
"A bomber. Or a would-be bomber. Apparently he tried, but it did not explode."
"Pan World America, you are cleared for immediate landing on runway 13R, winds north–northwest at twenty knots, altimeter two niner niner seven."
"Pan World, 13R, roger."
"All inbound aircraft to JFK, be advised there is an emergency in progress. Northwest on short final for 13R. Please expedite your landing, exit runway at first opportunity. All other aircraft in queue for 13R go around for reassignment."
"Northwest, roger, expediting approach."
"JFK, this is Pan World, we request emergency equipment on site."
"Roger, Pan World, we will meet you with emergency equipment."
Back in the cabin, the fire was out, and Rena was applying ointment to the burns on Mike Stewart's chest and arms. Yamaninan lay in the aisle, untreated and moaning quietly. He was naked because most of his clothes had burned off, and his chest, abdomen, arms, and face were charred black.
"Cabin crew, we're cleared for immediate landing. Prepare to deploy slide chutes," the pilot's voice said over the cabin loudspeakers.
The big 777 made a much steeper and faster landing approach than any of the passengers had ever experienced. The landing was hard, and immediately after touchdown the thrust reversers were engaged at full power, causing everyone to be thrown forward against their seat belts.
Once the plane turned off the active it proceeded but a short distance before coming to a stop. The cabin crew opened the doors fore and aft and deployed the sliding chutes. There was an orderly debarkation of all the passengers except Yamaninan, who was now being watched over by Mike Stewart and a male flight attendant.
Fire trucks, ambulances, and busses were already standing by.
Fort Rucker—Monday, March 19
Although Jake and Karin had each taken fifteen days' leave, they used only five days, spending most of it in Philadelphia, before returning to duty. Coming back from his leave, Jake found that, if anything, the conditions in the Army in general, and on the base in particular, had gotten even worse. Captain Greenly had not returned from his leave, and Jake didn't think he was going to. Captain Greenly and the three warrants who were assigned to EFT were all gone now, ostensibly on leave, though Jake had serious doubt that any of them would return. All four officers had completed their mandatory service and could leave the service simply by submitting their resignation papers, and Jake was reasonably certain they had either done so, or were going to.
That meant that Jake was not only chief of the environmental flight detachment, he was the only officer remaining. And now, in an effort to keep his men gainfully occupied, Jake was totally reorganizing Environmental Flight Tactics, the department he was responsible for. He was redoing the curriculum and rewriting all lesson plans, lesson objectives, and specific objectives, as well as reevaluating the course exams.
He went home late that night, planning to eat alone, primarily because Karin had night duty at the hospital.
After warming a can of chili, Jake took it and a can of root beer into the living room, where he turned on TV to watch the news.
President Ohmshidi announced today that, effective immediately, all banks in the nation with assets greater than twenty million dollars would be nationalized. CEOs and members of the Board of Directors of the affected banks will be asked to step down without any terminating compensation. The Federal Reserve will appoint government officials to run the banks, and any profits derived therefrom will accrue to the United States Government.
As a part of this proposal, all banking, savings, bond, and stock accounts are being inventoried, and a fair tax is being applied. An absolute limit of ten million dollars is being put on private wealth, and any amount of privately held funds greater than ten million dollars will be confiscated by the government. Persons with a net worth of between five and ten million dollars will be assessed a tax of seventy-five percent. Those people with a net worth of between one million and five million dollars are being assessed at a rate of fifty percent. There will be a twenty-five percent tax on all accounts between half a million and a million.
Those with a net value of from two hundred fifty to five hundred thousand will be taxed at ten percent and there will be no tax for those who have a net worth of less than two hundred and fifty thousand dollars.
Anyone who has less than one hundred thousand dollars will come under the president's new program of equalization. To those people, the government will be sending out checks within the week, totaling up to one hundred thousand dollars per check, the amount calculated to provide a baseline of at least one hundred thousand dollars for every American.
This is being done, the president says, to provide, fully and equitably, for all our citizens.
Jake did not have two hundred fifty thousand dollars, so he would not be subjected to a tax this year. He imagined there would be many people in the country who would welcome this tax relief, and many more who would welcome a government contribution that would elevate their net worth to one hundred thousand dollars. But Jake didn't feel good about it at all. This could not bode well. Where would be the incentive of the more successful and entrepreneurial people to build businesses, which provided jobs?
Finishing his bowl of chili, Jake went into the kitchen and opened the refrigerator door, then took out a block of cheese and carved off a thick slice. He grabbed a handful of crackers and another can of root beer, then returned to the living room to watch George Gregoire.
Hello, America.
I wish I weren't doing this show today. I wish I did not have to say to you, what I am going to say.
But I told you when I started this program that I would always question with boldness and I will always tell you the truth.
Well, I'm going to tell you the truth now, and it is something that I never wanted to say, even though it is exactly what I have been suggesting for three months now—ever since Ohmshidi took office.
I believe, in all sincerity, that this nation is now on a path to utter destruction. We are on a luge course, sliding downhill at ninety miles per hour, with no brakes, and with no barriers to hold us back.
My advice to you is to dig in, and hold on. While there is still food available, and while money still has some value, though its value decreases each day, start stocking up. Buy packaged foods and canned foods, foods that have a long shelf life. Make a survival bunker in your basement; fill it with food, blankets, water, clothing, and yes, guns and ammunition.
We are going down. Prepare for a very rough landing.
Tuesday, March 20
When Jake went into his office the next morning, he saw a lot of smiles on the faces of the lower grades. Although no checks had yet been issued, they were already spending their money.
"I'm going to get a new Mustang," one specialist was saying.
"Mustang, hell! I'm going to get new Caddy," a sergeant replied.
"Not me. I'm investing my money," a sergeant first class said.
"What makes you think you will be able to buy a Cadillac with one hundred thousand dollars?" Sergeant Major Clay Matthews asked. "Or for that matter, even a Mustang?"
"Cadillacs don't cost no hundred thousand dollars," the sergeant said.
"And for sure, Mustangs don't cost that much," the specialist put in. "I'm gettin' me a red convertible with white leather seats."
"Yeah," the sergeant said. "You know what? That don't sound half bad. Maybe I'll get a Mustang my own self, and save the difference in the money between that and a Caddy. Only I want mine to be white, with black leather seats."
"You're both crazy spendin' money like that. You ought to be like me, and invest it," the sergeant first class insisted.
"No," Clay Matthews said. "The truth is they have a better idea than you do. They are right about spending it as soon as they get it, because the way things are going, if you invest your money now, within six months it will be worth about half. If it takes that long."
"What are you talking about? I don't plan to speculate. I'll probably buy mutual funds. They will spread it out, and be very conservative."
"Jenkins, if you double your money in six months—say you run it up to two hundred thousand, or a quarter of a million, it won't make any difference," Clay said. "Two hundred fifty thousand dollars, six months from now, will be worth what fifty thousand is now. And fifty thousand now is worth what five thousand dollars was three months ago. My advice is to spend it as soon as you get it."
"Yeah," the other two said. "Come with us, we'll all three buy new cars."
"When is the last time you priced a new car?" Clay asked.
"I don't know. I've never had enough money to buy a new car before."
"You don't have enough money now, either. I have looked; the cheapest new car on the market today is one hundred twenty-five thousand dollars."
"What? That ain't right."
"Go online, find your dream car, then tell me what it costs," Clay invited.
"I'll do it," Jenkins said, sitting down at a nearby computer. He did an Internet search, found a car that the other two agreed they liked, then asked for the price.
"Holy shit! Two hundred and twenty thousand dollars?" the sergeant who had wanted the Cadillac said. "What's going on here?"
"Inflation," Clay replied. "Inflation like we've never had in this country before."
"Sergeant Major Matthews," Jake called. "Would you step into my office for moment?"
"Yes, sir," Clay said.
"Look on there, see what the hell kind of car we can buy," one of the men asked Jenkins as Clay went into Jake's office.
"Close the door," Jake asked. "And have a seat."
Clay closed the door, then took the seat Jake offered him.
"How are the men holding up?" Jake asked.
"They're gettin' a little antsy, what with nothing real to do," Clay answered. "Truth is I'm beginnin' to get that way myself."
"I know what you mean," Jake said.
"Major, do you know something I don't?" Clay asked.
"Like what?"
"I know that none of the officers are getting any flight time. None of the enlisted personnel who are on flight status are getting any flight time either. Are things ever going to get back to normal? Is the school going to reopen?"
"I don't know the answer to either question," Jake replied. "But if I had to guess, I would say no, things are never going to get back to the way they were, and no, I don't believe we will be making any more new pilots."
"Pardon my language, sir, but just what in the hell is going on? Does this new president have his head up his ass and locked so tight that he is going to be the ruin of us all?"
"I'm afraid that might be the case," Jake answered.
It was obvious that Clay was not expecting that answer, and he blinked in surprise. "You're serious, aren't you?"
"I'm very serious," Jake replied. He opened the middle drawer to his desk and pulled out a manila envelope. "This envelope is filled with signed requisition forms, DD-1195," Jake said. "I want you to take as long as you need to get every requisition processed and filled."
Clay pulled out some of the forms. "Whoa, twenty cases of MREs? Five cases of nine millimeter and five cases of .223 ammunition. Are we going on a field maneuver, Major?"
"As far as anyone else is concerned, we are."
"Ten barrels of JP-four. Why do you want that? Doesn't that normally come through the school?"
"I don't want any of this to go through the school," Jake said. "I don't want anyone to know anything about this. And if you are unable to get anything on this list by requisition, then I want you to get it in any way you can. I seem to remember that you are an expert at scrounging."
"And a water desalination device. A water desalination device? Major, you want to tell me what's going on here?" Clay asked.
"All right," Jake answered. "Clay, did you know that during the night, last night, the dollar was disconnected from the international money exchange?"
"What does that mean?"
"It means that the dollar is no longer the monetary standard for the rest of the world. Instead of saying that one dollar is equal to one and a half euros, the rate is now free to float. It might cost ten dollars for one euro, or one hundred, or ten thousand."
"That's not good, is it?"
"No, it isn't. There is a possibility, and I think a strong possibility, that this republic is going to come crashing down around us. And if it does, it's pretty much going to be every man for himself. Unless small groups get together for mutual benefit."
"I see. Do you have a such a group in mind, Major?"
"Not yet," Jake replied. "But when the time comes, I want you to be a part of it. If you are willing."
Clay stood up, saluted, then stuck his hand across the desk. "I would be honored," he said.
"Clay, until the time comes, this is between us," Jake cautioned.
"Right, sir." Clay put the forms back in the envelope. "I guess that, in addition to rounding up these items, I should also find a secure place to store them."
"I think that would be a very good idea," Jake said.
"I'll get right on it," Clay said.
The telephone rang and Jake picked it up. "Environmental, Major Lantz."
"Jake, have you seen the news this morning?" Karin asked.
"No, what has happened now?"
"Ohmshidi has turned Yamaninan over to the Islamic Republic of Yazikistan."
"What?"
"You have a TV in your office?"
"Yes."
"Turn it on. Ohmshidi is speaking now."
"All right, thanks." Jake held his hand up to stop Clay from leaving.
"What's up?" Clay asked.
"Wait. Before you go, you might want to see this," Jake said. He picked up the remote and clicked on the TV that was mounted on a stand high in the corner of the room. The president was talking.
By extending our hand in peace, by proclaiming to the people and the leaders of the Islamic Republic of Yazikistan that we mean them no harm, I am taking the first step in building a bridge of understanding between our two cultures. It is a bridge that I am certain will pay incalculable dividends.
While some of you might consider Abdullah Ibrahim Yamaninan a terrorist, to the people of Yazikistan, this brave man is a hero who was willing to give his own life for the cause that is so dear to his country. All of us cannot help but admire someone who has the courage and dedication to give that last full measure of devotion to his country and to his cause.
It is my sincere belief that this incident, which resulted only in injury to Yamaninan, offers us the perfect opportunity to end the hostility between us. Therefore I am returning Yamaninan to his country, along with a note of admiration for his courage and dedication. For too long now, there has been enmity between us, an enmity created by conflicting religious views. Now is the time for religious mythology to be assigned to its proper place so that secular humanity can rule our activities.
Thank you, and good night.
Ohmshidi's picture left the screen to be replaced by Carl Wilson, an anchorman for World Cable News.
This is Carl Wilson. We have just heard the president announce that Abdullah Ibrahim Yamaninan, the terrorist who tried to blow up an airliner over New York, will be returned to Yazikistan. In the studio with me now is Lawrence Prescott, former head of the Yazikistan office for the CIA. Mr. Prescott, your thoughts?
My thoughts ? I will be honest with you, Carl. If I shared my sincere thoughts with you, we would be taken off the air. If someone were to write a manual on what not to do when dealing with these people, this would be principle number one.
The Middle Easterner on the street sees negotiation of any kind as a sign of weakness. And this? Turning over a suicide bomber—or rather a would-be suicide bomber to the country that launched the attack, without any concessions? This isn't negotiation. This is surrender.
And, where do you think this will lead?
It can only lead to catastrophe. Look, Yazikistan has let it be known that they want a nuclear weapon, and they are willing to pay any amount of money to get it. The country is wealthy in oil money, and unlike America, or any western nation, the oil money goes, not to private investors, but directly to the government. If it costs them ten billion dollars to acquire nuclear weapons, they would be willing to spend it.
But, where would they find someone willing to sell the weapons to them? Aren't all the nuclear weapons closely guarded?
Are they guarded, Carl? There are some estimates that as many as one hundred nuclear weapons that once belonged to the Soviet Union are unaccounted for. Do not think for a moment that one or more of these weapons could not be bought if the price is right.
That is a frightening thought.
Carl Wilson looked at the camera.
Again, for those of you just tuning in, President Ohmshidi has just announced that Abdullah Ibrahim Yamaninan, the man who attempted to bring down Pan World America flight one zero three over New York City, is being returned to the Islamic Republic of Yazikistan, without any conditions.
We will keep you updated on the latest developments as they occur. And now, we return you to "Focus," our regular morning show.
"Major, I know that man is our commander in chief, and I need to show the proper respect to him," Clay said. "But he is a raving maniac."
"You'd better get started, Sergeant Major. I don't know how much time we have remaining," Jake said.
"Yes, sir, I will get right on it," Clay promised.
CHAPTER SIX
Moscow, Russia
When the Soviet Union collapsed, the emerging nation of Russia inherited a military designed to fight an all-out global war. At the end of the Cold War the Russian military was left with an air force that could no longer afford to fly its airplanes, a naval fleet that sat rusting in harbors, and an army in shambles. With no more clear-cut enemies in central Europe or on the Chinese border, the military-technical considerations that played a dominant role in Soviet force development and deployment throughout the Cold War period became obsolete. An army that was once one of the two biggest super powers in the world struggled in its brief, but bloody war with Chechnya.
Colonel Andre Yassilov, the commanding officer of a missile battalion on an army base near Moscow, was a victim of the chain of events in Russia. Yassilov's grandfather had been a hero of the Soviet Union who was personally decorated by Josef Stalin for bravery against the Germans during the Great Patriotic War. Colonel Yassilov's father was killed in Afghanistan and given a hero's funeral. Colonel Yassilov, who had served with honor and distinction in the war in Chechnya, had been a member of the army for twenty-six years.
But now things had changed. There was no pay for the military. Worse, there was very little food. The desertion rate of Yassilov's soldiers was higher than fifty percent and Yassilov couldn't blame them. At one time being an officer in the Russian armed forces meant having a position that not only commanded great respect, but also paid very well. Now, however, Yassilov, whose missile battery was equipped with SS-25 nuclear warheads, was forced to wait tables. The irony of it was extremely bitter. He had more firepower under his command than the total amount of explosives used in all of World War I and World War II combined—but he was waiting tables, groveling before diners for their measly tips.
Yassilov had swallowed his pride to work at the Gostiny Dvor restaurant, which was situated deep within a decorated garden on Volkova Street. The restaurant had a large dining hall that could seat eighty persons, a VIP hall, and a summer terrace. The interior of the restaurant was vintage Russian with timbered walls, massive dark-brown furniture, decorative windows with shutters, and handmade carpets. In the yard in front of the restaurant there was a fountain and a small waterfall. It was a favorite dining place for tourists, and had especially been so for Americans before recent events had almost completely stopped American travel.
It was here that Yassilov was first approached. One year earlier Yassilov would have never even considered discussing business with someone who made an offer to buy nuclear warheads. Had he been approached even three months ago he would have reported the contact to the proper authorities. But his personal situation had so deteriorated, and the amount of money the man offered was so large that it staggered the senses. It was sufficient money for Yassilov to return his own family to the economic level they had once enjoyed, and there would even be enough leftover to buy food and supplies for his men.
What Yassilov was being asked to do was a terrible violation of the oath he once swore, but, he reasoned, that oath should go both ways. If he owed allegiance to his country, then didn't his country owe allegiance to him?
The SS-25 missiles under his command had been destined for destruction under terms of the SALT treaty, which meant it would be fairly easy for Yassilov to comply with the request.
Yassilov dismantled the weapons as ordered, but he adjusted the inventory so that all were accounted for, though in fact he held back ten of the warheads. He sold the three warheads for a total of one million five hundred thousand Rubles. That was a small outlay for the man who bought the warheads. He had a buyer in Germany who would pay ten million euros for them.
The buyer in Germany contacted a Venezuelan arms dealer who agreed to pay one hundred million euros for the ten warheads. The Venezuelan paid the money without haggling because he had a customer in Yazikistan who would pay him five billion Venezuelan bolivars. The weapons never left Russia until the final deal was completed; then they were transported quickly and easily from Russia to Yazikistan in containers marked MEDICAL RADIOLOGY.
Tuesday, April 17
Hello, America.
Last month Mr. Ohmshidi made world headlines, incurring the wrath of most Americans and the incredulity of all the Western nations by returning the Muslim terrorist Yamaninan to Yazikistan. Even though he was still suffering from the burns of an unsuccessful attempt to detonate the PETN—not suffering enough, if you ask me—Yamaninan was placed in the backseat of President Rafeek Syed's personal open-top car and given a hero's parade through the streets of Kabrahn.
The open hand of peace Ohmshidi said he was extending to Yazikistan was met, not by an open hand, nor even by a fist, but by a swinging scimitar. As many as one million Yazikistanis lined the streets of Kabrahn, cheering loudly for their hero, and shouting such things as "Death to America!" "America the Satan!" and "Ohmshidi will go to hell!"
American interests all over the world are being attacked now. United Technomics in Paris was firebombed. In addition, every American news service in Europe has been attacked. Our embassies are under siege, and we have no means of protecting them, or American businesses overseas.
This is only the beginning.
That night Karin came to Jake's house, bringing dinner with her, and because she was expected, she let herself into the house. "Jake?" she called.
"I'm in the kitchen," Jake replied. "I saw you drive up, so I'm getting the root beers."
"Are you ready to see the Reds beat the Cardinals ?"
"Ha!" Jake replied returning to the living room carrying the two soft drinks. "In your dreams. The Cardinals have the Reds' number, and always have."
"What time does the game start?"
"At seven," Jake said, putting the drinks on the coffee table. "I've got it on the right channel. The clicker is on the lamp table right beside you. Just turn it on."
This is Carl Wilson with World Cable News. We are waiting for an address from the President of the United States. In the eighty-eight days since President Ohmshidi took office, he has given seventy-three televised speeches. He will be speaking from the Oval Office shortly, and we are told that the address will be only three minutes long.
"Damn, has that man ever seen a television camera that he wasn't in love with?" Jake asked. "What did you get?"
"Hot wings and potato logs. I was in the grocery store and walked by the deli. It smelled good, so that's what I got. Hope you approve."
"Oh, yeah, it looks and smells great. Now, if we could just get this idiot to stop going on TV every day."
"Supposedly he is only going to talk for three minutes. We may as well hear what he has to say," Karin said.
"Why? Whatever he says, it is just going to make matters worse."
We at World Cable News, along with all other television networks, have been given a transcript of the president's speech, but were told that we cannot say anything about it prior to his address. I can tell you this, however. It will be, to say the least, a stunning announcement. Afterward, we will discuss the address with our distinguished panel of news analysts.
"That's what we need," Jake said. "Another stunning announcement." Jake picked up a hot wing, separated it, and began eating.
The picture on the screen showed the president sitting behind his desk in the oval office. Behind him were two flags, the flag of the United States and a white flag, bearing what had been his campaign logo but had since replaced the flag bearing the presidential seal as the image of the Ohmshidi administration. It was a green circle enclosing wavy blue lines that represented clean water, over which was imposed a stylized green plant.
"I know you aren't supposed to hate," Jake said, "but every time I look at that man, I come as close to hating as you can get."
"Remember," Karin said, "that's our commander in chief you are talking about."
"How can I forget?" Jake asked with a growl.
"Shhh," Karin said. "He's about to speak."
"Whoop-de-doo," Jake replied.
Ladies and gentleman, the President of the United States, an off-camera voice said.
My fellow Americans. For too many years now, we have been dependent upon fossil fuels to meet our energy needs. This dependency has been the cause of nearly every problem we have faced, beginning in the late twentieth, and continuing into this, the twenty-first century. It has poisoned our environment, caused cancer and countless other health problems. It has destroyed our ozone layer, leading us toward irreversible global warming. It has created severe economic problems, and it has been the cause of international hatred and war.
For the last fifty years, there have been discussions of moving to a green economy with alternative, clean, and renewable energy as our nation's engine. And while other presidents before me have announced that as their goal, they have all failed.
I will not fail because I am taking a bold, and admittedly very difficult, step. It is, however, a step that I must take. I am, today, ordering an immediate cessation to all drilling and refining of domestic fossil fuels. In addition, we, as a nation, will no longer import fuels. We will have only that fuel currently extracted, refined, and in our inventory. When that is gone there will be no more. My analysts tell me that with strict rationing of the kind used during World War Two, our fuel supply should last about six months.
Now, while this may seem like a draconian step to many of you, it is, I believe, a way of spurring our scientists and engineers into committing to a twenty-four-hour-per-day, seven-days-per-week effort to find a sustainable alternative energy program. Will it be hydrogen? Will it be cold fusion? Will it be some scientific breakthrough that we have not yet imagined?
We of course have no idea as yet what this new source will be, but our future is exciting because I have absolute faith in our scientists to find a solution. Until then, all Americans will have to tighten their belts as we embark upon this great adventure together.
Thank you, and good night.
"He can't be serious!" Jake shouted. "He has lost his mind! He has finally lost his mind!"
"You have won me over, Jake," Karin said in a quiet, hesitant voice. "I think he has lost his mind."
"Do you know how much jet fuel we use in just one week at Fort Rucker?" Jake asked.
"I know it is a lot."
"We use three hundred thousand gallons per week. That is, we were using that when we were operational. Now if just we were using that much, how much fuel do you think our whole country uses? Everything, and I mean everything, is going to come crashing to a halt."
"How long do you think before that happens?" Karin asked.
"The last time you filled up, what did you pay for gasoline?" Jake asked.
"I don't know, around three twenty-five I think. Or something like that."
"You mark my words, tomorrow gasoline will be ten dollars a gallon, and that's only the beginning."
"I don't mind telling you, Jake; I'm getting a little frightened, now."
"Only idiots aren't frightened now," Jake said.
Thursday, May 17
In the weeks following the president's announcement that he was halting all acquisition of fossil fuel, either by domestic drilling, or importation, the price of gasoline began to increase, jumping at the rate of at least two dollars per day. The cost of fuel was beginning to be a problem for Jake and he was making a good salary. He couldn't help but wonder how others were dealing with it.
It was ten miles from Ozark to Fort Rucker and Jake drove it every day. This was Friday morning and, as he did every Friday morning, he stopped his two-year-old Volvo at the Busy Bee Quick Stop service station to fill his tank. Though this was normally a "fast in, fast out" stop, this morning he saw several cars waiting at each fuel island. This had become routine in the last few weeks, and Jake was prepared for it. He was in no particular hurry and he sat listening to Vivaldi's "Four Seasons" on the satellite radio as he waited.
"You son of a bitch! You pulled in front of me!" someone yelled to the driver of a car in the next line over. The shout was followed by the incessant honking of a horn that did not cease until a couple of policemen arrived.
"That asshole pulled in front of me!" the driver yelled to the police.
"Both of you," the police ordered, "out of line."
Grumbling, both the aggrieved, and the aggrieving driver were ordered to leave.
"Find somewhere else to get your gas," the policeman said. "And don't both of you go to the same station!"
Jake watched the two cars drive away. There was a time when he might have been amused by the little drama, but he had been seeing television reports of similar incidents all over the country. People were afraid, and the more frightened they got, the more uneasy the situation was becoming.
After a wait of about fifteen minutes, Jake pulled up to the pump and saw the price of gasoline, then gasped. It was thirty-six dollars per gallon.
"What?" he shouted. Thinking it might be a mistake, he checked some of the other fuel pumps.
"It's no mistake, sir," said a sergeant on the next island over, when he saw Jake checking the prices. "I stopped here yesterday and it was thirty-four dollars a gallon. I thought that was too much, but if we aren't going to get any new oil, this is just going to get worse. I should have bought gas yesterday."
"You had better fill your tank, Sergeant," Jake said. "At this rate, it could be fifty dollars a gallon or more by this time next week."
It cost Jake four hundred and thirty-two dollars to fill his tank. He was still frustrated when he reached his office. There were now more soldiers at Fort Rucker than there had been at any time since the Vietnam War, but because all training operations had stopped, except for normal housekeeping duties there was not one soldier who was gainfully employed. Jake knew that it could not last like this.
When Jake reached his office, Sergeant Major Matthews was waiting for him.
"Good morning, sir," Clay said.
"Sergeant Major. How are you coming on your requisitions?"
"I've added something to the list. I hope you don't mind."
"No, not at all. If you can think of something else we might need, by all means, acquire it if you can."
"I already have," Clay said. "I have twenty barrels of Mogas."
"You have twenty drums of gasoline?" Jake asked in surprise.
"No, sir, barrels, not drums. Drums hold only fifty gallons, a barrel holds fifty-five gallons. I figured it might be good to have."
"You figured correctly," Jake said.
"I know gas is expensive now, but I don't think we should use this until we have to," Clay suggested.
"I agree," Jake said. "We need to put it somewhere safe."
"I thought I would hide it in a hangar out at Hanchey Field."
"No, too many people out there. We need a more remote place than that."
"How about one of the stagefields?"
"Yes, excellent idea," Jake said. "And I know where to go with it. TAC-X. It's thirteen miles away, has four buildings, and is totally abandoned."
"All right, I'll get a truck from the motor pool."
"No," Jake said. "You would have to get a trip ticket for TAC-X and since it is no longer being used, that might arouse some suspicion. I think you would be better off renting a truck."
Jake wrote a check for two thousand dollars and handed it to him. "I hope this covers your expenses," he said. "But I would cash it immediately. And use it up as quickly as you can. The way the value of the dollar is plummeting, it may be worth only half as much this afternoon."
"I hear you," Clay said. "By the way, Captain Gooding is the POL Officer. If you would happen to get a telephone call from him, maybe you could cover my ass with a bit of a runaround.
"I'll do it," Jake said.
"Thanks."
"I'll leave it in your capable hands, Sergeant Major."
"I'd better go find a truck."
CHAPTER SEVEN
Dale County Truck Rental, Ozark, Alabama—Thursday, May 17
"You do realize that all I want to do is rent this truck, don't you? I'm not trying to buy it," Clay said to the proprietor. "And it is a local move, I'm not going anywhere with it."
"You'll have it back today?"
"I'll have it back by six tonight."
"Fifteen hundred dollars. And the gas tank had better be topped off."
"All right. You're robbing me blind, but I have to have a truck today."
"You got a beef, Sergeant Major, take it up with President Ohmshidi. It's his dumbass policies that have gotten us into this mess."
"Yeah, well, I can't argue with you there," Clay said. "That sonofabitch has been a disaster."
"Well, why didn't you tell me you hated Ohmshidi as much as I do? Tell you what. I'll take two hundred fifty dollars off. You can have the truck for twelve hundred and fifty."
"Thank you," Clay said.
When Clay drove through the Ozark Gate he was stopped by the MP.
"You'll have to get a visitor's pass for that truck," the MP said. "And I'll need to put down where you are going."
"I'm moving out of my quarters," Clay replied.
The MP entered the destination into his log, then handed Clay a visitor's pass with instructions to put it on the dash so it could be seen through the windshield. From there he drove to the POL center.
"I don't know, Sergeant Major," a specialist said. "I don't feel right about loading military fuel into the back of a civilian truck."
"What difference does it make what kind of truck you load it in?" Clay asked. "I have an authorized and approved requisition document."
"Maybe I should call Captain Gooding and ask him what I should do."
"Go ahead and call him if you want to. His name is right here on the requisition form," Clay said.
"I just don't feel right about putting the fuel onto a civilian truck," the specialist said.
"What would make you feel right about it?"
"Well, I mean, when you figure how much gasoline costs right now . . . I've got a leave coming up, but I can't go home because I can't afford the gas."
"How many gallons would it take you to get home?"
"About forty gallons."
"So, what if you had enough fuel to get home, plus say, oh, about fifteen gallons more so you could run around a bit when you got home?"
"That would be fifty-five gallons," the specialist said.
"Interesting coincidence, isn't it, that you need fifty-five gallons of gasoline, and that is exactly the amount that is in one of these barrels?"
"Yes," the specialist said. "Very interesting."
"So, are you going to help me to get my nineteen barrels loaded onto this truck or what?"
"Nineteen barrels?"
"Nineteen," Clay said.
The specialist smiled. "They are on pallets, five to a pallet. I'll get a forklift."
Clay pushed one of the barrels off one of the pallets. "Only four on this one."
"We'd better hurry," the specialist said, going toward the forklift.
Stagefield TAC-X
There are thirteen stagefields located around Fort Rucker. A stagefield is a facility that is somewhat remote from the main base, allowing student pilots to conduct flight and tactical operations there. TAC-X, or tactical operations training field X, was one of the thirteen, and though many Army aviators had trained here, it was no longer in operation.
When Clay approached the entrance to the stagefield, he saw that a double chain-link gate blocked the road. The gate was locked by process of a chain and padlock. A sign on the gate read:
U.S. GOVERNMENT PROPERTY UNAUTHORIZED ENTRY PROHIBITED!
Clay got out of the truck and, using a pair of bolt cutters, cut the lock. A moment later he swung the gates open and drove the truck through. Stopping the truck, he got out and closed the gates behind him, passed the chain back through, and reset the lock so that it looked as if it was still secure. Then he drove to the largest of the four buildings, this one a hangar, and went through the same process of cutting that lock and swinging open the hangar door.
Something scurried past his legs, startling him, and he let out a little shout until he realized that it was nothing more than a raccoon. He backed the truck into the hangar, then rolled the barrels down the tail ramp. It took less than an hour to off-load every barrel of gasoline, roll them over into the corner, and set them upright. When every barrel was off-loaded he covered them with an old tarpaulin. With the tarp in place, he went around picking up trash from the hangar, a solvent bucket, some paint cans, an old oil pan, a couple of wooden boxes, and some Plexiglas and sheet metal, which he placed on top of the tarp. His crowning achievement was finding six empty barrels, which he placed in front of his handiwork.
He examined the area when he was finished. Even if someone came into the hangar and looked around, they would have no idea that there was a little over one thousand gallons of gasoline here.
Clay closed the hangar doors, then locked them shut with his own padlock. Leaving stagefield TAC-X he did the same thing at the front gate, replacing the old lock with one of his own.
As he drove back to Ozark to turn in the truck, he called his daughter, who was a student about to graduate from Northwestern Louisiana University in Natchitoches, Louisiana. Although Clay had helped out as much as he could, she had held up her end by working as a waitress.
"Hello."
"Hi, Jenna. This is your dad."
"Hi, Daddy. I hope you are calling to tell me you can come to my graduation."
"Darlin', there's nothing I'd like more," Clay said. "But with the cost of fuel now—that is, when you can even get fuel—I just don't think I'll be able to. You can thank your president for that."
"I know you don't like him," Jenna said. "But that's because you haven't given him a chance. He is trying to do some things to make a real difference in the world."
"Yeah, like bringing all transportation to a halt."
"You aren't being fair. Mom and I are going to a pro-Ohmshidi rally tonight."
"Your mother still trying to save the world, is she?" Clay asked.
"Daddy, be fair. Just because you are a dinosaur doesn't mean you can't appreciate what Mom is trying to do."
Clay chuckled. "I will confess that your mother never met a cause she didn't support, or a movement she didn't join."
"And you never met a war you didn't love."
"I don't love war, darlin'."
"That would be hard to prove by me. You went to Iraq under the first President Bush, you went to Bosnia for Clinton, then two more times to Iraq and once to Afghanistan for the second President Bush. And you got medals for every one of them."
Clay lowered the phone from his ear and drummed his fingers on the dashboard for a moment. Why did Jenna have to sound exactly like her mother? Fortunately, she also looked like Carol, who was a beautiful woman.
"Daddy? Daddy, are you still there?"
Jenna's voice was tinny over the phone, and Clay raised it back to his ear. "I'm still here, darlin', but the traffic is getting a little heavier so I had better hang up. I'm very proud of you for graduating. And I love you, sweetheart."
"I love you too, Daddy. I just wish that you were a little more open-minded about things."
"Tell your mama I said hi," Clay said. "Bye."
"I will. Bye, Daddy."
Clay punched out of the call then dropped the phone back into his shirt pocket.
He thought of Carol, whose background was everything his wasn't. Whereas Clay's father was a Vietnam war veteran, Carol's mother had been a hippie, caught up in the free-spirit anti-war crowd. Carol, who was born in San Francisco, had no idea who her father was but, as she often assured Clay, she at least had the satisfaction of knowing that he was not a warrior.
There had been sparks between them from the moment they met, but underneath those sparks, or perhaps causing them, there was a very strong sexual attraction. In the end, though, the sexual attraction was not enough to save their marriage, and when Clay deployed to Iraq the second time, this time under George W. Bush, Carol left to protest that same war. In doing this, she was taking up where her mother had left off a generation earlier.
World Cable News—Thursday, May 10
In Washington today, Congress passed by acclamation President Ohmshidi's Water Resources Act, a comprehensive law that gives the federal government absolute control over all coastal waters, bays, rivers, lakes—whether natural or man-made—springs, creeks, canals, drainage ditches, and ponds. Under the auspices of this act those bodies of water that now lie on private land will be confiscated, and their use for any reason, whether watering livestock, fishing, boating, or drinking water, will be subject to federal government approval and taxation.
Congressman Hugh Langston of Alabama, who had led the fight against the Water Resources Act, protested the acclamation, claiming that the count was too close for a voice vote. Speaker of the House Nina Percy had Congressman Langston removed from the House floor.
In other news, the National Chamber of Commerce estimates that the amount of money taken from the U.S. economy as a result of the petroleum freeze, by virtue of lost revenue from the shortage of goods and services, as well as lost income from jobs that have been eliminated, to be in excess of five trillion dollars.
Despite severe rationing, and the steadily increasing cost of gasoline, our nation's supply of fuel has already reached the critical stage. Analysts are particularly concerned about those in the north who heat their homes with petroleum. If the winter is very severe, there will be extreme hardship throughout the Northeast.
CHAPTER EIGHT
Page and Canvas Bookstore, Fairhope, Alabama—Saturday, May 26
One hundred eighty-five miles southwest of Fort Rucker, Bob Varney sat behind a table in the Page and Canvas bookstore. The Page and Canvas was a popular and well-known bookstore among authors because they could always count on a good turnout for their autographing sessions. Bob was here to autograph his book Murder in Milwaukee. His action and adventure books were consistent bestsellers and he always did well at book signings. In addition, he was considered a local author in Fairhope, even though he actually lived thirty miles south in an area called The Dunes, on the Gulf Beach, next to historic old Fort Morgan.
There were at least twenty people waiting for Bob before he started signing. The problem was, there were only four books available.
"I'm sorry, Mr. Varney," Shirley Grace, the owner of the bookstore, said. "I had fifty books on order in plenty of time, but the distributor I have always worked with has gone out of business."
"Yes, that's been a problem over the last few years," Bob said. "The consolidation of distributors. Dozens of the regional distributors have gone out of business or sold out. Now there are only two or three major companies left."
Shirley shook her head. "There are none left," she said.
"What do you mean, none?"
"There are no book distributors left in business. Not one."
"Damn, I didn't know that. So, where are you going to get your books? Direct from the publisher?"
"I'm not going to get them from anywhere. As soon as this inventory is gone I'm closing my doors. She took in the books in her store with a wave of her hand. "The prices of the books are printed on the covers. I can't very well sell a seven-dollar book for fifty dollars, but at the current rate of inflation, that's what I would have to charge for them, just to break even."
"Tell me about it," Bob said. "I have three more books to do on this contract. I thought it was a good contract when I signed it—now what would have been a year's income will barely last a month. And my Army retirement and Social Security checks? They have been suspended, supposedly while DFAS reorganizes, and they haven't been restarted. Also, because I had more than one hundred thousand dollars in stock, I was not included in the one hundred thousand dollar giveaway."
Shirley wiped a tear away. "My grandfather started this store when he came home from World War Two," she said. "We have been in business since 1947. Ernest Hemingway, William Faulkner, Truman Capote, Eudora Welty, Willie Morris—they all signed books here. I can't believe it is all coming to an end."
"I'm so very sorry," Bob said. "I want you to know, Shirley, what an honor it has always been for me to sign here. I don't know where I'll be signing from now on, but I will very much miss this place."
"I don't know where you will be signing either," Shirley said. "I've talked to every bookstore owner in Baldwin and Mobile Counties. None are staying open."
"You're talking about the independents, though."
Shirley shook her head. "Not only independents but the big chain stores as well. By the end of this month, there won't be a bookstore left anywhere on the Gulf Coast, and I would be surprised if there were any left anywhere in America."
"When did all this happen?" Bob asked.
"Bob, I know you and the others down at Fort Morgan live almost like hermits. But you really should pay attention to what is going on around you. It's not only the bookstores; the entire country is coming down around us."
"I've been watching the news, but I never know what to believe. Even the news now is so slanted that it is difficult to discern the truth. The left-wing media says we are in a temporary recession and all will be well, while the right-wing media is all gloom and doom. I pass it all off as hyperbole and figure the truth is somewhere in the middle."
"It's not hyperbole. I wish it were, but it isn't."
"Maybe I should get out more," Bob said.
"No, you are probably better off holed up down there on the beach, away from everything. I would think that now is the time to just dig in and wait until this all blows over," Shirley said.
"I'll miss coming over here," Bob said. "And if you ever decide to reopen, let me know. I'll be your first guest author."
"If I ever reopen, yes," Shirley said wistfully. "It would be wonderful to think so."
As was his custom, Bob called his agent after the signing to tell him how the signing went.
"You have reached The Taylor Group Literary Agency. Please leave your name and telephone number."
Bob also knew Greg Taylor's personal cell number, so he called that.
"Hello, Bob," Greg answered.
"Hi, Greg. I forgot this was Saturday and I called the office," Bob said. "I didn't leave a message, but when you go in to the office Monday and check your missed calls, one will be from me, so you can just disregard it."
"I won't be going into the office Monday," Greg said. "No one will be. There is no office, there is no agency."
"What? What are you talking about?"
"You haven't checked your e-mail, have you?"
"No. What happened?"
"Ohmshidi is what happened," Greg said. "Bob, there is no agency, because there is no publishing. As of Monday morning, every publishing house in New York will be closing its doors."
"Every publishing house? You mean Kinston?"
"I mean Kinston, Berkline, Pelican; Pulman, Harkins and Role, Bantar; Dale, St. Morton—all of them, every publishing house in New York. The cost of fuel has shut down not only Ingerman, but every other distributor as well. And you know better than anyone that the book industry cannot survive without distribution. The only thing left now is online publishing. I tried to move the three books you have on contract with Kinston to Spindle press. Spindle agreed to take the books, but they are offering no advance, just royalties against the Internet sales. I told them I would talk to you, but to tell the truth, I don't think it's worth it. And since Kinston closed its doors, you are under no obligation to complete the three remaining books. I'd advise you to not to write any of them since you would be writing them for nothing."
Bob was silent for a long moment.
"Are you still there?"
"Yeah, I'm still here," Bob said.
"I wish I had better news for you."
"Yeah, so do I."
"It's been good working with you, Bob. I wish you all the best."
"Thanks, Greg. You too," Bob said. He punched the cell phone off and continued his drive to the Wash House Restaurant, where he was to meet his wife, Ellen, and their neighbors, James and Cille Laney, and Jerry and Gaye Cornett, from The Dunes at Fort Morgan.
Ellen was surprised when she saw Bob coming in.
"I thought you were signing books."
"I did sign them."
"Already? That was quick."
"There were only four books to sign."
"Only four books? That doesn't sound like Shirley. She always keeps a lot of your books on hand."
"Not today."
"You should call Gary right now, and tell him that she only received four books. He could probably find out what happened," Ellen said. Then, to the others, she added, "Gary Goldman is Bob's editor. He's been with Bob through three publishing houses, Bantar, Berkline, and now Kinston."
"I can't call Gary because there is no Kinston publishing anymore," Bob said, morosely.
"Oh, Bob! No!" Ellen said. "Kinston has gone out of business? But they are one of the strongest houses in New York. Well, is Greg going to try and put you somewhere else?"
"Greg closed the agency, and there is nowhere else."
"What—what do you mean?" Ellen asked in a small, frightened voice.
"There's no Bantar or Berkline or St. Morton," Bob said. "There are no publishing houses anywhere, and no distributors."
"If that is the case . . ." She let the sentence die, unable, or unwilling to finish it.
"If that's the case, I am out of work," Bob said, finishing the sentence for her.
"And with your retirement income halted," Ellen added. "Oh, Bob, what are we going to do?"
"We're going to do what every other American is doing," Bob said. "We are going to find some way to survive. And we can start by having lunch."
"I don't know," Ellen said. "Maybe we shouldn't be spending money eating out until we know where this is all going to go."
"We may as well spend the money," Bob said. "I have a feeling it will be worth about half as much tomorrow.
"This is getting bad for everyone," Jerry said. "I was talking to my broker yesterday; he said things are going to hell in a hand bucket on Wall Street. It's the total opposite of economic crisis in the past. Instead of the Dow going down, it is going up every day. But even though the Dow is now more than three hundred percent higher than it was when Ohmshidi was sworn in, its real value, according to my broker, is about one-fourth of what it was."
"I don't write books and I don't have any money in the stock market," James said. "All I have is my retirement from the power company, and it hasn't gone up one dime since all this began. Hell, my monthly income used to be enough to enable Cille and me to live comfortably. Now, it is going to take an entire month's retirement check, just to pay for this meal."
"James, maybe we shouldn't eat here," Cille said. "Maybe we should save our money, and go back home."
"Why?" James asked. "Bob is right. Whatever we spend today is probably worth twice as much as what it will be tomorrow. This may be our last meal out, so I say let's enjoy it."
Monday, May 28
Bob and Ellen Varney owned a house on the beach and a condo in St. Louis. They had a condo in St. Louis because they had a son who lived there with his wife and son. Every summer they rented out the beach house, earning enough rental income to pay for their St. Louis condo. The idea was that when they got too old to be able to live in the beach house, they would sell it, then move to St. Louis and live there full time.
Normally by this time of the year their house would be rented and they would already be in St. Louis. But so far not one person had booked their house for the summer. In fact, Sunrise Properties, who handled the rental for them, confided to Bob that not one of the thirty-six houses they managed had been rented.
Bob was sitting on the sofa in his living room, watching television. His wirehaired Jack Russell, Charley, was on the sofa with him, lying up as close to Bob's leg as he could get. Bob was rubbing Charley behind his ears.
Bob watched World Cable News almost exclusively. His son, who was considerably more liberal in his thinking, had often teased Bob about watching the most conservative of all the cable news channels, but Bob believed, sincerely, that WCN was the most accurate in their reporting. Besides, WCN had George Gregoire, and Gregoire was Bob's favorite commentator. But Gregoire did not come on until six o'clock, and it was just a little after five, so Bob was watching the Evening News Report with Sherman Jones.
You are looking at pictures of the many Liberty Party rallies held across the country today. Although the Liberty Party has neither national organization nor officers, they have sprung up since the election of Ohmshidi to make their feelings known. This rally, held in Chicago, had well over two hundred thousand in attendance. Similar rallies have been held in Cleveland, Philadelphia, New York, and Houston so far, and they are planning a large rally in Washington, D. C. But if you didn't watch WCN, or didn't know someone, personally, who attended one of the rallies, you wouldn't know anything about it. Not one of the other networks has carried so much as one minute of news pertaining to the Liberty Party rallies.
Well-known conservative talk show host Royal Peabody spoke at the rally in Houston.
The picture moved from the rally in Chicago to the one in Houston. There were several signs on display:
Impeach The Foreign Imposter
We Need Fuel
Fuel Now
Royal Peabody was standing behind a podium on a flatbed trailer as he addressed the crowd.
We are the heart and soul of America; we are the voice of the people. Some are mocking us, saying that we are in the pocket of a political party, but I say no, a thousand times no! We are beholden to no political party or ideology other than the principle of freedom, common sense, and the right to pursue happiness.
You know what would make us happy now? Fuel!
Peabody shouted the word, and it came roaring back on two hundred thousand voices. "Fuel!"
There are literally hundreds of billions of barrels of recoverable oil in the Bakken range, and nearly that much oil in Anwar. In addition we have more usable coal than the rest of the world combined, to say nothing of our huge gas reserves.
Ladies and gentlemen, our nation is collapsing around us, while our salvation is before our very eyes. We have enough energy to last for one thousand years without importing so much as one drop of oil. We have forty trillion in pre-Ohmshidi dollars worth of energy.
We were at the beginning of a monetary windfall that would put to shame anything we have ever experienced before—then we elected Ohmshidi. My friends, Ohmshidi promised us change, and he has delivered on that promise. We have changed from boom to bust. Ohmshidi's misguided policies, his insane order to halt all drilling and refining, even the importation of fossil fuel, has snatched financial disaster from the jaws of economic boom.
"Supper's ready," Ellen said, and Bob muted the sound as he and Charley went into the kitchen. Though they had a dining room, they ate there only when they had company. When it was just the two of them, they ate across from each other at a small table in the kitchen.
"Bob, what's going to happen to us?" Ellen asked.
"Nothing. Except we will probably spend the summer here, instead of going up to St. Louis as we normally do. With the cost of fuel it would be foolish to go up there for no reason. Besides, if it actually comes down to a condition of survival, I think we could survive better here, than in St. Louis."
"It is going to come down to that, isn't it?" Ellen asked. "A condition of survival."
"I wouldn't have said this six months ago, but yes, I believe it is."
"Are you afraid?" Ellen asked.
"No."
Ellen smiled wanly, then reached across the table to put her hand over his.
"Good," she said. "As long as you are not afraid, then neither am I."
"I think we need to start getting ready, though."
"Getting ready, how?"
"You know how. Just like we do when we are getting ready for a hurricane. The only difference is, this time we are going to have to be prepared for a much longer time than we ever had to with any hurricane."
"We've got the freezer nearly full now."
Bob shook his head. "The freezer won't do it," he said. "When it goes, everything is going, including the electricity."
"But we've got our own generator, and one hundred-pound propane tank."
"Which, if we run it full time, will last us for about two weeks. I believe we are looking at a year of being totally on our own."
"A year?" Ellen gasped.
"Or longer," Bob said.
In the living room they could hear the TV still going.
A suicide bomber blew himself up today in Grand Central Station in New York. Nineteen were killed and at least thirty more were injured. That is the fourth terrorist attack in the continental United States in the last twenty days, bringing the death toll total to eighty-six.
President Ohmshidi lodged a strong protest with the government of Yazikistan, but President Rafeek Syed dismissed the protest as the meaningless whining of a nation of kafirs, or unbelievers.
CHAPTER NINE
Base Hospital, Fort Rucker—Friday, June 15
"Hello, Colonel Chambers," Karin said, putting on as cheerful a front as she could. The patient, Colonel Garrison J. Chambers, a veteran of World War II, was ninety years old. One week earlier he had cut his leg on a piece of rusty, corrugated tin. That cut had gotten infected and the infection was spreading. He should have been given penicillin, but there was no penicillin available.
Karin once read that honey had been used for hundreds of years to treat infected wounds and though the doctor told her she was being foolish when she suggested they try it, he had finally come around. Chambers's wound was being treated with honey.
Karin removed the bandage and looked at the wound. It might have been her imagination, but she believed she was seeing some improvement. She began cleansing the area around the wound.
"How does it look?" Colonel Chambers asked.
"It's looking good," Karin said.
Chambers lifted his head and looked down at his leg. "Captain, if you think that looks good, you are definitely a woman who isn't turned off by ugly. And that, my dear"—he pointed to the purple, puffy wound—"has a serious case of ugly."
"Oh, it's all in the eyes of the beholder," Karin said.
"Hmm, where were you seventy years ago when I needed a woman who could overlook ugly?"
Karin laughed. "I'll bet you were a fine-looking young officer," she said. "I read in your records that you spent some time in Paris immediately after the war."
Karin knew that Colonel Chambers, as a company commander in the 101st Airborne, had also jumped into France on D-Day, and had been at Bastogne during the siege, where he was awarded the Silver Star and a Purple Heart.
"I was in Paris, yes."
"Now, be truthful, Colonel," she said, as she used peroxide-soaked cotton balls to dab gently around his wound. "Did you, or did you not have your share of beautiful young French ladies?"
"Ahh, you do bring up memories, my dear," Colonel Chambers said. "I seem to recall that there is one particularly pretty young lady who always sits down at the end of the bar at the Parisian Pony. Lovely thing she is, high-lifted breasts, long, smooth legs. I hope I'm not embarrassing you."
"Not at all, I'm enjoying the description," Karin replied.
Colonel Chambers was quiet for a moment. "I can see her now. She is so beautiful. Or was, I should say. My Lord, Chantal would be in her late eighties now. All of them. Every young woman I knew there. The soldiers too, the men who served with me, and under me. They were all so young then, and when I think of them, I remember them as they were, not as they must be now." He grew pensive.
"All memories are like that, Colonel."
"I suppose they are. If nobody has told you before, Captain, getting old—what is the term the young people use? Oh, yes, sucks. Getting old sucks."
"Yes, but consider the alternative," Karin said.
Colonel Chambers laughed out loud. "Good point, Captain, good point," he said.
"Tell me, my dear, when I get out of here, would you be too terribly embarrassed to have dinner with an old man?"
"Embarrassed? Not at all," Karin said. "I would love to have dinner with you."
"That is, assuming there is a restaurant still open somewhere by then. I've lived under eighteen presidents ; none have frightened me as much as this one does." He reached up to take Karin's hand in his. "On the one hand, I am glad I am so old because I don't believe living under this president is going to be very pleasant. On the other hand, I defended this country for many years, and I almost feel as if it would be an act of betrayal on my part if I were to die now, and leave this mess behind me."
"Say what you want about yourself, Colonel, but don't ever feel that you have betrayed your country in any way."
"I notice by the lack of a ring that you aren't married. But do you have a young man?" Colonel Chambers asked.
"I do," Karin said. "But he wouldn't mind at all my having dinner with you."
"Ohh," Colonel Chambers said. "Darlin', when a pretty young girl says she will have dinner with you, and then says in the same breath that it won't matter to her beau, that's when you know you are getting very old."
Karin laughed, then leaned over and kissed him on the forehead before she left.
Environmental Flight Tactics
Jake was sitting at his desk reviewing the new curriculum, lesson plans, and objectives, as well as a new SOP, wondering what he could come up with next to keep the men busy. Sergeant Major Clay Matthews tapped lightly on the door to his office, then pushed it open.
"Major Lantz?"
"Yes, Sergeant Major, come on in," Jake said, pushing the written material to one side. "Have a seat," he offered.
"How are the new lesson plans?" Clay asked.
"They are good," Jake said. "They are surprisingly good. I just wish we could get the training going again so we could implement them. What's on your mind?"
"I thought you might like to know that I have everything on your list that you asked for."
"Including fuel? Jet fuel, I mean. I know you got the gasoline last month."
"Yes, sir, I got the jet fuel."
"I'm impressed," Jake said. "How did you do that?"
"I had General von Cairns sign for a fifty-barrel emergency reserve."
"And you convinced him to do that?"
"Not exactly," Clay said. "Turns out that Specialist Roswell, who works down at HQ, can sign the general's name as well as the general can. He signed an 1195 for me."
"You didn't tell him what this was about, did you?"
"No. I convinced him that I was going to sell it on the black market and give him half the profit."
"I don't know, Sergeant Major," Jake said. "If there is no sale and he doesn't get his share, it could cause us some trouble."
Clay shook his head. "Not really, sir," he said. "First of all, he's not going to be able to say that he signed the general's name without getting himself in a lot of trouble. And secondly, I have already sold fifteen barrels for five thousand dollars per barrel. I'm going to give Roswell the entire seventy-five thousand and tell him that's half."
"So, we have thirty-five barrels left?"
"No, sir, we have fifteen barrels left. I told the POL sergeant that the general had really only requested thirty barrels, but I changed the number. That way he could have twenty barrels to do whatever he wants with. That expedited the operation and it also kept him from making any telephone calls back to the general to verify the requisition."
Jake chuckled. "I'm glad you are working for me, instead of against me."
"I would never work against you, Major," Clay said. He smiled. "I might be a thief, but I am a thief with honor."
"I can't argue with that, Sergeant Major."
"Oh, and I got a desalination device. Hand pump, not power. I figure we may not always have power."
"Good move," Jake said, just as the phone rang. He picked it up. "Environmental Flight Tactics, Major Lantz. All right, thank you. I'll be right there."
Hanging up the phone, he ran his finger down the scar on his cheek for a moment. "That was the general's office," he said. "He's issued an officers' call. I hope it's not . . ."
"It has nothing to do with our scrounging, sir," Clay said. "It is something else. Something entirely different."
"Do you know what it is?"
"Yes, sir, I think I do."
"What is it?"
"I'd rather not say, sir," Clay said. "I think you should hear it from the general himself."
"All right, I will," Jake said, standing up and reaching for his black beret. "You did well, Clay. You did damn well."
"Thank you, sir."
Jake learned that this officers' call involved more than half of the officers on the post, including not just department heads and unit commanders, but all staff and faculty, hospital personnel, officers of the TO&E units, and even those officers who had returned from overseas and were now at Fort Rucker awaiting further assignment. The officers' call was held in the post theater, and every seat was filled.
"Gentlemen, the commanding general!" someone called, and all stood.
"Seats, gentlemen, seats," General von Cairns said as he strolled briskly onto the stage. Walking up to the podium, he tapped twice on the microphone, and was rewarded with a thumping sound that returned through the speakers and reverberated through the auditorium.
"Gentlemen, I will make this announcement short and sweet. I will take no questions afterward because I have no further information. I'm afraid you will just have to hear the announcement, then wait and see where it leads.
"This morning I was informed by the Department of the Army that, effective within the next thirty days, there will be a seventy five percent reduction in force in both officer and enlisted ranks. In short, gentlemen, one month from now, three out of four officers and men throughout the entire army, will be gone.
The theater rang with loud shouts of shock and dismay.
"No!"
"What the hell is going on?"
"What is the cutoff? What if we are within months of retirement?"
"I don't know that there is a cutoff," von Cairns said. "There was no mention of it. And, gentlemen, I'll be honest with you; I don't know that there will be any more retirements. I know that those who are currently retired have not received their retirement checks for some time, now."
"This isn't right!"
"This is the president's idea?"
"So I have been told," von Cairns replied.
"What's wrong with this man? Is he insane?
"This so-called president is destroying the Army. And with it, the nation," Colonel Haney said.
"Somebody needs to drop-kick that foreign son of a bitch from here back to Pakistan!" a chief warrant officer said.
"Gentlemen, please," von Cairns said, lifting his hands to request order. "I can only tell you what I know. And I know that this was soundly protested by the Department of Defense. By the way, disabuse yourself of any thought that the Army is taking this hit alone. Because this order does come directly from the president, it pertains to all branches of the service, active, reserve, and guard.
"I'm sorry, gentlemen, I wish I had better news for you. This meeting is dismissed."
As the officers filed out of the theater, still stunned from the general's announcement, Jake saw Karin standing with another group of nurses. He didn't want to intrude so he started to walk away, but, seeing him, she hurried over to join him.
"That was quite a shocker, wasn't it?" she asked as she fell in beside him.
"I'll be one of the first to leave," Jake said.
"What makes you think so? You have an exemplary military record, outstanding OERs, combat time, not only combat time but combat command time, with a Distinguished Flying Cross, Bronze Star, not to mention a Purple Heart."
"Karin, I have not had a student in over three months, and I have not flown so much as one hour in all that time. I can't see the Army paying me to sit behind a desk and drum my fingers. At least you have been kept busy."
"True, we have been busy at the hospital, but eighty percent of our workload has been with retirees and veterans, and we heard that hospital service for non-active duty personnel is about to be stopped. Nurses don't have any special protective status, and you heard what the general said. Seventy-five percent of us are going to be gone."
"Do you have any patients depending on you, this afternoon?"
Karin thought of Colonel Chambers. She had cleaned and doctored his wound this morning, and he was resting comfortably when she left. She had no other patients who were at a critical stage.
"No, not really," she said.
"Then don't go back to the hospital. Take the rest of the day off and let's go out to Lake Tholocco."
"What? You mean now? Just leave?"
"Why not?" Jake asked. "What are they going to do? Kick us out of the Army?"
Karin laughed. "You are right. What are they going to do?"
Leaving the theater, Jake drove not to the lake, but to the post commissary. When he parked the car, Karin started to get out, but Jake reached out to stop her.
"No, you stay here and listen to music," he said. "I'll be right back out."
"Are you sure? I know how you hate to shop. I'll be glad to help you."
"I may not want you to see what I'm buying."
"Oh, a mystery? I like mysteries."
Ten minutes later Jake came through the checkout line. When his purchase was rung up, it came to six hundred and fifty-two dollars.
"What, and no cents?" he asked with a sarcastic growl.
"We don't mess with anything less than a dollar anymore," the clerk replied with a straight face.
"Tell me something," Jake said. "How do the lower-ranking enlisted personnel handle this?"
"They don't. Since the one hundred thousand dollars everyone was given ran out, I haven't seen anyone below E-6 in here."
"Where do they go? As expensive as it is, groceries are still cheaper in the commissary than they are off post."
"Tell me about it," the clerk said. "I work here, but because I am a civilian, I can't buy here. And the truth is, I couldn't afford it if I could."
Jake returned to the car carrying all his purchases in one sack. He set the sack on the floor behind his seat.
"I thought you were going to listen to music. What is that noise you are listening to?" he asked.
"Top Dollar."
"That's not music."
"What do you call music?"
Jake punched a button to switch to a classical station.
"There you go," he said. "'Emperor Waltz.' That's music."
Lake Tholocco is a six hundred fifty-acre lake located entirely within the confines of Fort Rucker. The lake officers' club was at one time a favorite hangout for the young bachelor officers, but when the Army and the other branches of service did away with officers' clubs, the lake club lost some of its cachet.
The lake was still a popular place to go though, with swimming, skiing, boating, and even fishing. There were also several rustic cabins around the lake in an area known as Singing Pines, and Jake pulled up in front of one of them.
"You have a key to this cabin?" Karin asked.
"Yes."
"You mean you planned to come out here? This wasn't a spur-of-the-moment thing?"
"Karin, you know me well enough to know that I plan everything," Jake answered.
There was a beautiful view of the lake from the cabin, and, because none of the other cabins were occupied, there was a great deal of privacy.
Getting out of the car, Jake reached into the back to retrieve his commissary purchases.
"When do I learn what you bought?" Karin asked, also exiting the car.
"Two steaks, two baking potatoes, two prepared salads, a loaf of French bread, garlic, salt, pepper, two root beers, and a bottle of cabernet sauvignon. You wanted romantic, I'm giving you romantic."
"Is a secluded cabin on the lake a part of that romantic scene?" Karin asked.
Jake smiled at her. "Yeah, it's the biggest part."
Jake made use of the charcoal grill outside the cabin to cook the two steaks. There was no charcoal available, but there were pine cones, and they made a suitable substitute. Karin put the potatoes in the oven and the salad and drinks in the refrigerator. That done, she came back outside to stand with Jake.
"I wish I had known we were coming to the lake, I would have brought my bathing suit in to work this morning."
"You can always skinny-dip," Jake suggested.
"You tease, but as deserted as the lake is now, I believe I could do that. I know it is still duty hours, but there is almost always someone here. Where is everyone?"
"I wouldn't be surprised if half the people on the base were gone by now," Jake said.
"Gone where?"
"On their way home, wherever that is. Karin, did you know that Clay Matthews knew about this even before the general announced it?"
"How did he know?"
"It's the NCO underground," Jake said. "I learned a long time ago that NCOs and even the lower-ranking enlisted personnel tend to find out things a lot faster than the officers do. I doubt, very much, that there was one enlisted man on this base who did not know what the general was going to tell us, even before he told us this afternoon. There's no telling how many of them have packed up and left today."
"You mean they've already gotten their RIF orders?"
"No orders, they just left"
"Deserted, you mean."
"Who deserted, Karin? Did they desert the Army? Or did the Army desert them?"
"Yes, I see what you mean. When you think about it, I don't believe they deserted the Army, and I don't believe the Army deserted them," Karin said. "It is Ohmshidi who has deserted us all."
Jake lifted up one of the steaks to examine it; then he looked back at Karin. "That's about the size of it," he said.
They had an early dinner, then sat out front and watched the sun go down over the lake.
"Karin, have you given any thought to what lies ahead?" Jake asked.
"I try not to. Every time I think about it, I just get more frightened. Have you thought about it?"
"Yes, I have given it a great deal of thought. It's like flying, and always anticipating the worst so you can be prepared for what might happen. I believe we must think and plan ahead."
"Plan and think about what?"
"Survival."
"I know what you mean. Unemployment is now at thirty-six percent, or so they say. And I wouldn't be surprised if it is much higher than that. If we are riffed, it is going to be very hard to get a job in civilian life."
"I'm not talking about getting a job," Jake said. "Very soon there aren't going to be any jobs anywhere, for anyone. I'm talking about survival as in staying alive—the kind of survival when there is a complete and total collapse of civilization."
"Jake!" Karin gasped. "You don't mean that, do you?"
"I do mean it," Jake said. "Believe me; it cannot go on like this. I believe our republic is going to have a complete breakdown."
"If that happens what will we do? What will anyone do?"
"I've already started," Jake said. "Actually, I started a couple of months ago, but I didn't say anything to you about it then, because I didn't want to worry you. As if you didn't have enough sense to be worried, just by listening to Ohmshidi."
"You already started what?" Karin asked.
"Do you remember when I told you I was going to see to it that you and I would survive? I am putting together a team," Jake said. "A team of survivors."
"By team, you mean there will be others?"
"Yes."
"Who do you have in mind?"
"Sergeant Major Clay Matthews, for one. I've already got him getting things ready. He is the kind of person I will be looking for, people who possess skills that can contribute to the survival of the team. Like your medical skills."
Karin hit him on the shoulder. "What? You mean the only reason you want me is because I'm a nurse? And here, all this time, I thought you wanted me because I am me," she said in mock mortification.
Jake laughed, and pulled her back to him. "Well, yes, that too," he said. "The steaks are done."
Jake returned to the grill and started picking up the steaks when he saw Karin just staring out at the lake.
"Karin, what is it?" he asked. "What's wrong?"
"This," she said, sweeping her hand across the lake.
"What? I think it's very nice here. Don't you like it?"
"Oh, I love it, Jake, but have you considered that this might be our last time like this? I mean, if things really do get worse, much worse, if we actually do have to go into a survival mode, something like this will be only in our memory."
Jake leaned over and kissed her, lightly. "Then, let's make this a good memory," he said.
"Yes."
They were just finishing their meal when Jake glanced at his watch. "It's about time for the news," he said.
"Are you sure you want to watch? I mean all the news is so depressing now."
"Do I want to watch? No, I don't," Jake said. He sighed. "But I don't think we have any choice. I am now morbidly obsessed with watching this man to see what he is going to do next. It's like being unable to turn away from a train wreck."
They moved over to the sofa, where Jake opened the wine, then poured each of them a glass. Handing a glass to Karin, he sat down beside her, then clicked on the TV.
Thirty-one people were killed and at least one hundred eighty injured when riots broke out in Louisville today, protesting the lack of fuel, [the announcer said.] This is the fifteenth riot this month, resulting in loss of life. The worst was the riot last week in Detroit, where one hundred and fifty-three were killed, and two hundred and forty-seven injured. The amount of damage done to private property as a result of the riots is incalculable, due to the fact that there is no longer a measureable value to the dollar.
In Cleveland, a spokesman for the national trucking industry said that if an allocation of fuel is not set aside and administered just for trucks there will soon be a nationwide shortage of food. Such a shortage is already occurring in the smaller towns across the nation where trucks are their only source of supply. Governor Coleman of Missouri has asked that a state of emergency be declared for the small town of Advance, which he says is completely out of food.
On another front, the national commissioner of baseball announced today that after the games this Sunday, all league play will be suspended. Since the president ordered the fuel embargo, airline operations are at the lowest they have been at any time since 1931 when air travel was still a fledgling industry. The airlines have said they could not guarantee Major League Baseball that they could offer enough flights to maintain the team schedules. Plans are underway to have a truncated World Series between the two teams with the best record in the National and American Leagues. The commissioner admitted during a press conference that the future of Major League Baseball is uncertain. The winner of this series may well go down as the last World Series Champion in history, he said.
The National Football League and the NCAA have already announced that there will be no football season this fall, and are unable to make any predictions on the future. As these games are played a week apart, they might be able to continue their schedule by train, though Amtrak has cut its operations by more than half.
A NEWS BREAK notice flashed on the screen.
We have this just in. President Ohmshidi has asked for network time to make an announcement. We go now to the Oval Office in Washington, D. C.
Ohmshidi was staring at the camera, obviously waiting for the signal to proceed. A slight nod of his head indicated that the signal had been given, and he began speaking.
My fellow Americans. When I was elected president, I inherited a nation that was in chaos. Because of the reckless policies of the previous president, the gap between the haves and the have-nots had widened precipitously. We were seeing big business run amok, and greedy banks foreclosing on hardworking Americans. My predecessor made no real effort to stem our insatiable lust for oil; indeed he sat idly by as our environment was destroyed and global warming increased. In addition, by fighting unnecessary and unjust wars in which innocent women and children fell victim to American bombs, he made this country the most hated in the world.
I began immediately to tend to these shortfalls. My first step was in ordering the return of all American troops, and I am happy to report that no longer is the uniform of an American soldier seen anywhere beyond our borders. I also ordered a cessation to any new oil as a way of forcing scientists to develop a new and sustainable source of energy. And though our nation is going through a period of belt-tightening and some hardships, we are weathering this storm together, knowing that there is a bright and shining future ahead.
Last night, in a midnight session, I asked Congress to pass the Enabling Act. The act passed by an overwhelming majority, though I am sorry to say that the opposing party proved, once again, to be the party of no, because not one of them voted for it. To bring about the fundamental change the people of this country voted for, and to deal with the inherited problems I have enumerated, it has become necessary for me to ask for increased presidential powers. The Enabling Act gives me those powers.
You may ask, and well you should, what is the Enabling Act? Under this act the roles of the president and Congress are reversed. As it is now, I can propose a bill, but I must wait for Congress with all its petty jealousies and bickering to act upon it. The sheer number of people, egos, and differing ideas make it almost impossible to get a bill through, and by the time it does come through Congress, there are so many compromises and amendments that it might bear scant resemblance to the bill I had proposed. This is unacceptable. In this time of crisis, a crisis I inherited from my predecessor, action must be decisive and immediate. We wait for the dawdling outcome of Congress at our own peril. When the bill reaches my desk I may ratify it by affixing my signature to the document, or I might refuse it by veto. If I veto the bill it will require a two-thirds majority to override that veto.
Under this new act this system will be reversed. I will replace Congress in the order of bringing a bill into law. There will be no need for bickering, compromise, amendment, or vote. Any proposal I make will become law as soon as I declare it so, unless it is vetoed by Congress, and that will require a two-thirds majority of both the House and Senate.
A new era of government has begun and I, Mehdi Ohmshidi, promise you efficiency and progress such as this country has never seen before.
In conclusion, let me share with you three more decisions I have made.
First, I have ordered that there be a one hundred percent stand-down in our Strategic Defense Initiative, better known as Sky Wars. I want every anti-missile missile to be withdrawn and destroyed. Secondly, I hereby order the immediate and unilateral dismantling of all our nuclear weapons in any guise, whether they be delivered by rocket, aircraft, or any other means.
And finally, I am, today, closing down the FBI, the CIA, the Secret Service, and the Homeland Defense Agency. I am doing this because of all the questionable activities that have been carried on by these agencies over the years, from the persecution by the FBI of innocent citizens during the long, dark, and oppressive years of the Cold War, to the torture and murder perpetrated by the CIA, to the illegal spying and violation-of-privacy acts conducted by Homeland Security. I am creating a new agency, answerable only to me. This new agency shall be known as the State Protective Service, or, the SPS. This new agency will have, in addition to their other many responsibilities, the task of protecting your president. Therefore the Secret Service will no longer be required.
So as not to taint the SPS with any of the misdeeds of the former agencies, no one who was a member of those agencies will be authorized to wear the coveted uniform of the SPS.
I believe that the steps I have taken so far—the pullback of all U.S. military from overseas, the seventy-five percent reduction in the number of personnel in our armed forces, the dismantling of all offensive and defensive missiles, and yes, even the rejection of fossil fuels—have created an environment of peace that we will be able to enjoy for years to come.
Thank you, and good night.
Karin hit the mute button on the remote. "He can't do that, can he?" she asked, shocked by what she had just heard. "He's assuming dictatorial powers, and he can't do that. Surely the Supreme Court will stop it."
"Of course they will stop it," Jake said. "He's gone way too far now. Maybe this will all turn out for the good."
"Good? How can it be good?"
"I can't see any way that the Supreme Court will let this stand. They will stop it."
"Do you think so?"
"Yes, they have to. How can they not?" Jake asked. He leaned his head back and closed his eyes. "Or, maybe not," he added. "It may no longer be our problem."
"Why do you say that?"
"If we activate the survival team, we will be cutting ourselves off from the rest of the country."
"I know."
"I said I was choosing all the members of the team based upon their particular skills. You know what my particular skill is?" he asked.
"You have a lot of skills," Karin said. "You are one of the most accomplished men I know. You are rated in how many aircraft? You have instrument ratings, you are a qualified flight instructor, you know electronics, you are a whiz with the computer."
Jake started laughing, and he laughed so hard that tears came to his eyes.
"What is it? What did I say?" Karin asked.
"Think about it," Jake said.
Karin thought for a moment; then she shook her head. "Oh, wow," she said. "Am I really that dumb? If there is a total breakdown, none of that will matter, will it?"
"My skill is the way I spent my youth," Jake said. "I never rode in an automobile until I was eighteen years old. You went to Lancaster with me, you know what my life was like. I was raised without electricity, without running water, without telephone, radio, or television. We farmed with mules, and what the mules couldn't do, we did with our bare hands and muscle. The first time I ever sat in an aircraft was when I went to flight school. I know how to live in a world that never even entered the twentieth century, let alone the twenty-first century."
"When we went to Pennsylvania for you to visit your family, I confess that I felt a little sorry for those people, dressed in plain clothes and riding around in a horse and buggy," Karin said. "I thought of how deprived they are. But now it holds a strange attraction for me. I could almost see myself living that life."
"Under circumstances like we are facing today, yes, the Life is very seductive," Jake said.
Karin smiled. "Seductive," she said. "Yes, you were very masculine, very seductive in your plain clothes."
"Who seduced who?" Jake asked, returning her smile.
"That isn't fair. You can't blame me. I was disoriented, not thinking straight, discombobulated."
"Really? You mean you just tolerated me in bed?"
"I had to. You outrank me," Karin teased.
"So, you will do anything I order you to do?"
Karin leaned into him, shut her eyes, and raised her lips to his, but stopping just short of a kiss.
"Captain Dawes, reporting for duty, sir," she mumbled.
Jake did not close the distance between their lips. Puzzled, Karin opened her eyes and looked at him.
"Haven't you heard? We are an all-volunteer army now," Jake said. "If we go any further, you are going to have to volunteer."
"Shut up and kiss me—sir," Karin said, pressing her lips against his.
"I need you, Karin," Jake said. "I don't think I've ever needed you more than I need you at this moment."
Karin stood, then started toward the bedroom. "Give me a moment."
"Make it a quick moment," Jake replied, his voice husky with desire.
CHAPTER TEN
Base hospital, Fort Rucker—Monday, June 18
"Captain Dawes?"
Looking up from her computer, on which she was filing a report, Karin saw Sergeant Julie Norton. Julie was a clerk in the office of the hospital commander.
"Yes, Sergeant Norton," Karin said, smiling at the young black woman. Julie was twenty-two years old and two years earlier, had been first runner-up in the Miss Georgia beauty contest.
"I thought you might like to know that Colonel Chambers just died," Julie said in a sad voice.
"I was afraid of that," Karin said. "He beat the infection, but then pneumonia set in, and he couldn't beat that. Pneumonia is hard to fight when you are young and strong. His body was weak; I'm surprised he lasted as long as he did."
"Yes, ma'am. The doctor said the only reason he lasted as long as he did was because of the way you took care of him," Julie said.
"It's a shame," Karin said. "He was such a fascinating old man. And so pleasant."
"Did you know he listed you as his next of kin?" Julie asked.
"What? No, I had no idea. Why would he do something like that?"
"I don't know, either. Maybe it's because he doesn't have anyone else. I pulled his records after the doctor told me he had died, just like I do for everyone when they are being released, either discharged, or by dying. His wife died last year and their only son was killed in Vietnam."
"Oh, what a lonely man he must have been," Karin said.
"He left this letter for you."
Julie handed a sealed envelope to Karin. Opening it, she removed the letter. Despite the colonel's age, the penmanship was bold and very legible.
Dear Captain Dawes,
I want to thank you for the loving care you gave me during my time here in the hospital. The sad thing is, I know I will not survive this stay. It is not sad for me. I have lived more than my prescribed years, and I am ready to shuffle off this mortal coil. But it may be sad for you, because you invested so much of your time and effort in tending to an old man.
As you may already know from a perusal of my records, I was at Bastogne in December of 1944. I was a company commander for one of the forward units. The German commander sent a note to General McAuliffe demanding the surrender of the Americans. The general sent back a note that angered the German commander, and the German commander threw that note away.
I found that note and have kept it ever since. I am leaving that note to you.
Sincerely,
Garrison J. Chambers
Col U.S. Army (ret)
Karin looked back into the envelope and saw another piece of paper, folded into a square, the paper browning.
"Oh, my God, this can't be real," Karin said.
Karin removed the brown piece of paper, then opened it up. There were only nine words written on the paper, only one of which was the body of the note. But that one word had come down through American history as a symbol of duty, honor, and country.
To the German Commander
NUTS!
From the American Commander
The Dunes, Fort Morgan, Alabama—Monday, June 18
Bob Varney was standing on the beach looking out at the now deserted offshore gas drilling rigs. Until Ohmshidi halted all domestic drilling, the rigs were ablaze with lights each night as for twenty-four hours a day, every day; they pumped gas from the rich deposits just off the Alabama Gulf Coast. Now the rigs were dark and deserted.
Charley was busily digging up sand crabs. When he found one he would jerk it up out of the hole, then throw it. More often than not he would watch the crab scurry away quickly, then go to another hole to start the process all over again.
"Nothing has changed for you, has it, Charley Dog? As far as you are concerned, your world is still Ellen, me, this beach, and the sand crabs. You're not worried that I'm not writing anymore, or that my Army retirement and Social Security checks have stopped."
Charley came over to Bob and reared up, putting his front two feet on Bob's legs. Bob reached down and rubbed him behind the ears for a moment.
"Go find another crab," he said, and Charley took off on his mission.
Although normally at this time of the year the beach would be crowded with summer people, it was empty now. Bob wasn't surprised. Vacations cost money and with the cost of gasoline today—that is, when you could even find gasoline—it would be prohibitive for any family to make the long drive.
There were twenty-two houses in The Dunes compound. The houses that sat right on the beach were all huge, multimillion-dollar homes. Normally rented in the summer time, they were all empty now, as was every other house in The Dunes, except for the three houses that were occupied by permanent residents. Bob was a permanent resident and his house was on the third row, approximately three hundred yards from the beach.
In addition to the houses, there were two seven-story condominiums, The Dunes and The Indies. Not one unit in either of the two large condominiums was occupied. One mile farther down was Fort Morgan, an historic old fort that was built just after the War of 1812.
Bob turned toward the surf and shouted at the top of his voice. "It is I, Robinson Crusoe. Where is everyone?"
Charley came running back to him and looked up with a quizzical expression on his face.
"You think I'm losing my mind, do you, Charley Dog?" Bob asked. He shook his head. "No, I'm not losing my mind. I'm just losing my will.
"My will to what? Survive? No, I'll not lose my will to survive. I survived three combat tours as a helicopter pilot in Vietnam. If I could deal with long strings of green tracer rounds coming toward me, to say nothing of air-bursting flak, then I can damn well deal with what we are facing now.
"I think.
"Come on, Charley, I'm tired of walking. Be a good dog and lay a couple of turds for me so we can head back to the house."
Almost as if responding to Bob's request, Charley hunkered down to do his business. Once completed, he looked at his deposit as if proud of it, then came over to Bob to await the treat that was his reward for performance. Bob gave him a treat, then dug a hole in the sand and pushed Charley's effort into it. He used to pick it up in a plastic bag, then drop the bag in the trash can, but that was when Baldwin County was still picking up trash.
Due to fuel concerns, the letter from Baldwin County waste disposal said, we will no longer be making our regular trash pickup. We apologize for any inconvenience this may cause.
"You apologize," Bob said when he got the notice. "Well, as long as you apologize, I'm sure it will be all right."
They walked back on the boardwalk between the USA and Dreamweaver houses to the golf cart that was parked on the road. Charley ran to the cart, then jumped up on the seat.
Seeing Charley do this, exactly as he had done for the last ten years, Bob felt a lump in his throat. How he envied Charley's ignorance of the fact that everything was coming down around them.
He glanced at his watch. If he was going to get back home in time to watch George Gregoire, he was going to have to hurry.
Hello, America.
Last Friday we heard Mehdi Ohmshidi take the unprecedented act of declaring himself dictator of this nation. He did this with something called "the Enabling Act." Let me give you a history lesson. On March 23, 1933, the German Reichstag met in Berlin to consider passing a law that would end democracy in Germany, and establish the legal dictatorship of Adolph Hitler. This act was called the Enabling Act. It did pass, Hitler became dictator of Germany, and we all know what happened.
Is it just ironic coincidence that Ohmshidi chose the same name?
But, not to worry. The Enabling Act is clearly a violation of the United States Constitution, so the Supreme Court will overturn it. Right?
Not so fast. We now know that within moments after Ohmshidi made his announcement Friday night, the Supreme Court did meet in emergency session to consider the constitutionality of this new law.
What did they decide?
They decided nothing. They couldn't decide because before they even convened, federal agents descended upon the Supreme Court and took every justice of the Supreme Court into what is being called protective custody.
Who did this ?
Not the FBI, not the CIA, not the Homeland Security. By dictatorial fiat, those agencies no longer exist. No, the arresting officers, we are told, belong to the newly organized agency, the State Protective Service.
A spokesman for the SPS has stated that the justices are not under arrest, but have been moved to an undisclosed site for their own safety. While there, the spokesman added, the justices will participate in a conference during which the details of the Enabling Act will be discussed.
America, we are seeing, before our very eyes, the total destruction of our republic. Misguided voters, thinking it would be cool to vote a naturalized American into office, flocked to the polls to show the rest of the world what an unbiased and open-minded nation we are. They voted for this man without really knowing anything about him.
And now, we are about to pay the piper.
The telephone rang and Ellen answered it.
"Hello? Oh, Tim, hi, sweetheart. Yes, he's here. Okay, just a minute."
Ellen brought the phone over to Bob. "Tim wants to talk to you," she said.
"Hello, Tim, what's up?" Bob asked.
"You were right, Dad," Tim said. His voice was low and obviously strained.
"Right about what?"
"About everything," Tim said. "How could I have ever been such a fool to vote for this man?"
"If it is any consolation to you, you aren't the only one. He is president because more people voted for him than against him. That's the way democracy works. Or at least, that is the way it used to work. After his Enabling Act, and some of the other things he's done, I don't know."
"Dad, I've pulled all your money out of the market," Tim said. "Mine too."
"Why?"
"I've been in this business for ten years," Tim said. "I can read the signs. The way things are going, the stock market isn't going to last another year. I'm not sure it's going to last another month."
"But the market keeps going up," Bob said.
"Yes, it keeps going up, but the real value is plummeting. I'm going to do an electronic transfer of the money to your bank."
"How much is it?"
Tim chuckled. "It's a little over one and a half million," he said.
"Whoa, that's pretty good, isn't it? Last time I checked it was a little under three hundred thousand."
"I wish I could say that it was good, but the only thing it means is that money is losing its value faster than we can keep track."
"Tim, wait, don't do an electronic transfer," Bob said. "I won't be able to get it out of the bank. I can't draw out any more than ten thousand dollars at a time."
Tim laughed. "You haven't been keeping up, have you?"
"What do you mean?"
"The banks are no longer observing that limit. The value of the dollar has decreased so far that the FDIC insurance is virtually worthless now. That means that, though the Fed still has authority over the banks, they no longer have any leverage. The banks can do whatever they want and it doesn't matter."
"Tim, you know how far we are from town. Tell me truthfully, is the money even worth going into town for?"
"I don't know, Dad. I wish I could answer that. But if I were you, I would go into town and take it out, then buy as much as you can with it. The more you have in real property, such as food, bottled water, fuel, anything you can think of that you might need—and that you can actually find on the market—the better off you will be. The problem now is there are less and less goods and services still available."
"How are you doing?" Bob asked. "I mean, you are a broker, if the stock market really is going to go belly up, where does that leave you?"
Tim laughed, but it was a harsh, humorless laugh. "Dad, it leaves me in the same boat as everyone else in the country—up shit creek without a paddle."
"Are you, Pam, and Jack going to be all right?"
"You remember that place we bought on Lake of the Ozarks?"
"Yes."
"I've filled the SUV with survival gear. There's fresh water and game there. We're heading there tomorrow."
"Keep in touch with us," Bob said.
"I will for as long as I can," Tim said.
"You better talk to your mom now," Bob said. "You don't have to repeat all this. I'll fill her in on it later."
"Thanks, Dad."
Bob handed the phone over to Ellen. "He wants to talk to you," he said.
Because Ellen had been listening to one side of the conversation, she knew that whatever it was wasn't pleasant, and her voice broke when she took the phone. "Hello?"
Fort Rucker—Monday, June 18
"My God," Jake said. "Do you know what you have here?" He was holding the McAuliffe note Karin had given him.
"Yes, it's the note General McAuliffe sent to the German commander," Karin said.
"It is a piece of American history," Jake said. "All the more important now that our history is being taken from us."
"It's amazing that Colonel Chambers held on to it all these years," Karin said. "It had to be worth a lot of money."
"I would say, conservatively, it was worth more than a million dollars back when a million dollars actually meant something. But I'm not surprised that he held on to it. He was there, so I'm sure that, to him, this note was worth more than any amount of money."
Karin nodded. "I didn't know him that long," she said. "But, from what I did know of him, I would say yes, he was that kind of a man."
"For someone who didn't know him all that long, you certainly made an impression on him," Jake said.
"Not nearly as much as the impression he made on me," Karin replied.
There was a light knock on the door and Sergeant Major Matthews stuck his head in.
"Excuse the intrusion, ma'am," Clay said. Then to Jake, "Major, I thought you might like to know that Sergeants Dagan, McMurtry, Jenkins, Pounders, and Vivian are gone."
"I'm surprised they stayed around as long as they did," Jake said. "Did you check with Staff and Faculty Company? Are they being reported missing on the morning report?"
"There is no morning report, Major. There is no first sergeant, there is no company commander. Nobody knows where Captain Poppell is. No one has seen him since the announcement of the RIF."
"I hope Dagan and the others get home all right," Jake said. "If that is where they are going. How many people do we have still reporting for work every day?"
"I'm not entirely sure, but I would say about twenty, sir."
"Twenty out of an authorized strength of seventy-two. Actually, that's better than I thought it would be."
"Yes, sir, well, I reckon they are pretty much like me, they don't have any place else to go."
"Are the mess halls still feeding?"
"A couple of them are. The consolidated mess is still serving meals."
"That's good. If the men are going to stay around, they should at least have someplace to eat."
"Yes, sir, that's pretty much what I think as well. When do you think they are going to start sending the RIF orders down?" Clay asked.
"From the looks of things, they aren't going to need to send any orders down. Looks to me like the reduction in force is taking care of itself."
"Yes, sir, I would say that as well," Clay answered. "It almost makes you wonder if this isn't the way they planned it in the first place."
"Sergeant, in order to plan something, one must have enough sense to anticipate the outcome. It is clear to me that nobody in Washington, in or out of uniform, has that kind of sense."
Clay laughed out loud.
"I didn't say that for a joke, Clay, I said it as a matter of grave concern."
"Yes, sir, I know that, Major. But I figure that about the only way we are going to get through all this is if we learn to laugh at the stupidity."
Jake chuckled, and nodded his head. "You may have a point there, Sergeant Major. You may indeed have a point."
"Ma'am," Clay said before he withdrew.
"Sergeant Major, wait a moment," Jake said. "You come from an old Army family, don't you?"
"My Dad was in Korea and Vietnam, my grandpa was in World War Two, my great-grandpa was in World War One, and my great-great-grandpa was with Custer. Actually, he was with Benteen during the fight, or else I wouldn't be here."
"Then with that kind of background, you might appreciate this," Jake said. He handed the browned piece of paper to Clay.
Clay looked at it, then glanced up at Jake and Karin. "Is this real?" he asked.
"As far as we know, it is," Jake said.
"This is the note that McAuliffe sent to General Freiherr von Lüttwitz," Clay said. "I thought Colonel Chambers had it."
"You knew Colonel Chambers?" Karin asked, surprised by Clay's comment.
"Knew? You mean he has died?"
"Yes, this morning."
"I didn't know that," Clay said. "But yes, I knew him. My dad and my grandpa both served with him. He retired before I came into the Army, almost thirty years ago, but I remember him well. He was a fine old gentleman."
Karin showed Clay the letter Chambers had written before he died.
"Good for you, Captain," Clay said after he read the letter. "I can't think of anyone who would deserve it more."
CHAPTER ELEVEN
The Dunes, Fort Morgan—Wednesday, June 20
Bob wasn't able to get very much with the money that was transferred to his account. It wasn't that he didn't have enough money, though certainly what he did buy cost more than he could have possible imagined just one month ago. One pound of dry beans cost one thousand dollars; a five-pound bag of flour was fifteen hundred dollars.
Bob had more than a million dollars to work with, and he didn't mind spending it because he was sure it would be worth half as much the next day. What limited his purchases was not money, but availability. Most of the stores in Gulf Shores, and in Foley, had closed, and the few that remained open had less than ten percent of their normal items on the shelves.
When they returned to their house they loaded everything into the elevator to take it up to the kitchen.
"Look at that," Bob said, pointing to the groceries. "What we bought today cost more than the total amount of my last contract, and it doesn't even cover the whole floor of the elevator."
"At this rate, we are going to run out of money within a month," Ellen said.
"It won't matter."
"Of course it will matter. What do you mean it won't matter?"
"Ellen, one month from now we'll be using hundred-dollar bills as toilet paper."
Bob helped Ellen put away the groceries; then he sat on the couch and picked up the remote. Charley jumped up beside him.
When the TV screen came up there was a huge letter O in the middle of the light blue screen. The O was green, with three horizontal, wavy blue lines at the bottom. Above the blue wavy lines was a green plant that looked for all the world like a marijuana plant.
The O went away, and the camera showed President Ohmshidi sitting at his desk in the Oval Office. There were some changes in the Oval Office from the last time Ohmshidi had made a public address—which was yesterday. The changes were immediately apparent. His desk was flanked, left and right, by two muscular and unsmiling black men, both members of the SPS. They were wearing forest-green uniforms with SPS gold collar pins; the two S letters, rendered as lightning bolts, were separated by the letter P, which resembled a one-sided hatchet. Ohmshidi O logo armbands were around their left arm, and they stood at parade rest, staring unblinkingly straight ahead.
The American flag was missing. In its place was a white banner, which, because of the way it was hanging, did not display all its components.
My fellow citizens, as you know, in the past few days, since the establishment of the Enabling Act, the nation has been curious as to how the Supreme Court would act. I am pleased to report that this issue has now been resolved in my favor. All nine sitting justices have submitted their resignation and I have appointed a new court.
This new court has unanimously approved the Enabling Act, as I knew they would. They have also ruled favorably upon other actions I have already taken, and will now explain to you.
During my campaign, I had a logo developed for me that reflected my belief in this country, not what it is now, and certainly not what it has been in its odious past, but in the greatness that lies before us as we complete our fundamental change. As you recall that logo is my initial, the letter O in green. Inside this green circle are wavy blue lines that represent clean water, and a stylized green plant that symbolizes not only a green, clean world, but new growth.
I'm going to ask the two officers of the SPS who are standing here to display one of the two banners that are behind me now.
The two men removed one of the banners from its stand, then spread it out so it could be seen. The logo Ohmshidi had described was in the center of the flag.
This symbol, placed upon a pure white field, will be the new flag of our nation, proclaiming to the world that we are a nation of peace and a nation that safeguards our environment. For far too long, the red, white, and blue stars and stripes flag has, in song and story, represented us as a bellicose nation, too eager to go to war at the slightest, or even perceived, provocation.
I am ordering today that, with immediate effect, this new flag replace the Stars and Stripes as our national standard. I am further declaring that the display of the old flag, or any previous national symbol, such as the representation of an eagle, on public or private property, to be declared a seditious act. The wearing of a flag pin on the lapel is also prohibited. Accordingly, I have given orders to the SPS to arrest anyone who displays the old flag so that they may be brought to justice. Further, singing of the song "The Star-Spangled Banner" is hereby prohibited, and violators will be prosecuted. Any newspaper that publishes an article in opposition to this act will be shut down, and the author of the article, as well as the publisher of the newspaper will be arrested. Talk radio and opinionated television commentators are here and now cautioned that public protest over this will be regarded as an act of sedition and they will be arrested. In addition, any radio or television station that carries this seditious programming will be shut down, confiscated, and given to those citizens who are loyal to me, and to the new paradigm I am bringing about. This is absolutely necessary if we are to have a clean break with our troubled and misguided past.
And finally, I have changed the name of our country to the New World Collective. This is in keeping with my determination to make our nation a beacon to the rest of the world—a leader in peace, progress, and real equality for all humankind. I am ordering all government documents from henceforth to represent not only our new national symbol, but also our new name.
As I am sure you will understand, a new nation will require a new constitution. Accordingly I have declared the constitution of the nation once known at the United States of America to be null and void. I am having a new constitution written, one that will insure an equitable distribution of wealth among all its peoples, and one that will take into account the necessity of efficiency of government, by providing the president with absolute authority.
And now I make this promise to you. I will work tirelessly to make this new nation succeed, but I cannot do it alone. Much sacrifice will be required from you, so I ask all of you to do your part, and to report to the authorities anyone you see who, by word or deed, commission or omission, may be undermining the authority of your president.
I will close this broadcast with the words and music of the new national anthem, as sung by the Children's Choir of the Tranquility School of Baltimore.
Thank you, and long live the New World Collective.
The camera moved then to a group of children all wearing choir robes with the new O symbol upon their chests, singing what was to be the new national anthem. As they sang, the words rolled across the screen.
Unbreakable New World Collective
Our people loyal and true
To Ohmshidi, our Leader
We give all honor to you.
(Chorus) Glory to our great leader
May he remain right and strong
The party of the faithful
Ohmshidi to lead us on!
In the New World Collective,
We see the future of our dear land
And to the Ohmshidi banner,
In obedience shall we stand!
(Chorus) Glory to our great leader
May he remain right and strong
The party of the faithful
Ohmshidi to lead us on
"Jesus, Ellen, I never thought I would live to see something like this happen. This must be the way the Germans felt when they realized what Hitler was doing to them."
"Surely this can't go on," Ellen said. "Someone will stop it."
"Who?"
"I don't know. Congress? The Supreme Court?"
"He controls Congress, they don't control him. And he got rid of the Supreme Court and replaced it with his own court. And, you heard him say it yourself: he has declared the Constitution to be null and void."
"Then we will vote him out," Ellen said.
"You are assuming there is going to be another election," Bob said.
"He can't stop the elections, Bob. The people won't let him."
"Ohmshidi stopping the next election isn't the problem," Bob said. "The problem is we will no longer be a nation by the time the next election is due."
Fort Rucker—Thursday, June 21
At Fort Rucker the next morning the Stars and Stripes flag was run up the flagpole; then a cannon shot was fired and the bugle call for Retreat Ceremony was played. Normal procedure for retreat was for all soldiers, wherever on the base they may be, to stop what they were doing. If they were driving, they were required to stop alongside the road, get out, face the flag even if they couldn't see it, and salute.
That was exactly what was happening now, though many wondered if there was some sort of mistake. Retreat was at the end of the duty day, not at the beginning.
On the parade ground as Retreat sounded, a soldier slowly, and stately, lowered the Stars and Stripes. Then, very deliberately, and with as much dignity as could be mastered, two soldiers folded the flag into a triangle shape, so that only the blue field showed, without even a trace of the red. The flag was presented to a sergeant, who then presented it to General Clifton von Cairns. After presenting the flag, the sergeant took one step back and saluted. Von Cairns stuck the folded flag under his left arm, then returned the sergeant's salute.
"Sergeant, dismiss the detail," the general ordered.
"Shall we hoist the new flag, sir?" the sergeant replied.
"No. Dismiss the detail."
A broad grin spread across the sergeant's face. "Yes, sir!" he said, proudly. Then he did a smart about-face and called out, "Retreat detail dismissed!"
General von Cairns walked back into the headquarters building and into his office. Once inside his office, he closed the door, opened the drawer of his desk, took out a bottle of whiskey, removed the cap, then turned it up to his lips. He had long ago quit using a glass.
Base hospital, Fort Rucker—Wednesday, June 28
Colonel Ruben Sturgis, MC, the hospital commander, called his staff together. At one time there were twenty doctors, forty nurses, and sixty enlisted personnel on duty at the hospital. Today there were two doctors, three nurses, and one sergeant present for the meeting.
"Dr. Urban, you are the chief surgeon now, so this comes under your bailiwick. Effective immediately we are to provide no more care to retired personnel, nor to those who are qualified under VA," Colonel Sturgis said.
"What are we to do with those we have now?" Dr. Urban said.
"Discharge them," Sturgis said.
"Colonel, we have three in intensive care. If we discharge them immediately, they will die before nightfall."
"What is their prognosis?" Sturgis asked.
"I'll let Dr. Presley answer that," Urban said.
"Not good," the younger of the two doctors said. "The truth is, I doubt any of them will live to the end of the week. They are all three in extremis, and we simply don't have the medication to treat them."
Colonel Sturgis drummed his fingers on the table for a moment, then nodded. "Alright, keep them. Discharge the ones that we can, and admit no one new."
"Colonel, we have no orderlies left," Julie said.
"How many enlisted personnel do we have left?" Colonel Sturgis asked.
"I'm the only one."
"It isn't just the enlisted personnel," Karin said. "As far as I know, we are the only nurses left." Karin took in the other two nurses with a wave of her hand.
"How many patients do we have now?"
"We have seven," Julie said. "Four retired, two VA, and one active duty."
"What is the condition of the active-duty patient?"
"I took out his appendix yesterday," Dr. Presley said. "I was going to release him this afternoon."
"Release everyone, except the three who are in ICU," Sturgis said.
"Alright," Dr. Urban said.
Sturgis pursed his lips, then let out a long breath. "Just so you know, I have submitted my retirement papers. That was just a formality, I don't expect DA to act on them. Hell, I'm not even sure there is a DA anymore. I'm leaving tomorrow morning, no matter what. And if I leave, I don't intend to hold any of you here. Chances are we aren't going to even have an army within another month, if we last that long."
"We were going to ask you about that," one of the nurses said. "Linda and I were planning on leaving tomorrow."
"I'm going as well," Dr. Presley said.
"Will no one be here for the three ICU patients?" Sturgis asked.
"I'll stay until the end of the week," Dr. Urban said.
"I'll stay as well," Karin said.
"I'm not going anywhere," Julie said.
"Look, I'll stay too if you need me," Sturgis said. "I feel bad about deserting you at a time like this."
"We can handle it, Colonel," Dr. Urban said. "Hell, there's nothing to do but watch them die anyway."
Sturgis looked at what was left of his staff, then nodded. "I don't know where we are going from here," he said. "But it has been a privilege to work with you. All of you."
CHAPTER TWELVE
Wednesday, July 4
Hello, Americans.
Today is Independence Day. For two hundred and thirty-four years, our nation honored this historic occasion. Even when our country was young, it was a cause for joy and celebration. In the great cities and small towns, parades were held, patriotic music was played, there were barbecues and fireworks, and baseball games.
When you think about it, Baseball was America, wasn't it? Babe Ruth, Stan Musial, Willie Mays, Hank Aaron, Derek Jeter.
George Gregoire paused for a long moment, his voice choking. He wiped a tear, then continued.
But—those days are no more.
It isn't just no more barbecues, no more fireworks, no more baseball. America itself, is no more.
When I first warned you of the danger we were facing under the evil, and yes, evil is the only word I can use to describe this tyrant, this evil Ohmshidi, I prayed long and hard that I would be wrong. But I wasn't wrong. In fact, if I made any mistake, it was in not being forceful enough.
Thomas Jefferson once said: "The tree of liberty must be refreshed from time to time with the blood of patriots and tyrants." Winston Churchill said: "If you will not fight for the right when you can easily win without bloodshed; if you will not fight when your victory will be sure and not too costly; you may come to the moment when you will have to fight with all the odds against you and only a small chance of survival. There may even be a worse case: you may have to fight when there is no hope of victory, because it is better to perish than to live as slaves."
My fellow Americans—yes, I said Americans, not World Collectives—that time has come! I am calling upon all Americans to rise up against the despot Ohmshidi!
The Stars and Stripes appeared on screen, with the music of "The Star-Spangled Banner." The flag was replaced with scenes of U.S. Air Force jets flying in a diamond formation; that was replaced with a Navy destroyer at sea; and that was replaced with Army tanks rushing across a desert.
Then, suddenly the music stopped and the screen went black. After a moment, a placard was placed on the screen.
By order of Mehdi Ohmshidi,
Supreme Leader, New World Collective,
Broadcasting on this network has been
suspended for engaging in acts of sedition
Fort Rucker—Monday, July 16
Although it had been some time since the president ordered a seventy-five percent reduction in force, no RIF orders had come down. That was understandable as there was almost a complete breakdown at all levels of the military, and the Pentagon was no longer issuing orders. Fort Rucker was a mere shadow of itself, practically a ghost town now, with only a few hundred soldiers still present for duty.
At this point, "Present for duty" was nothing more than an entry in the morning report, or it would have been if company clerks were still filing morning reports. But morning reports were no longer being filed because there were very few company clerks remaining and many of the clerks who did remain had no first sergeants or company commanders to validate the reports. Four fifths of the buildings on the base stood vacant, the classrooms and training facilities were empty, entire companies of the TO&E units were gone, and the base headquarters was just a shell with no more than two or three officers and NCOs still reporting for duty.
Those soldiers who were still reporting to their duty station did so as a matter of habit, and because they had nowhere else to go, or nothing else to do. They tried to hang on to a semblance of the lives they had before all this happened by coming to "work" though all they did was play hearts, bridge, poker, and blackjack. They gambled hundreds of dollars on every card, losing or winning with aplomb because, increasingly, money was losing its meaning. Most of the lower-ranking soldiers who did remain on the base did so only because the Army was still supplying them with quarters and food.
But even that was not a guarantee. The mess halls had not had a new delivery in the last two weeks, and the post was running critically low on provisions. Also there were few cooks remaining so, more often than not, the preparation of the food was being done by the soldiers themselves.
The stimulus package of one hundred thousand dollars issued by Ohmshidi, who now called himself Supreme Leader, to individuals to "jump-start" the economy, had long ago been used up. Those who cashed their checks immediately realized some benefit. Those who deposited their checks in the bank had money on paper, but not in reality, as a cascading closure of banks all across the country left much of the deposited money in limbo and unaccounted for.
By now, it made little difference what the money was worth anyway, as there was a steadily decreasing availability of goods and services. Automobile factories had shut down long ago, including foreign car companies, but the auto industry wasn't the only production stopped. No longer was there any major manufacturing of any kind, from aircraft, to household appliances, guns, and furniture, to clothing. In addition, food-processing plants had stopped so that no canned, frozen, or packaged food was being produced, and the food remaining in the nation's inventory was being used up at an alarming rate.
The value of stocks plummeted so far that there were no viable stocks remaining, and the market stopped all trading. Gasoline was rationed to five gallons per family, per week. The posted cost of gasoline, mandated by the government, was two hundred dollars per gallon, but the rationing and the cost were meaningless, as there were fewer and fewer service stations that actually had gasoline. Those few stations that did have gasoline would no longer sell for any amount of money, but would exchange it for something tangible. Bartering had become the new medium of exchange, and farmers and gardeners who had eggs, chickens, pigs, and vegetables became the new wealthy.
Unlike the Great Depression of the 1930s, when people who had cash were able to weather the storm, money meant nothing in this economy. Millionaires, and those billionaires who had managed to hide their money from Ohmshidi's "equalization" confiscation, discovered that it was all for naught. Those with assets in cash, stocks, and bonds, were totally wiped out.
Eventually all transportation came to a halt—the airlines halted operations, trains quit running, trucks stopped rolling. The interstate highway system had no traffic, though that wasn't to say that it had no cars. It had become the final resting place for millions of cars. There was a forced egalitarianism among automobiles, whether new and luxurious, or old and austere; they contributed equally to the national graveyard of vehicles, sitting alongside each other, abandoned right where they had run out of gas.
The trailers of the abandoned trucks had all been forced open and emptied of whatever cargo they might have been carrying. The state police no longer patrolled the roads and highways and, on those interstates not blocked off by parked cars, drivers who had gasoline, and who were foolish enough to waste it, could drive over one hundred miles per hour without worrying about a traffic ticket.
At Fort Rucker, as at nearly every other military base, the post exchange, commissary, and clubs were all closed. Aircraft sat unattended on the flight lines of all five Fort Rucker airfields. There was no traffic of military vehicles, and even the MPs, the few who remained, stayed in their quarters, or reported to the office only out of a sense of habit. Like the city and state police all across the nation, law enforcement was nonexistent.
The Daleville and Ozark gates were unattended, which meant the post could be entered by anyone, military or civilian, and there were increasing numbers of civilians wandering around the base to see what they could take, scavenging without opposition from any of the soldiers, most of whom were now scavenging for their own survival.
Nearly all of the local power-generating companies in and around Fort Rucker had gone off-line from their own resources, but had, so far, managed to maintain service by drawing electrical power from other grids around the nation. However, the condition was so precariously balanced that any unprogrammed surge could have catastrophic results. Fort Rucker was not affected because it was generating its own power, sufficient for the fort's use, though not enough to help the neighboring towns.
Tuesday, July 17
At the North American Electric Reliability Council, in an attempt to balance the electrical usage with decreasing fuel allocations, a switch was thrown, temporarily diverting so much power into the Ohio-based First Energy power lines that the system became overloaded. At the same time the warning system was short-circuited so that the monitors were unaware that a problem was developing.
1900 Zulu: A 689-megawatt coal plant in Eastlake, Ohio, went out of service.
2006 Zulu: A 345-kilovolt power line tripped off, putting strain on a neighboring line.
2032 Zulu: The power overload on the neighboring line caused it to sag and go out of service.
2115 Zulu: Two more 345-kilovolt lines failed within five minutes of each other.
2130 Zulu: Two 345-kilovolt lines in Michigan tripped off. A coal-fired plant near Grand Haven, Michigan went off-line.
2141 Zulu: A coal-fired power plant in Avon Lake, Ohio, went out of service.
2155 Zulu: The nuclear reactor in Perry, Ohio, shut down automatically, after losing power.
2200 Zulu: Systems from Detroit to New Jersey and Canada, including all of New York City, shut down.
That was followed by a massive power outage that first covered the northeast, then cascaded across all the power grids in the entire nation, so that by 2315 Zulu, or 6:15 P.M. Eastern Daylight Time, the entire nation from Canada to Mexico, and from the Atlantic to the Pacific, was blacked out. Only the tiniest communities that were not a part of the national grid, and who still had enough fuel to operate their generators, still had power. The loss of power took almost all television and radio broadcasts off the air. It also interrupted all communication between Fort Rucker and the Department of the Army. There was no longer any telephone, wire, or even Internet connection between Fort Rucker and the Department of Defense. Although the on-base telephones were still working, they were limited to on-base calls only. There was no longer any military cohesion. The few soldiers who remained wandered around the base without direction or purpose.
The Dunes, Fort Morgan—6 :14 P.M. CDT, Tuesday, July 17
Bob Varney was sitting on the couch in his living room watching the local news from the one television station that continued to operate in Mobile. How different the news was now with only one announcer and one camera. No longer were any of the features presented where the TV station would find people who had contributed to the news, either by participation in some newsworthy event, or by some odd little quirk that would often elicit a chuckle from the viewers. Those features had been eliminated because there was no longer enough fuel to allow the news reporters to go into the field.
Supreme Leader Ohmshidi announced today that he is asking the other nations of the world to come to our aid in this time of national crisis. He reminded the leaders of the other nations that we have always been quick to supply food, medicine, and humanitarian aid to other nations when they were in need. Now, according to Ohmshidi, the total mismanagement of the previous administration has made it difficult to implement his policies, resulting in a nationwide food shortage. The nation of Yazikistan, once our enemy, has announced that it will send three shiploads of food—proof, Ohmshidi says, that his policy of negotiation with the Muslim nations is paying dividends.
"Jesus," Bob said in disgust. "We are begging now. Can you believe that, Ellen? Our country has been reduced to the point of begging. I never thought I would live to see this day."
"You shouldn't watch the news," Ellen said. "It gets you so upset that I'm afraid you might have a stroke."
"There is no baseball to watch anymore and the only thing else on the satellite channels—that is, the ones that are still broadcasting—are reruns. And there are virtually no commercials because there are so many companies out of business. Who would have ever thought that I would long for the time when there were commercials?"
In Mobile today—
The television suddenly went black.
"Bob, the power just went off," Ellen said from the kitchen.
"It will probably be back on in a minute or so," Bob replied. "There's no storm or anything, no reason why it went off."
"Why don't you start the generator, at least long enough for me to finish cooking supper?"
"I hate to waste the gas," Bob said. "Hurricane season is here. You know how it was with Ivan and Katrina. We were without power for two weeks."
"Yes, but I also know how hard it is to get food now," Ellen replied. "If I don't keep cooking this now, it won't be any good."
"All right, I'll start the generator."
The generator was all the way downstairs, run by a large tank of propane gas. Bob started it, then came back upstairs.
"Stove going again?" he asked.
"Yes, thanks."
"Maybe I can watch the rest of the news now." He turned the TV back on, but the local television channel was black.
"Damn, they must be out of power in Mobile too. That's funny; we don't normally get power outages that go that far."
Bob swept through the channels but got nothing from any of them.
"I wonder if something happened to our satellite dish?"
"Try the radio," Ellen suggested.
Bob turned on the radio and got nothing on either the AM or the FM band.
"Damn, there's nothing on the radio either, not a thing. I'm going to try satellite radio."
Bob had a satellite radio upstairs in his office. It was there for two reasons: one, because it was the best place for his antenna, and the second, because he liked to listen to classical music as he was writing. As he started upstairs, Charley ran up the stairs ahead of him, then got into his position under the desk.
"I know, I know, I haven't written in a couple of weeks have I, Charley?" Bob said. "But I'm going to get back to it, I promise. Now I'm just going to see what I can find on the radio."
He found nothing.
"Something has happened," he said to Ellen as he came back into the kitchen. "It's not normal that every radio and television station be off the air."
"It's not normal?" Ellen asked. "Tell me anything that has been normal since Ohmshidi took office."
Charley reared up to put his two front paws on Bob's leg. Bob picked him up and Charley began kissing him.
"Charley," Bob said. "Charley is normal."
"Supper is ready," Ellen said, putting the plates on the table.
"Fried chicken. Looks good," Bob said.
"That's the end of our chicken," Ellen said. "And I don't know if there will be any more. As you recall, when we were in the store it was practically empty. I got as much as I could because I think it's only going to get worse."
"I think so too," Bob said. He picked up a drumstick. "But I intend to enjoy it while we have it."
CHAPTER THIRTEEN
Fort Rucker—Wednesday, July 18
In the Post Headquarters, General von Cairns sat in his swivel chair staring out through the window at the open parade ground and empty flagpole. He took another pull from his bottle of whiskey.
"You drink too much," his wife had told him. "Ever since you came back from the war, you drink too much."
"It helps me forget."
"Forget what?"
"What I want to forget."
Those discussions about drinking were the beginning of the problems that only got worse in his marriage. Then, two years ago, Kitty left him. But because their daughter was already married and gone, the breakup wasn't that traumatic.
Von Cairns looked over at his bookshelf, filled with various artifacts that cataloged his time in the Army. There was a photo of him in his football uniform at West Point. Their record with Navy for the time von Cairns was a cadet was three losses and one tie, that game having been played before the NCAA had put in the tie-breaker system. Von Cairns had helped preserve the tie by intercepting a midshipman's pass in the end zone. Among the other items were a bent tail rotor from a helicopter crash he had survived, a captured Iraqi flag, silver cups from half a dozen stateside assignments, and a black, polished, knobbed shillelagh called a "Garryowen stick."
His first assignment out of West Point had been to the Seventh Cavalry, now at Fort Stewart, Georgia, but then he was stationed at Conn and Ledward Casernes in Schweinfurt, Germany. The Seventh Cavalry, "Custer's Own," had the nickname "Garryowen" and all the officers carried a swagger stick, the Garryowen stick.
There was something wonderful about being alive then, the tradition of being a part of such a storied unit, the pure joy of being a young single officer on flight status—they were still flying Hueys then—and being in Germany itself, the beer houses, the restaurants, and the outdoor markets, especially the flower markets. Germany was where he met Kitty, who had been, at the time, a schoolteacher in the American dependent children's school.
He remembered the first time he ever saw her. It was in the officers' club and she was sitting at a table with one of the other teachers. She had long dark brown hair and a wave of it kept falling across one eye, causing her to have to push it back. He thought she was absolutely beautiful, and he went over to introduce himself to her.
"I am Lieutenant Clifton von Cairns, and after I take you out to dinner tonight, you are going to write to your mother and tell her all about me."
Kitty laughed at him, but she did go to dinner with him that night, and several nights thereafter until the other teachers and other officers would always include the two of them for any planned event.
They were so in love then. They went to Paris together, to London, and to Rome. They were married in Germany, first a German civil marriage, because the U.S. has no reciprocal marriage agreements with foreign countries, then in the post chapel, exiting under arched sabers. They swore to each that their love would last forever. But that—like everything else—was gone.
"Where are you now, Kitty?" von Cairns asked, speaking aloud, but quietly. "What are you doing this minute? Are you recalling our days together in Germany? Do you have any memories of that time? Are all your memories bad ones?" he continued, speaking softly. "I still love you, you know. I can't tell you that. This is not the time for that. But I do hope that you have found happiness."
Von Cairns turned up his whiskey bottle for another drink, but there was none left. And the way things were now, there was not likely to be any more.
Back in the BOQ, Lieutenant Phil Patterson was contemplating his situation. He rarely reported to work now; General von Cairns had told him there was really no need for him to come in. He had nothing left to read, and nothing left to do. There was no radio to listen to, no TV to watch. Newspapers and magazines were no longer being published.
How differently things had turned out from what he had imagined when, as a boy, he had dreams of going to West Point. He remembered that June day of graduation, the sense of accomplishment, and the pride he had in receiving his commission. Part of his dream had been to fly, and he fulfilled that as well, though it had now been over four months since last he flew.
He was frightened. Not the kind of fear one might get when his life is in imminent danger. There was, with that kind of fear, also a charge of adrenalin, an awareness of life, an excitement.
This fear was intense, mind-numbing, and paralyzing. He had no idea what was going to happen, but he knew it was going to be bad. Very bad. And worse, he knew there was absolutely nothing he could do about it.
He had been thinking about it for several days now, and had even made preparations. Today he was going to leave.
Lieutenant Patterson drove down to the Headquarters Building and went into General von Cairns's office. He wasn't going to ask the general if he could leave; he was going to tell him that he was going to leave. It wasn't something he was proud of, but it was something he was going to do, and he knew that, realistically, there was nothing the general could do about it.
He saw that the general's chair was turned around and, as he often did, the general was staring through the window at the empty parade ground outside. He had been doing that more and more lately. And drinking—Patterson saw the empty whiskey bottle on the desk, just above the open drawer. At one point he wondered if he should make some comment about the general's drinking, being very subtle about it. But he was just a lieutenant, and such a thing wasn't his place.
"General," Patterson said. "General, sir, I'm sorry sir, I wanted to stay and do all I could, but, to be honest, I don't think there is anything left here I can do anymore. In fact, I don't think there is anything anyone can do. My mom is alone back in Arkansas, and I haven't heard from her in two weeks. I'm worried about her, and I'd like to go home. If you want me to just take a leave, I will do so. But, the way things are now, well, I hope you understand." Patterson paused, waiting for the general to turn around and reply.
"General?" Patterson said again.
When the general still didn't turn around, Patterson walked around the desk.
"Sir?"
Patterson stepped up to the general's chair, then gasped out loud. The general's head was back against the headrest and lolling over to one side. In his hand was an Army-issue M9 pistol. There was a black hole in his forehead, but a surprising lack of blood.
"I guess I'll just go," Patterson said in a quiet voice.
As he left the general's office and walked through the cavernous HQ building, he did not encounter a single soul.
His car was full of gasoline and he had two five-gallon cans in the trunk. He hoped that would be enough to get him to Blytheville, Arkansas.
Fort Rucker—Friday, July 20
Jake was playing a game of solitaire when the phone rang.
"EFT, Major Lantz."
"Major, this is Mr. Tadlock."
"Yes, Chief, what can I do for you?"
"I'm calling you from the General's office."
"Okay, put him on."
"Uh, no sir, I can't do that."
"Oh?"
"The reason I'm calling is I wonder if you could come over here."
"I guess I can. I hate to waste the gas, but if the general needs me for some reason . . ."
"It isn't the general who needs you, Major. I need you. Or at least, I need your advice. The general is dead."
"What?"
"Looks like he committed suicide. He's sittin' here in his chair and there's a pistol in his lap."
"Where's Lieutenant Patterson?"
"I don't know, Major. He wasn't here when I got here. Fact is, there's not a soul in the whole damn building."
"I'll be there as quick as I can."
Jake hung up the phone, then called out into the outer office. "Sergeant Major, are you out there?"
"Yes, sir," Clay answered.
Jake walked out into the outer office and saw Clay just getting up from his desk.
"I'd like you to come to the general's office with me if you would," he said.
"Damn! He hasn't caught on about the fuel requisitions, has he?"
"No. Ed Tadlock just called. The general is dead, Clay. Tadlock said he shot himself."
"I'll be damned. I thought von Cairns had more sand than that."
Chief Warrant Officer-3 Edward Tadlock was waiting just outside the door to the Post Headquarters building when Jake and Clay arrived in Jake's Jeep SUV.
"I waited out here," Tadlock said. "I don't mind telling you, it's creepy as hell in there."
"How do you know it was a suicide?" Jake asked. "Did he leave a note?"
"No, there was no note. But the pistol is still in his hand."
"Let's have a look."
The three men went back inside the building, which, as Tadlock had said, was completely deserted.
"I'm taking off," Tadlock said. "I'm going to Missouri. I own a small farm there, I'm going back to work it. My wife and kids are already there, waiting for me."
"Do you have enough fuel to make it all the way to Missouri?"
"I'm driving a diesel, and running it on jet fuel. I bought thirty gallons extra from someone. I didn't ask any questions as to where he got it."
"Well, good luck to you, Chief," Jake said.
When they stepped into the general's office, he was still sitting in his swivel chair, facing the window that looked out over the parade ground.
"I left him just the way I found him," Tadlock said.
Jake walked around to get a closer look at him. He shook his head. "Damn," he said. "He was a good man. I hate to see this."
"Ohmshidi killed him," Tadlock said. "Yeah, von Cairns may have pulled the trigger, but Ohmshidi killed him."
"I can't argue with that," Jake said.
"So now the question is, what do we do with him?"
"Does he have any next of kin?" Clay asked.
"He's divorced, I know that," Tadlock said.
"He has a daughter somewhere," Jake said. "If we looked through all his things, we could probably find out where she is. But then what? The way things are now, what could she do with him?"
"We can't leave him here," Tadlock said.
"Let's bury him out there on the parade ground, under the flagpole," Clay suggested.
"Damn good idea, Sergeant Major, damn good idea," Tadlock said.
Clay went to the general's quarters to get his dress blue uniform and he and Jake dressed the general, including all his medals. While they were doing that, Tadlock rounded up as many officers and men as he could, including seven men who would form a firing squadron to render last honors, and one bandsman who agreed to play taps.
Now the general lay in a main-rotor shipping case alongside a grave that three of the EM had dug. There were over fifty men and women present, in uniform, and in formation. The general was lowered into the grave, and Jake nodded at the firing team. The seven soldiers raised their rifles to their shoulders.
"Ready? Fire!"
The sound of the first volley echoed back from the buildings adjacent to the parade ground.
"Ready? Fire!"
Rifle fire, which, during his life, the general had heard in anger, now sounded in his honor.
"Ready? Fire!"
The last volley was fired, and those who were rendering hand salutes brought them down sharply.
The bandsman, a bespectacled specialist, raised a trumpet to his lips and with the first and third valves depressed, played taps.
Jake thought of the many times he had heard this haunting bugle call, at night in the barracks while in basic training, and in OCS. He had also heard it played for too many of his friends, killed in combat or in aircraft accidents.
The young soldier played the call slowly and stately, holding the higher notes, gradually getting louder, then slowing the tempo as he reached the end, and holding the final, middle C longer than any other note before, he allowed it simply and sadly to . . . fade away.
CHAPTER FOURTEEN
The Dunes, Fort Morgan—Friday, July 27
Many of the houses at The Dunes had their own generators as a result of the long power outages following hurricanes. The problem was that at full usage the propane tanks could last for no more than a week. By rationing the usage, Bob was able to get ten days of service; then he, James, and Jerry began taking propane tanks from the unoccupied houses and using them for their own needs.
In addition to taking the propane tanks, they also took the food that many of the houses had in their freezer and pantries They ate the frozen food first so they would not have to keep it cold for too long—and at the beginning of their ordeal, they ate very well: steaks, roasts, lobster, shrimp, and fish.
For better utilization of the food the three families decided to take all their meals together.
"What is this? Leg of lamb?" Jerry asked as he cut into the meat.
"Yes. It came from Dr. Kelly's freezer," James said.
"You can say what you want about Dr. Kelly, the man does have good taste," Bob said. "That's also where we got the prime rib, isn't it?"
"And the lobster," James added.
James Laney did not have a college education, but he had worked himself up from general handyman to plant manager of the Cobb County Electric Cooperative in Georgia. He accomplished this by mastering every job in the plant and even though he was now retired, he kept himself busy by taking care of the houses of the absentee owners: doing electrical repair, plumbing, carpentry, and painting. Bob and Jerry had "elected" him mayor of The Dunes. The election was only partially in jest; James knew all the absentee owners, and had a key to every house.
James had married his wife Cille when he was seventeen and she was fifteen, and though everything seemed stacked against their marriage succeeding, they had celebrated their fiftieth wedding anniversary earlier this year.
Jerry Cornett was a retired salesman and now the consummate outdoorsman who had fished and hunted around the world. He did his hunting with bows that he made himself, and was a champion bowman, having won so many bow shooting contests that the magazine Bowhunt America pronounced him "America's Senior Champion Archer." Jerry's wife, Gaye, was a retired hairstylist.
Until the collapse of the economy, Bob Varney had been a successful novelist. Success had come late for him—he had written over three hundred books, but most were ghostwriting projects for others. And while his ghostwriting paid well, it left him with a sense of frustration. He saw many of the books he had authored become bestsellers, but he was unable to take advantage of them. Then, two years ago he convinced his editor to let him write his own books. Those books had done very well and at long last, let him build his own career.
Ironically, the career that had taken him so long to build was now nonexistent due to the collapse of the U.S. economy.
Bob was also a retired Army officer who had done three tours in Vietnam as a helicopter pilot. It had been forty years since he last flew, and he often watched, with a bit of nostalgia, the service helicopters pass over his house as they flew out to the offshore gas rigs.
Because of his writing and research, he was practically a walking encyclopedia, fascinated by trivia, and he knew diverse facts ranging from the number of words in the King James Version of the Bible—181,253—to the temperature of the sun at its core—twenty-seven million degrees. And he was a Custer expert who could name every officer who made that last scout with Custer, including those who were with Reno and Benteen, and tell what eventually happened to them.
Bob's only other talent was in cooking, but he excelled in that, his specialty being coming up with very satisfying meals by utilizing what was available. Bob's wife, Ellen, was a retired schoolteacher, who, in her youth, had taught in Point Hope, Alaska. That remote experience was coming in handy now that the small Fort Morgan Dunes group was isolated from the rest of the world.
Hampton Roads, Virginia—Saturday, July 28
The United States had a long record of helping other countries in time of need—from the earthquakes in Haiti to the Indonesian tsunami, going all the way back to the Marshall Plan that helped Europe recover from World War Two. Because of that, countries around the world, many of whom had been helped by the U.S. at one time or another now rose to the occasion offering aid to the stricken nation that had fallen from such great heights.
Three cargo ships, the Abdul, the Kishwan, and the Sahil, their containers filled with food and goods, left Yazikistan bound for the United States. All three were flying proud banners proclaiming:
Friendship from the Islamic Republic of Yazikistan
After crossing the ocean together, the three ships parted company off the coast of the U.S., then sailed on for three separate destinations: Boston, New York, and Hampton Roads.
As the three ships separated, Ariz Khabir Jawwad, who was on the Kishwan bound for Hampton Roads, went to the container that was labeled RICE.
Buried in the rice inside this container was a Russian SS-25 nuclear warhead, with an effective charge of one hundred kilotons, which was eight times more powerful than the bomb that was dropped on Hiroshima. There was also a similar nuclear warhead on each of the other two relief ships.
Jawwad recalled the briefing he and the other two martyrs for Allah had received, prior to being dispatched on their mission.
"We expect the initial casualties from the three bombs to exceed one million in number. That figure does not include those deaths that will come indirectly from disease or other longer-term fatalities. The overall effect of an attack on this scale is particularly numbing. Anyone trying to flee will find themselves travelling through contaminated areas. The pollution of water supplies, destruction of homes, and general devastation will result in secondary problems with disease. Radiation reduces the body's ability to fight off illness. The final number of deaths from the martyrdom would then be much higher."
Jawwad thought of the others on board this ship, and the two other ships. They were completely unaware that one of the cargo containers held a nuclear bomb. When Jawwad detonated the bomb it would not only kill the American infidels, it would also kill everyone on board this ship and most of them were, like Jawwad, Muslims, true members of the faith. In addition, as Jawwad had been a crew member of the Kishwan for the last three years, he had made many close friends among the other crew members on board. When he had mentioned to his trainers that innocent Muslims would die, he was told that he would be doing them all a favor.
"They will be welcomed into heaven as martyrs who died for the faith. Worry not about them."
The Kishwan dropped anchor in Hampton Roads at 9 A.M. local time. It was a coordinated arrival; the Abdul would be in New York and the Sahil would be in Boston at exactly the same time. This had been carefully arranged so the mercy missions of the three ships would achieve maximum exposure in the world press.
Jawwad checked his watch. It had been agreed by the three martyrs that they would detonate the bombs at exactly nine-thirty. Jawwad watched the minute hand reach the number six, and waited fifteen seconds until the sweep hand reached the number twelve.
"Allah Akbar!" he shouted.
"Jawwad, why do you . . ."
The light was so brilliant that people as far away as Richmond saw it. A huge fireball rolled out across the water, touching the land on each side. The heat inside the fireball was hotter than the sun and the ship was vaporized.
At the same time the bomb at Hampton Roads was detonated, the bombs in New York and Boston also exploded.
Concurrent with the nuclear blasts in the U.S., one nuke was detonated in Spain, one in France, one in Germany, and one in Great Britain. Three nuclear-tipped missiles were launched into Israel.
The governments of Iran, Iraq, Libya, Syria, Afghanistan, Lebanon, and Saudi Arabia received a message from Caliph Rafeek Syed demanding that they recognize the Greater Islamic Caliphate of Allah, with him as the grand caliph. Every government contacted acquiesced to the demand.
Fort Rucker—Monday, July 30
Although it had been two days since the nuclear bombs had been detonated on American soil, Jake Lantz knew nothing about them. General von Cairns had been dead for almost two weeks, and not only had the Department of the Army not appointed a new base commander, Jake was convinced that DA didn't even know General von Cairns was dead. For that matter, Jake wasn't entirely convinced that there even was a Department of the Army anymore.
Lieutenant Colonel Dave Royal, who was director of the Aviation Maintenance Course, had contacted Jake shortly after the funeral, asking if Jake knew of anyone remaining on the base who might outrank him. When Jake said he did not, Colonel Royal asked if Jake thought he should assume command.
"I think it is entirely up to you, Dave," Jake replied. "The truth is, I know and you know there is nothing left to command. Most of the officers and nearly all of the men are gone now. The only people left behind are those who have no other place to go."
"I think you are right," Colonel Royal said. "But I'm not one who likes loose ends, so if you have no objection, I think I will assume command."
That was ten days ago, and Jake had heard nothing from Lieutenant Colonel Royal since that time, and wasn't even sure Royal was still on the base. And that brought Jake to the moment of truth. He knew that there was no longer any reason for him to stay. His Army career had come to an end, and it was time for him to activate his survival team.
Pulling a yellow tablet from his desk, Jake began studying the names he had written. He had personally chosen everyone whose name was on the tablet, using three things for his governing criteria. First, each must possess a skill set that would be beneficial to the team as a whole, since he was convinced that they would be completely self-contained for a long time. Second, they must all be single. He had nothing against married people, but he was going to limit this team to no more than eight members, and spouses would be excess baggage. Third, he had to know them personally, and he had to like them.
The first name on his list was Karin Dawes.
CAPTAIN KARIN DAWES
If you had asked any of Karin's childhood friends about her, they would all say that the last thing any of them ever expected was for Karin to wind up in the Army. Would they think of her as a nurse? Perhaps. All agreed that she was someone who tended to look out for others, but nobody expected her to become an Army nurse. And certainly no one ever thought of her being in a combat zone, though not because anyone questioned her courage or her physical stamina. Karin had always been very athletic.
Karin was a distance runner and a cheerleader, both in high school and at the University of Kentucky. She continued to run after graduating from college and last year she came in first place among all women in both the Atlanta and the Chicago marathons.
While in college, Karin fell in love and planned to be married upon graduation. But only three days before the wedding, Tony Mason, her fiancé, was killed in a car wreck. Karin was so distraught by the event that she joined the Army. She was sent to Afghanistan immediately after her training, and there, she encountered a captain with severe wounds to his face, side, and leg. The wound in his leg got infected and there was a strong possibility that it would have to be amputated, but Karin made a special effort to tend to the wound, keeping him dosed with antibiotics, keeping it clean and aspirated, and treating it with antiseptic, sometimes spending the night in the room with the patient in order to attend to it. The captain's leg was saved. Dr. Metzger told the wounded officer, Captain Jake Lantz, that Karin had saved his leg.
It was not entirely by coincidence that when they returned to the States, they both wound up at Fort Rucker, Alabama. Karin thought she would never feel about another man as she had felt about Tony, but Jake Lantz had changed her mind. She would marry him in a minute if he would ask her, and if he didn't ask her, she might just ask him.
SERGEANT MAJOR CLAY MATTHEWS
Tactical officers in Officer Candidate School can be particularly brutal on the officer candidates, and if the TAC officers find some peculiarity, they can make it very hard. When the TACs learned that Jake had been raised Amish, they used it as a tool to try to break him. They called him "Amish boy" and asked if he knew how to drive a car, or turn on a light.
Sergeant Clay Matthews was a TAC NCO, and one day when the TAC officers were enjoying a field day with Jake, Sergeant Matthews called the young officer candidate to one side, telling the others that he would "get the candidate straight." When he was sure that the TAC officers were out of earshot, Clay spoke to Jake in a quiet and reassuring voice.
"Son, don't let these TACs get under your skin. They are paid to be assholes and they take their jobs seriously. But I've been watching you, and the truth is you've got more sand than anyone in this class, including the TACs. Stick with it; you are going to make a fine officer, one I would be proud to serve under."
Jake did graduate, then went almost immediately to college. After he got his degree he went to flight school, and after flight school he did serve with Sergeant Matthews in Iraq. During their service together, he took special pleasure in submitting Sergeant Matthews's name for the Silver Star when the sergeant pulled four of his fellow soldiers from a burning Humvee, then killed the six militants who had attacked them.
For the moment, Clay Matthews, who was forty-eight years old and divorced, was the noncommissioned officer in charge of Environmental Flight Tactics, Jake's department. At least he had been the NCOIC while EFT still existed.
Jake valued dependability above all else in the people he worked with, and Clay was the most dependable man, soldier or civilian, Jake had ever known. If you asked Clay to take care of a job, you could put that job out of your mind, because Clay would take care of it. He would be, Jake was sure, his most valuable asset in setting up a survivalist team. With him, Jake felt confident of success. Without him, it would be iffy at best.
SERGEANTS JOHN DEEDLE AND MARCUS WARNER
Deedle and Warner were helicopter mechanics who worked on the flight line. In all the time Jake had been in the Army he had never run across any who were better at their job, or any two men who better complemented each other's work. And their skills weren't limited just to aircraft maintenance. The two men were among the most resourceful he had ever met.
At twenty-four, Deedle was the younger of the two. By general consensus, John Deedle was the best mechanic on the base. Some swore he was the best in the Army. He was quick to diagnose problems, knew all the special tools and parts, and could, if asked, even though it was against Army procedures, disassemble and rebuild, fashioning parts if need be to fit any engine, transmission, or rotor component.
Marcus Warner was twenty-six and had been married, but was now divorced. He had no children and his wife had since remarried, so he had no obligations that would detract from his being a team member. Marcus Warner's mother had been born in Italy and immigrated to the U.S. when she was twelve. He grew up speaking Italian and English, but demonstrated a flair for languages early in his life. After he joined the Army he went to its language school at Presidio. He was fluent in Italian, Spanish, Portuguese, French, and German. But, since the Army had no particular need for any of those languages, he applied for, and was accepted into, the helicopter maintenance course. He had pulled two tours in Iraq, and was now a technical inspector for the school's fleet of aircraft.
SERGEANT FIRST CLASS WILLIE STARK
Willie Stark was thirty years old. An avionics specialist, Stark could practically build a radio from scratch. He not only knew all the inner workings and hidden mechanisms of any radio, he could read and send messages in Morse code. Stark, whom others referred to as an "electronics geek," was a wizard around radios and computers, but was too shy to have much experience with women. He had never been married, and had never even had a serious relationship with a woman.
SERGEANT DEON PRATT
Deon Pratt, twenty-five, was a powerfully built black man who was an instructor in the Escape and Evasion course at Fort Rucker. The consummate warrior, Deon was skilled in hand-to-hand combat. He was also an expert in firearms and explosives. Deon had won a Silver Star in Afghanistan for killing fifteen enemy fighters and rescuing, under fire, his captain and first sergeant who had been wounded and were trapped beneath a collapsed building. Jake had thought long and hard about including a combat expert, but he realized that if there was a complete breakdown of civilization, Sergeant Pratt would be a good man to have on his side.
SERGEANT JULIE NORTON
Sergeant Norton was Karin's recommendation. She worked in the hospital with Karin, primarily as a clerk, but the beautiful twenty-two-year-old black woman was an organizing genius, an efficiency expert who had taken the post hospital from a barely functioning mess to one of the best-organized hospitals in the Army.
Julie was from a mill town in Georgia, and when she was still in high school, she volunteered to work in the poorest sections, holding fund-raisers to buy food, tutoring the young, helping single mothers to cope. One day playing softball with some of the children, she saw two thugs attack an older woman who had just cashed her Social Security check. While others stood by doing nothing, Julie took the ball bat and drove both of them off before they could get the money.
So far, Jake had not shared his idea of forming a survival team with anyone except Karin and Clay Matthews. He was very familiar with the people he was considering, respected all of them, and knew that they respected him. And although he had not yet asked them to join him, he was certain that they would come if asked. In fact, he was so certain, that he had not considered anyone else.
The ringing phone surprised Jake. As there was no business being conducted anywhere on the base, the telephones had been quiet.
"Environmental Flight, Major Lantz," he answered, mouthing the Army answering formula of name, rank, and unit, even though none of it mattered any longer.
"Jake? Have you heard?" Even in Karin's quiet voice, Jake could hear the fear coming through.
"Have I heard what?"
"There were three nuclear bombs detonated day before yesterday, one in Boston, one in New York, and one Norfolk, Virginia."
"What! How do you know?"
"Dr. Urban has a shortwave radio. He picked up a radio station in Canada."
"And the Canadian station said there were three nukes here?"
"Yes, but it wasn't just here. There were bombs in England, France, Germany, Spain, and Israel too."
"What are the casualties here?"
"I don't know, I haven't heard. But it has to be several million. Jake, the worst has happened, the absolute worst."
"Come to my office, Karin. Bring Sergeant Norton. I'm going to call the others. I think it is time we assembled our team."
CHAPTER FIFTEEN
Jake had invited everyone on his team to come to his house for dinner. Since it was getting increasingly more difficult to come by food, the meal would have been incentive enough, even if they hadn't been interested in his idea of a survival team.
Jake served Meals Ready to Eat—or MREs. He also had several containers of fresh water, having taken that precaution before the water stopped running. There was no electricity in his house, but there were several candles on the dining room table so there was adequate light. He had two Coleman lanterns, but he wasn't using them because he wanted to be sparing of the fuel.
His house was ten miles from the base and he wasn't sure if all would be able to find transportation, but was relieved when he saw that Deedle, Warner, and Stark had arrived with Clay in his Jeep Liberty. Deon Pratt and Julie Norton arrived with Karin in her Camry.
"Oh, a candlelight soirée," John Deedle teased when they came into the candlelit house. "Been a while since I went to one of them."
"Ha," Marcus Warner said. "I can just see you at a candlelight soirée."
"I'm sorry that I can only serve MREs," Jake said as he passed out the little packets.
"Major, I haven't eaten since yesterday morning," Deon Pratt said. "To me, this is like a Thanksgiving feast." Pratt tore into the packet, emptied the food on his plate, then began wolfing it down, ravenously.
Jake held up his hand. "That's another thing," he said. "The Army is no more, so I am no longer a major. I am Jake, to all of you."
"Now, Major—I mean, Jake, for an old lifer like me, that's goin' to take a little gettin' used to," Clay Matthews said. "I've been in this Army, man and boy, for thirty-one years."
"It ought to be easier for you than anyone else, Clay," Jake teased. "I remember when I was an officer candidate and you used to chew my ass."
"Yes, sir, well, sometimes your ass needed chewin'," Clay replied, and the others laughed.
"I know that Clay, Marcus, Willie, and John all know each other. Karin, you and Julie know each other. Deon and I know each other, so what I'd like is for you all to get acquainted while you are enjoying the delicious meal that I spent all day preparing for you. After that, I'll tell you why I've gathered you together, and what I think we should do. I'm going to need two things from you. I'm going to need your willingness to participate, and I'm going to need some input as to how best to accomplish what we need to do."
"I'll get us started," Clay said. Clay was just over six feet tall, with a square jaw and a flat-top haircut. Dark at one time, his hair was now gray. "My name is Clay Matthews. I am a fourth-generation soldier, born at the base hospital at Fort Benning, Georgia. However, there is absolutely no truth to the rumor that I was issued on a DA form 3161."
The others laughed.
"I've known the major—that is, Jake—for almost twelve years, and he is as fine an officer as I ever served with. He has shared with me what he has planned, and I am happy to be a part of it, but I'm getting a little long in the tooth, so I just hope I can pull my own weight."
Clay sat down.
"Clay, you were only about a year younger than you are now when you pulled four men out of a burning Hummer just before it exploded. I'm pretty sure you can pull your weight with us," Jake said.
"I'll go next," John Deedle said. Deedle was very short, and had a rather large nose on his narrow face. He had heavy brows and deep-set, dark eyes. "I'm a mechanic, and the way Sarge—uh, that is Clay—was born to the Army, I was born to maintenance. My dad owned a mechanic shop—he worked on cars, tractors, trucks, anything that had an engine. When I was no more than five years old, my dad would hold me by my heels and dangle me down into engine compartments because I was small and could get to nuts and bolts."
Willie Stark held up his hand, and Clay nodded.
Willie Stark was average height and weight with red hair and light blue eyes, which seemed slightly enlarged behind his glasses. "I'm Willie Stark, I'm in avionics—that is, I was in avionics until the Army had no further use for me. I'm also what you might call a computer geek, but there isn't a lot of call for that right now, either. But I do like radios, all kinds of radios, and I built my first complete radio with I was twelve years old. Before that, I built a crystal radio set."
"What's a crystal radio?" Julie Norton asked.
"It's a radio that doesn't use power, doesn't need a battery or electricity. It gets its power from the radio waves themselves."
Willie sat down, and Julie stood. Julie was tall and willowy, a black woman with blemish-free skin that was the same shade of golden brown as a piece of toast. She had huge, almond-shaped, very dark eyes under long black lashes, a physical attribute that had served her well in the beauty contests she used to enter.
"I'm not sure why I was invited to join this group, except maybe because I know Captain—uh, that is, Karin Dawes. And if that is the only thing that got me in, then I'm thankful for it. About the only thing I'm good for is keeping track of things. And since we no longer have anything to keep track of, not even that talent is worth anything now."
"I think just the opposite is true," Jake said. "It is when things are very scarce that we need to keep track. And if we are going to succeed, we are going to have to be organized. Karin tells me that you are the best at that of anyone she has ever seen."
"Yes, sir, well, I guess I get that from my mama. She worked for Mr. Simmons at the mill, kept all his books for him, kept his schedule too. He wound up making her his second in charge. Oh, did that make my grandmother proud. My grandmother had never been anything but a maid and here was her daughter, my mama, bossing around white men who used to boss my grandmother. Anyway, Mama started me out by making me keep my own things neat and in order, and it just sort of grew from there."
"You are my friend, Julie," Karin said. "But more than that, you are someone we are going to need."
"I second that," Jake said. "Marcus? Or do you prefer Mark?" Jake asked.
"I've always gone by Marcus."
"All right, Marcus, what do you bring to the table?"
"Hola, mi nombre es Marcus Warner. Soy mecánico de aviones, y hablo seis lenguas," Marcus began. When he saw the others looking at other in confusion, he laughed. "I just told you my name, explained that I was an aircraft mechanic, and could speak in four languages. I spoke to you in Spanish, but could have spoken to you in French, Italian, German, or Portuguese. I don't know that we will run into anyone from any of those other countries, but if we do, we will be able to talk to them."
"Deon?" Jake said.
Deon Pratt was dark-skinned with wide shoulders, powerful arms, and a flat, rock-hard stomach. "My field is weapons, explosives, hand-to-hand combat," Deon said. I'm a warrior without a war. Hell, I'm a warrior without even an army. But I'm glad to be with you guys, and I'll do whatever I can to make this thing work."
Karin was last to introduce herself, stressing only that, as a nurse, she would do what she could to keep everyone healthy.
"Thanks," Jake said when all had finished. "Now I suppose it is time for me to tell you what I have in mind, and I can say it in one word. Survival."
"I'm all for surviving," John said. "Where do we start?"
"What I propose now is that we spend the next several hours rounding up as much survival gear as we can. As soon as we are ready, we'll leave this area and set up a survival base."
"Where are we going?" Clay asked.
"Fort Morgan."
"Where?" Deon asked.
"It's an old fort, built long before the Civil War, down at Mobile Point, a little spit of land that separates Mobile Bay from the Gulf of Mexico. You may have heard the term, 'Damn the torpedoes, full speed ahead.' Fort Morgan is where Admiral Farragut was when he ran under the guns of the fort, and through the mines that blocked the entrance into Mobile Bay."
"What's at the fort now?" Willie asked.
"Now, there is nothing there but old casements and stone walls. But inside the fort is a rather large area of arable land, probably the only place down on the beach that has real soil, rather than sand. We could grow a substantial garden there."
"Sounds good to me," Julie said. "I use to love to work in my grandma's garden."
"I've chosen Fort Morgan for a number of reasons," Jake said. "Number-one reason is perhaps, the most obvious. It is a fort, and as conditions deteriorate even further in the country, I believe there are going to be armed bands of hooligans preying on anyone and everyone they think might have something they could use. As we are going to be well off, relatively speaking, we would be a prime target for such groups. The fort will provide us with protection.
"And as I said, there is arable land inside the fort where we can grow vegetables. Plus, there will be plenty of fish, and there is a considerable amount of game, from rabbits to possum to alligators."
"Alligators?" Marcus said. "Alligators as game?"
"Fried alligator tastes . . ."
"Like chicken, right?" Willie interrupted, and the others laughed.
"Wrong," Jake said. "It's a lot better than chicken."
"How are we going to get there?" John asked. "What I mean is, how will we get enough gas to drive that far? For that matter, if there are going to be these roving gangs you are talking about, I'm not sure I want to drive through them. I saw what IEDs could do in Afghanistan, and there we were in up-armored Humvees, or even armored personnel carriers. I wouldn't want to face one of them in a car."
"We aren't driving down, we're flying. We'll take one of the helicopters from Fort Rucker. I don't think the Army will miss it."
"We may have a problem there, sir—I mean, Jake," Marcus said. "For the last month, people have been stripping the helicopters down pretty good. I know for a fact that we don't have one flyable ship on the entire base."
"I know that too," Jake said. "But that's where you and John will earn your keep. We're going to take parts from as many aircraft as we need in order to get one that is flyable."
"What about fuel?" Marcus asked.
Jake chuckled. "Clay, you want to handle this?"
"Several weeks ago, when the major saw this coming, he asked me to find some way to put a little fuel aside as an emergency. I have fifteen fifty-five-gallon drums of JP-4 hidden away."
"Wow! That will top off the tanks, and give us one hundred ten gallons extra," John said.
"That's right," Jake replied.
"I have a question," Deon said. "There are already a lot more civilians roaming around the base than there are soldiers—looking to see what they can rip off. If they see us building a helicopter, they are likely to give us some trouble."
"We'll find a secure hangar to build the helicopter," Jake said. "And, from this day forward, we will wear no uniforms. That way if anyone sees us messing around out there, they'll think we are no different from them, we are just trying to find something to trade."
"May I make a suggestion?" Deon asked.
"Of course you can."
"Wherever this hangar is, I think I had better provide a little security."
Jake smiled. "I was hoping you would say that."
"You going to move out onto the base, Jake?" Clay asked. "Because if you don't, this drive back and forth to town is going to get long, and use up what gasoline you have left"
"I am going to move onto the base. In fact I think we should all bivouac together in the hangar. But before we leave town, I suggest we go on a scavenger mission. We need to round up as much useful supplies as we can."
"From what I have heard, people have been looting stores for the last three weeks. I doubt there is much left," Willie said.
"Mostly they have been stealing canned goods, packaged foods, that sort of thing. Some of them have even been stealing TV sets, though God only knows why since there is no more television," Jake said. "We need to be a little more discriminating on what we take."
"Like what?" John asked.
"I've made a list," Jake replied. Opening a drawer on the bar that separated the kitchen from the dining room, he began reading from the list. "A multi-tool knife, compass, flashlight with the windup generators, first-aid kits, blankets, matches, a lighter or lighters if we can come up with them, sunscreen, mosquito repellent, whistles and signal mirrors, nylon cord—that we can get from parachutes, and we'll use the parachute canopies as well. I already have two parachutes laid aside. Anyone have something they would like to add?"
"How about MREs?" Deon asked.
Jake smiled. "Are you enjoying these?"
"Yeah, but I'm weird, I always have liked MREs."
"Good, because we have fifteen cases of them down in the basement. That's one hundred eighty meals, which will last us for a little under two weeks if we have no more than two meals per day."
"What happens at the end of that two weeks? Marcus asked.
"Fish, game, whatever wild vegetables we can come up with. If we haven't found a way to be self-sustaining within two weeks, then we probably wouldn't make it anyway," Jake said. "Any other suggestions?"
"How about some tablets and pencils?" Julie asked.
"Yes, very good idea."
"How about toilet paper?" Karin added. "I mean this roughing it goes only so far."
"Toilet paper is good," Jake said. "But no matter how much of it we round up, there is going to come a time when we go through it. Then we'll have to come up with something else."
"Well, hell, Jake, if we can't find enough food, maybe we won't even be needing any toilet paper," Marcus said, and the others laughed.
"I'd rather face the problem of having to find toilet paper," Clay said.
"Me too," Willie agreed.
"If we are going to maintain this helicopter once we get down to Fort Morgan, we are going to need a full mechanic's toolbox, as well as some special tools," John said.
"If I have guessed right about you, you have your toolbox hidden away somewhere," Jake said.
John smiled. "Yeah, I do."
"We'll also be needing some replacement parts, especially filters, gaskets, and so forth," Marcus added.
"And at least one machine gun to mount in the doors," Deon said.
"You're right, an M240 machine gun might not be a bad thing to have once we get down to the fort," Clay said.
"Where are we going to find those?" Marcus asked. "Weapons were about the first things to go—a lot of soldiers sold them to civilians."
"I know where there is an untapped armory," Deon said. "M240s, M16s, pistols, ammunition. I don't know anything about repairing helicopters, and I don't know if we are going to have enough food to sustain us. But I know for a fact that we will be heavily armed."
"Good man," Jake said.
"What about radios?" Willie asked.
"I've been anticipating this for a while now," Jake said. "I have two Midland Radios, XT511 base camp emergency crank radios with GMRS two-way radio technology—AC, DC or hand-crank charger for BATT5R batteries."
"Wow, you went all out, didn't you?" Willie said.
"I'm reasonably sure there will be some, at least, shortwave radio out there. If there is, we will need to access it and, perhaps, to communicate with others like us."
"What about money?" Marcus asked.
Jake shook his head. "Money is only as good as the government that backs it," he said. "And as of now, we have no government."
"Gold coins?"
"No. Gold might be good if there was a viable market. We not only don't have a government, we don't have a market. When you go 'shopping' tonight, just take what you find, but don't take it from anyone else unless you barter something for it. I am all for our survival, but I'm not ready to put someone else at risk."
"When do we get started?" Clay asked.
"We start now," Jake replied. "We'll go into every store and abandoned building in town. Also, the mall. Clay, you want to pass out the arms?"
Clay nodded, then walked over to the corner to an olive-drab B-4 bag. "I don't have as big an armory as Deon is talking about, but I've got enough for us to get by right now." Opening the bag, he pulled out an M9 pistol and held it up. "I have one of these for each of us," he said. All eight pistols have full, fifteen-round magazines, plus I have five hundred additional rounds of nine millimeter ammunition."
"Carry these pistols with you," Jake said, "but use them only if it is absolutely necessary to defend yourself."
"I've never fired one of these," Julie said as she was given the pistol.
"It's easy enough," Deon said. "Here is the safety. When the safety is off, all you have to do is pull the trigger."
"I also have four flashlights and four small RadioShack special two-way radios. The small limited-range two-way radios will serve us well for keeping touch with each other. And four is all we will need, as we will deploy in two-man teams," Jake said.
"Good idea. And it is probably safer that we deploy in two-man teams," Deon said.
"Thanks for the endorsement," Jake said. "Karin, you go with me, and Julie, I suggest you go with Deon. The rest of you team up however you want." Jake looked at his watch. "It is nineteen thirty hours now, we will rendezvous back here at twenty-four hundred. Any questions?"
"Yeah, I have one," John said. "When we get to the checkout counter, do we ask for plastic or paper bags?"
The others laughed.
"Major, these radios are pretty standard," Willie said. "That means that anyone who wants to can listen in. I suggest that we adopt call signs, rather than use names."
"Good idea, Willie. How about you assign them?"
"Okay. You'll be Vexation Six."
"Negative," Clay spoke up quickly. "Six designates the commander. Anyone out there listening who has ever been in the Army will recognize that in a heartbeat. I think we should keep the call signs as innocuous as we can."
"Yeah," Willie said. "You are right. Okay, how about Mickey Mouse one through four?"
"Alright. Jake, you and Karin are Mickey Mouse One; John, you and Marcus are Mickey Mouse Two; Deon, you and Julie are Three; and Clay and I will be Mickey Mouse Four," Willie said.
"Let's test them out," Jake suggested.
The four radios were checked; then radios, pistols, and flashlights were tucked into pockets.
"Back here at twenty-four hundred," Jake reminded them as they left on their appointed rounds.
CHAPTER SIXTEEN
By the time Jake and Karin arrived, there was very little left of the Wal-Mart Supercenter on South 231. The doors had been smashed in, and Karin started to step inside, but Jake held out his hand to stop her.
"Wait," he whispered. "Let's make sure nobody else is in here."
The two stood quietly just inside the store for a long moment. The store was so dark that they couldn't see two feet in front of them, which meant that if anyone was here they would have to be using a light, and the light could be seen.
They saw no light, and they heard no sound. After waiting about a minute, Jake turned on the large flashlight he was carrying.
"I think we've got it all to ourselves, such as it is," Jake said.
As the moved deeper inside they could see that what merchandise did remain was scattered around on the floor. There was a large yellow smiley face next to a sign that said SHOP WAL-MART.
Jake moved the light back and forth on the floor so they could see to pick their way through without tripping over anything.
Though the food products, clothing, and small utensils had been well cleaned out, the large-ticket items, TVs, etc., remained. Under ordinary circumstances, this would have been strange, but because there were no television stations broadcasting anywhere in America, at least as far as Jake knew, seeing the TV sets still sitting on the shelves wasn't at all surprising. However many of the TV sets had been smashed, not incidentally, but purposely, as an expression of anger and frustration.
Over each empty aisle in the food store were signs that told what product had once been there. Now the signs were little more than a tantalizing tease.
COOKIES, CRACKERS, CHIPS, AND SNACKS
RICE, BEANS, SPAGHETTI, NOODLES
SOUPS, CANNED MEAT
COLD AND HOT CEREALS
COFFEE, SOFT DRINKS
"Soft drinks," Karin said. "Do you think . . . ?"
"I bought the last root beer they had when they were still doing business," Jake said.
There was not one food item remaining anywhere in the store. Not even bulk, uncooked items, such as rice, flour, or beans.
In the book and magazine section, there were several soft-cover books scattered around on the floor.
"Let's grab as many of these as we can," Jake suggested. "Without TV or radio, I expect reading will be about our only source of entertainment."
"Good idea," Karin replied. "What do you like?"
"Westerns, action stories, just about anything, I guess. I think we are far beyond the ability to be choosy."
"Look, tablets and pencils," Karin said, scooping up several of them from the same aisle as the books. "This will please Julie."
As they moved on through the store, Jake saw a box underneath a turned-over stocking shelf. Pushing the shelf out of the way he saw that the box, though not completely full, had at least ten packages of "sandwich cookies, peanut-butter filling."
"Whoa, now this is going to be a treat," he said, stuffing the cookies down into the large, canvas bag.
"Where to now?" Karin asked.
"Let's go to the garden shop," Jake suggested.
Amazingly, the garden shop was virtually untouched. There, Jake found a wheelbarrow, which he loaded with a couple of watering cans, spades, rakes, and dozens of packets of seed from half a dozen vegetables. Here, too, he found insect repellent and he put as many cans as he could into the wheelbarrow.
"Wait," Jake said, stopping at one shelf. "These are the seeds we want."
"What do you mean? What's wrong with what we have?
"These are non-hybrid seeds. I can't believe there are so many of them."
"What are non-hybrid seeds?"
"Almost all the vegetables we see today are hybrids. Hybrid vegetables make the best vegetables, but they can't be counted on to produce seed that will reproduce. For that you need seeds in their original form. That's what this is."
Jake scooped up several packets, getting much more seed than he would need.
"If we can stay alive until these seeds produce, we'll be in good shape," Jake said.
Karin laughed. "Oh, great. All we have to do is stay alive? Yes, I'm for that."
Finally, with a completely stuffed B-4 bag on top of the filled wheelbarrow, Jake and Karin stepped back through the smashed doors and started across the nearly vacant parking lot toward Jake's Volvo.
Jake saw a pickup truck parked next to his car, and he knew, at once, that the pickup truck driver was either siphoning, or about to siphon, gas from his car. He heard a loud, crunching sound, and realized that the driver had not started yet because he had been held back by the locked cover over the gas cap.
Jake set the wheelbarrow down and ran quickly toward his car. The gas thief had a tire iron and was trying to pry up the cover. He was so intent on breaking into the gas tank that he had not seen Jake approach.
"Mister, I paid an arm and a leg for that gasoline and I don't intend to stand by and watch you steal it," Jake said.
Jake's voice startled the would-be thief, and he glanced up at Jake with a wild look in his eyes. He raised the tire iron he was using over his head.
"Stay back, Major," he said, remembering Jake's military rank. "Stay back or I'll lay your head open."
"You recognized me," Jake said. "Are you a soldier ?"
"I was. But there ain't nobody a soldier no more, not even you," the wild-eyed young man said. "And you bein' a major don't mean jack shit to me no more. So you just stand over there—sir." He slurred the word sir, setting it apart to show his disdain. "And soon as I drain your tank, I'll be on my way."
Jake pulled his pistol and pointed it at the young man. "Son, you need to learn not to bring a tire iron to a gunfight. Now my recommendation to you is that you climb in your truck and you drive away. Otherwise I'll just have to shoot you."
Seeing the gun in Jake's hand, the young man's demeanor changed. No longer belligerent, he lowered the tire iron he had been using to pry open the gas-cap cover.
"All right, all right, I'm goin'," the young man said, holding one hand out in front of him, palm facing Jake as if by so doing, he could hold Jake off. He glanced at the right rear quarter of Jake's car. The paint was badly scratched and dented where he had been working to open the gas-cap cover. "I, uh, I'm sorry I messed up your car."
"Don't talk anymore," Jake said, coldly. "You piss me off every time you open your mouth. Just shut up, get in your truck, and drive away from here."
The young man threw the tire iron into the back of his truck, hurried around to the driver's side, got in, and drove away.
"Come on, Karin!" Jake shouted. "Let's get out of here."
By midnight everyone had returned to Jake's house and they put their acquisitions together to see how well they had done.
Clay and Marcus scored two five-gallon cans of gasoline. Jake didn't ask where, or how, they got it.
Deon and Julie returned with fifty pounds of flour, ten pounds of sugar, twenty pounds of rice, twenty-five pounds of dried beans, and five gallons of cooking oil.
"Where did you find this?" Jake asked. "I can't imagine any grocery store or warehouse still having any of this left."
"We got it from a VFW kitchen," Deon said.
"Whoa, good thinking."
"It was Julie's idea."
"My aunt used to work as a cook in the VFW back in Georgia," Julie said. "I know her kitchen was always well stocked and I thought there was a chance that nobody would think to look there."
"We also got this," Deon said, pulling something out of a sack. It was a bullhorn and he held it up to his mouth, then pulled the talk trigger.
"Jumpers in the air, you have a sixty-knot wind coming from your right!"
"Whoa!" Clay said, laughing. "That's a hell of a wind to be jumping into."
"Maybe for a leg," Deon teased. "Not for an airborne troop like me."
"I'm glad you came up with that thing," Jake said.
"Why, what are we going to use it for?" Karin asked.
"You heard Deon. What if we see some paratroopers in the air? We might have to give them directions."
The others laughed.
"Okay, you guys did well. You did very well in fact," Jake said. "So now, I suggest we spend the rest of the night here, then go out to the post in the morning. Our first order of business will be to find a hangar we can secure; second will be to find a helicopter we can put into flying condition."
The next morning the eight gathered for breakfast in Jake's dining room, again eating MREs though, as Jake explained, these were from a broken case and not part of the fifteen cases he had for their survival supplies.
Karin looked around the dining room, gray walls set off by a large seascape painting, a dark blue carpet, and off-white upholstered chairs.
Jake saw where she was looking and he reached out to put his hand on hers. "You are thinking about this room and how we decorated it together, aren't you?"
"Jake, will we ever come back here?"
"I don't know," Jake replied. "I know that's not what you wanted to hear, but I have to be honest with you. I truly don't know if we will ever be able to come back or not. And if we do come back, what will we find?"
Karin nodded. "I know," she said. "And I'm okay with it."
Jake squeezed her hand, then looked over toward Willie.
"Willie, what do you say we crank up one of these radios and see if we can pick up any news on the shortwave bands?" he suggested.
"Good idea, yes, let's see what's out there," Clay agreed.
Willie cranked the radio for one minute; then he turned it on and started sweeping through the frequencies.
"Getting carrier waves," he said. "That's good."
"What does that mean?"
"That means there are some transmitting stations that are still up, just nobody talking on them right now."
Marcus continued to turn the dial until he picked up a woman's voice. She was clearly on the edge of panic.
"Someone, anyone," she was saying. "Can anyone hear me? This is Yellowbird. Can anyone hear me?"
Willie keyed the microphone.
"Yellowbird, this is Mickey Mouse. Over."
"Mickey Mouse, oh, thank God! There is someone out there!" The woman practically shouted in her excitement.
"Where are you, Yellowbird? What is your status? Over."
"I'm in Portsmouth, Virginia, real close to where the bomb went off."
"Are you safe?" Marcus asked.
"Safe? What is safe? We weren't hurt by the bomb, but I don't know about the radiation. We are so close."
"You say we. Who is we?"
"My husband, our two children, my brother and sister-in-law, and their three children."
"Are there others around?"
"Nobody that we want to associate with. There are a lot of men wandering around outside, shouting and breaking into houses and cars. We've heard screams and shooting. I've been trying to contact the police, but haven't been able to do so."
"What you need to do is get out of there," Willie said. "There are no police."
"How do you know there are no police? Oh!" In the background, Marcus could hear loud voices and the sound of shooting. "Can you hear that? Why don't the police come?"
"Yes, ma'am, I can hear it. But you can't count on the police. There has been a complete breakdown of all government agencies including the police. Do you have a car? And if so, do you have gasoline in your car?"
"I . . ." There was a long pause before the woman came back on the air. "My husband says I shouldn't answer that."
Now everyone was huddled around the radio listening to the woman's terrified voice from the other end, hearing, also, the shouts and the shooting.
"Your husband is correct, ma'am, you shouldn't tell," Willie said. "And I apologize for asking. But my advice to you is this. If you have a car with fuel, pack as much food, water, blankets, matches, and other such items as you might have, then get as far away from there as you can. The farther away from people you are, the safer you will be. Over."
There was a long silence, and Willie keyed the mic to speak again. "Yellowbird, do you read me? Over."
Still no reply.
"What happened?" Julie asked. "Why doesn't she answer?"
"I don't know." Willie keyed the mic again. "Yellowbird, if you can read me, pack as much food, water, blankets, matches, and other such items as you have, then get as far away from there as you can. Over."
"Good advice, dipshit," a man's gruff voice replied. "But the little lady and her family won't be needing it now. Over," he added with a malevolent laugh.
Willie did not respond. Instead he pinched the bridge of his nose and shook his head. "Damn," he said quietly.
"See if you can find a broadcast somewhere," Jake suggested. "I mean a real news broadcast."
"Yeah," Willie said. "There's nothing we can do for Yellowbird."
Willie continued turning the dial, picking up whistles, static, sidetones, and carrier waves.
El gobierno Mexicano ha cerrado la frontera para impedir a Estadounidenses de inundar nuestro país.
"What is he saying?" Jake asked Marcus.
Marcus chuckled. "How is this for irony? He is saying that the Mexican government has closed its borders to keep Americans from flooding into their country."
"See if you can find something in English," Clay said.
Willie turned the dial again, finally picking up an English broadcast.
. . . broadcasting over this shortwave frequency in the hope that there are some people out there with shortwave radios who can hear us.
"Hey, that's George Gregoire," Jake said. "I recognized his voice."
"Damn, I used to watch him," Clay said. "I thought they drug him off and killed him."
"Evidently not. Let's see what he has to say," Jake said.
Everyone drew close to the radio.
Hello, America.
I can't tell you where I am. As I'm sure you know, I am now a wanted man. I never broadcast from the same place twice for fear of the SPS homing in on my radio signal. And to be honest with you, I don't even know if there is an SPS anymore.
I have a small group of dedicated people with me and they have been tuning in to shortwave broadcasts from around the world. We do that so we can keep you up to date on what is going on.
It has always been my belief that the peril we know is much less dangerous than the peril we don't know. And, as of now, this is what we know.
There were three nuclear bombs detonated on our soil. One in New York, one in Boston, and one in Virginia. These were not small bombs, certainly not the "suitcase" bombs that were, for so long, the stuff of novels and action movies. The bombs were huge, and the devastation is great. It is believed that the bombs were smuggled into the country inside large cargo containers on board container ships.
There has been no word from any official of the New World Collective government, that is assuming that there is a government. We don't know if the supreme leader of the New World . . ."
Gregoire paused in mid-broadcast for a moment; then, with a sigh, he continued.
To hell with that New World Collective nonsense, he said. I intend to refer to our country, or what is left of it, as the United States. And I think I am perfectly safe in doing so, since if we do have any government left, they are totally impotent now. We have tried to make contact by shortwave with anyone in Washington, D.C., who could give us some information on the status of Ohmshidi—indeed, the status of our country. They say that there is a silver lining to every cloud. It is hard to find one to this cloud, but if there is, it is that the government, and if I may be so bold, Ohmshidi, are no longer functioning.
As some of you may have heard on our initial broadcast, the United States was not the only nation to suffer these brutal nuclear attacks. Much of Europe seems to be in chaos right now, though they are not as bad as we are. We still don't know anything about Israel, other than the fact that they were hit by at least three, and maybe more, nuclear missiles.
America, is there any question as to how and why all this has happened? For nearly a century now, going back to the First World War, America has been the bulwark of freedom and democracy. We defended Europe in the First World War, we freed Europe in the Second World War, and we stood at their side during the long, frightening days of the Cold War.
But the first thing Ohmshidi did when he took office was pull all American soldiers back from their overseas assignments. Then he systematically disarmed us, while at the same time destroying our nation from within. Without a strong America, there is nothing left to stand between the world and the evil that would engulf the world.
I cannot but hope that there are groups of you hearing my voice now, groups of you who have taken the necessary steps to survive. And, once survival is assured, it is my hope that we will come together again, reclaim our nation, and once more be a united country under the Stars and Stripes.
And now, a word about who we are, and why and how we are broadcasting. As I am sure everyone within the sound of my voice is aware, there are no longer any television networks, or even television stations that are broadcasting. Those of us who are continuing with the shortwave 'casts are no longer employed, nor are we being compensated. But we do this because we are newsmen and women, first, last, and always. We do it because we must. And if providing news to a shattered people who are desperate for information serves a mission greater than ourselves, then we are compensated enough.
I am signing off now, but will broadcast again sometime tomorrow. I'm sorry I can't be more specific as to the time, but for now, I must err on the side of caution.
This is George Gregoire saying, good night, America, and God bless us all.
CHAPTER SEVENTEEN
"Do we have a name?" Julie asked.
"A name?" Clay replied.
"This group," Julie said. "Do we have a name?"
"I don't know. I hadn't really thought of a name," Jake said.
"We've got to have a name," Julie said.
"All right. Do you have a suggestion?
"Yes, as a matter of fact, I do have a suggestion," Julie said.
"What?"
"I think we should call ourselves Phoenix."
"Phoenix?" John asked.
"Julie explained. "I mean the United States of America, the country we all took an oath to serve, is dead as far as most people are concerned. Except us—we're going to make it rise from the ashes."
"Yeah," Deon said. "I like it."
"Phoenix," Clay said. He nodded. "It does have a ring to it."
"I like it," Karin said.
"Me, too," Marcus added.
"Alright," Jake said, smiling. "Hereinafter, we will be known as Phoenix."
"Well then, in that case we should change our radio call sign from Mickey Mouse to Phoenix," Willie suggested.
"I agree," Jake said. "From now on my call sign is Phoenix One."
"I've got somethin' else to bring up," Clay said.
"Now's the time to do it," Jake replied.
"Major—I mean, Jake, I know you said we aren't in the Army anymore. But the truth is, while we were in the Army, we had a standard operating procedure. And even if we aren't in the Army, I think we still need some structure. I mean, all you have to do is look at what's going on all around us now to know that we must have some SOP. I know you don't want to be a major anymore, but how about you taking charge, as a civilian, of our group?"
"We are all together in this," Jake said. "I don't want to presume."
"You wouldn't be presuming, and I agree with Clay," John said. "We do need some SOP, and you are the one who started Phoenix, so I think it only makes sense that you be our leader. We can still remain on a first-name basis." John smiled. "I sort of like calling officers by their first names."
"I concur," Marcus said. "Jake should be our leader."
"Count me in," Deon added.
Willie, Julie, and Karin quickly added their own support for the idea.
"Alright," Jake said. "I accept. Now, what do you say we get back out to the post and get busy?"
"Go out to the post and get busy? Jesus, give the man a little authority and he goes all power mad on us," John said.
The others laughed.
The Dunes, Fort Morgan—Tuesday, July 30
"Ellen, where is my typewriter?" Bob Varney asked.
"It's in the very back of the storeroom off your office," Ellen said. "Way in the back. Why do you ask?"
"I'm going to write," Bob said.
"I really . . ." Ellen started to say that she really thought it would be a waste of time, but she stopped in midsentence. She had lived with this man for over forty years and she knew him inside out. And she knew that he needed to write, and if truth be told, she needed it as well. She needed a sense of continuity to her life, and having her husband write books, whether they were ever published or not, was that continuity.
"I really think that is a good idea," she said.
Bob leaned over to kiss her. "Thanks for not trying to talk me out of it," he said.
It had been almost thirty years since Bob Varney last used a typewriter, but he had kept his old Smith-Corona portable all those years, keeping it in good shape, and keeping it in fresh typewriter ribbons. Retrieving it from the back of the storeroom, he opened the case, then blew and brushed the dust and cobwebs away. That done, he rolled two pieces of paper into the typewriter, using the second page as a pad against the platen because when he took typing in high school his typing teacher, Miss Sidwell, had told her students to do that.
Using the lever, he counted down eleven double-spaces before he typed:
Lilies Are for Dying
by
Robert Varney
Chapter One
John Hughes had what is called a very structured personality. Every morning he had one soft-boiled egg, a dry piece of toast, and half a grapefruit. He drove to work by the same route every day, and crossed the intersection of Greer and Elm at exactly the same time. That's why he was passing Elmer's Liquor Store just in time to see Elmer being shot.
"Charley, listen to this and tell me what you think," Bob said to his dog. He read the opening paragraph aloud. "Is that a grabber?"
Charley was lying under the desk with his head on Bob's foot. This was the normal position for writer and dog when a book was in progress. But that was the only normal thing about the setup. Bob was writing this book on a typewriter, and he knew this book was going nowhere.
His agent had told him that he need not waste his time writing any of the three books that remained on his contact, but his agent didn't understand. Bob didn't write because it was his job, Bob wrote because he had to write.
He returned to the book, listening to the tap, tap, tap of the keys, remembering that sound from years ago and, oddly, being comforted by it, as if it could take him back to another time and another place when things were as they should be.
As he continued to write through the morning, the pages began to pile up on the right side of the typewriter, and he remembered that as well, recalling the sense of satisfaction he got from watching the pile of pages grow. He had mentioned to his father once how he enjoyed watching the pile of pages grow, and his father, who had been a farmer, compared it to watching a crop being "made," as in "Are you making any cotton?" It's funny, Bob didn't realize until now, how much he missed watching the pile of pages grow. Seeing the word-count number increase at the bottom of the computer screen was never the same thing.
From his office he could see the Gulf through the front windows and Mobile Bay through the back windows. He saw a boat about a mile offshore and figured it must be a fishing boat. Was he catching fish to eat? Or to barter? Probably a little of both, he decided.
"It is good to see you writing, again," Ellen said, coming up behind him and putting her hands on his shoulders.
"You do realize that it is a complete exercise in futility, don't you?" Bob asked.
"Not futile," Ellen said. "It doesn't matter that it isn't going anywhere, it is restoring a sense of balance to our lives. It gives the illusion that everything is as it was, and I need that. We need it."
Bob lifted her hand to his lips and kissed it. "We were born twenty years too late," he said.
"Why do you say that?"
"If we had been born twenty years earlier, we would more than likely be gone by now, and we would have left the world while it was still sane."
"What's going to happen to us, Bob?"
"Nothing," Bob said. "We're going to ride it out and, in the long run, we'll be okay. Just don't be planning any trips to New York or Chicago. Or even into Gulf Shores," he said.
"Maybe I'll start my romance novel," Ellen said.
"Ha! You've been saying you were going to write a romance novel for the last forty-five years."
"I know, but other things kept coming up," Ellen said. "This time I'm going to do it, for sure. I've got a bunch of yellow tablets and a bunch of pencils. And the time to do it."
"Good for you," Bob said. "You start it. If you need help, just ask."
Fort Rucker—Wednesday, August 1
Jake and Karin were the last two to leave Ozark and head out to Fort Rucker, the others having left two days earlier. They were halfway to the post when they saw a pickup truck with a trailer, crossways on the road, blocking any possibility of passage.
"I wonder what this is?" Karin said.
The truck and the trailer were both filled with furniture, bedding, boxes, barrels, and crates.
"Looks like someone is trying to move all their belongings," Jake replied as he stopped the car and put it in park. "My guess is they were trying to turn around and got hung up with the trailer. I'll see if I can help."
Getting out of the car, Jake started toward the pickup truck. That was when someone stepped around the front of the truck. The man was wearing black pants, a black T-shirt, and a black headband. He also had a holstered pistol strapped to his belt. That didn't concern Jake—many had taken to wearing pistols since the total collapse of the republic. Jake was also carrying a pistol, but it was under the flap of his shirt and so not immediately visible.
"Can I help you?" Jake asked.
"Oh, yeah, you can help me," the man replied.
"What can I do for you?"
"Well, it's like this. You see this truck? It don't have enough gas in it to even get me back to Ozark. But seein' as you was drivin' your car, it looks to me like you do have gas. So what I'm goin' to do is, I'm gonna give you a can and a rubber hose." He put his hand on his pistol and patted it a couple of times. "And what you are going to do for me is siphon out all the gas that's in your tank and fill this can."
Jake pulled his pistol and pointed at the man. "No, I don't think so," he said.
"Whoa, I didn't know you were carrying," the man said, holding both hands up, palms facing Jake.
"Apparently not. Now, I'm going to ask you real nice to get that truck off the road and out of my way," Jake said.
To Jake's surprise, the man dropped his hands and chuckled. "You don't seem to understand what's at stake here," he said. "I don't know if that pretty little woman back there is your girlfriend or your wife, but if you don't do what I told you to do, my friend is going to put a bullet through her head.
Jake turned back toward his car and saw that Karin was now out on the road, standing just in front of a man who was holding a pistol to her head. This man, like the one who had confronted Jake, was wearing black pants, a black T-shirt, and a black headband.
"Better do what my friend says, mister, unless you want to see this woman's brains on the highway."
"You would shoot an innocent woman over a can of gasoline?" Jake asked.
"Oh, yeah, you can count on it," the man replied.
"I'm sorry, Jake," Karin said. "He must have been lying in the ditch alongside the road. I didn't see him come up."
"Let her go," Jake said, pointing his pistol at the man who was holding his gun to Karin's head.
"Ha! Is that pistol supposed to scare me?" the man replied. "You're a good sixty feet away from me—I'm only about six inches away from your woman. You really think you are good enough to shoot me, without hitting her?"
"How about those Kentucky Wildcats?" Jake asked.
"Say what?" the man with the gun replied.
"I like the cheerleaders," Jake said.
"Man, are you crazy or what? Can't you see I've got your woman here? Now are you going to fill that gas can or . . ."
At first Karin was confused by Jake's comment; then she smiled as she knew exactly what he meant. Suddenly Karin did a backflip, vaulting completely over the head of the man who was holding a gun on her.
"What the . . . ?"
That was as far as the gunman got because as he turned toward Karin, Jake took his shot. Blood and brain matter spewed out from the entry wound in the temple.
Because Jake had turned to take his shot, his back was now to the man standing in front of the pickup truck.
"You son of a bitch!" the man yelled.
Jake whirled back on the would-be gasoline thief, shooting him between the eyes even as the man was bringing his own pistol up.
"Are you all right?" Jake called back to Karin.
"Yes," Karin replied. She looked down at the man who had been holding his gun on her; then she walked up to Jake. "It took me a second to figure out what you were saying."
"You figured it out quickly enough. You did well."
"Were they soldiers, do you think?" Karin asked.
"There are no soldiers anymore," Jake answered.
Karin knew that Jake did not want to think that he had shot two men who may have, just recently, served in the same army with him, so she didn't press the issue any further.
"I'll get the truck out of the way," Jake said. Climbing in behind the wheel, he turned on the key and saw that the gas gauge didn't even come up to the E mark.
"I hope there's enough fuel to get it off the road," he said. He hit the starter and the engine kicked over. He drove it off the road and down into the ditch. Then, exiting the truck, he climbed back up to the road.
"What are we going to do with them?" Karin asked.
"What do you want to do with them?"
"I don't know. Somehow it doesn't feel right to just leave them both lying in the middle of the road."
"All right. I'll get them out of the road," Jake promised.
Grabbing one of them by his feet, Jake dragged him down into the ditch and left him by the truck. Then he returned and did the same thing to the other man. He started to go through their pockets to see if they had any identification, but stopped short because he realized that he didn't want to know who they were.
Returning to his car, he slid in behind the steering wheel and glanced over at Karin. She looked a bit queasy so she reached over to put his hand on hers.
"You did well," he said again.
"Jake, has it come to this?" Karin asked. "Is it going to be dog-eat-dog?"
"I'm afraid it is," Jake said. "But dogs run in packs. And we have our pack now."
Karin smiled, wanly. "Phoenix," she said.
"Phoenix," Jake repeated.
Three miles farther on Jake and Karin saw two people lying alongside the road. Jake slowed down enough to get a good look at them. It was an old man and a woman.
"Jake, stop." Karin said. "I have to check on them."
Jake stopped and Karin, getting out of the car, hurried over to the couple. She squatted down and felt for a pulse in each of them. "The woman is dead, but the man is alive," she said.
Both the man and the woman had been shot.
"Sir, what happened?" Jake asked.
"The sons of bitches took my truck," the old man said, straining to talk. "They took my truck and trailer. Had all our belongin's on it."
Jake looked at Karin and she shook her head, to tell him that the man didn't have long left.
"They shot Suzie," the man said. "Then they shot me. Dressed like pirates they were, all in black. The sons of bitches." He coughed a couple of times, then took one last rattling breath.
Karin tried his pulse again.
"He's dead," she said.
"Those murdering bastards," Jake said. "If I was feeling any twinge of regret before, I don't now."
"What are we going to do with them?" Karin asked. "We can't just leave them here like we did the other two. After what the other two did, they can lie out on the road until the buzzards pick them clean as far as I'm concerned. But these folks are innocent. They didn't do anything to bring this on."
"I've got an entrenching tool in the car," Jake said. "We sure can't give them anything like a proper burial, but you're right. We don't need to leave them out here, exposed to the elements."
Forty-five minutes later Jake and Karin stood over a fresh mound. Jake buried them both in the same grave. It kept him from having to dig two graves, but he was fairly certain they would have wanted to be buried together anyway. Before he buried them, he took the old man's billfold. It had pictures, a driver's license, but no money. But it didn't matter that there was no money, since money was worthless anyway. And at least now, he knew who they were.
"Mr. Theodore Fuller, Mrs. Suzie Fuller, I don't think I'll be able to find any of your next of kin to let them know what happened to you, but I hope there is some comfort that you didn't leave this world without someone knowing your names," Jake said. "I'm sorry your lives ended up like this. On the other hand, you did go out together, and you'll be together for all eternity now. And truth to tell, with what the rest of us are facing, you may well be the lucky ones."
Karin reached over and squeezed Jake's hand.
"You might be right," she said. "They might be the lucky ones."
"Come on, the others will be worrying about us. Let's go see if John and Marcus have found a helicopter we can put back together."
CHAPTER EIGHTEEN
The Ozark gate at Fort Rucker was unmanned, but it had been unmanned for several days now. There was, however, still a sign attached to the MP shack that read VISITORS MUST OBTAIN PASS.
After going through the gate they passed several abandoned vehicles, and as they moved farther onto the base, Jake was surprised to see that many of the base housing units were still occupied. But then, as he thought more about it, it wasn't surprising at all. These houses were home to the married soldiers. Like the houses in Ozark, Enterprise, and Dothan, these houses had no electricity or water, but they did provide shelter. Where else would they go?
One man walking alongside the road saw the blue post sticker on Jake's windshield indicating that he was an officer and automatically saluted. Just as automatically, Jake returned the man's salute; then, as he passed, he looked in the mirror as the man continued his long, lonely walk.
"Oh," Karin said quietly, and glancing toward her, Jake saw a tear sliding down her face. "Where will he go? What will he eat? We are trained to look out for our troops but—oh, Jake, I feel so helpless."
Jake squeezed her hand again. "We are looking out for six of them," he said. "That's a start."
"They are looking out for us, just as much," Karin replied.
"That's true."
When he turned off Hatch Road onto Hanchey Field Road, Jake felt a sudden twinge of melancholy. So many times over the years of his multiple assignments to "Mother Rucker," he had made this same turn on the way to log flight time. At first glance Hanchey Field didn't look too much different from how it always did. More than one hundred helicopters were parked at the huge heliport. However, a closer examination of the helicopters showed that nearly all had been stripped of anything of value. Many were missing rotor blades—others had been stripped clean of sheet metal so that nothing but the skeletal frame remained.
"I don't know," Jake said as they drove toward the hangars. "From the looks of things, I'm not sure we can find enough to assemble even one flyable helicopter. Give them a call. We ought to be in range."
Karin keyed the two-way radio. "Phoenix Base, this is Phoenix One, over."
There was a moment's delay; then they heard, "Phoenix One, this is Phoenix Base. Where are you?"
"We're on the field."
"Phoenix One, do you remember Dewey Alain and the foam generator?"
Jake smiled, then nodded. "Tell him yes."
"I remember," Karin said.
"That's where we are. Call when you approach. I'll open the doors and let you in."
"Phoenix One, out," Karin said. She looked over at Jake, who was still chuckling. "What is this about the foam generator?"
"We had a hangar fire drill once," Jake said. "Sergeant Alain was supposed to simulate hitting the big red button that would activate the foam generator and flood the hangar. But he didn't simulate, he actually did it, and the hangar was filled waist high with foam. It took two days to clean it all up."
Karin laughed.
"Wait, that's not the half of it," he said. "Two weeks later, there was a report of survey done for the damage, and the inspection team wanted to know what happened. Sergeant Alain explained about the fire drill, told them his job was to simulate hitting the foam button. 'But I didn't simulate it,' he said. 'I actually hit it. '" Jake laughed. "Then he . . ."
"Don't tell me, he hit it again?" Karin asked, laughing.
Now Jake was laughing so hard that tears came to his eyes. "If I'm lyin' I'm dyin'," he said. "He hit it again."
Jake started toward the hangar. "That's the one," he said.
"Phoenix Base, hit the foam generator."
"Ha, he told you," John replied. The hangar door started up, and as they came closer, they saw Marcus pulling the chain to raise the door. As soon as they were inside, the door went back down.
Jake parked his dark gray Volvo next to Clay's red Liberty. As he exited the car he looked at the Blackhawk helicopter they had selected. At first glance it looked as if he could climb in, light the engine, and pull pitch, but he knew that looks were deceiving. "It looks good," he said. "What does the logbook say about it?"
John shook his head. "We have no idea. The logbooks are all stored on the mainframe server and with the server down, we have no way of accessing the records."
"There is something to be said for hard-copy logbooks," Jake said.
"You got that right. But we are doing a very thorough periodic inspection, so we are finding and correcting all the faults."
"Do we have any idea as to the total hours?"
"Onboard system says twenty-seven hundred and fifteen hours, but of course, that's just the airframe. We have no idea of the total hours on the engine, engine components, transmission, or rotor system."
"There has been a lot of cannibalization on this aircraft," Marcus added. "Neither engine has a fuel control. Igniters are missing on engine number two. No filters anywhere, engine or transmission. We're missing a pitch change link on the main rotor."
"Why would anyone take a pitch change link?" Jake asked. "What on earth would they use it for?"
"Beats me. But we've got almost a hundred helicopters to draw from, I'm pretty sure we'll find one we can use.
"Have any trouble coming in?" Clay asked.
Jake and Karin exchanged looks.
"You did, didn't you?" Clay said. "What happened ?"
Jake told about their encounter on the road with the two young men who had stopped them.
"There's going to be a lot more of that," Clay said. "Especially if anyone finds out how much fuel we have."
"But we have jet fuel, don't we? What good would that do in a car?" Karin asked.
"A gasoline engine will not run on jet fuel, but a diesel engine will," Jake explained.
Clay pointed to his Jeep Liberty. "That will not only run on jet fuel, it is running on jet fuel."
"I was going to suggest that," Jake said. "I am just about out of gasoline, so any running around we have to do in the next few days is going to have to be in your vehicle."
"We could go out to TAC-X," Clay suggested.
"No, not yet," Jake said. "We'll keep that as an emergency supply. There may come a time when we will be in desperate need of it."
"You're probably right," Clay replied.
The others expressed some curiosity in what Jake and Clay were talking about, but none of them asked any questions, and neither Jake nor Clay made any attempt to satisfy their curiosity.
Saturday, August 4
"I need some Kleenex," John said. "At least six boxes."
"Six boxes? Wow, you must have some runny nose," Deon said.
"Not for my nose. We aren't going to be able to find any new filters, so I'm going to have to make some. Kleenex tissues will work."
"Alright," Clay said. "I'll check the commissary and the PX. If I can't find any there I'll run in to town."
"Better try Enterprise or Dothan. I know there aren't any in Ozark," John said. "We looked yesterday."
Clay started toward his Jeep SUV.
"Deon, go with him," Jake said. "After what Karin and I ran into on the way out here today, there's no telling who or what might be waiting for you."
"All right," Deon said. He walked over to the wall where their weapons were, picked up two M-16s, then put them both back down. Instead, he picked up an M-240, a machine gun.
"You plan on starting a war?" Julie asked.
"No," Deon said. "I don't plan to start one, but if I happen to get into one, I damn sure plan on winning it."
"I like the way that man thinks," Clay said.
Although they passed several abandoned vehicles on the way in to Dothan, and even more once they reached the city, they did not run into any trouble. Seeing a Winn-Dixie on Westgate Parkway, Clay pulled into the parking lot, weaving around abandoned cars and trucks. He pulled all the way up onto the wide sidewalk in front of the store.
"Let's see what we can find in here," Clay suggested.
The inside of the store was a jumbled mess—overturned shelves and counters, broken glass, empty boxes, shredded paper, and signs that mocked with their cheery false promises.
Winn Dixie Brings You the Freshest Produce!
"We aren't goin' to find anything in here," Deon said.
"Doesn't look like it," Clay admitted. "But we may as well give it a try. If there is anything, it will more than likely be over here," he said, pointing toward a sign that said PAPER PRODUCTS.
The two men looked through the residue under the sign. Suddenly Deon raised up and pointed. "Is that what I think it is?" he asked.
Clay looked in the direction Deon was pointing. There, on the floor, was a huge pile of currency in denominations from ones to one-hundred-dollar bills.
"How much do you think is there?" Deon asked.
"I don't know. A couple hundred thousand dollars, probably."
"It's just lying here. You know there've been hundreds of people that have picked through this store since everything collapsed, but nobody took any of the money."
"You want to take some of it?" Clay asked.
"No. What good is it?"
"There you go, that's why nobody has taken any of it," Clay said. "We better try someplace else. We aren't going to find anything here."
The next place they stopped was a Bruno's store, and there they found several boxes of a house-brand tissue that were water damaged.
"What do you think?" Deon asked.
"I think this will have to do."
John had only asked for six boxes, but they found ten boxes that looked to be in pretty good shape, figuring that the extra boxes might make up for any that were too damaged to be of use.
On the way back to Fort Rucker, they saw a pickup truck parked across the road in a blocking position. There were four armed men, and three of them were pointing their weapons at them, while the fourth held up his hand in a signal for them to stop.
"Uh-oh," Clay said.
"Stop here," Deon said.
"What for? We're goin' to have to face them so we may as well see what they want."
"You know damn well what they want," Deon said. "They either want this vehicle or the fuel. Stop here; let me get up on top."
Deon reached in the back and picked up the machine gun; then he got out of the jeep and climbed up on top. He loaded the weapon, chambered a round, lay down on top of the car facing forward, braced his feet on the top railing, then hit his hand on the roof.
"Let's go!" he called.
Seeing that the Jeep wasn't going to stop, the three men with weapons—two rifles and a pistol—began firing.
Deon opened fire with the machine gun and the two men with rifles went down. The one with the pistol threw his weapon on the ground and put his hands up.
"Drive on through!" Deon called down to Clay.
Clay accelerated, then left the road to go around the block when he got there.
"You boys have a nice day now, you hear?" Deon shouted as they drove by.
The two who were still on their feet glared back, but said nothing.
"Oh, yeah, this will work just fine," John said when he saw the tissues they had brought back. "I'll have these filters better than new. Marcus, how are you coming on the hydro mechanical unit?"
"I've about got the HMU up and running. I've got the variable geometry actuator, the compressor inlet sensor, and the high-pressure fuel pump installed," he said.
"Good, good, we're cookin' with gas now."
"Cooking with gas? I thought this helicopter used jet fuel," Julie said.
"It does. When we say cooking with gas we mean . . ."
Julie started laughing.
"I think Julie is pulling a couple of legs," Karin said.
For the next several minutes John and Marcus worked on the helicopter engine while the others watched and handed them tools when asked. Deon took a pair of binoculars and an M-16 with him, then went up into the tower where he would have a panoramic view of the entire airfield.
"Phoenix," he called down a few moments after he left. "There are some people over in the far northeast corner of the field."
"What are they doing?"
"Looks like they are trying to find a helicopter that still has some fuel."
"Keep an eye on them, but don't do anything unless they start something."
"Roger."
That night they scheduled a guard detail. John was first on, and he went up into the tower to keep watch while the others spread out their sleeping bags. Willie cranked up one of the radios and after some searching, found a broadcast. The voice they heard was that of Ohmshidi.
"Damn, you mean that son of a bitch is still around?" Clay asked.
To my fellow citizens of the great commonwealth of the New World Collective, I send my greetings, and my assurances that I am well, and I am working very hard to restore order and hasten the recovery.
As I am sure all within the sound of this broadcast know, there were three nuclear detonations upon our soil. While I believe my peace overtures to the Islamic nations were bearing fruit, I neglected the danger from within. It is my sincere belief that the bombs were detonated not by foreign enemies but by domestic terrorists incited to do so by the seditious broadcasts of George Gregoire. I have declared Gregoire to be an enemy of the state, and hereby grant to any citizen who comes in contact with Gregoire the authority to shoot him on sight.
This may seem like a drastic measure, but under our current situation, drastic measures are allowed—indeed, even required.
Since the nuclear attack against us, I have, under the authority granted me by the Enabling Act, taken additional action to insure our security. So that there may be efficiency of operation, I have dissolved the government. Congress, as you once knew it, no longer exists, nor does the Supreme Court. Under the aegis of the Enabling Act I have taken on the total responsibility of the government. I assure you this is only a temporary condition and as soon as order is restored and I am assured that recovery is well underway, I will authorize new elections to replace the Congress. I will then, on a gradual basis, return some of the authority I have assumed to the newly elected body, which shall be known as the People's Collective. I have also dissolved the entire military, from the chairman of the Joint Chiefs of Staff, down to the lowest-ranking private. If you are in the military, and you hear this broadcast, you may take this as your authority to leave the post, base, or ship to which you are assigned. You are free to go home, or to anyplace you wish. I thank you for your service. All police and military authority now rests with me, and me alone, to be carried out by the SPS units, which I will be expanding.
Because of the unrest that is rampant throughout the country, and as a matter of personal security, I will tell you that I am no longer in New World City, the place once known as Washington, D. C., but have established the capital in a location that, for now, I shall not disclose. But, as you can tell by this broadcast, the government, through me, is still functioning, and that means that the NWC is still a member nation in the world of nations.
Thank you, and good night.
CHAPTER NINETEEN
"Holy crap, can you believe that?" Willie said. "He has actually dismissed the entire Army."
"It is a stupid and empty gesture," Jake said. "He destroyed the Army with his incompetent meddling, so there is no more Army anyway. He is just saying this to make it appear as if he is still in charge."
"Who were the idiots who voted for this man?" Clay asked. "How could we have possibly elected someone like this?"
"When more than half the voting public got their news and opinions from stand-up comedians, what did you expect?" Jake asked.
"I don't know," Clay replied. "But I sure didn't expect anything like this."
"Always before, when our country got into serious straits someone would say something like, 'We are Americans, we've been through hard times before. We got through that, and we'll get through this,'" John said. He shook his head. "But I don't think we have ever been through anything like this."
"Anybody else broadcasting?" Jake asked.
"There's nothing left on this frequency but carrier wave," Willie said. "Let me see what else I can find."
El huracán hará la recalada en Point de Mobile antes de medianoche esta noche. Es ahora una categoría cinco, pero disminuirá en la fuerza antes de la recalada a categoría tres . . .
"I'll see if I can find something in English," Willie said.
"No, wait! It's a hurricane!" Marcus said, holding his hand out to stop Willie.
"A hurricane? Where? When?"
"It's in the Gulf now, should make landfall by midnight tonight at Mobile Point."
"Mobile Point? Isn't that where Fort Morgan is?" John asked.
"Yes," Jake said. "Marcus, what's the strength, did he say?"
"It's a cat five now," Marcus said. "They are saying it should be a three by landfall."
"Whoa, that's quite a storm."
"Does that change our plans any?" Deon asked. "I mean, do we still plan to go to Fort Morgan?"
"Fort Morgan has been there since 1834," Jake said. "I imagine it has gone through its share of storms. We're still going. But we are going to get a lot of rain and wind here tonight, so we need to be ready for it."
Willie continued to search the dial, stopping when he heard English being spoken. It was not only English, it was English with a British accent.
This is the BBC World Broadcast on 5.110 megahertz. Now, here is the news.
Authorities in Southampton have stopped looking for survivors and are now cordoning off the entire area to prevent entry into the contaminated area. The final count of casualties is believed to be a little over one hundred fifty thousand dead, with another one hundred thousand injured. There is no way of knowing the number of people who will ultimately die of radiation exposure, though it is thought to be very high.
There is little news out of Germany or Spain, we know only that they, too, had nuclear bombs explode in their countries. France reports an expected death total of one and one half million. So far there has been no news of any sort from Israel—indeed, we do not know if the country yet exists.
Caliph Rafeek Syed has announced the formation of the Greater Islamic Caliphate of Allah, composed of the former nations of Iraq, Iran, Syria, Libya, Lebanon, Saudi Arabia, and Yazikistan. That association, if true, would constitute a greater power than all of Europe, and what is left of the New World Collective.
We now know that every bomb that exploded in Europe, as well as the three that exploded in the New World Collective, were delivered on board ships. The Greater Islamic Caliphate of Allah has taken credit for the blasts, and has declared that the suicide bombers, as well as the unsuspecting crewmen of all three ships, are martyrs and now in paradise.
Supreme Leader Ohmshidi is no longer in New World City, but continues to make pronouncements from undisclosed locations. In his latest broadcast, he announced the dissolution of Congress and the Supreme Court, though such action is meaningless at the present, since the United States, or the New World Collective as the country is now known, has ceased to exist as a nation.
In addition to the loss of life and extreme damage wrought by the three atomic bombs, the rest of the former United States is in far more dire straits than any of the European Union countries. For all intent and purposes, the U.S. no longer has a functioning government. It has no viable currency, which doesn't matter as there is no longer a commercial enterprise in operation within the borders of what was once the United States. All transportation has come to a halt and the supply of food is becoming very critical in all areas of the country. Electricity and running water have been terminated, as has all telephone service. There are no broadcast facilities in operation, not even at the local level, though we have picked up some shortwave transmissions, including those of one-time right-wing television host, George Gregoire. There are no policemen on duty anywhere in the entire nation, nor is there martial law. There can be no martial law because there is no military.
At a NATO meeting yesterday, it was decided that those nations who have not suffered a nuclear attack would make an attempt to send food and supplies to the stricken nations as soon as possible. There was also some discussion as to whether NATO should send a contingent of military to the NWC to help in restoring and maintaining order in the affected nations, though no action was taken.
Prime Minister Corey Wellington said yesterday that while he is confident that the United Kingdom will survive the single nuclear bomb blast that occurred on British soil, he fears that the nuclear explosions in the United States, on top of what the nation was already experiencing there, will make recovery impossible. He is personally saddened by the terrible conditions of what was "once the preeminent superpower of the world."
"The United States was always the leader in, and set the standard for, aid to countries in need," the prime minister said. "How sad it is that such a powerful and benevolent nation could have been brought to its knees by an inept and ultimately destructive president. We cannot, and we will not turn our back on our American cousins."
This is BBC and this has been the news.
"Damn," Clay said. "It looks now like the whole world is going to hell in a handbasket."
"What happened has happened," Jake said. "But we can't look back now. We have to keep our eyes on what is in front of us. We have to survive."
"Survive, then what?" Karin said. "What do we do then? Do we just live out the rest of our lives in isolation?"
"You heard Gregoire. There will be others like us," Jake said. "We will establish contact with them."
"And once we establish contact, then what?"
"We're not ready yet to decide then what," Jake said. "As I said, our first duty is to survive. And we do that by facing one challenge at a time, one day at a time."
The Dunes, Fort Morgan—Saturday, August 4
James Laney stood on the roof of The Indies, a seven-story condo built ten years earlier for people who wanted a vacation beach home for themselves, and also for those who wanted investment rental property.
Not one unit was occupied now, as many had been taken over by the bank in foreclosures. Even those units that were owned outright stood empty now because there was no fuel available for the owners to come down, nor incentive to do so, since there was neither electricity nor running water.
A few minutes earlier James had climbed the stairs to the top of The Indies in order to have a better view of the Gulf and, more importantly, the sky over the Gulf.
"We've got a hurricane comin'," he had announced that morning to Jerry Cornett and Bob Varney. "I can feel it in my ankles and in my knees."
Since James had accurately forecasted both Hurricanes Ivan and Katrina, Jerry and Bob took him seriously. They were sitting at a table on the deck behind James's house when he came back.
"Did you see anything?" Jerry asked.
"No," James said. "But I know damn well one is out there."
"What are we going to do?" Bob asked.
"What can we do? If we combined what fuel we have left and all crowded into the same car, it would not be enough to get us away from the storm," James said. "And even if we did we wouldn't have enough fuel to come back here. We don't have any choice. We are going to ride it out."
"I thought we said that, after Katrina, we weren't going to ride another one out," Bob said.
"Yeah, we did say that," James agreed. "But I don't think we have a choice now."
"James is right," Jerry said. "We don't have any choice."
"Jerry, your house is right down on the front line," James said. "You and Gaye might be better off coming here to stay with Cille and me."
"Yeah, I think we will. What about you, Bob?"
"We're a little farther back from the beach than James, even," Bob said. "We'll ride it out there, then come back over here after it passes."
Because there was no functioning TV or radio, Bob and the others of the little group had no news on the hurricane with regard to either its strength, or its name. They took it upon themselves to name the storm, calling it Hurricane Ohmshidi, declaring that no matter how strong it was, it couldn't possibly do more damage than the president had already done.
Bob convinced Ellen that they, and Charley, should ride out the storm in their minivan. "It presents less of a surface to the wind," he explained. "Besides, if you can drive a car at a hundred miles an hour on the highway, it seems reasonable to assume that the car can withstand one-hundred-mile-per-hour winds."
The wind started increasing in strength at about six o'clock that evening, getting progressively stronger until midnight. Bob, Ellen, and Charley were in their Toyota Sienna, looking through the windshield onto the street in front of their house. The wind was howling like the engines of a jet airliner sitting on the end of the runway just starting its takeoff run, and the minivan was buffeted about like an airplane flying through rough air.
The rain that pelted the windshield made it very difficult to see, because each drop of rain was filled with sand that had blown up from the beach. When they could see, they saw roofs from houses, balconies, outside steps, and large pieces of wood tumbling by in front of them. Bob had parked under his house, so the van had some protection from the tumbling debris by the large doubly braced stilts upon which his house set.
At midnight the eye passed over them and everything stilled. With no rain nor wind, Bob flashed on the lights so they could see. The street was piled high with wreckage from houses that had fallen before the storm.
Bob had a small, handheld, two-way radio. He depressed the talk button. "James, do you hear me?"
"Yeah, I hear you."
"You folks making out all right over there?"
"We're fine," James said. "But there is water all the way up to the back of my property line. There's water from here all the way down to the Gulf. How are you folks doing?"
"The car is getting buffeted around quite a bit, but other than that we are doing fine."
"You can always come over here if you want."
"No, we've come this far, we'll ride the rest of it out. Fact is, I don't want to go outside now, anyway, because the wind is picking up again."
"All right, we'll see you in the morning."
When the rain started again, Bob put his seat back down.
"What are you doing?" Ellen asked.
"I'm going to sleep."
"In this? How can you sleep in this?"
"What else is there to do?" Bob asked.
Charley had been sitting on Ellen's lap, but he jumped over onto Bob and lay down on top of him. He was shaking badly.
"You don't need to be afraid, Charley Dog," Bob said. "You aren't going to get wet or blown away."
"I hope that's true for all of us," Ellen said.
Bob reached up to take her hand. "It could be worse," he said.
"How could it be worse?"
"This could be ten years ago when my mother and your mother were still alive, and they could both be in the backseat."
Ellen laughed. "You're right," she said. "It could be worse."
Though the noise of the storm and the wind continued unabated for at least seven more hours, from midnight until seven o'clock the next morning, Bob went to sleep. He didn't wake until Ellen shook his shoulder.
"What is it?"
"The storm has stopped," Ellen said.
"Good," Bob said. He put his seat back up. "Did you get any sleep?"
" No."
"Why not?"
"Someone had to stay awake."
"Why?"
"I don't know why," Ellen said. "It's just that somebody needed to stay awake."
"I appreciate your dedication to duty," Bob said.
The rain had stopped, and they could see, but the wind, while no longer at hurricane strength, was still blowing very hard. However, the wind had stilled enough that large pieces of debris were no longer flying by.
"Let's go over and see how the others fared," Bob suggested.
Walking was difficult, but by leaning into the wind, they were able to stay on their feet. Charley could not stand up against it, and was rolled up by the wind, so Bob had to carry him. When they reached James's house, they saw that the water had come up to the very edge of his property. Every other house in the compound, at least those that remained standing, were in water that was halfway up the stilts upon which all the houses were mounted. They were surprised to see two women with James and the others.
James introduced them as Sarah Miller, who was twenty-one, and Becky Jackson, her aunt. Though she was Sarah's aunt, Becky was only twenty-three.
"They were in the Carpe Diem house," James explained. "That's where they rode out the storm last night."
"Whoa, Carpe Diem is under water up to the first floor, isn't it?" Bob asked.
"Yes," Becky said.
"I'll bet it was a frightening night."
"I've never been so afraid in my life," Sarah said.
"We thought we were the only ones out here, until Mr. Laney came over in his boat to get us this morning," Becky said.
"What in the world were you doing there?"
"We had a gift shop in Mobile, but when everything started going bad, we had to close our shop. Then things got a little dangerous there. My folks have this place down here so we came down, thinking it would be safer," Becky said.
"Of course, without TV or radio, we had no idea we were coming right into the middle of a hurricane. We got here yesterday morning, the hurricane hit last night," Sarah added.
"You're lucky the house didn't blow away," Cille said. "So many of them have."
"Nineteen in The Dunes alone," James said. "Including Jerry's house."
"I'm sorry, Jerry," Bob said.
"It could've been worse," Jerry said. "Gaye and I could have been in it."
"I guess that's right."
"Wait until you see the front of The Indies condo," James said. "The entire front wall came down last night. It looks like a giant dollhouse. You can see into every unit in the building."
"And we're going to have to live with this a long time," Bob said. "It's not like it was with Katrina and Ivan when everyone started rebuilding right away."
"At least we don't have to worry about the power coming back," Jerry said. "Because it never is coming back."
"I'd love to know what's going on in the world," Bob said. "It would be nice if they've impeached Ohmshidi. Surely, by now, those idiots in Washington have figured out that this idiot is the one who totally destroyed our economy."
"We don't have a government anymore," Becky said.
"You can say that again. I mean if they are just going to sit by and watch Ohmshidi destroy us without doing a thing to stop him . . ."
"No, I mean seriously. Since the three atom bombs, Ohmshidi has dissolved the government."
"Three atom bombs? Government dissolved? What are you talking about?" Bob asked.
"And where are you getting all this information?" Jerry added.
"There is a man in Mobile who lives close to us, who has a shortwave radio," Becky said. "Boston, New York, and, I think Norfolk, were all hit with nuclear bombs."
"Son of a bitch!" Jerry said. "It's not enough I've lost my house. We're losing our entire country."
"Losing our country?" James said. "Sounds to me like we have lost it."
"Well, there's one good thing," Bob said.
"What's that?"
"It can't get any worse."
Despite himself, Jerry laughed. "And that's a good thing?"
"Any port in a storm."
"Oh, don't talk any more about storms," Becky said. "The one we had last night was enough."
"Well, we all made it through, so what do you say we have some breakfast?" Jerry suggested.
"Sounds good to me," Cille said.
"How about biscuits and gravy?" Bob asked.
"Biscuits and gravy? You've got biscuits and gravy?" Becky asked. "How? I mean, where did you get it?"
"You just don't know my talent," Bob said. "I can make something out of nothing."
Ellen laughed. "He's telling you a big one. But he is good at making do. And since nearly every house out here had a well-stocked pantry and there is nobody here anymore, we have sort of inherited it. Flour, cornmeal, canned vegetables, condensed milk, and several canned meat products."
"What about water?"
"Every house out here has a hot water tank. Some of them have two tanks," James said. "We're okay on water for a while."
"What happens when you run out?"
"As you may have noticed, last night we get a lot of rain. We'll build a catchment and storage system when we need it."
A little less than an hour later all eight sat down to a bountiful breakfast of biscuits, gravy, and fried Spam.
"Oh," Sarah said as she took her first bite. "This is good!"
"Life is good," Bob said. He chuckled. "Or, at least as good as it can be under the present circumstances."
CHAPTER TWENTY
Fort Rucker—Sunday, August 5
Although Fort Rucker wasn't hit with the full force of the hurricane, there were sustained winds of sixty miles per hour as well as torrential rains, and the wind and rain pounded the walls of the hangar throughout the long night. When Jake and the others emerged from the hangar the next morning they were surprised at the amount of damage the storm had done. Because none of the helicopters on the field had been tied down, many of them had been overturned, or pushed into other helicopters nearby.
"I'm glad we found a helicopter when we did," John said. "Look at that mess. I doubt there is one airframe left that could be used."
"Yes," Willie agreed. "It's a good thing we moved the zero-seven-seventeen inside."
"The only thing we have left is to connect the servo, right?" Jake asked.
"That's it."
"How long will that take?"
"No more than twenty minutes to connect it and bleed it," John said. "Then it'll be ready for a test flight."
"Why test fly it?" Marcus asked. "Let's just climb on it and go."
"No," Jake said. "If this thing is going to fall apart in the air, I'm going to be the only one on board. I won't do a complete, by-the-book test flight, but I do want to make sure everything is okay before I risk anyone's life but my own."
Half an hour later, Karin and Julie pulled on the chain to raise the door as the men pushed the helicopter out of the hangar and onto the tarmac. Although the U-60 is a twin-engine aircraft, they had made the decision to disconnect the left engine from the freewheeling clutch at the transmission. That way they were able to cannibalize it for the other engine. It meant that the helicopter would have less power and would fly slower with less of a payload, but it also meant it would use just over half as much fuel.
"All right, Jake, why don't you climb in and see what we have?" John said.
"You don't mind if I do a pre-flight, do you?" Jake asked.
"Have at it," John said.
Jake did a thorough walk-around inspection, checking all fluid levels, as well as the rotor system for any loose or missing items. He looked at the blades to make certain there was no damage or separation of the laminated surfaces. After that, he started checking for leaks in the engine, transmission, gearboxes for the tail rotor drive train, hydraulics, and blade grips. Everything checked out.
After the walk around, Jake climbed in to the right seat to start his cockpit checks. He put on his SPH4 flight helmet, then plugged in the radio jack. Once he had a good start with the engine and rotor coming up to flight idle speed quite nicely, he checked to see that all of the gauges were in the green. Smiling, he gave John a thumbs-up.
"Alright!" John said, returning the thumbs-up salute.
After the engine was running, Jake went through a number of other checks, including turning on and checking the inverters, checking the generator, and turning off the hydraulic system to check control travel, then turning it back on to insure that there were no stuck valves that might bring trouble, including even a hydrostatic lock, which could result in a loss of control.
It was now well over four months since Jake had sat behind the controls of a functioning helicopter, and he felt a strong sense of satisfaction at being there again. With the rotor blades spinning at full speed, he moved the cyclic around to check the rotor plane. It dipped exactly as it was supposed to, and he felt no falloff of rotor control as a result of John's jury-rigged pitch change link.
Automatically, he set the radios to departure frequency, then keyed the transmit switch before he realized he had nobody to contact. He released the radio transmit switch, and pulled up on the collective pitch control, causing the helicopter to lift from the ground. He stabilized it, then pulled the collective and pushed the cyclic forward. The helicopter took off easily and he climbed to five hundred feet as he passed over the edge of Hanchey Field. He did a complete circle around the field, looking down at the hangar and the little group of people who were gathered anxiously, awaiting his return.
He had just started shooting his approach to the tarmac right in front of the hangar when he saw a pickup truck coming quickly up Hanchey Road. He expedited the approach, sat down, then killed the engine.
"How did it go?" John asked, running over and sticking his head in through the open window.
"The flight was perfect," Jake said. "But it looks like we're about to have company. Deon!" he called.
"What's up?" Deon asked, sprinting over to him.
"There is a pickup truck full of men coming this way, fast," Jake replied. "It may not mean anything, but there is no sense in taking a chance."
"I'll get a machine gun up in the tower," Deon said.
"Good idea. But don't do anything unless you get word from me," Jake said. "Or unless they start shooting."
Deon nodded; then, grabbing the M-240 and an ammo box, he hurried up the outside steps into the tower.
"What is your plan?" Clay asked.
"Get everyone into the hangar, but have them armed, just in case," Jake said.
"Are you going to stay out here?"
"Yes. I intend to see what they want."
"I'm going to stay with you."
"No need," Jake said.
"I'd feel better. I'm going to be standing right beside you with an M-16."
"All right," Jake said. "If you say so."
Clay hurried back into the hangar, then returned with the rifle just as the red Dodge Ram pulled onto the airfield and started toward them.
The pickup truck approached at full speed.
"What the hell?" Clay said. "They plan to run us down!"
Clay raised his rifle to fire, but there were two men in the back of the truck, with their rifles resting on the top. They opened fire and Clay was hit.
"Clay!" Jake shouted and reached for him.
"No! Get out of the way!" Clay yelled, and he shoved Jake hard, knocking him down, but getting him out of the way of the onrushing truck. Even as Jake hit the ground, he heard a sickening thump when the truck ran Clay over. The driver slammed on the brakes, then swung the truck around. In the meantime, the two gunners opened up on Jake. Jake was still on the ground and he rolled hard to his right as the bullets ricocheted off the blacktop just beside him. Deon opened up with the M-240 and Jake saw the tracer rounds streaming into the truck. Then he saw the driver lose control, and crash into the helicopter. Both helicopter and truck exploded into a huge ball of fire. Neither the driver, nor the two shooters got out.
Jake moved quickly to check Clay, but jerked his face away when he saw that Clay's head had been smashed by one of the wheels. Steeling himself, he turned back to his longtime friend, removed Clay's shirt, and spread it over his head. There was no need for either of the women to see this, and even though Karin was a nurse, this was more than anything she would ordinarily see.
Deon came down from the tower as the others came out of the hangar.
"What happened?" Karin said, then seeing Clay lying on the ground, his head covered by his shirt and blood pooling underneath, she gasped. "Oh my God," she said.
"Damn," Willie said. "The sergeant major got through Iraq and Afghanistan, only to have some homegrown bastard kill him."
There were some secondary explosions from the burning truck and helicopter.
"Think anyone survived that?" Marcus asked.
"I hope they did," Willie answered. "I hope the sons of bitches are roasting alive. They may as well get a taste of what it's like before they go to hell."
"I doubt anyone survived," Jake said.
"Including the helicopter," Deon said. "We're going to have to start over from scratch."
"Scratch is right," John said. "Because there isn't anything left we can build from. We are, as they say, SOL."
"What about one of the other airfields?" Marcus suggested.
"No good," Jake said, shaking his hand. "One of the last things we did while the post was still functioning was move all the UH-sixties to Hanchey. If we can't put together another one from what we have here, we are going to have to come up with another plan."
"A Chinook?" Marcus suggested.
"I don't think so," John said. "They've got more parts than a Blackhawk; it'll be harder coming up with all we need for them than it was for the zero-seven-seventeen."
"What are we going to do with Sergeant Major Matthews?" Karin asked.
"We're going to bury him," Jake said.
"Where? How? There is nothing here but blacktop and cement," Marcus said.
"There's real ground behind the hangar," Jake said. "And an old warrior like Clay would probably want to be buried on an Army base, even if there is no Army anymore. If a couple of you will help me carry Sarge around back."
"We'll get him," Marcus said, nodding toward the others.
Willie and Marcus took Clay's arms, Deon and John took his feet. Jake, Karin, and Julie followed them around the side of the hangar.
The rain from the storm made the ground soft, so it took no more than half an hour to dig a grave for Clay. With his blood-soaked shirt still wrapped around his head, they lowered him gently into the grave.
"I wish we had a flag to drape over him," Karin said.
"We do!" Deon said with a big smile. "I saw one while I was up in the tower. I'll go get it."
"You knew the sergeant major a long time, didn't you, sir?" Marcus asked. For the moment, the rank and military courtesy seemed appropriate.
"Yes," Jake answered. "I don't think I would have gotten through Officer Candidate School without him."
A moment later Deon returned with the flag. It was a storm flag rather than a garrison flag, so it was considerably smaller.
"It's not big enough to drape over him," Julie said.
"We'll fold it into the triangle, then put it in his hands. He will like that," Jake said.
"I've often wondered why we fold a flag like that," Julie said. "It has to be symbolic of something."
"It is," Jake said. "Folded properly, it takes exactly thirteen folds, two lengthwise and eleven triangular. That represents the thirteen original states."
Willie and John folded the flag into the triangle.
"Now," Jake said, holding out the flag. "This triangle resembles a cocked hat, representing every solder, sailor, marine, and airman who has ever served. And finally, you can see that there are now only four stars visible. Those four stars stand for 'In God we trust.'"
"I'll put the flag in his hands," Deon offered, and taking the flag from Jake, he leaned down over the open grave and placed the flag so that Clay was holding it over his heart, with his hands over it. It was almost as if the sergeant major was actually holding on to the flag.
"I'm going to have a little service for him, if you don't mind," Jake said.
"Who would mind?" Deon asked. "I think it is entirely appropriate."
"So do I," Karin said.
Jake lowered his head, and the others did as well. "Dear Lord," Jake began. "We commit into your keeping Sergeant Major Clayton Bertis Matthews the Third. Clay served his country and his fellow man with honor and valor. He took up arms to defend all that good men find dear: life, liberty, and the pursuit of happiness. He lived his life according to those ideals, and, moreover, he imparted that dedication and his wisdom to others. He was my mentor, my friend, and my strength. We leave him here now, secure in the knowledge that you hold him in the palm of your hand. Amen."
"Amen," the others repeated.
As they left, the others, one at a time, passed by Clay's grave. Willie and Marcus both signed themselves with the cross.
CHAPTER TWENTY-ONE
By the time they returned to the front of the hangar the fire had burned out and all that was left was the smoldering wreckage of the truck and helicopter. No longer red, the truck had rusted out in the flames. The tires had been burned off and the aluminum wheels were no more than molten slag. Inside the truck a charred body was draped over what was left of the steering wheel. The two gunners had been thrown forward over the cab of the truck and their charred remains lay in the blackened residue of what had been the helicopter.
"How much fuel was on board?" Jake asked.
"Unfortunately, we had it topped off," John said.
"That leaves us just under four hundred gallons. If we can put another one together, we won't be able to top off the tank, but we'll have enough fuel to get to where we are going."
"If we can put another one together," John said. "I'm going to take Clay's Jeep and drive around the field to see if I can find something we can use to start over."
"I've been thinking," Jake said.
"Well, Major, that's why the Army pays you the big bucks," Deon said, and the others, including Jake, laughed.
"John, while you are looking at the other helicopters, I suggest that the rest of us build some hasty fortifications of some sort. That way if this happens again, we'll be ready for them."
"Good idea," Marcus said.
"I'm glad you think it's a good idea," Jake said. "Because now that I have suggested that, I have no idea what we can use for the fortification."
"There are ten fifty-five-gallon drums over in the hangar next door," John said. "They are empty, but if we put dirt in them . . ."
"Yes," Jake said interrupting him. "We did that in Iraq, built up around our Quonset huts. It worked well."
"You'll have to cut the tops off to get the dirt in," John said. "I've got a hacksaw and some blades here."
"Won't you need that if you find something out on the line?" Jake asked.
"If it isn't something I can take off using a wrench or a screwdriver, then it's not likely to be anything we can use. Take the hacksaw."
"All right. Let's get started," Jake said.
"A suggestion, sir?" Marcus offered.
"Any suggestion is welcomed."
"As soon as we get the top off one of the barrels, I suggest that one of us saw, while the rest of us fill the empty with sand."
"Good idea."
An hour later they had only two barrels filled with sand. Jake raised up from digging and wiped the back of his hand across his forehead.
"You know, when I said that we did this in Iraq—I forgot. There was no 'we' to it. We hired local contractors for the job."
"Yeah, I was wondering where you got that 'we,'" Deon said.
By nightfall, they had the ten barrels in a V shape in front of the personnel door. Back inside, they wondered aloud whether or not John would find anything they could use.
It was almost an hour later before John returned, and when he came into the hangar, the expression on his face told everything.
"We're stopped cold," John said. "There is nothing left that is salvageable."
"Do you mean to say that out of a hundred or more helicopters, that you can't put together one that is flyable?" Jake asked.
"I'm not saying that," John said. "But what I am saying is that there is so little salvageable remaining on each of the aircraft that it would take days, maybe weeks, to put one together. The biggest problem is with the airframes. Those that haven't been destroyed by all the scavengers are too badly damaged by the storm. I wish I had better news for you, but I don't."
"What do we do now?" Karin asked.
"What about the museum?" Deon asked.
"The museum? What about the museum? What are you talking about?" John asked.
"During the Vietnam War my dad was a door gunner on a Huey. There is a Huey on display at the museum just like the one he was on. I've seen it a dozen times—it looks like it's ready to fly."
"I wonder if the engine and transmission are in it." John said.
"They are," Jake said with a wide smile. "I remember reading an article in the Flyer last year about when it was brought to the museum. It was landed out front, then moved inside."
"How long has it been there?" John asked.
"I don't know, twenty years, maybe a little longer."
"And it was flyable when it arrived?"
"Yes, in the article I read, they had a big ceremony about it. The pilot who flew it in was one of the last Vietnam veterans still on active duty. Do you think you could make it flyable?"
"We could come a hell of a lot closer with it than I can with anything that's out here," John said. "All of the parts should still be there, but after all this time there will be dried-out bushings, filters, gaskets, and so forth. We'll have to rehydrate them, if we can."
"Question is, how do we get it here?" Marcus asked.
"We'll get it here in Clay's Jeep," Jake said.
"What? You can't get a helicopter in that Jeep."
"We could if we took the body off. Then we could set the helicopter on the Jeep's frame. The tail cone will stick out, but it's on skids, not wheels, so we can tie it down securely without worrying about it falling off."
"Yeah!" Deon said. "Damn right."
"Thing is, I hate doing that to Clay's car," Marcus said.
"Believe me, Marcus, I knew Clay better than anyone here. And if Clay were still alive, he would be the first one to say do it," Jake said.
"Yeah," John said. "I think he would too. All right, let's get this buggy stripped down."
It was a tired bunch who ate their supper that night, but before they turned in, they drew little slips of paper upon which times were recorded, the times determining who and when they would pull guard duty.
The next morning, with nothing left of Clay's Jeep but the frame, the men drove over to the Army Aviation Museum. Like all the other buildings on the fort, the museum had been vandalized and stripped of anything that could be construed to be of value. But the display of a Huey, depicting an LZ in Vietnam, was still intact. John opened the cowl and took a quick glance at the engine.
"All right!" he said. "Looks like nobody has messed with it. I think we've got a shot at getting this thing going!"
The hardest thing was going to be getting the helicopter loaded onto the back of the Jeep, but anticipating that, they had brought a crane and pulley system from the hangar and, after half an hour getting everything rigged up, John climbed up onto the engine deck and screwed a lifting eye onto the top of the mast. This was exactly the kind of lifting eye that was used by the aircraft recovery teams when they were sent in to pick up a downed helicopter on the battlefield.
When everything was rigged up, they began cranking on the crane and pulley system until the helicopter was lifted from the place it had occupied on the display for nearly twenty years, then swung over to the Jeep frame and lowered. The skids were lashed in place, and everyone but Deon, who was driving the Jeep, climbed into the helicopter for the drive back to Hanchey Field.
Once they had the helicopter in the hangar, John began a more thorough examination.
"Damn!" he said. "How did I not see this?"
"What?"
"We're missing a drag brace."
"How important is that?" Deon asked.
"Not all that important, if you don't mind throwing a rotor blade," John said.
"Anything out there we can use?" Jake asked.
John shook his head. "No, they are very precise."
"John, isn't there a Cobra helicopter there in the museum?" Marcus asked.
A huge smile spread across John's face. "Yes! And they share the same rotor system!"
"Let's go back."
"Deon, go with them," Jake said. "It's getting a lot more antsy out there. I don't know what they might run into."
"All right," Deon said. "Willie, the M-two-forty is still in the tower. How about you going up there and keeping an eye open while I'm gone?"
"Good idea," Willie said.
John, Marcus, and Deon climbed back onto what was left of the Jeep, as Willie went back up into the tower. That left Jake, Karin, and Julie alone in the hangar.
"If you don't mind, I'm going to see if we can get anything else on the radio," Julie said.
"You know how to do it?" Karin asked.
"Oh, yeah, I've been watching Willie."
Julie turned the crank to build up the power; then she turned the radio and started moving the dial through all the frequencies.
. . . establish contact. We have to be very careful in this, because the IRE, the Islamic Republic of Enlightenment, has their spies everywhere. No doubt they are monitoring this very transmission. Well, I've got news for you, IRE, there are millions of us out here. We've been knocked down, but we aren't knocked out. We have survivalist groups coalescing all over the country and the time is going to come, and soon, when we get together and reconstitute the United States of America.
To my fellow American patriots, find safe ways to contact each other, make yourselves strong, grow in strength, until we are able to join together as one unbeatable band. Until then, this is General Francis Marion of the Brotherhood of Liberty, and I'm using that term in its most generic sense, because we welcome our sister patriots with open arms. And in the Brotherhood of Liberty, men and women, black, white, Asian, American Indian, Protestant, Catholic, Christian, Jew, freedom-loving secularists, we are united, we are strong, and we will be victorious. I am asking you to grow strong, hold on, and wait until that glorious day when we will take our country back. God bless America!
Oh, do you think that's real?" Karin asked.
"I don't know if it is for real or not," Jake said. "But his name is obviously false."
"Why do you say that?"
"Because Francis Marion was in the American Revolution. He was the first guerrilla fighter."
"Are we going to try and make contact with them?"
"We'll play that by ear," Jake said. "For now our primary objective is to survive."
CHAPTER TWENTY-TWO
When John, Marcus, and Deon returned a couple of hours later, they began unloading parts.
"We took everything we might possibly need," John said. "We got both drag braces, the dampers, the pitch change links, the igniters, the fuel pump and fuel control, the hydraulic pump, the servos, the tail rotor gearboxes, the hangar bearings, everything we could get."
"We'll have this sucker up and flying in no time," Marcus said confidently.
"No trouble, I take it?"
"Not really," Deon said.
"What do you mean by not really? Was there trouble or not?"
"The place is crawling with scavengers," Deon said. "And they are getting frustrated."
"How do you know they are getting frustrated?"
"They've been setting fire to all the buildings." He looked at Karin and Julie. "The hospital is burned down," he said.
"Why would they burn the hospital?" Julie asked.
"My guess is they were after drugs, if not for themselves, to use in barter," Deon said. "But you know, for sure, that there were no drugs of any value left in the hospital when it was abandoned."
"You're right," Karin said. "As a matter of fact there was nothing left even before the hospital was abandoned. For the last two months, the strongest thing we had was aspirin."
"There are others like us out there," Julie said, happily.
"I'm sure there are," Deon said.
"No, I mean for real, like us," Julie said. "We heard them on the radio."
"Did you?"
"They call themselves the Brotherhood of Liberty," Julie said.
"They've asked us to join them," Karin added.
Deon and the others looked directly at Jake. "Are we?" Deon asked.
"Are we what?"
"Are we going to join them?"
"Maybe, someday," Jake said. "If they are legitimate."
"Why would you think they would not be legitimate ?"
"The way I look at it there is a fifty–fifty chance that it is legitimate. It could either be set up by the SPS to reel in the revolutionaries, or it could be legitimate. When the time comes, we may check it out. But this is not the time."
Wednesday, August 8
Because the helicopter had been flyable when it was put on display in the museum, work on it proceeded much faster than it had on the original Blackhawk. They had a little trouble with the drag brace because, as it turned out, the chord of the blade was a little wider than they thought, which meant the drag brace was a little longer, so they had to compensate by repositioning it slightly.
They replaced the drag brace and a couple of hanger bearings on the tail rotor driveshaft. In addition, they removed every gasket, seal, and filter, soaked them all in solvent, then oil. In that way they were able to reconstitute all but two. And those two they were able to replace by reworking gaskets they found on the helicopters that were still out on the airfield. Finally, they found a battery from one of the helicopters on the field that they were able to install, with some adjustment, into the Huey.
Finally they had everything put back, and were about ready to use the positioning wheels on the skids to roll the helicopter out of the hangar when a call came over the radio from Deon.
"Yeah, Deon, go ahead," Marcus said.
"We've got company coming," Deon said. "And it doesn't look like any social call. They came up on motorcycles, but they left them on the other side of the field. They are armed, and they are moving toward the hangar in combat advance."
"How many are there?" Marcus asked.
"A shitwad load," Deon said.
"That many?"
"At least."
"Alright, grab your weapons. Let's get outside and into position," Jake called to the others. He took the radio from Marcus. "Deon, stay alert, but keep out of sight as much as you can."
"Roger."
Jake and the others rushed outside, then took up positions behind the V of sand barrels. "John, you take the right end of the V. Willie, you take the left end. Marcus, you and I will have the point. Ladies, one of you on each side," Jake directed.
Jake waited until everyone was in position; then he raised his head just above the barricade and brought the bullhorn to his lips.
"Those of you coming across the field. Turn around and go back. Do not come any closer," he said over the loudspeaker.
"What have you got in the hangar?" someone shouted back.
"Nothing that concerns you. Turn around and go back."
"You got gasoline in there?"
" No."
"You're lyin'! Let us take a look."
" No."
"I don't believe you. I think you've got gasoline in there, and we're going to take it."
"I told you, we don't have gasoline. Look around out on the field, take what you want, but do not come any closer. This is your last warning. If you come closer, we will shoot."
The answer this time was a rifle shot. The bullet whistled by just overhead, then punched through the hangar wall.
"Turn around and go back!" Jake said over the bullhorn. "There's nothing here for you."
This time two of the scavengers fired.
"Jake, I have the shooters in sight," Deon said. "Permission to fire?"
Two more shots were fired by the scavengers, and one of the bullets hit the top of a barrel, and sent a little shard of steel into Marcus's face.
"Damn, I'm hit!" Marcus said.
Jake looked at him, then laughed. "You've cut yourself worse, shaving," he said. He picked up the little radio. "Deon, fire at will," he said.
Deon opened up with the M-240 from the top of the control tower. Jake could see the tracer rounds slashing down, and he heard one of the scavengers let out a yell of pain.
This time there were more than a couple of rounds fired—several scavengers opened fire, and some had M-16s, as evidenced by the automatic fire. They started maneuvering toward the hangar and as they worked their way forward, Jake counted at least twelve.
The Phoenix group was outnumbered, but they had position, and with Deon and the M-240, superior firepower.
The firefight was intense for several minutes; then it died off. It was quiet for a moment. Then Jake and the others could hear the motorcycle engines start. A moment later, they could hear the Doppler effect of motorcycles riding away.
"They're gone," Marcus said.
"Maybe," Jake replied. He leaned his rifle against one of the barrels, then pulled his pistol. He thumbed the magazine out, checked it, then slid it back into the handle. "But I'm going to have to find out."
"You going out there alone?" Marcus asked.
"Yes, no sense in risking more than one of us."
"You're not going out there alone," John said. "If something happens to you, we're all up shit creek. You're the only one who can fly this thing."
"John's right," Deon said. Deon had come down from the tower to join the others. "We can't risk you. I'll go out."
"You can go out with me," Jake said. "But I'm going out."
"Pulling rank on us, are you—Major?" John asked, coming down hard on the last word.
"I told you, we don't have rank," Jake started to say; then he paused. "All right, mea culpa. Deon, do what you have to do."
"Want company, Deon?" John asked.
"No. No offense, but you are a wrench turner. I can do better if I don't have you to worry about."
John smiled. "Okay, Rambo, fine by me. I was just putting on a brave front for the ladies."
Deon came back after about fifteen minutes with his report. "Six dead, one wounded."
"How badly is he wounded?" Karin asked.
"He was hit in the thigh, but I don't think he's going to die."
"He could if he gets an infection. Or at the minimum, lose his leg. I'd better go take a look."
"Why?" John asked. "Half an hour ago the son of a bitch was trying to kill us."
"He's probably a soldier, John, just like us," Karin said. "If the situation was normal, you would pass him in the PX and never blink an eye."
"Yeah, you're right," John said.
"I'd better go with you," Deon said.
"Wait until I get my kit."
The wounded scavenger looked to be in his late twenties. He was wearing BDUs, but there was no rank visible. He was sitting up, holding a belt tourniquet around his leg.
"No," Karin said. "You don't want to use a tourniquet unless you are unable to stop the bleeding by direct pressure. Otherwise you could get tissue damage. Let me take a look."
"You a doctor?"
"What difference does it make to you who she is?" Deon asked. "Half an hour ago you were trying to kill her. Now she's here to help you, though why she is willing to do that beats the hell out of me."
Karin removed the tourniquet and looked at the wound. "I've got to get the bullet out," she said.
"How are you going to do that?"
"I'm going to pull it out," she said, removing a forceps from the kit she had brought with her. She stuck the forceps down into the wound until she came in contact with the bullet. Then, grabbing the bullet, she pulled it out.
"Damn, it hurt more coming out than it did going in," the wounded man said.
"Good," Deon said. "If it was up to me you'd be dead now. So if you're goin' to live, I at least want you to hurt some."
"The problem is going to be if any of the cloth from your pants went into the hole with the bullet," she said.
"How are you going to know?"
"I'm going to look for it," she said. She took another instrument from her kit that looked like an oversized pair of tweezers from her kit. She put this down into the wound and clamped it shut. "Ahh, feels like I got something."
Pulling the tweezers out, she saw a small piece of cloth clamped between the arms.
"Good," she said.
By now the bleeding had stopped and Karin took out a bottle of alcohol. "This is going to hurt a little," she said.
"It already hurts," the wounded man said.
Karin poured alcohol onto the wound.
"Damn, damn, damn!" the wounded man said, shutting his eyes and wincing in pain.
Karin used a cotton ball to clean the wound. Then, she soaked a second cotton ball in alcohol and stuffed it into the bullet hole. Finally, she wrapped a compression bandage around the wound and secured it tightly.
"Don't take this off for at least seventy-two hours," she said. She stood up. "I'm ready to go back," she said to Deon.
"What? Are you just going to leave me here?" the wounded man asked.
"I've done all I can for you," Karin said.
"But what do I do now? Where do I go?"
"You can go anywhere you want," Karin said. "And if you keep the wound clean, it should heal without any difficulty."
"I left my bike on the other side. I can't walk. Will you bring it to me? It's the green . . ." He paused and looked over toward the body of one of the other scavengers. "I mean it's the Purple Honda VTX-1800," he said. "Only thing is, Cootie, over there, has the keys."
"All right," Deon said.
"Deon, you know that isn't his, don't you?"
Deon shrugged. "What difference does it make now?"
Karin chuckled. "I guess you're right."
"Hey," the wounded man said. "Thanks for patching me up."
Karin nodded, but said nothing.
"Listen, 'cause you helped me? I'm going to tell you something. They'll be back. And now they know about the machine gun and they know you've built a barricade in front of the hangar. They'll be back, and this time, there will be a lot more of them. They know you have fuel."
"We don't have any gasoline," Karin said.
"Doesn't matter. You've got jet fuel, and it'll trade just as well. I hear you're building a helicopter in there."
"Where'd you hear that?" Deon asked, coming back with the keys.
"Word gets around. If I was you, I'd get out of here as soon as you can."
"Thanks for the warning," Karin said.
"Yeah, well, I guess I owe you," the wounded man said.
Deon jogged back to get the motorcycle as Karin returned to the hangar. She heard the motorcycle start just as she reached the fortifications.
Deon waited until the wounded man drove off. Then he returned to join the others. "Did you tell them what he said?" he asked Karin.
"Yes."
"I think we're all set now," John said. "You've never flown a Huey before, have you, Jake? You think you can fly it all right?"
"Have you ever worked on a Huey before, John?"
"No, I never have. But the principles of maintenance aren't that much different. A helicopter is a . . ." John stopped in midsentence, then smiled. "Okay, I get your point. Have at it, Jake, your chariot awaits," John said, holding his hand out invitingly, toward the helicopter.
"I hope this jury-rigged battery works," Jake said. "Cross your fingers that we don't get a hung start or hot start."
He checked battery voltage and placed the starter-generator switch in the starter position, turned on the main fuel pump, then opened the throttle to a point just below the flight idle detent. He pulled the starter trigger on the pilot's collective pitch control and heard the igniters pop in his earphones as the engine started spooling up, monitoring his gauges closely. He was gratified to see everything move into the green. With a big smile, he gave a thumbs-up to those waiting outside. His test flight, which was nothing more than a sweep around the airfield, went well. He saw the one motorcycle going up Hatch Road, and he saw no one coming toward them. He landed and killed the engine.
"Let's get it loaded and get out of here," he said.
CHAPTER TWENTY-THREE
The Dunes, Fort Morgan—Wednesday, August 8
"Lookie here," James said as he, Bob, and Jerry were poking around in the refuse left by the hurricane.
"What?" Jerry said.
"This a solar panel power setup. Or, what's left of it after the storm."
"You think we can reconstruct it?" Bob asked.
"Let me see what's here," James said. "We've got the panels. We're going to need to poke around and see if we can find everything else we need."
"What would that be?" Jerry asked.
"We need a current regulator, something that will keep the batteries from overcharging, or draining in case the current tries to run backward."
"Like a reverse current relay between a battery and a generator?" Bob asked.
"Yes. And if we're going to use it in a house, we'll also need a converter that changes DC to AC."
"Why do we even need to mess with it?" Jerry asked. "I mean, as long as we've got propane, we have power."
"What happens when we run out of propane?" Bob asked. "Where are we going to get more?"
"Oh, yeah, I see what you mean. Okay, let's see if we can find everything that you need," Jerry said.
"Problem is," James said, "this setup will only power one house."
"No problem. When the time comes and we've run out of propane, we'll just choose the house we want, and we'll all move in together," Bob said.
"Or build us a new house," James said. "One that will accommodate three families, and make maximum use of the electricity we can generate this way."
"Whoa, I don't know. I'm not much at house building," Bob said.
"James and I will build it. You can be the gofer," Jerry suggested.
Fort Rucker—Wednesday, August 8
With all their survival gear on board, Jake pulled pitch and the helicopter took off. As they passed over the golf course they saw two people playing golf, and Jake laughed.
"What is it?" John asked. John was in the left seat and, like Jake, was wearing a flight helmet. Jake keyed the mic switch to the first indent, which was intercom. "I always heard that come hell or high water, a committed golfer was going to play. Look down there."
Jake made a circle around the Silver Wings Golf Course, and the two players waved up at him.
As they flew south toward the Gulf, they saw, as they expected, the highways littered with cars and trucks. Then he heard Karin's voice on the intercom. She was also wearing a headset and was plugged in to the crew chief's section.
"Jake, look down there," Karin said.
When Jake looked around, Karin pointed to something on the ground. There, below, in white paint on the grass behind a farmhouse were the words:
HELP
We Need Food
Medical Care
"Can't we land so I can see if I can help?" Karin asked.
"They probably need food more than they need medical care," Jake replied.
"Maybe we could give them a case of MREs," Karin suggested.
"It's all right with me if you can get the others to agree," Jake said.
Karin discussed it with the others, then a moment later keyed her mic again.
"Everyone else says it is okay," she said.
"All right, we'll land and see what we can do for them."
Jake circled back, then started his descent. The rotor blades popped loudly as they cavitated down through their own rotor wash.
Just as he was flaring out to land, two men ran of the house. Both were carrying M-16s and they began shooting.
Jake terminated the descent, pushed the cyclic forward, and jerked up on the collective. The helicopter leaped up over the house; then he lowered pitch, flying nap of the earth and putting the house between them and the two gunmen on the ground.
He continued flying at a low level, popping up just high enough to clear the ground obstacles. Finally, when he was more than two miles away, he climbed back up to altitude.
The blades were now making a whistling sound and the helicopter had picked up a slight vertical bounce.
"Damn," John said. "We've got a whistle and a one-to-one vertical."
"Yeah, I've been through this before," Jake said. "We took a round, or maybe a couple of rounds, through the rotor blades."
"They didn't really need help, did they?"
"No. They were using it to lure a helicopter down so they could do to them exactly what they tried to do to us."
"Yeah, but, you've got to wonder just how many helicopters there are flying around right now," John said.
"Can't be too many, I don't think," Jake said. "As far as I know we are the only ones to fly out of Rucker in the last six weeks, and I can't think of anyplace that would be more likely to launch a helicopter than us. Bless their hearts, we were about their only chance and they blew it."
John laughed so hard that tears began rolling down his face, and only Karin, who also had a headset, knew what he was laughing about. The others looked at her quizzically, and she tried to explain what they were laughing at, but it wasn't the same.
The flight down was almost two hours in length, and all along the route they saw vehicles abandoned on the road, towns with their business districts deserted, and burned-out houses. They also saw several bodies, some alongside the roads, some on the streets of the town, and others lying out in open fields.
"There it is, the Gulf," Jake said.
Before them the Gulf spread from horizon to horizon, blue and sparkling in the sunshine. Jake turned right and they saw a long peninsula stretching toward the west. The peninsula was quite narrow and it separated the Gulf of Mexico from Mobile Bay.
"Where is this place we're going?" John asked.
"It's at the very end of this peninsula."
"It's an island," Karin said. "They call it Pleasure Island."
"How can it be an island? It isn't surrounded by water," John said.
"Yeah, technically it is," Jake said. "We just crossed the intercoastal canal. That cuts the peninsula off from the mainland and makes it an island."
Jake dropped down to about five hundred feet, then flew right along the surf. As they looked out onto the very expensive houses along the beach they could see the extent of damage from the recent hurricane. At least one out of every three houses was completely destroyed, and half of the ones still standing were damaged by degrees from light to severe. The farther west they flew, the greater the damage.
"Look at these houses," Karin said. "They are million-dollar-plus houses."
"Yeah," Jake said. "And that was back when a million dollars meant something."
The Dunes, Fort Morgan
Bob Varney was on page one hundred thirteen of his work in progress. One thing he did miss about the computer was the word-count feature. He estimated that he was just under twenty-five thousand words, using his old method of counting at two-hundred-twenty words per page.
After he finished the book, he would have to go back through the first draft as he did in the pre-computer days, marking up typos, and making editorial adjustments. That would require retyping the entire manuscript. It was hard to believe that he wrote his first one hundred books that way.
Suddenly Bob heard a sound that took him forty years back in time. For a moment, he wasn't in the third-floor office of his beach home, he was in the BOQ at Tan Son Nhut in Saigon. He was typing on this very typewriter, and passing overhead was a UH-1 helicopter returning to the base.
Bob had been hearing helicopters pass overhead for the last ten years, but he had not heard anything like this in a long time. The UH-1 has a very unique sound. Put any Vietnam veteran in an open field, blindfold him, and fly ten helicopters overhead but only one Huey, and the veteran will be able to tell which one is the Huey.
"Damn, Charley, that's a Huey!" Bob shouted. Getting up from his desk, he stepped onto the front deck and saw, flying at an altitude of about five hundred feet over the surf, a U.S. Army UH-1D helicopter, in the muted colors that were used for such aircraft in Vietnam.
"Choi oi," Bob said, using the Vietnamese expression that covers everything from mild surprise to total shock. "It is a Huey! What is it doing down here?"
Charley looked at him quizzically.
"You don't understand, do you?" Bob said. "Well, here's the thing, see. I used to fly that very same kind of helicopter. Oh, this was long before you were born, Charley Dog. And the thing is, this is definitely an Army helicopter, but the Army doesn't fly them anymore. And, even if they did, what is it doing down here?"
Bob watched the UH-1D as it headed west, still over the surf; then he saw the pilot make a climbing turn out over the water, gaining another two hundred feet or so as he continued the turn through two hundred seventy degrees. Then the helicopter started descending.
"Where is he going, Charley?"
Bob ran back into the house. "Ellen!" he called. "A Huey!"
"What?"
"Didn't you hear the helicopter?"
"Yes. What about it? Isn't it going out to one of the offshore rigs?"
"No, it wasn't a civilian helicopter. It's Army. It was a Huey, a UH-1D model."
"Maybe the Army sent someone down here to check on the damage from the hurricane."
"This isn't the Army."
"I thought you just said that it was."
"I meant it was an Army helicopter, but not the kind they use now. This was a Huey, like I flew in Vietnam. Only it's been thirty years since the Army flew them, so who is flying it, and what is it doing down here?"
Fort Morgan
Jake circled over Fort Morgan so everyone could get a good look at it.
"Wow," John said. "It's shaped like a star."
"Yes, that's the way forts were built then."
"What is it doing here?"
"It has been preserved as an historical landmark," Jake said. "This very fort is considered the finest example of military architecture in the Western Hemisphere."
"It sure looks old."
"It is old. Well over one hundred fifty years old. It was a very important fort during the Civil War."
Jake made a wide circle out over the sea, nearly as far out as the old Mobile Point Lighthouse; then he turned back toward the fort, setting up a 500-feet-per-minute rate of descent. He made a long, shallow approach to the fort until he cleared the walls. Then he arrested his forward motion, moved into a hover, and finally made a vertical descent for the last fifty feet. The rotor wash blew up bits of grass and some of the sand that had been blown in by the recent hurricane. He touched down, then killed the engine, and the blades coasted down until they stopped.
For a long moment after the aircraft sat down and the rotors stopped, nobody got out. They all sat there in awe and silence, listening to the descending hum of the instrument gyros as they spun down.
"And this is where we are going to live?" Julie asked.
"This is it. You can consider this place your home sweet home," Jake said. He loosened his seat and harness. "What do you say we step outside and have a look around?"
The ground inside the fort was amazingly green with well-nourished grass. The sandstone walls were high, and gray, and foreboding looking. There were several sally ports that led from the field to the outside of the walls, and into the walls of the fort itself. They looked around for a bit. Then Jake pointed to one of the sally ports. "What do you say we take a look in there?" he suggested.
They walked across the grass, then stepped up onto the paving stones and walked in to the arched passage way. Once out of the sun, it became much cooler. Coming off the sally port and branching off to one side was a large, all-brick room.
"What do you think this was?" Willie asked.
"My guess would be that it was an ammo casement," Jake said.
"Casement? You mean like casement windows?" Julie asked.
"No, a casement is also a secure room for storing arms and ammunition."
"Secure, huh? Well, you can't get much more secure than this," Marcus said. "Look at this thing."
They left the first casement and began exploring the others. The brick rooms were large, dank, and dark, lit only by the door off the sally port and tiny slits that were more for ventilation than for light.
"This looks like a dungeon," Karin said. "I can almost close my eyes and see someone hanging from the walls."
"Oooh," Julie replied with a shiver. "That's a pleasant thought."
"Ah, hang a few pictures, put up some curtains, throw a carpet on the floor, and we'll have it looking really homey in no time," Jake said, and the others laughed.
"Is this really going to be our quarters?" Marcus asked.
"Until we can come up with something better," Jake replied. "And we can grow a garden. Did you see how green the grass is?"
"It's too late to plant anything now, isn't it?" Willie asked.
"I don't think so. It doesn't get cold down here until mid-December. That leaves us almost five months, if we get the garden in soon."
"We'll get it in soon," John said. "What else do we have to do?"
"I'd like to take a look at that gun on top," John said. "It doesn't look like a Civil War cannon to me."
"It isn't," Jake said. "It's an eight-inch coastal artillery gun."
"What's it doing here?"
"This fort was manned until the beginning of World War Two," Jake said.
"Ha! Did they think the Germans were going to attack us by ship?" John asked.
"They did attack us by ship. Or at least, by submarine. There were fifty-six ships sunk in the Gulf by German submarines. Some of the people who lived on the coast then could actually watch the attacks."
"So, that big gun up top was used, huh?"
"No, it would have only been good against surface vessels, and if any of the German surface ships came into the Gulf, they never came close enough to the coast to be seen. Only thing to come this close were the submarines, and there were an awfully lot of them."
When Jake and the others went back out into the open area of the fort, they were surprised to see three men standing there by bicycles. All three men looked to be in their late sixties and none of them were armed.
"Who are you?" Jake asked.
"My name is Bob Varney, and we were about to ask you the same thing."
"I'm Jacob Lantz," Jake said.
"What are you doing here, Jake?"
"What business is it of yours what we are doing here?" Willie asked.
Jake held his hand out. "They aren't armed, Willie. I don't think they mean us any trouble." He turned back to Bob Varney. "It got a little too difficult to stay where we were, so we decided to come down here. We're going to live here for a while."
"You're going to live here? In the fort?" one of the men with Bob asked.
"Yes. Who are you?"
"My name is James Laney. Bob, Jerry Cornett, and I live here, with our wives."
Jake looked surprised. "You live here, in the fort?"
"No, not in the fort. Just back down the road a ways, to the first bunch of houses."
"It looked to me like nearly all the houses were destroyed by the hurricane," Jake said.
"They just about were."
"Are you the only ones out here?"
"We always were the only ones who lived here full time," Jerry said. "Then, when the economy got so bad, nobody else could afford to come down here, so for the last few months we've been here all alone."
"Do you have any idea about what's going on in the rest of the country?" Bob asked. "Since there is no radio or TV, we are completely in the dark. But we heard that there were some nuclear bombs dropped."
"They weren't dropped," Jake said. "They were smuggled in on board cargo ships."
"Damn. So, what's being done about it? Are we under martial law?" Bob asked.
Jake shook his head. "You would have to have a military in order to have martial law," he said. "And thanks to Ohmshidi, we no longer have a military. We are all military—or at least as close to the military as you are going to get. Also you have to have a government to declare martial law, and we no longer have a government."
"And you are?" Bob asked.
"Sergeant—that is, John. John Deedle." John introduced the others in the group.
"We call ourselves Phoenix," Julie said.
"Good name."
"Bob's an author," James said.
"Bob Varney—yes, Robert Varney," Jake said. "I've read some of your books."
"Don't hold that against me," Bob said with a self-deprecating smile.
"No, I enjoyed them. Seriously. You are a very skilled writer."
"Perhaps, but you can understand, I am sure, that of all the skills needed for survival, a writer's skill contributes the least."
"What are you planning to do here?" James asked.
"We are establishing our base here," Karin added.
"Firebase Phoenix," Bob suggested.
Karin smiled broadly. "What a great name!" she said. "Yes, this is Firebase Phoenix."
"All right, so tell me. What happened to the president? What happened to Congress?"
"As far as Congress is concerned, Ohmshidi dismissed Congress, and the Supreme Court, so they were already long gone," Jake said. "And as far as the president is concerned, that cowardly bastard is hiding out somewhere."
"Or he's dead," John said.
"No such luck," Deon put in.
As the others continued conversing among themselves, Bob walked over to the Huey and looked at it.
"What do you think?" Jake asked.
"I thought the Army retired the last Huey thirty years ago."
"They did. We stole this one from the museum."
"And put it back in flying shape. You must have some pretty good maintenance men."
"They're the best."
Bob looked up at the rotor head. "That's not the original drag brace. But evidently it held, all right. What is it? Off a five-forty rotor head?"
"A five-forty rotor head?" Jake asked.
"This is a D model. The C model and the Cobra both have five hundred forty rotor systems. Slightly different."
"Yes, we took the drag brace from a Cobra," John said.
"That's what I thought. Looks like you did a good job of adapting it."
"You seem to know your helicopters," Jake said.
"Well, I certainly know the UH1-D model," Bob said. "And probably just about every other one you saw in the museum."
"Were you an Army aviator?"
"Yes, I have a little over six thousand hours, thirty-six hundred hours in one just like this. I flew three tours in Vietnam, and I taught maintenance test flight procedures at AMOC in Fort Eustis. Chief Warrant Officer-four, retired."
Jake extended his hand again. "Major Lantz, and I didn't get a chance to retire. Ostensibly I'm still on active duty since I never got any orders relieving me. We are all in the same boat. Karin is a captain, an Army nurse, the rest are sergeants."
"Welcome to Fort Morgan," Bob said. "It'll be good to have neighbors again."
CHAPTER TWENTY-FOUR
When Bob, James, and Jerry returned to The Dunes after their visit to Fort Morgan, they found that all the women in James's house were in a condition of shock and fear. Becky was lying on the couch and the other women were gathered around her. She had a black eye and a split lip.
"What happened?" Bob asked.
"We've been robbed," Ellen said.
"Robbed? How did that happen?"
"Four armed men came through the gate," Ellen said. "When they found out we were living here, they broke in to the house and took as much as they could carry from the pantry and from the freezer."
"They also took the propane tanks," Cille said.
"We have no food or electricity," Gaye added.
"What happened to Becky?"
"We don't really know," Ellen said. "She had gone with Sarah, so they weren't here when the men came. It wasn't until after the men left that we found her."
"You found her?"
"She was down by the swimming pool. She was lying in the road and at first we were afraid she was dead. But she came to, and we helped her back. We asked her what happened, but she can't remember."
"What's the last thing you remember?" Bob asked.
"Walking on the beach with Sarah," Becky said.
"Where is Sarah now?"
"I don't know," Becky said. "I don't even know how I wound up at the swimming pool. One minute I was walking on the beach, and the next minute Ellen and Cille were asking me how I feel."
"We are a little worried about Sarah," Cille said.
"Maybe more than a little worried," Gaye said.
"I'll go look for her," James offered.
"Was there anyone down at the fort?" Ellen asked.
"Yes. And that reminds me," Bob said. "Becky, why don't you come with me? One of the people in the helicopter that landed down at the fort is an Army nurse. I think maybe she should look you over."
"You don't need to do that."
"You were knocked out, weren't you?"
"I guess I was. I don't know why else I would be lying in the road next to the swimming pool."
"That means that, at the very least, you have a concussion," Bob said. "I think it would be good for her to look you over."
"All right."
"If you and James are both going to be gone, I think I should stay here with the women," Jerry said.
"I think that's probably a good idea. Why don't you go over to my house and look in the top drawer of the chest of drawers in the bedroom. I have a P-thirty-eight pistol there, with a full magazine."
"Right," Jerry said.
"Becky, do you feel up to walking down to the golf cart? We'll take that down to the fort."
"I think so," Becky said. She sat up, but when she tried to stand, she fell back on the sofa. "I don't know, maybe I can't."
"Tell you what. We'll set you in a kitchen chair," Bob said. "I'll hold the back and Jerry can take the legs. We can get you down the stairs that way."
Cille brought a chair over and they helped Becky onto it. Then the two men carried her downstairs, all the way to the golf cart. She was able to move from the chair to the golf cart on her own.
"I think I've got enough charge left to get us there and back," Bob said as he put the cart in gear.
"Jake!" Deon called down from the top of the wall. "It's that warrant officer and some woman, coming up on a golf cart."
"Wave them on in," Jake said.
Deon was standing on the north parapet of the fort and he waved at the two on the golf cart, inviting them in.
Jake and Karin went out to meet them.
"Hi, Bob what brings you back so quickly?" Jake said. Then he noticed Becky. "Good Lord, what happened?"
"She doesn't know, exactly," Bob said. "But while we were here visiting, some men broke into our place. They took all our food and our propane. After they left, the women found Becky unconscious on the road near the house. I remembered that the captain here was a nurse."
"Yes, I am," Karin said. "A nurse, I mean. None of us have rank anymore."
"I was wondering if you would take a look at her."
"I'd be glad to. Becky, is it?"
"Yes, ma'am," Becky replied.
"I'm Karin. Bob, why don't you get down and let me drive her into one of the casements so we can have a little privacy."
"All right."
"You say you were robbed?" Jake asked, after Karin drove off.
"The women were, yes. They took almost all of our food."
"Look, we don't have much, but we could let you have a case of MREs. That's not going to last you very long, though."
"We've got a couple of people who are pretty good at hunting and fishing," Bob said. "We'll get along all right."
"They're good at it, you say?"
"Yes, very good. With wild game and fish, and seaweed for our vegetable, we'll survive."
"I wonder if you would be interested in something," Jake said.
"What is that?"
"Moving in with us. I know that you are probably much more comfortable in your own house right now, but I'll be honest with you, Bob. If you have been holed up here for the last six weeks or two months, you don't have any idea what is going on out there. You are likely to have more unwanted company, and they might do more than just rob you. If you are with us, you would at least be safe."
"That is something to consider," Bob said. "Let me take it up with the others. If it were just Ellen and me, I would take you up on it in a military minute. But I don't want to abandon the others."
"I understand," Jake said. "Talk with the others and see what they say."
"I can see the advantage we might have in moving in here, for security purposes," Bob said. "But what would you get out of it?"
"The more of us there are, the more security that is for all of us," Jake said. "And, to be honest, I'm intrigued by the fact that you say James and Jerry are very good at hunting and fishing. We have enough food and MREs to last us a few weeks, somewhat less if you join us; then we are going to have to supply our own food. I see there is chickweed here. That's edible. And, as you say, seaweed."
"Across the road there are some scuppernongs," Bob said.
"Scuppernongs?"
"It's a wild grape. It's good raw, and it makes a really good jelly."
"We've brought seeds," Jake said. "And they are pure seeds, not hybrid. I plan to get a garden in as soon as I can. If it stays warm long enough, we'll have some homegrown vegetables in six to eight weeks."
"Sounds good. What have you got?"
"Several kinds of beans, peas, corn, beets, carrots, cucumber, radish, spinach, cabbage, lettuce, tomato, onion, bell peppers, jalapeno peppers, watermelon, and cantaloupe. I also have a dozen potatoes that we can use the eyes from."
"I can make a great possum stew with potatoes, onion, carrots, and jalapeño peppers," Bob said. "I can't wait until the garden grows."
Karin drove the golf cart back out into the central area then. Becky was crying, quietly. Bob looked at her quizzically, but Karin got out of the cart and signaled for Bob to step away from the others.
"What is it?" Bob asked.
"She was raped," Karin said.
"Raped?" Bob gasped. He looked back toward her. "Those bastards!"
"She doesn't remember it, but while I was examining her, she told me she felt pain there. She isn't wearing panties under her dress, though she insists that she had them on when she and the other girl left. She isn't wearing them now, and there are abrasions and semen residue."
"Damn," Bob said.
"She's very upset, as I'm sure you can understand. She's likely to go through some serious depression over the next several days, at least until she learns whether or not she was made pregnant."
"The poor girl," Bob said. "Oh, what about her dizziness?"
"She had a concussion, that is for certain," Karin said. "But I don't think she had a skull fracture. I treated her lacerations. If she keeps the wounds clean, they shouldn't give her any trouble."
"Thanks, Karin," Bob said. He walked back over to the golf cart, then spoke to Jake. "I'll talk to the others soon as I get back," he said. "For now, we have a girl missing. James is out looking for her."
Becky was quiet for half of the trip back from the fort. Then she spoke, so quietly that Bob could barely hear her.
"Don't tell the others," she said. "Don't tell the others I was raped. I'm so ashamed." She began to cry.
"Becky, you have absolutely nothing to be ashamed of," Bob said.
"I don't know whether I do or not," Becky said. "I don't even remember it. God in heaven, how can you be raped and not even remember it?"
"It is traumatic amnesia," Bob said. "It's not that uncommon. I've experienced it myself."
"But you won't tell? Promise that you won't tell."
"I won't tell," Bob promised.
When Bob and Becky returned, James was back with Sarah safely in tow.
"She was all the way down to Ponce de Leon," James said. "She had no idea we were looking for her."
"How is Becky?" Sarah asked.
"She'll be fine," Bob said. "The nurse doesn't think she has a skull fracture. But she did have a concussion and doesn't remember a thing, so don't be pestering her with questions."
All the women clustered around Becky, anxious to do whatever they could for her. While they were talking, Bob brought up Jake's offer to the two men.
"I don't know," James said. "It's going to be hard to get Cille to leave this house."
"Not as hard as you think," Cille said. "What are you talking about?"
"It's about time we let them in on it as well," Bob said. "We sure aren't going to make this decision without them.
"You got that right," Jerry said. "Go ahead, tell them."
"Ladies," Bob said, "I have a proposition for you."
Bob told them then who was down at the fort. He told them also that he trusted the men and women he had met down there.
"I trust them too," James said. "There are some folks that, as soon as you meet them, you just know what kind of people they are, and these are good people."
"Here's the thing," Bob said. "They have invited us to come down there and live with them."
"What?" Gaye said. "You mean live there, in the fort? Have you ever been there?" Gaye wrapped her arms around herself and shivered. "Even in the summertime it's cold there. Nothing but brick, stone, and cement."
"Yes," Bob said. "That's because it is a fort, literally, and that's what makes it attractive. There won't be any repeat of what happened here."
"If you ladies decide that you are willing to do this, I'll make you a promise," James said. "I'll build us some place comfortable for us to stay. There is enough building material scattered around here from the hurricane that I don't expect it will be very hard to do."
"I'll say this," Ellen said. "I sure don't want a repeat of what happened to us this morning."
"I'm for it," Cille said.
"If we have a vote in this, I'm for it as well," Sarah said.
"Of course you have a vote," Bob said. "Right now you are one of us."
"Becky? Becky?" Sarah asked. "What do you say?"
"I'm for it," Becky said.
"Gaye, it's up to you," Ellen said.
"No, it isn't up to me," Gaye said. "If we are voting on this thing I have already been outvoted."
"I guess you have been," Bob said. "But I really don't want to force anyone into anything that they don't want to do."
"You aren't forcing me," Gaye said. "Everything you are saying makes sense. If we are down there, inside a fort with a bunch of other people, then it has to be a lot safer there than it would be if we stay here. I'm for going as well."
"Alright, let's load up James's truck with whatever we think we might use."
"Are we going to take any of the lumber, or building material?" Jerry asked.
"What do you think, James?" Bob asked.
"I don't think so, yet," James said. "We'll just take whatever personal items we are going to need on this trip. It's going to take several trips to get what we'll need to start building."
CHAPTER TWENTY-FIVE
After Bob Varney and his group moved down to Fort Morgan they, along with Jake and the Phoenix group, began to build a place to live, using as their scheme a motel-like plan. Jake and James drew up the design of one long, single-story structure divided into individual cabins.
It was decided that each of the three married couples would have their own cabin, Karin and Julie would share a cabin, Becky and Sara would live together, and Jake and Deon would be roommates, while John, Marcus, and Willie would share the final cabin.
They began the structure by using one of the massive stone walls of the fort as their back wall. Next, they built a floor that extended twenty feet out from the wall, and stretched one hundred and five feet long. After they finished the floor, they put in a wall at each end and separated the floor into seven compartments, each protruding fifteen feet out from the stone wall of the fort. This left five feet for a front porch. Next they put on the front wall, with a door and window for each unit, then a roof, with a chimney from each unit. Finally they built a fireplace in each room.
Throughout the entire construction project, James and Jake worked very well together, James, because he was a natural handyman, and Jake, because such work was a product of his youth. He had built many wood-frame structures, even as late as last March when he had gone back home to visit his folks, and helped in a barn raising.
While the men were building their quarters, Bob and the women put in the garden. Ellen was a particularly good gardener, as was Julie, and they took charge of the layout and planting.
Fort Morgan—Wednesday, August 22
So far all of the building material came from the scrap lumber and residue left by Hurricane Ohmshidi hauled down to the fort in James's truck. But midway through the process James announced that he didn't think he had enough gasoline to make another trip.
"We can make our own gas," Bob suggested.
"Ha! How, by eating a lot of beans?" John asked.
The others laughed.
"No," Bob said. "During World War II a lot of people converted coal, charcoal, or wood, to a gas that would power their vehicles. John, you are a good mechanic and James, you can do just about anything. You two could build a gas converter."
"I don't have any idea what you are talking about," James said.
"No problem. I'll show you how to do it."
"You?" Ellen said. "Bob, you can barely change a lightbulb."
"That's true," Bob said, "but when I was in the Army, I taught aircraft maintenance. I couldn't do it, you understand, but I could teach it. And I could teach John and James how to make a gas converter. It isn't all that efficient, but it will work.
"Could it work well enough to drive the truck up to Fort Rucker?" Jake asked.
"Yes, I suppose so. But why would you want to go up there?" Bob asked.
"Because I know where there is a thousand gallons of gasoline."
"What?" Marcus asked, surprised by Bob's comment. "Where?"
"Clay hid nineteen barrels of Mogas in a hangar at TAC-X."
"TAC-X? A stagefield?" Bob asked.
"Yes, at Samson. Do you know it?"
"I know where Samson is," Bob said. "Don't know TAC-X. I know TAC-Able."
"TAC-Able? Don't you mean TAC-Alpha?" Willie asked.
"You are dealing with an old soldier here, Sergeant. Before the phonetic alphabet changed in 1956, the letter A was Able. Able, Baker, Charley, Dog, Easy, Fox, George, Haystack, Item, Jig, King, Love, Mike, Nan, Oboe, Papa, Queen, Roger, Sugar, Tare, Uncle, Victor, William, X-ray, Yoke, Zebra."
"Wow," Willie said. He laughed. "And you still remember all that?"
"When you are my age, Willie, you will still remember the current phonetics."
"That's interesting, but let's get back to this thousand gallons of gasoline at TAC-X," John said. "How do you know it is there?"
"Sergeant Major Matthews put it there," Jake said. "Just for something like this."
"Yeah, but what good does it do us up there, if we can't go get it?" Marcus asked.
"We can go get it, if Bob really can tell John and James how to build a machine that will convert wood into gasoline," Jake said
Bob shook his head. "Not gasoline," he said. "Wood gas. When you burn wood, or charcoal, or coal, or any other fuel, if you don't completely consume the fuel, it will emit a gas that is combustible. Sometimes, for example, if the mixture in a gasoline engine is too rich, the exhaust gas can be burned. That's what happens when you see flame coming out of the exhaust pipe."
"Yeah, I've seen that," John said.
"All we need to do is construct a unit that will create this gas, then pipe that gas into the fuel injector on James's truck. The fuel injector will introduce the gas into the cylinders, and that will run your engine."
Using a hot-water tank for the combustion chamber, John and James, following Bob's instructions, built the gasification device in one afternoon. While they were building the device, Jake, Marcus, Willie, and Deon made charcoal, as it would be much easier to handle than wood. By evening they had the truck equipped with the gasification device, charcoal made and loaded aboard, and they were ready to go.
"With this thing taking up so much room back here, I doubt we can get all nineteen barrels loaded onto the truck," James said.
"You won't need this thing anymore," Bob said, patting the side of the gasification machine. "Once you get there, you can dump it, and put regular gas in your car."
"Will the truck still run on gasoline then?" James asked. "Or will this have messed up the engine?"
"It'll still run," Bob promised. "Remember you haven't run anything through your system but gas."
"Yeah, I can see that," John said.
It had grown dark by the time they finished building, and testing the device. So they gathered around a cinder-block cooking stove, oven, and grill for a supper of grilled snapper, the fish provided by Jerry Cornett.
Fort Morgan—Thursday, August 23
Jake and John left in James's truck, leaving before dawn the next morning. Jake took John in case they had mechanical problems with the truck on the way. They left James behind because he was too valuable in building their quarters. The others stayed back simply because there wasn't enough room in the truck for them.
A drive that, under normal conditions, would have taken no more than three and a half hours, took seven hours and they arrived at TAC-X at noon. The bad thing was that the power generated by the gasification machine was so inefficient that they could not go over thirty miles per hour. The good thing was, they saw no traffic during the entire one-hundred-and-seventy-five-mile trip. They also encountered no obstacles to their journey.
"You think the gas is still here?" John asked as they approached the locked gate.
"I think so," Jake said. "Otherwise I think the lock would be broken."
"Unless somebody changed the lock," John suggested.
"Yes, I hadn't thought of that. This is the key Clay gave me. If this key fits the lock, then yes, I think the gasoline will still be here."
Jake slipped the key into the lock, then was gratified when it opened easily. "Alright," he said.
Jake pushed the gate open as John drove the truck through. Then he shut the gate behind them and relocked it. That way if someone happened by, they wouldn't notice anything different.
When they reached hangar three, Jake unlocked it and opened the door, then closed it once John drove inside. The closed door made the hangar darker, but there was enough light, filtered through the dirty panes of the high-mounted windows to allow them to see what they were doing.
John and Jake moved the empty drums out of the way, removed the trash from the tarp, then peeled back the tarp to disclose nineteen barrels. The top of each barrel bore the broken lettered stenciling, in yellow.
GASOLINE, AUTOMOTIVE COMBAT MIL–G–3056
FOR USE IN ALL MOBILE AND STATIONARY SPARK-IGNITION ENGINES
"John, we have just struck gold!" Jake said, happily. "Let's get them loaded.
The first thing they did was get rid of the gas generator that had brought them up here; then they put gasoline into the truck. They did that by rolling one of the barrels up into the bed of the truck, then using a hose to siphon gas from the barrel and let it flow into the gas tank. It worked well, though it was much slower than it would have been at a normal gas pump.
When the truck was fueled, they rolled the remaining drums onto the back of the truck. There was only room to get fifteen barrels loaded so, reluctantly, they left four barrels behind, including the one they had drawn the fuel from for the truck. John started to put the tarp back over the remaining barrels, but Jake stopped him.
"We might be better off spreading that tarp over our load," he said. "If somebody sees us driving a truck with all these barrels, that would be like having a big sign that says, 'We have gas, please take it from us.'"
"Yeah," John said. "I see what you mean."
"The problem is going to be holding the tarp down," Jake said.
"Not a problem," John insisted. "We'll tie it down with safety wire."
"Where are we going to get safety wire?"
"I'll show you," John said. He walked over to the hangar wall, then stuck his hand down behind the cross brace, and pulled out a spool of safety wire.
"Damn, how did you know it would be there?" Jake asked.
John laughed. "Before they closed this stagefield I came out here to do some first-echelon maintenance. I hid the safety wire there so that if I came out to do any more maintenance I would have it. But they closed this field and I never came back."
"That was against regs, wasn't it? Wasn't the safety wire supposed to be returned?" Jake asked.
"So, take a stripe from me," John said, and both men laughed.
With the safety wire it was easy to secure the tarpaulin, so within twenty minutes they left, locking both the hangar and the gate behind them.
They made it all the way back to Fort Morgan Road before they ran into trouble. John was driving, and he stopped when they saw that the road in front of them was blocked off by a pile of old refrigerators. There were at least ten men sitting on top or standing on the ground around the blockade. All were armed.
"Oh, oh," John said. "That doesn't look good."
"No, it doesn't," Jake replied.
"What do we do?"
"Nothing for the moment. They aren't mobile—we are."
"A lot of good that does us. This is the only road. We're blocked here."
"I wonder how far we are from Fort Morgan," Jake said.
"That mileage marker says seven," John said.
"Let's see if we can raise Deon and Willie. It'll be a long hike for them, but if they come up behind the blockade, we might be able to force our way through."
Jake cranked in some power to the radio, then picked up the mic.
"Phoenix Base, this is Phoenix One, do you copy?"
"Phoenix One, this is Phoenix Base, over."
Jake recognized Karin's voice.
"Phoenix, we've run into a little trouble here. We are at the seven-mile marker on Fort Morgan Road. There is a barrier across the road in front of us, as well as several armed men."
"Are you in danger?" Karin asked, and Jake could hear the concern in her voice.
"Not immediate danger," Jake replied. "We are mobile and they aren't. The problem is, we can't get through. We may need a little backup."
"Phoenix One, what do you propose?" This was Deon's voice.
"It'll be a long hike for you, but maybe if you came up from behind, put pressure on them, we could get through here."
"Will do. Out."
Jake put the microphone down and continued to stare through the windshield at the barrier in front of them.
"What do we do now?" John asked.
"We wait."
Fort Morgan
"Marcus, Willie, grab a weapon. We have a long hike in front of us," Deon said. Deon picked up the M-240 machine gun.
"You don't have to hike," James said. "We have plenty of bicycles down here."
"Yeah, that's right, you do, don't you?" Deon said.
"You can ride a bike if you want to," Bob said. He was looking at the Huey. "Or, we could bring some heavy firepower down on them."
"What do you mean?" Deon asked.
Bob pointed to the helicopter. "Marcus, when you left Hanchey Field, did you have a full load of fuel?"
"Yes, sir," Marcus replied.
"You flew straight here, didn't you?"
"Yes, sir."
Bob smiled. "Then you've got at least a hundred and fifty miles' operating range remaining," he said.
"What are you thinking, Mr. Varney?" Deon asked.
"Didn't you tell me your dad was a door gunner in Vietnam?" Bob replied.
"Yes, sir, he was. He was with the sixty-eighth."
Bob nodded. "Sixty-eighth? Ahh, Top Tiger," he said. "Good outfit. How would you like to be a door gunner?"
"Wait a minute, are you saying you are going to try and fly that thing?" Marcus asked.
"I'm not going to try, I'm going to do it," Bob said.
"Oh, Bob, no!" Ellen said. "You haven't flown in almost forty years!"
"I hadn't ridden a bike in a longer time than that, but first time I got on one down here, I was able to ride, wasn't I?"
"That's not the same thing and you know it," Ellen said.
"Sure it is. The only difference is, I have a hell of a lot more time in one of these than I ever did on a bicycle. What about it, guys? You want to try it?" Bob asked.
A huge smile spread across Deon's face. "Hell yes," he said. "Let's go!"
Deon and Willie began putting their weapons in the helicopter while Marcus untied the blade. Bob did a walk-around pre-flight inspection, and as he climbed up onto the deck to examine the rotor head, it was as if he had done this just yesterday. He took another look at the jury-rigged drag brace; then he jumped down and started to get in.
"Wait a minute," Jerry said. "I'm going too." Jerry was carrying a bow and a quiver of arrows.
Willie laughed. "Do you plan to scalp them too?" Marcus laughed as well, but Deon held out his hand. "No, wait," he said. "I've got an idea. Wait here for a moment."
Deon jumped out of the helicopter and ran into the nearby casement. When he came back he was carrying something.
"I'll be damned," Marcus said. "That's a good idea."
"What's a good idea?" Willie asked. "What are you talking about?"
"That's C-four," Marcus said.
Deon got back into the helicopter. "Hand me a few of your arrows," he said.
Jerry complied, and Deon wrapped the C-4 plastic explosive material around the arrows, then, using a knife to cut off the arrow heads, replaced them with blasting caps. He did four arrows that way.
"Now," he said. "When we go in, we are going to go in heavy."
"Clear!" Bob shouted as he pulled the starter trigger.
CHAPTER TWENTY-SIX
"Is that a helicopter I hear?" John asked.
"It is, yes," Jake said. "I hear it, but I don't see it."
"It's close," John said. "Look, they hear it too." John pointed to the men who were standing by the barricade. They could be seen searching the sky and talking to each other, obviously looking for the helicopter.
Suddenly a Huey popped up just over the roof of the houses along the beach. "Damn! That's our Huey!" John said. "Who the hell is flying it?"
Jake laughed. "It has to be Bob," he said.
The helicopter did a quick pass by the barricade, and Jake saw an arrow streaming down.
"What the hell? He's shooting arrows at them?"
There was a loud explosion where the arrow hit, the blast big enough to throw several of the refrigerators around.
Jake laughed out loud. "C-4!" he said. "They've put C-4 on the arrows!"
The helicopter made another pass. This time Jake and John could see tracer bullets coming from the cargo door. There was also a second arrow fired, and another explosion.
Some of the men at the refrigerator barricade started shooting back at the helicopter, but the M-240 in the cargo door of the Huey was too much for them and those who weren't killed began running. The Huey chased down the runners and fired again, until the area was completely cleared of any would-be bandits.
"That old man can handle it, can't he?" John said.
"Phoenix One, this is Goodnature, do you copy?"
"Goodnature?" Jake replied.
"It was my call sign in Vietnam. I figured I may as well use it again," Bob said.
"Roger, Goodnature," Jake answered.
"If you can negotiate the barricade, I'll fly cover for you back to base," Bob said.
"We're on our way," Jake said.
"If you can't get through, get out and move one or two aside," Bob said. "Do not get off the road—if you do you'll get stuck axle deep in the sand."
Starting the truck, John drove up to the barricades, then stopped. "We're going to have to go around," he said.
"No," Jake replied. "Bob lives down here, so I'm sure he knows what he is talking about. We're going to have to push a couple of the refrigerators out of the way."
John put the truck in neutral, and he and Jake got out and started pushing refrigerators aside until they had opened a path big enough for the truck to get through.
"I think we can do it now," Jake said.
"Yeah, we've got it made in the shade," John said with a happy laugh.
Jake heard the solid thunk of the bullet hitting John. Blood and brain detritus erupted from the wound on the side of John's head and he fell toward Jake.
Jake caught him, and held him up.
"John! John!" he called.
John didn't reply, and as Jake made a closer examination of him, he realized that he was dead.
Another bullet whistled by Jake's head, coming so close that he could hear the pop of as it snapped by his ear. The only weapon Jake had with him was the pistol he was wearing on his belt. He jumped behind the refrigerator then tried to look for the shooter. The only thing he knew was that the shot had come from the row of houses nearest the beach.
Jake could hear the radio in his truck, but he couldn't get to it.
"Phoenix, this is Goodnature. Keep your head down."
A moment later the Huey passed low overhead with the machine gun firing from one cargo door and an M-16 from the other. It zoomed up over the beach houses, then made a circle back. This time there was no fire coming from the helicopter.
"Phoenix, target neutralized. You are clear to proceed," Bob's voice said.
Jake waved at the helicopter; then he picked John's body up and put it in the cab of the truck. Starting the truck, he managed to pick through the rubble and residue until he was on the other side of the barricade. Now, with an empty highway, and the truck running on gasoline so that the engine was at full efficiency, he opened up. Running at eighty miles an hour, it was all the Huey could do to stay with him, until he pulled in to the fort.
The Huey landed shortly after Jake arrived.
"Jake!" Karin said, running to him as he stepped down from the truck. She threw her arms around him in gratitude over his survival, but her joy over seeing Jake was mollified when she realized that John was dead.
Ellen greeted Bob and Gaye greeted Jerry with equal enthusiasm as they stepped out of the helicopter.
"I was so worried about you," Ellen said.
"Why were you worried? Not one bullet came close."
"I wasn't worried about you getting shot. I was worried that you might not be as good a pilot as you think you are."
"Trust me, Ellen," Jake said. "I've never seen a helicopter handled better. For an old man, he did damn well."
"Poor John," Karin said.
"That's two that we've lost," Jake said. "Clay and John. I don't plan on us losing anyone else."
Fort Morgan—Thursday, November 22, Thanksgiving Day
In the next three months after John was killed, the men and women who called themselves Phoenix turned the fort into a comfortable and sustainable community. The garden was productive and Jake, harking back to his Amish background, led the others in canning food. They had acquired the jars necessary by raiding the many empty houses up and down the beach.
The motel was complete and comfortable, each unit equipped with a fireplace for warmth in the coming winter. They also had electricity, James having installed solar panels complete with batteries, a charge controller, and inverters. That gave them heat and electricity, and with the desalination device, they had an unlimited supply of water. Bob had even brought his TV down and hooked it up to the same satellite dish he had used back at his house. The others had teased him about it, but he said it comforted him to see it, because it gave the illusion of normalcy.
Jerry and James kept the little community well stocked with fish, fowl, and game, and on this Thanksgiving Day, they were preparing a feast equal to any they had ever enjoyed before.
"Before this, Thanksgiving only really meant two things to me," Bob said. "Food and football."
"You got that right," Deon said.
"But this year I believe it is more meaningful to me than it has ever been in the past. I mean, when you think about it, our Thanksgiving here is not all that different from the first Thanksgiving. We can truly be thankful that we have survived this long and now are at the point where we are not only surviving, we are thriving."
"Yeah, well, there is the food too," Willie said. "Those geese smell so good my stomach is really growling."
"We have a lot to be thankful for," Jake said. "So if nobody minds, I'm going to say a blessing before we eat."
"I think that would be a very good idea," Bob said.
"Jake," Becky said, "maybe you could add an extra blessing."
"Sure, I'd be glad to," he said. "What is it for?"
Becky looked over at Karin. "You can tell them," she said. "I've been thinking about it, and praying about it, and I'm all right with it."
"Are you sure?" Karin asked.
Becky nodded. "I am sure," she said.
Karin smiled. "Folks, we are going to be getting an addition to our little community. Becky is pregnant."
"What?" Sarah asked. "Who is the father?"
"I don't know," Becky replied.
"You don't know?"
"Before anyone asks, I already knew about it, and I'm not the father," Marcus said. He was sitting next to Becky, and he reached out to take her hand in his.
"Do you remember the time when the men came and robbed you, and you found Becky unconscious on the road?" Karin asked. "She was raped. This pregnancy is the result of that."
"And you are going to have the baby of a rapist?" Sarah asked, incredulously.
"Sarah, you can't be just a little pregnant," Becky said. "I'm three months pregnant—it has only one possible conclusion."
"You can abort," Sarah said.
"I don't want an abortion."
"But the baby's father is a rapist."
"And I am its mother," Becky said. "You cannot hold a child guilty for the sins of its parents. I've thought long and hard about this, and I've kept it secret all these months. The way I look at it, this child, be it boy or girl, is the first link to my future. I will have this child."
"And we will be here for you," Ellen said. She looked at Sarah. "Won't we?"
Sarah broke into a smile, then went over to give her aunt a hug. "Yes," she said. "We will be here for you."
Gaye and Cille brought the two geese out. They were golden brown and aromatic. The table was also set with carrots and peas, mashed potatoes and biscuits, and scuppernong jelly.
"Jake, before you give your blessing, I have something for you," Karin said.
Jake looked at Karin with an expression of surprise and concern. "You do?"
Karin laughed. "Don't get nervous, it's not what you think," she said, and the others, who by now knew of the depth of commitment between Jake and Karin, laughed as well.
"What is it?"
She went into the little cabin that she shared with Julie, then came out with something wrapped in a sweater. Smiling broadly, she opened up the sweater to show what she was holding.
It was a can of root beer.
"Ahhh!" Jake said. "I haven't had a root beer in three months! I love you! I could kiss you!"
Karin laughed. "James found it in one of the houses," Karin said. "So if you are going to kiss anybody, you need to kiss James."
"Okay, James, I'll kiss you too," Jake said. He held up his finger. "But there will be no tongue."
The others laughed again.
"A handshake will do," James said.
Jake took the root beer from Karin, kissed her, then carried it over to the table and put it down lovingly by his plate.
"I'm sorry there's not enough to go around for all of you," he said.
"That's all right, Jake. I've gotten used to drinking water," Deon said. "It's a lot healthier for you anyway."
Bowing his head, Jake began the blessing.
"Unser himmlischer Vater—Our Heavenly Father, I ask that you bless these wonderful people today for their generous hearts, helping hands, and loving souls. And we thank you for the women who prepared the meal that will sustain us through this day. Segnen Sie dieses Essen zu unserer Verwendung, und wir zu Ihrem Dienst—bless this food to our use, and ourselves to thy service. Amen."
He had included the German phrases in the blessing because it was an Amish prayer, the same prayer Mr. Yoder had prayed when Jake had gone back home to help build the barn. He wondered for a moment about his parents, and added a silent prayer for them.
"Now, let's eat," he said, popping open the can of root beer.
After a very satisfying meal, Bob went back into his and Ellen's quarters. As he did every day, he turned on the TV, then hit search.
Suddenly the search stopped and a picture filled the screen.
"Hey!" he shouted. He ran to the door and stuck his head out. "Hey! We've got TV!"
"What?"
"We've got TV!" Bob shouted again.
The others came running down to Bob's apartment ; then all of them crowded in, finding seats anywhere they could.
At the moment the screen was blue, with the words STAND BY.
The standby card went away to be replaced by a picture. The man in the picture, short blond hair, cherubic face, slightly pudgy, was familiar to them all.
"It's George Gregoire!" Jake said.
Hello, America.
This is a simulcast over shortwave radio, satellite radio, and satellite TV. That's right, I'm back on TV, though the size of my TV audience is probably less than a cable access show of the joys of scrap-booking. Nevertheless, I am extremely proud of our little group of technicians who have managed to put our show up on the bird so that those with electricity and a TV can see us.
First, I will bring you up to date on the latest news we have been able to gather.
It appears that the so-called Islamic Republic of Enlightenment holds only Washington, D. C., and Detroit. The fact that they hold our capital city has given them a great deal of cachet in the rest of the world, but we, here in America, know that they are unable to expand beyond that. Already there have been isolated and uncoordinated raids against the Enlightened ones, none of which, at this point, have been much more than a nuisance. In the meantime the Enlighteneds' atrocities against our people, especially the women and children, continue.
Ohmshidi is alive and well, somewhere, we know not where. From time to time he will make a radio broadcast to rally his base.
Really? Rally his base?
Tell me, friends, does he even have a base any longer? I think not. I think that once we reestablish control, put decent Americans back to work, and reconstitute our government, we will then have time to find Ohmshidi and bring him to justice for all the crimes he has committed.
That means we have much to do, America, and since last I spoke to you much has been done. I have been contacted on 5110LSB by several groups of brave Americans who have banned together to fight this evil that has come into our midst. I will not disclose at this time how many groups I have been in contact with, where they are, or what their strength is. I will only say that for the first time since Ohmshidi was elected president, I am feeling optimistic.
It is my sincere belief that there are many, many more groups than have yet made contact with me, so I feel that, even though Americans made the colossal mistake of sending an arrogant, incompetent fool to the White House, those same Americans are now prepared to rectify that mistake.
During those days when we existed as a democratic republic, we often heard one party or another—whichever party was out of office—adopt the political slogan, "Take back America."
Well, my friends, this is no longer empty political rhetoric. This is a real, and vital, battle cry. And I urge you, with all that is in my being, to hold yourselves in readiness until we can coalesce as a mighty revolutionary army to do that very thing.
Now you may well ask the question, from whom are we to take back our country? Is it from Ohmshidi and his State Protective Service? Or is it from the roving bands of brigands and thugs, people from among us who prey upon the weak and helpless, Americans by birth, but not by any moral code that we all hold dear?
The answer, my friends, is we must be prepared to do battle with all of them. I urge those of you who are watching this program, and those of you who are within sound of my voice, to establish contact with the Brotherhood of Liberty, join forces in this new revolution.
"What do you think, Jake?" Deon asked. "Should we contact them?"
"How would we do that?" Jake asked.
"He said he could be contacted on fifty-one ten LSB," Willie said. "I think we could get through. We've got a pretty good antenna system here now. I have it attached to the top of the lighthouse tower.
"Alright, give it a try."
"This is Phoenix, calling on fifty-one ten LSB. Does anyone read me? Over."
Willie released the switch and listened, but got no response, so he tried again.
"This is Phoenix, calling on fifty-one ten LSB. Does anyone read? Over."
"Phoenix, this is Firebase Freedom. Over."
"Hey, we got someone!" Willie said.
Everyone gathered around the radio then to listen.
"Phoenix, this is Firebase Freedom. Go ahead."
"Tell him we are trying to make contact with the Brotherhood of Liberty," Jake said.
"Firebase Freedom, we are trying to make contact with the Brotherhood of Liberty. Over."
"For what purpose, Phoenix? Over."
Jake reached for the microphone, and Willie handed to him. "Firebase Freedom, we want to discuss mutual goals," he said. "Over."
"This is a different voice, over. Identify yourself."
"This is Phoenix." Jake paused for a moment, then looked at the others. "Six," he added. "Do you copy? I am Phoenix Six."
"All right," Deon said. "The six is back." He, Marcus, John, and Willie gave each other high fives.
"What is six?" Cille asked.
"It means the commanding officer," Bob said with a broad grin. "I'm glad to see that some things haven't changed since I retired."
"Roger, Phoenix Six," Firebase Freedom responded. "If you don't mind, I would like to authenticate."
"How are we going to do that without an SOI?"
"I authenticate silent seven."
Jake released the mic button and looked at the others. "Does anyone know what the hell he is talking about?"
"Respond, faithful five," Bob said.
"What?"
"Respond with 'faithful five,'" Bob repeated.
Jake keyed the mic again. "Faithful five. I say again, faithful five."
"Welcome home, brother," Firebase Freedom said.
"Tell him welcome home," Bob said.
"Welcome home," Jake said.
"Maintain contact, Phoenix. As things develop, we will keep you apprised. Over."
"Will do. Out," Jake said.
Jake looked at Bob. "Okay, so tell me what the hell I was talking about."
"He was talking about the eleven general orders for sentry duty," Bob said. "Silent seven is the seventh general order—I will speak to no one except in the line of duty. Faithful five is the fifth general order—I will quit my post only when properly relieved. Of the eleven general orders, they are the only two that have nicknames."
"Eleven general orders? What are you talking about?" Deon asked. "There are only three."
Bob chuckled. "Maybe there are only three now," he said. "But there were eleven when I was in the Army, and evidently when Firebase Freedom was, as well, since he is a Vietnam vet."
"How do you know he is a Vietnam vet?"
"Welcome home," Bob said. "We never got a welcome home, so that is sort of a code we use with each other."
"How come you never got a welcome home?" Willie asked.
"It was the way the war was fought, and the way we deployed," Bob said. "We went to Vietnam as individuals, we came back as individuals. Oh, there would be one hundred fifty to two hundred others on the plane with you, but you didn't know any of them. When you arrived at Oakland, you would be greeted with jeers and curses and thrown objects by the war protestors, so you just sort of kept your head down and kept walking.
"You would be pulled out of the jungle on Saturday, and on Monday you would go down to the local burger joint in your hometown. You would look around at the others and know that, physically, you were here, but in your mind, and in your soul, you were still back in Vietnam.
"Only about thirty percent of those who served in Vietnam are still alive, and today we are all brothers, and we greet each other with the welcome home that we never received."
Deon got up from where he was sitting, and walked over to Bob. He stuck out his hand. "Welcome home, brother," he said. "And I don't mean brother because I was in Vietnam—I mean brother, because I consider us all here to be brothers, and sisters."
Following Deon's comment, Marcus turned to Becky. "Damn, I hadn't really been thinking of you as a sister."
Others in the group, well aware of the growing relationship between Marcus and Becky, laughed.
"I think Deon meant it figuratively," Bob said.
"I wonder what the future holds for us," Karen mused.
"What future?" Julie asked. "We don't have a future."
Jake looked over at Becky, who was sitting in a chair Willie had pulled up for her.
"Sure we do," he said. "The child Becky is carrying is part of our future. And when you think about it, Adam and Eve started with a lot less than we have."
"Whoa! I hope you aren't planning on any of us adding anything to the population," Jerry said, taking in himself and the other two older couples with a wave of his hand.
Jake laughed. "Well, we aren't exactly alone," he said. "We've already established radio contact with some others just like us. I've no doubt that we will join them soon."
"Then what?" Willie asked.
"Then?" Jake replied. He raised his arm over his head and made a fist. "We take back America."
The others, as one, repeated his words.
"Take back America!"
PINNACLE BOOKS are published by
Kensington Publishing Corp.
119 West 40th Street
New York, NY 10018
Copyright © 2011 William W. Johnstone
All rights reserved. No part of this book may be reproduced in any form or by any means without the prior written consent of the publisher, excepting brief quotes used in reviews.
PUBLISHER'S NOTE
Following the death of William W. Johnstone, the Johnstone family is working with a carefully selected writer to organize and complete Mr. Johnstone's outlines and many unfinished manuscripts to create additional novels in all of his series like The Last Gunfighter, Mountain Man, and Eagles, among others. This novel was inspired by Mr. Johnstone's superb storytelling.
If you purchased this book without a cover, you should be aware that this book is stolen property. It was reported as "unsold and destroyed" to the publisher, and neither the author nor the publisher has received any payment for this "stripped book."
This book is a work of fiction. Names, characters, businesses, organizations, places, events, and incidents either are the product of the author's imagination or are used fictitiously. Any resemblance to actual persons, living or dead, events, or locales is entirely coincidental.
PINNACLE BOOKS and the Pinnacle logo are Reg. U.S. Pat. & TM Off.
ISBN: 978-0-7860-2864-1
| {
"redpajama_set_name": "RedPajamaBook"
} | 7,515 |
Victoria, Australia : Savdek Management Pty. Ltd., 2014.
Casebook of Barnaby Adair novels.
"With her husband, amateur-sleuth the Honorable Barnaby Adair, decidedly eccentric fashionable matron Penelope Adair is attending the premier event opening the haut ton's Season when a body is discovered in the gardens. A lady has been struck down with a finial from the terrace balustrade. Her family is present, as are the cream of the haut ton--the shocked hosts turn to Barnaby and Penelope for help. Barnaby calls in Inspector Basil Stokes and they begin their investigation. Penelope assists by learning all she can about the victim's family, and uncovers a feud between them and the Latimers over the fabulous shoes known as Lady Latimer's shoes, currently exclusive to the Latimers. The deeper Penelope delves, the more convinced she becomes that the murder is somehow connected to the shoes. She conscripts Griselda, Stokes's wife, and Violet Montague, now Penelope's secretary, and the trio set out to learn all they can about the people involved, and most importantly the shoes, a direction vindicated when unexpected witnesses report seeing a lady fleeing the scene--wearing Lady Latimer's shoes. But nothing is as it seems, and the more Penelope and her friends learn about the shoes, conundrums abound, compounded by a Romeo-and-Juliet romance and escalating social pressure ... until at last the pieces fall into place, and finally understanding what has occurred, the six intrepid investigators race to prevent an even worse tragedy."--Page 4 of cover.
Criminal investigation -- England -- London -- 19th century -- Fiction.
#1 New York Times bestselling author Stephanie Laurens brings you a tale of mysterious death, feuding families, star-crossed lovers--and shoes to die for.With her husband, amateur-sleuth the Honorable Barnaby Adair, decidedly eccentric fashionable matron Penelope Adair is attending the premier event opening the haut ton's Season when a body is discovered in the gardens. A lady has been struck down with a finial from the terrace balustrade. Her family is present, as are the cream of the haut ton--the shocked hosts turn to Barnaby and Penelope for help.Barnaby calls in Inspector Basil Stokes and they begin their investigation. Penelope assists by learning all she can about the victim's family, and uncovers a feud between them and the Latimers over the fabulous shoes known as Lady Latimer's shoes, currently exclusive to the Latimers.The deeper Penelope delves, the more convinced she becomes that the murder is somehow connected to the shoes. She conscripts Griselda, Stokes's wife, and Violet Montague, now Penelope's secretary, and the trio set out to learn all they can about the people involved and most importantly the shoes, a direction vindicated when unexpected witnesses report seeing a lady fleeing the scene--wearing Lady Latimer's shoes.But nothing is as it seems, and the more Penelope and her friends learn about the shoes, conundrums abound, compounded by a Romeo-and-Juliet romance and escalating social pressure...until at last the pieces fall into place, and finally understanding what has occurred, the six intrepid investigators race to prevent an even worse tragedy.A pre-Victorian mystery with strong elements of romance. A novel of 76,000 words.An entry in The Casebook of Barnaby Adair series, the events in this short novel occur between the events described in the Casebook novels, THE MASTERFUL MR. MONTAGUE and the upcoming LOVING ROSE: THE REDEMPTION OF MALCOLM SINCLAIR.
Stephanie Laurens was born in Sri Lanka, which was at the time the British colony of Ceylon. Her family moved to Melbourne, Australia, where she eventually received a Ph.D in biochemistry. She and her husband moved to London for four years where they worked as research scientists. They returned to Melbourne where she worked in the field of cancer research and eventually ran her own research laboratory. She began writing romance novels as a hobby, but due to her success she became a full-time novelist. Her first book, Tangled Reins, was published in 1992. Her other works include the Cynster Family series, the Cynster Sisters series, and the Bastion Club series.
Stephanie's book's, By Winter's Light and The Lady By His Side made the New York Times bestseller list. | {
"redpajama_set_name": "RedPajamaC4"
} | 376 |
The Performing Arts Workers' Equity (PAWE) was a small trade union in South Africa. It had a membership of only 365, but was affiliated to the Congress of South African Trade Unions. It merged with the Musicians Union of South Africa (MUSA) to form the Creative Workers Union of South Africa (CWUSA).
References
External links
PAWE at the COSATU.
Information on PAWE at UNESCO
Defunct trade unions in South Africa
Entertainment industry unions
Trade unions based in Johannesburg | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 4,949 |
Tout est sous contrôle (titre original : ) est un roman humoristique à suspense, une parodie de roman policier et d'espionnage, écrit par Hugh Laurie et paru en 1996.
Résumé
Un homme de 36 ans, Thomas Lang, est engagé pour tuer un autre homme. Ce qu'il refuse de faire et, tentant de prévenir cette personne du danger, il se rend chez l'homme mais au lieu de tomber sur lui, il fait une mauvaise rencontre. Après s'être débarrassé d'un gêneur, Lang se retrouve nez à nez avec Sarah Woolf, la fille de la « censée victime ». Les ennuis commencent…
Personnages principaux
Thomas Lang : ancien officier de l'armée qui s'en est retiré et qui travaille aujourd'hui pour de menus travaux correspondant à ses aptitudes. Vivait une existence simple avant l'arrivée des Woolf dans son existence. Il a pour habitude de faire de la moto et de boire à outrance.
Sarah Woolf : fille d'Alexander Woolf, cible de départ de Lang. Elle rencontre Thomas au début du roman. Leur relation est une suite de rencontres ponctuelles. Le héros se persuade sans cesse qu'il n'est pas amoureux d'elle.
David Solomon : policier au service du Ministère de la Défense britannique et ami de longue date de Thomas Lang, ayant participé avec lui notamment à une campagne en Irlande du Nord.
Pat-Rohnim Murt (Naihm Murdah en version originale) : principal antagoniste.
Autour du livre
Hugh Laurie a utilisé un pseudonyme lorsqu'il a soumis son manuscrit à son éditeur, afin que sa notoriété en tant qu'acteur n'influe pas sur la décision d'être publié ; son éditeur l'a cependant ensuite convaincu d'utiliser son nom réel dans le but d'augmenter la visibilité de son livre.
Le roman n'a été traduit dans d'autres langues qu'après que son auteur Hugh Laurie a acquis une plus grande notoriété en tant qu'acteur grâce à son rôle dans la série télévisée Dr House ; il sort ainsi en 2006 en espagnol, en 2008 en polonais et en 2009 en français, soit plus de dix ans après sa sortie originale. Il est donc difficile d'en saisir la subtilité et l'originalité dans son contexte (l'avant 11-Septembre).
Une adaptation cinématographique était en cours en 2001, elle devait être produite par Russell Smith et John Malkovich pour United Artists / MGM, sur un scénario d'Hugh Laurie lui-même. Le projet a cependant dû être abandonné à la suite des attentats du 11 septembre 2001, l'intrigue traitant d'un complot organisé par les États-Unis, mettant en scène un attentat terroriste pour justifier une politique sécuritaire et militaire ultérieure.
Une suite intitulée The Paper Soldier a été envisagée. Le site Amazon.com prévoyait sa sortie au et l'avait même proposée en précommande. Néanmoins, alors que la date approchait mais que le livre n'était toujours pas sorti, l'agent d'Hugh Laurie a précisé qu'il s'agissait d'une erreur et que la rédaction n'était même pas encore commencée. Amazon a ensuite prévu la date de sortie au . Cependant, cette suite n'a jamais été éditée.
Éditions
En anglais
1996 (UK) : ; relié : 352 p. ; broché : 340 p.
1996 (UK) :
1997 (UK) :
1997 (US) :
1998 (US) :
2000 (UK) :
2004 (UK) :
En français
2009 :
En espagnol
2006 :
En polonais
2008 :
Références
Roman britannique paru en 1996
Roman policier britannique
Roman d'espionnage britannique
Roman humoristique
Littérature parodique
1996 en littérature policière | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 8,773 |
Blanche Feillet-Hennebutte, née le à Paris, et morte le à Biarritz, est une peintre, dessinatrice et lithographe française.
Biographie
Blanche Feillet est la fille de l'artiste Pierre Jacques Feillet et de Hélène Pernotin.
Comme sa sœur Hélène, Blanche apprend le dessin et la peinture auprès de son père et de son grand-père Pernotin. La famille séjourne plusieurs années à Madrid, avant de s'installer définitivement à Bayonne en 1834.
En 1844, Blanche Feillet épouse l'éditeur Charles Henry Firmin Hennebutte. Blanche et sa sœur Hélène fourniront les illustrations de plusieurs guides touristiques publiés par Charles Hennebutte.
En 1857, Blanche est la directrice de l'école de dessin et de peinture de Bayonne.
Œuvre
Blanche Feillet expose au Salon de 1841 une Vue d'un moulin près de Saint-Jean-Pied-de-Port, puis, en 1848, un Naufrage dans les rochers de Biarritz près de Bayonne et une Vue de l'entrée de la Barre de Bayonne.
L'essentiel de son travail est tourné vers le dessin et la lithographie : elle réalise de nombreuses vues du Pays basque, qui, transcrites en lithographie, viennent illustrer les guides touristiques édités par son époux.
De nombreux dessins et estampes de Blanche Feillet et de sa sœur Hélène sont conservés à la médiathèque de Bayonne et au Musée Basque.
Notes et références
Annexes
Bibliographie
Gilbert Desport, Répertoire des peintres et sculpteurs du Pays basque, Biarritz, Atlantica, 2005.
Henri Jeanpierre, « Un peintre romantique bayonnais, Hélène Feillet », in Bulletin de la Société des Sciences, Lettres et Arts de Bayonne, 1964, , .
Liens externes
Exposition virtuelle sur l'œuvre des deux sœurs
Peintre français du XIXe siècle
Peintre française
Dessinatrice française
Lithographe français du XIXe siècle
Graveuse française du XIXe siècle
Artiste lié aux Pyrénées-Atlantiques
Naissance en novembre 1815
Naissance à Paris
Décès en septembre 1886
Décès à Biarritz
Décès dans les Basses-Pyrénées
Décès à 70 ans | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 6,388 |
In this interview, we speak to Beth, who is 22 and studying for a degree in English Literature. In her spare time, she plays racket sports, and likes to help out by volunteering.
Beth started experiencing depression in her early teens, and since then has struggled with concentrating during her school and university studies. In these videos, she describes how finding helpful and understanding healthcare professionals has made a big difference to her recovery. | {
"redpajama_set_name": "RedPajamaC4"
} | 1,496 |
5 things black people can do to succeed (despite Obama)
"…When the Obama administration is over, black people will have lost ground in every single leading economic indicator category." — Tavis Smiley, liberal black radio talk-show host.
The black unemployment rate is 12.5 percent. According to Census Bureau figures, the poverty rate for blacks increased from 12 percent in 2008 to 16.1 percent last month; and median income declined by 3.6 percent for white households to $58,000, but fell 10.9 percent to $33,500 for black households. Where's Obama's pen and phone when we need it?
A recent viral video showed blacks in Chicago (Barack's adopted hometown) expressing their frustration with Obama. One man said if Obama can't help create jobs, he should "cut off his presidency and quit!" For blacks who are fed up with false hope, here are 5 things they can do to succeed (despite Obama):
1. Seek God first (not necessarily in church). The church has always played a central role in the lives of black Americans. Dr. Martin Luther King Jr. was a product of that tradition. Unfortunately, misguided preachers who've been called by their mama—not by God—today head most black churches. You don't have to go to church to find God. To put God first means you pray quietly every morning and read Scripture so you can develop a mind that will, "Seek first the kingdom of God, and his righteousness; and all these things shall be added unto you."—Matthew 6:33
2. Get married and don't have kids out of wedlock. The high illegitimacy rate among blacks is a relatively new trend. As late as 1950, black women were more likely to be married than white women, and only 9 percent of black families with children were headed by a single parent. Today the black out-of-wedlock birth rate is an alarming 72 percent!
Growing up on a plantation in Alabama during the 1950s, even though the laws were against us (Jim Crow, etc.), blacks, for the most part, loved God, worked hard, and got married. Nowadays, the two-parent black family is almost extinct. Yet you don't hear civil rights groups addressing this crisis. Why not?
Black men must stop making babies out of wedlock. Black women need to stop being promiscuous. As Heritage Foundation scholar Robert Rector has noted, "Illegitimacy is a major factor in America's crime problem. Lack of married parents, rather than race or poverty, is the principle factor in the crime rate."
3. Get over your blackness. After the assassination of Dr. King in 1968, the civil rights movement was co-opted by the Socialist wing of the Democratic Party. So-called 'leaders' like Jesse Jackson, the NAACP, the Congressional Black Caucus, and Al Sharpton, with the aid of the media, convinced blacks that color trumps character. While their 'leaders' have profited, the average black is wallowing in despair.
These black 'leaders' promote racism, welfare, and public education for the masses, while their kids attend the finest private schools and get great jobs. I often wonder why blacks don't question how these people became so successful if there is so much "systemic racism"? The hatred that most blacks harbor toward whites is impeding their spiritual and economic growth.
4. Get an education. The late, great Booker T. Washington, founder of Tuskegee Institute, was spot-on with his practical approach that incorporated classroom learning with working with the hands (trades). My nonprofit Organization BOND (Brotherhood Organization of A New Destiny) has started a private school—BOND Academy (grades 1-12, for boys & girls) modeled after Washington's approach.
In my work with families and with schools, I know that public education—especially in the inner cities—is awful! Half of the black male high school students drop out. Far-Left teachers attempt to indoctrinate the remaining students.
Despite the obvious breakdown, the Obama Administration is working to kill the implementation of Louisiana's tuition voucher program as they did in Washington D.C. These programs provide low-income families with the financial resources to enroll their children in the school of their choice. More black parents should support school choice or home school!
5. Work hard and start a business. The Obama administration is blocking opportunities and jobs. One example is the administration's delay in approving the Keystone XL oil pipeline. The $7 billion pipeline would carry oil from tar sands in western Canada to refineries along the Texas Gulf Coast, creating thousands of jobs. Blacks should be flocking to states like Texas and North Dakota to seek out these jobs.
In the 1930's and 40's blacks had thriving businesses. Blacks need to become entrepreneurs again.
Twenty-four years ago, I started a nonprofit organization (BOND)- after God allowed me to see that my problem was not due to the white man holding me back, but rather because of unresolved resentment and anger I had toward my parents. I wanted to share this news, which was the catalyst for starting BOND. This month BOND is celebrating its 24th anniversary and we've never received a dime of government money!
The above things don't require an executive order from Obama—but they do require commitment and hard work. If black Americans did these 5 things, they can change their destiny and do well despite this president.
Rev. Jesse Lee Peterson is the founder and president of BOND (Brotherhood Organization Of A New Destiny), and BOND Action.
For presidents, more than entertainment
ING Sees Q4 Bank Earnings Triple
Barstool Sports founder Dave Portnoy raises $25 million and counting for struggling small businesses
Notice: Trying to get property 'post_excerpt' of non-object in /home/tlf238f/public_html/wp-content/themes/jawn/parts/post-single.php on line 11
Black Americans can change their destiny and do well despite this president.
Rev. Jesse Lee Peterson
"???When the Obama administration is over, black people will have lost ground in every single leading economic indicator category." — Tavis Smiley, liberal black radio talk-show host.
The black unemployment rate is 12.5 percent. According to Census Bureau figures, the poverty rate for blacks increased from 12 percent in 2008 to 16.1 percent last month; and median income declined by 3.6 percent for white households to $58,000, but fell 10.9 percent to $33,500 for black households. Where???s Obama???s pen and phone when we need it?
A recent viral video showed blacks in Chicago (Barack???s adopted hometown) expressing their frustration with Obama. One man said if Obama can???t help create jobs, he should ???cut off his presidency and quit!??? For blacks who are fed up with false hope, here are 5 things they can do to succeed (despite Obama):
1. Seek God first (not necessarily in church). The church has always played a central role in the lives of black Americans. Dr. Martin Luther King Jr. was a product of that tradition. Unfortunately, misguided preachers who???ve been called by their mama???not by God???today head most black churches. You don???t have to go to church to find God. To put God first means you pray quietly every morning and read Scripture so you can develop a mind that will, ???Seek first the kingdom of God, and his righteousness; and all these things shall be added unto you.??????Matthew 6:33
2. Get married and don???t have kids out of wedlock. The high illegitimacy rate among blacks is a relatively new trend. As late as 1950, black women were more likely to be married than white women, and only 9 percent of black families with children were headed by a single parent. Today the black out-of-wedlock birth rate is an alarming 72 percent!
Growing up on a plantation in Alabama during the 1950s, even though the laws were against us (Jim Crow, etc.), blacks, for the most part, loved God, worked hard, and got married. Nowadays, the two-parent black family is almost extinct. Yet you don???t hear civil rights groups addressing this crisis. Why not?
Black men must stop making babies out of wedlock. Black women need to stop being promiscuous. As Heritage Foundation scholar Robert Rector has noted, ???Illegitimacy is a major factor in America's crime problem. Lack of married parents, rather than race or poverty, is the principle factor in the crime rate.???
3. Get over your blackness. After the assassination of Dr. King in 1968, the civil rights movement was co-opted by the Socialist wing of the Democratic Party. So-called ???leaders??? like Jesse Jackson, the NAACP, the Congressional Black Caucus, and Al Sharpton, with the aid of the media, convinced blacks that color trumps character. While their ???leaders??? have profited, the average black is wallowing in despair.
These black ???leaders??? promote racism, welfare, and public education for the masses, while their kids attend the finest private schools and get great jobs. I often wonder why blacks don???t question how these people became so successful if there is so much ???systemic racism???? The hatred that most blacks harbor toward whites is impeding their spiritual and economic growth.
4. Get an education. The late, great Booker T. Washington, founder of Tuskegee Institute, was spot-on with his practical approach that incorporated classroom learning with working with the hands (trades). My nonprofit Organization BOND (Brotherhood Organization of A New Destiny) has started a private school???BOND Academy (grades 1-12, for boys & girls) modeled after Washington???s approach.
In my work with families and with schools, I know that public education???especially in the inner cities???is awful! Half of the black male high school students drop out. Far-Left teachers attempt to indoctrinate the remaining students.
Despite the obvious breakdown, the Obama Administration is working to kill the implementation of Louisiana???s tuition voucher program as they did in Washington D.C. These programs provide low-income families with the financial resources to enroll their children in the school of their choice. More black parents should support school choice or home school!
5. Work hard and start a business. The Obama administration is blocking opportunities and jobs. One example is the administration???s delay in approving the Keystone XL oil pipeline. The $7 billion pipeline would carry oil from tar sands in western Canada to refineries along the Texas Gulf Coast, creating thousands of jobs. Blacks should be flocking to states like Texas and North Dakota to seek out these jobs.
In the 1930???s and 40???s blacks had thriving businesses. Blacks need to become entrepreneurs again.
Twenty-four years ago, I started a nonprofit organization (BOND)- after God allowed me to see that my problem was not due to the white man holding me back, but rather because of unresolved resentment and anger I had toward my parents. I wanted to share this news, which was the catalyst for starting BOND. This month BOND is celebrating its 24th anniversary and we???ve never received a dime of government money!
The above things don???t require an executive order from Obama???but they do require commitment and hard work. If black Americans did these 5 things, they can change their destiny and do well despite this president.
Written By Rev. Jesse Lee Peterson | {
"redpajama_set_name": "RedPajamaCommonCrawl"
} | 5,660 |
\section{Uncertainty-Aware Neural Networks}
\label{sec:approaches}
\input{figures/approaches_table}
While regular DNNs, also called \emph{Point-Prediction DNNs (PPNN)}, predict a single scalar value for every output variable, from the perspective of uncertainty quantification it would be preferable if the DNN calculated a distribution over possible output values, which would allow the extraction of the network's confidence in the prediction.
In this section we discuss the various DNN architectures which allow the calculation of such distributions:
first, we provide an overview on \emph{Bayesian Neural Networks (BNN)}, which are the standard approach in the machine learning literature about uncertainty quantification.\footnote{
For a more precise discussion of BNN, we recommend the comprehensive and usage-oriented article by Jospin \textit{et al.}\xspace ~\cite{Jospin2020}.
}
Second, we discuss other approaches, which typically have some theoretical disadvantages when compared to BNNs, but showed good results in practice.\cite{Ovadia2019}
\subsection{Pure Bayesian Neural Networks}
BNNs are neural networks in which weights are probabilistic, instead of scalar as in PPNN, and are represented as probability density functions.
To train a BNN, first, a prior distribution $p(\theta)$ over weights $\theta$ has to be defined.
Then, given some data $D$, the posterior distribution $p(\theta|D)$, i.e., the trained BNN is inferred using Bayes rule:
\begin{equation}
p(\theta|D) = \frac{p(D|\theta) p(\theta)}{p(D)}
= \frac{p(D|\theta) p(\theta)}{\int p(D|\theta) p(\theta) d\theta}
\end{equation}
Since weights are probabilistic, any output of the BNN is also probabilistic and thus allows a statistically well-founded uncertainty quantification.
Besides that, these BNNs in their pure form have other advantages over PPNN, among which:
they are robust to overfitting~\cite{Neal1992} and allow to distinguish between epistemic and aleatoric uncertainty~\cite{Jospin2020}.
BNNs in the pure form presented above quickly become intractable due to the large number of integrations required.
There has been a long search for mechanisms which make BNN easier to compute, with papers dating back to the early 1990s~\cite{MacKay1992, Neal1992}.
To do so, some approaches make additional assumptions, e.g., a normal distribution of the weights, which may cause the BNN to lose some of the advantages listed above.
Still, despite many advances in the field, even recent approaches are considered not to scale sufficiently well~\cite{Ilg2018} and are not yet widely used in practice~\cite{Jospin2020}.
Other, more scalable approaches have been shown to outperform Bayesian approaches~\cite{Ovadia2019} on practical benchmarks.
\subsection{MC-Dropout based Bayesian Neural Networks}
In their very influential paper\footnote{More than 2300 citations in the four years since its publication, according to \href{https://scholar.google.com/}{Google Scholar}.}, Gal \textit{et al.}\xspace~\cite{Gal2016} proposed to approximate BNNs though regular PPNNs.
Many PPNNs use \emph{dropout layers} for regularization. During training, in a dropout layer each neuron activation can be set to 0 with some probability $p$. This helps to avoid overfitting and can lead to better model performance in general~\cite{Srivastava2014}.
In their paper, Gal \textit{et al.}\xspace have shown that a PPNN trained with dropout enabled can be interpreted as a BNN,
due to the variation induced by the randomized dropout layers.
While traditionally the dropout functionality is disabled at prediction time, a dropout-based BNN keeps the random dropping of neuron activations
enabled even during predictions and calculates an output distribution by Monte-Carlo sampling of the results in multiple randomized predictions.
This approach is thus often referred to as \emph{MC-Dropout}.
Despite its high popularity, MC-Dropout is not without criticisms:
Osband \cite{Osband2016} claimed the inability of MC-Dropout to capture every type of uncertainty and
several papers have shown that it can be outperformed by other approaches~\cite{Lakshminarayanan2017, Ovadia2019}.
Also, despite the fact that \textit{"the forward passes can be done concurrently, resulting in constant running time"} \cite{Gal2016},
concurrent processing may not be possible in resource-constrained environments, making MC-Dropout clearly slower than a single prediction of the corresponding PPNN.
\subsection{Deep Ensemble Neural Networks}
Lakshminarayanan \textit{et al.}\xspace proposed a fundamentally different approach to quantify uncertainty called \emph{Deep Ensembles}, or \emph{Ensemble} for short, i.e., a collection of multiple \emph{atomic models} for the same problem,
differing only in their initialization and trained independently.
For every input, a prediction would be made on every atomic model.
The predictive distribution could then be inferred from these samples.
While deep ensembles are not inherently BNNs, it is possible to interpret them as BNN after applying some minor changes to the parameter regularization~\cite{Pearce2020}.
Nonetheless, even in their plain form, they have been shown to outperform MC-Dropout on the task of uncertainty quantification~\cite{Lakshminarayanan2017, Ovadia2019}.
Deep Ensembles are, compared to a single PPNN, slow to train if executed sequentially and memory intensive if used concurrently.
This may prevent the use of ensembles in some constrained environments.
A variety of improvements and modifications have been proposed for Deep Ensembles (Ilg \textit{et al.}\xspace \cite{Ilg2018} provide a good overview of them).
\subsection{Point Predictor Classifiers}
DNNs used for classification typically have a \emph{softmax} output layer, including one output neuron per class.
The output for each class is between $0$ and $1$, with the sum of all outputs being exactly $1$.
These outputs are often interpreted as probability distributions over the classes
and are used for network supervision by means of the following quantifiers:
\begin{termdefinition}[Max-Softmax (SM)] The highest softmax score is used as confidence quantification
(also referred to as \emph{Vanilla}~\cite{Ovadia2019} or \emph{Softmax Prediction Probability}~\cite{Berend2020} quantification).
\end{termdefinition}
\begin{termdefinition}[Prediction Confidence Score (PCS)~\cite{Zhang2020}]
The difference between the two highest softmax outputs is used as confidence quantification.
\end{termdefinition}
\begin{termdefinition}[Softmax-Entropy (SME)]
The entropy over the outputs of the softmax layer is used as confidence quantification.
\end{termdefinition}
These quantifiers are often criticized as a poor approach to uncertainty quantification:
As opposed to BNNs and BNN approximations, PPNN based quantifiers are not theoretically well founded,
and can be shown to severely overestimate a network's confidence~\cite{Gal2016}.
As a further disadvantage, such a PPNN based approach can not be directly applied to regression problems.
\subsection{Inferring Prediction and Uncertainty from Samples}
In MC-Dropout and Deep Ensembles, samples need to be aggregated into a point prediction and uncertainty quantification,
and the literature provides a variety of quantifiers able to do so.
In this Section, we first discuss the quantifiers applicable to regression problems, and then the ones for classification problems.
\paragraph{Regression Problems}
In their proposition of MC-Dropout, Gal \textit{et al.}\xspace~\cite{Gal2016} propose to use the average of the observed samples as prediction,
and their predictive variance as uncertainty:
\begin{termdefinition}[Predictive Variance]
The \emph{predictive variance} is the sample variance over the observed samples plus the inverse model precision.
\end{termdefinition}
The inverse model precision is constant for a given model.
Thus, for the purpose of network supervision, where strictly monotonic transformations of uncertainty quantification scores
do not change the supervisor performance, using predictive variance is equivalent to using the sample variance or the sample standard deviation.
An alternative approach was proposed by Lakshminarayanan \textit{et al.}\xspace\cite{Lakshminarayanan2017}.
For their deep ensembles, they propose that the atomic models are adapted s.t. they have a second output variable for every regression output which predicts the variance~\cite{Nix1994}.
The uncertainty of the ensemble can then be quantified by averaging these variances.
\paragraph{Classification Problems}
The following quantifiers are proposed to derive an overall prediction and uncertainty:
\begin{termdefinition}[Mean-Softmax (MS)]
The overall prediction is the class with the highest sum of softmax scores over all samples
and the corresponding confidence is the average softmax score of this class over all samples.
\end{termdefinition}
MS has been proposed and is often used with Deep Ensembles. It is thus also called \emph{ensembling}\cite{Jospin2020}.
Three other quantifiers have been proposed to be used with MC-Dropout\cite{Gal2016a, Gal2016, Michelmore2018}:
\emph{Variation Ratio (VR)}, which is defined as the percentage of samples for which the overall chosen class is \textbf{not} the class with the highest softmax output; \emph{Predictive Entropy (PE)}, which measures the average amount of information in the predicted distributions;
and the \emph{Mutual Information (MI)} between a prediction and the models posterior (see
Gal, 2016~\cite{Gal2016}, Section 3.3.1 for a precise description and for examples of these three uncertainty quantifiers).
\section{Supervised Neural Network Assessment}
\label{sec:assessment}
Network supervision can be viewed as a binary classification task:
\emph{malicious samples}, i.e., inputs which lead to a misclassification (for classification problems) or to severe imprecision (in regression problems) are positive samples that have to be rejected.
Other samples, also called \emph{benign samples}, are negative samples in the binary classification task.
An uncertainty based supervisor accepts an input $i$ as a benign sample if its uncertainty $u(i)$ is lower than some threshold $t$.
The choice of $t$ is a crucial setting, as a high $t$ will fail to reject many malicious samples (false negatives) and a low $t$ will cause too many false alerts (false positives).
Differently from standard binary classification tasks, the choice of $t$ for network supervision cannot rely on the optimization of an aggregate metric that accounts for both false positives and false negatives, such as the F1-metric, because negative samples are either completely unknown at training time, or, in case some negative samples are known, they cannot be assumed to be representative of all unknown execution conditions that will give raise to uncertainty at runtime.
Hence, the choice of $t$ is solely based on the false positives observed in the validation set.
In practice, given the uncertainties measured on the validation set, $t$ shall ensure that at runtime under similar conditions only an acceptable \emph{false positive rate} $\epsilon$ is expected to occur~\cite{Stocco2020}.
Existing metrics allow the individual assessment of the supervisor's performance separately from the assessment of the model performance\cite{Henriksson2019}. However, such measurements do not take into account the interaction between the two: since the output of the model is not used by the DLS when the supervisor activates the fail safe procedure ($u(i) \geq t$), it does not make any sense to evaluate the performance of the model in such scenarios. For this reason, we propose a new approach for the joint assessment of model and supervisor, which we call \textit{supervised metrics}. In the next two sections we first summarize the state of the art metrics for the separate, individual assessment of model and supervisor, followed by a description of our proposal of a new joint assessment approach.
\subsection{Existing Metrics for the Individual Assessment of Model and Supervisor}
There are well established metrics for the individual assessment of performance of a model $m$. These are based on some \emph{objective function} $obj(I,m)$, such as \emph{accuracy (ACC)} (for classifiers) or \emph{mean-squared error (MSE)} (for regression models), computed on a test dataset $I$.\footnote{We assume that $obj$ has to be maximized. The extension
to an $obj$ that should be minimized is straightforward. Also note that $obj$ does not have to be the same function used to optimize the DNN during training.}
Classically, the supervisor's performance would be assessed individually using performance metrics designed for binary classifiers. For a given $t$, the available metrics include \emph{true positive rate (TPR)}, \emph{false positive rate (FPR)}, \emph{true negative rate (TNR)} and \emph{false negative rate (FNR)}, \emph{$F_1$} score and \emph{accuracy (ACC)}.
To use these metrics with regression problems, an \emph{acceptable imprecision} would have to be defined, allowing to divide inputs into benign (negative cases) and malicious (positive cases).
Alternatively, the effect of the predictions on the overall DLS system could be monitored to only treat inputs leading to system failures as malicious ones~\cite{Stocco2020}.
There are also existing, classical metrics to assess a binary classifier independently of the threshold $t$.
For instance, the \emph{average precision score (AVGPR)} computes
the average of the precision values obtained when varying $t$, weighted by the recall measured at $t$.
Another popular threshold independent metric is the \emph{area under the receiver operating characteristic curve (AUROC)}\cite{Stocco2020, Berend2020, Michelmore2018}.
When individually assessing the performance of a supervisor, AVGPR should be preferred over AUROC as, amongst other advantages \cite{Davis2006}, it is better suited for the unbalanced datasets~\cite{Saito2015} typically observed during malicious input detection.
Threshold independent analysis of the supervisor in a regression problem is straightforward and can be done using point-biserial correlation between the quantified uncertainty and the observed prediction error, given some objective function (e.g. MSE).
Independent analysis of model's performance, considered in isolation without supervision, and of the supervisor's performance, again in isolation, results in measurements that do not capture the overall goal of the interaction between supervisor and model under supervision: ensuring high model performance on the samples considered safe by the supervisor, while keeping the amount of samples considered unsafe as small as possible.
To capture such goal, we propose novel metrics for joint model and uncertainty quantification assessment in a supervised DLS.
\subsection{Supervised Metrics for the Joint Assessment of Model and Supervisor}
When considering model and supervisor jointly, we can still use the objective function used to assess the model in isolation, but we evaluate it in a supervised context:
given a test set $I$ and an objective function $obj(I,m)$ for model $m$ with uncertainty quantifier $u$ and supervisor threshold $t$, the \textit{supervised objective function} $\overline{obj}(I,m)$ is defined as:
$$
\overline{obj}(m,I) = obj(\{i \mid i \in I \text{ and } u(i) < t \})
$$
i.e., the objective is applied only to the subset of inputs which is accepted by the supervisor.
By decreasing $t$, assuming that the cardinality of the resulting subset of inputs remains big enough to calculate a statistically significant $\overline{obj}(I)$, we may generally get higher values of the supervised objective function $\overline{obj}(I)$.
However, such high values of $\overline{obj}(I)$ are likely associated with a high false alarm rate of the supervisor.
Thus, any $\overline{obj}(I)$ should always be regarded in conjunction with the acceptance rate $\Delta_u(I)$ of the supervisor:
$$
\Delta_u(I) = \frac{\mid \{i \mid i \in I \text{ and } u(i) < t \} \mid }{\mid I \mid}
$$
Similar to the popular $F1$ score, the following combination of these two metrics allows to capture the effectiveness of the collaboration between supervised model and supervisor:
\begin{termdefinition}[S-Score]
The $S_1$-Score measures the harmonic mean of a supervised objective function, normalized between zero and one, and the supervisors acceptance rate as
$$
S_1(m,u,I) = \frac{2}{\frac{obj^+}{\overline{obj}(I,m) - obj^-} + \Delta_u(I)^{-1}}
$$
\end{termdefinition}
where $obj^-$ and $obj^+$ are the lower and upper bounds used for normalization of the objective function.
For classifiers, if accuracy is the objective function, $obj^- := 0$ and $obj^+ := 1$. For regression problems, or more generally for unbounded objective functions, $obj^-$ and $obj^+$ have to be estimated empirically (e.g., based on the empirical distribution of the objective function values), independently from $m$ and $u$.
The $S_1$ scores weights $\Delta_u(I)$ and $\overline{obj}(m,I)$ equally.
Equivalent to the popular $F_1$ score, other $S_\beta$ scores can be used, where $\beta > 0$ is the weighting parameter~\cite{Rijsbergen1979}:
$$
S_\beta(m,u,I) = (1 + \beta^2)\cdot \frac{\frac{\overline{obj}(I,m) - obj^-}{obj^+} \cdot \Delta_u(I)}{(\beta^2 \cdot \frac{\overline{obj}(I,m) - obj^-}{obj^+}) + \Delta_u(I)}
$$
\section{Background}
\label{sec:background}
In this section, we discuss the different root causes of DNN faults which can be understood as \emph{types of uncertainty}
and define the task of DNN fault prediction as a problem of uncertainty quantification.
\subsection{Sources of Uncertainty}
We distinguish between two types of uncertainty; uncertainty caused by a sub-optimal DNN model and uncertainty caused by randomness in the prediction target.
A detailed discussion of these types is provided by Kendall \textit{et al.}\xspace \cite{Kendall2017a}.
\begin{termdefinition}[Epistemic Uncertainty]
Epistemic uncertainty is caused by the sub-optimal training or configuration of the \emph{model}.
\end{termdefinition}
Epistemic uncertainty is sometimes also referred to as \emph{model uncertainty}.
There are many possible reasons for epistemic uncertainty, such as
insufficient training data, which does not represent the entire possible input space,
sub-optimal training hyper-parameters and inadequate DNN architecture.
In theory, epistemic uncertainty could be avoided,
provided good enough training data and optimal model configuration.
However, finding such optimal training configurations and data is impossible in most real world applications,
as real input spaces, as well as the space of the possible hyper-parameters and architectural choices, are typically too large.
The second type of uncertainty, which not even an optimal training set and model configuration can avoid, is called \emph{aleatoric uncertainty}:
\begin{termdefinition}[Aleatoric Uncertainty]
Aleatoric uncertainty is the uncertainty present in the true (unknown) distribution we are making predictions about.
\end{termdefinition}
Thus, aleatoric uncertainty can be seen as randomness, ambiguity or entropy in the prediction target.
When predicting a random event, even an optimal model will make wrong predictions.
As aleatoric uncertainty is independent of the model, but instead depends on the predicted data, it is also referred to as \emph{data uncertainty} or \emph{irreducible uncertainty}.
Aleatoric Uncertainty can be further distinguished between \emph{homoscedastic uncertainty},
where the uncertainty applies to all data, and \emph{heteorscedastic uncertainty},
where the uncertainty is more prevalent amongst some subsets of the data.
\input{figures/uncertainty_types}
Figure \ref{fig:uncertainty_types} provides a visual example of the difference between aleatoric and epistemic uncertainty.
The input to a classifier DNN is the image of a handwritten digit to be recognized.
Figures \ref{fig:u_type_nominal_1} and \ref{fig:u_type_nominal_4} illustrate regular inputs with low uncertainty.
Figure \ref{fig:u_type_aleatoric} is an example of a figure with high heteroscedastic aleatoric uncertainty:
Clearly, the image shows either a 1 or a 4, but it is impossible to say with certainty which one the writer intended.
Figure \ref{fig:u_type_epistemic_corrupted} illustrates a common reason for epistemic uncertainty:
A small perturbation of the image background, if not present in the training data, may lead the model to
be incapable of predicting the correct label.
Similarly, \ref{fig:u_type_epistemic_roman} shows an unexpected input leading to epistemic uncertainty.
While the true label is unambiguously 4, a model which was not trained on roman number representations will not be capable
to make a correct prediction.
\subsection{Uncertainty Quantification}
Ideally, instead of predicting a single value as output, a DNN should calculate a probability density function for regression problems or a likelihood for every outcome in a classification problem.
As such, every outcome would have its uncertainty quantified (e.g., by the variance of the output probability distribution).
We will discuss models capable of calculating such outputs in Section \ref{sec:approaches}.
However, for the scope of this paper, we consider a less general formulation of uncertainty quantification,
which is sufficient for network supervision~\cite{Riccio2020}:
\begin{termdefinition}[Uncertainty Quantification]
Uncertainty Quantification (UQ) is the task of calculating a scalar metric
strictly monotonically increasing in the likelihood (for classification tasks) or severity (for regression tasks)
of a deviation between the DNN prediction and the ground truth, given a particular input and the DNN used to do the prediction.
\end{termdefinition}
We are thus limiting our interest to the correctness of the chosen prediction,
as opposed to the distribution of all possible predictions in the output space.
The supervisor will reject inputs for which the uncertainty is above a certain threshold.
Consistently, we define \emph{confidence} as the opposite of uncertainty,
s.t. a confidence metric is supposed to strictly monotonically decrease in the likelihood or severity of a prediction error.
\section{Case Studies}
\input{figures/big_table}
\label{sec:case_studies}
We assess the uncertainty quantification capabilities of point predictors, deep ensemble and MC dropout using different quantifiers.
We intentionally focus our study on uncertainty quantifiers which can be applied to traditional and widely used DNN architectures, and we exclude those implementing the pure Bayesian form of uncertainty estimation, since they require the adoption of dedicated architectures where the network weights encode a probability distribution, not just a scalar value.
This restriction, in combination with \tool, allows developers to measure uncertainty at minimal effort,
given a traditional DNN.
The \textit{goal} of our empirical evaluation is to assess the usefulness of the uncertainty quantifiers supported by \tool when used as supervisors, as well as to collect lessons learned that practitioners can follow when applying \tool to their DNNs.
\subsection{Research Questions}
We consider the following research questions:
\noindent
\textbf{RQ\textsubscript{1} (effectiveness):}
\textit{How effective are supervisors at increasing the supervised model's performance?}
This is the key research question of our empirical study, since the main hypothesis behind supervisors is that they can prevent usage of a model when its performance is predicted to be low. Hence, we expect an increase of the supervised model's performance $\overline{obj}$ as compared to the unsupervised one $obj$.
\noindent
\textbf{RQ\textsubscript{2} (comparison):}
\textit{Is there a supervisor and quantifier type which yield optimal performance across subjects and across alternative choices of the uncertainty threshold?}
We consider three types of uncertainty estimators, Point Predictors, MC-Dropout and Ensemble, and several uncertainty quantifiers, respectively [SM, PCS, SME], [VR, PE, MI, MS], [VR, PE, MI, MS] (see Section~\ref{sec:approaches}). We want to investigate whether any combination of estimator type and quantifier dominates all the others in terms of $S_1$-score. To investigate how performance changes with the uncertainty threshold $t$, we consider different acceptable rates $\epsilon$ of false positives on the nominal data and we compute the threshold $t$ that ensures such FPR on the validation set, so that we can compare alternative estimators/quantifiers at equal FPR on the validation set of the nominal data.
\noindent
\textbf{RQ\textsubscript{3} (sample size):}
\textit{How many samples are required in stochastic and ensemble models to get reliable uncertainty quantification?}
Since the main cost associated with the usage of MC-Dropout and Ensemble is the collection of multiple samples for each individual prediction, we want to understand what is the minimum sample size that ensures good performance of each different supervisor. In particular, we study the convergence of supervised accuracy to its asymptotic value as the sample size is increased.
\noindent
\textbf{RQ\textsubscript{4} (sensitivity):}
\textit{How sensitive are supervisors to changes in their main hyperparameters?}
With this research question we want to understand whether the choice of hyperparameters is critical to get optimal performance, or on the contrary if they can be chosen in the neighbourhood of the optimal choice with minor impact on the resulting performance of the supervisor.
For Point Predictors, we consider the number of training epochs as the main hyperparameter;
for MC-Dropout, the number of training epochs and the number of samples; for Ensemble, the number of training epochs and the number of atomic models. We measure the standard deviation of the supervised objective function (e.g., supervised accuracy) in the neighbourhood of each hyperparameter choice, so as to identify the regions where such standard deviation is low.
\subsection{Subjects}
We use the following classification problems as case study subjects, aiming to increase diversity and practical relevance.
\begin{description}
\item [Mnist\cite{LeCun1998}] Classification of hand-written digits, formatted as small grayscale images.
This is the most popular dataset in machine learning testing~\cite{Riccio2020}, and a relatively easy problem, where even simple models achieve high accuracy.
We took the DNN architecture from a Keras tutorial~\cite{KerasMnistModel}.
\item [Cifar10\cite{Krizhevsky2009}] Classification of colored images into ten different classes. It is also very popular in DLS testing~\cite{Riccio2020} and it represents a more challenging task than Mnist.
We use the model architecture proposed in the Brownlees Cifar10 tutorial~\cite{BrownleeCifarModel}.
\item [Traffic~\cite{Serna2018}] Classification of images of European traffic signs~\cite{Segvic2010, Bonaci2011, Mathias2013, Timofte2014, Belaroussi2010, Stallkamp2012, Grigorescu2003, Larsson2011}.
The different sources the data was collected from, combined with the fact that the dataset is unbalanced and many images are of bad quality, reflect a quite realistic, high-uncertainty setup. Since traffic sign recognition is a core component of self-driving cars, this is also a very interesting case study from the software and system engineering point of view.
The model architecture we use was proposed alongside the release of this dataset~\cite{Serna2018}.
\item [Imagenet~\cite{Deng2009} (Pretrained)] Image classification problem with as many as 1,000 classes.
We use eight pre-trained \emph{Efficientnet} models~\cite{Tan2019}.
As for this subject we rely on pre-trained models (which include dropout layers), we can test them only
as MC-Dropout and Point Predictor models, but not as Ensembles.
\end{description}
\subsection{Experimental Setup}
Except for the pre-trained ones, models were trained for 200 epochs.
After every epoch, we assessed the models' performance on both a nominal and an out-of-distribution (OOD) dataset, for every quantifier.
To do so, we used Mnist-c~\cite{Mu2019} as OOD test set for Mnist and the color-image transformations
proposed by Hendrycks~\cite{Hendrycks2018} to generate OOD samples for the other subjects.
We used three different thresholds, calculated on the nominal validation set to ensure the lowest possible FPR above $\epsilon$, with $\epsilon\in\{.01,.5,.1\}$ respectively.
To measure the sensitivity to the number of samples, quantifiers of deep ensembles were assessed with every number of atomic models between 2 and 50,
Similarly, MC-Dropout was assessed on every number of samples between 2 and 100.
Counting atomic models individually, this procedure required the training of 153 DNNs,
and the calculation of 2'121'600 DNN predictions\footnote{Predictions were cached, such that for the evaluation of different sample sizes and different numbers
of atomic models, previous predictions could be re-used.}.
Due to the high workload, the training and prediction processes were distributed on three different workstations using Windows or Ubuntu and four different GPUs (one workstation had two GPUs).
\tool was used for training, prediction and uncertainty quantification.
\footnote{Replication package available at:\\ \href{https://github.com/testingautomated-usi/repli-icst2021-uncertainty}{github.com/testingautomated-usi/repli-icst2021-uncertainty}}
\subsection{Results}
We organize the analysis of the results obtained in our experiments by research question.
\subsubsection{RQ1 (Effectiveness)}
An overview of our results is provided in \autoref{tab:big_table}.
Due to space constraints, the results for $\epsilon=0.05$ are omitted in the table and the values for the 8 different Imagenet models are averaged.
The full set of results can be found in the online replication package.
Our results suggest that all supervisors lead to supervised accuracies $\overline{ACC}$ which are at least as high, but typically much higher than the accuracy $ACC$ of the unsupervised model.
Thus, supervisors are effective.
The effectiveness is particularly strong on the OOD datasets:
For example, on Mnist, where the unsupervised point predictor has an accuracy of 74\%,
the supervised accuracy at $\epsilon=0.1$,
is above 95\% with most supervisors.
In other words, a DLS using an unsupervised model will experience six times more faulty predictions than the unsupervised one.
Also notable are the results on the nominal Mnist dataset, where even simple supervisors based on point predictors turn an unsupervised accuracy of 96\% (at $\epsilon = 0.1$) into that of a nearly perfect predictor
while still accepting around 88\% of the inputs.
\begin{tcolorbox}
\textbf{Summary (RQ1)}: \textit{All tested supervisors are, in general, effective at increasing the supervised accuracy compared to the unsupervised accuracy.}
\end{tcolorbox}
\subsubsection{RQ2 (Comparison)}
\input{figures/ranks_table}
If we look at the $S_1$-Score in \autoref{tab:big_table}, it is apparent that there is no uncertainty quantifier which outperforms all the other ones on every subject/dataset and for every threshold ($\epsilon$).
To allow for an overall comparison of the quantifiers, we computed the average ranks of the quantifiers when ordered by $S_1$-Score. Results are shown in \autoref{tab:rank_order}, where $N$ indicates the number of data points on which average ranks have been computed. While there is no absolute dominant supervisor, i.e., the ideal choice of supervisor remains problem dependent,
Ensembles can be considered the overall best performing supervisors (in line with existing literature\cite{Ovadia2019}), because when they do not have the lowest rank, they still have quite low rank values. Actually,
even without supervision Ensembles often achieve higher accuracy than Point Predictors and MC-Dropout based models.
Interestingly, in most cases the sophisticated quantifiers PE and MI, grounded on information theory,
do not perform better and often perform worse than the simpler quantifiers VR and AS.
Despite the theoretical disadvantages of Point Predictors, in our experiments
this simple approach outperformed the theoretically well founded MC-Dropout. So, in practical cases as those considered in our experiments, Point Predictors may represent a good trade-off between performance and computational cost.
\begin{tcolorbox}
\textbf{Summary (RQ2)}: \textit{There is no dominant supervisor, i.e., no supervisor which performs best for every test subject, data source and threshold. Ensembles are ranked generally well across subjects and thresholds, while Point Predictors offer a valuable trade off between performance and execution cost.}
\end{tcolorbox}
\subsubsection{RQ3 (Sample size)}
\input{figures/sample_size}
We find that for both MC-Dropout and Ensembles the relatively low number of 20 samples
is already sufficient to get a similar supervised accuracy as with a much higher number of samples.
This is shown in \autoref{fig:ss_influence} for the MS quantifier and $\epsilon = 0.1$.
20 samples, while still higher than what's recommended in related literature~\cite{Ovadia2019},
is probably small enough to be used in practice in many applications.
The other quantifiers behave similarly, with one notable exception, which is the VR quantifier: VR can only take a finite set of discrete values, which can be easily shown to be equal to the number of samples.
Hence, a low sample size makes it impossible to set thresholds well fit to the target $\epsilon$, because thresholds are correspondingly also discrete and limited to the number of samples. So, to achieve the target FPR $\epsilon$ precisely, we might need substantially more than 20 samples.
\begin{tcolorbox}
\textbf{Summary (RQ3)}: \textit{A few (\texttildelow20) samples are enough to get good supervision results with most quantifiers (VR represents an exception, due to the discretization of the values it can take).}
\end{tcolorbox}
\subsubsection{RQ4 (Sensitivity)}
\input{figures/heatmaps}
The number of training epochs and the number of samples/atomic models can be visualized as a 200 (number of epochs) by 100 (number of samples) or 50 (number of atomic models) grid.
To assess the sensitivity of supervisors, we calculate the standard deviation (\textit{std}) of $\overline{ACC}$ within a 5x5 filter applied to this grid. In this way, we account for the variability of the supervised accuracy in a 5x5 neighbourhood of each hyperparameter configuration.
The higher the std, the higher the sensitivity to small hyperparameter changes in the neighbourhood.
We also calculate the average $\overline{ACC}$ in such 5x5 neighbourhood.
The result is a pair of heatmaps, an example of which is shown in \autoref{fig:heatmaps}.
This figure suggests that hyperparameter sensitivity negatively correlates with $\overline{ACC}$:
Hyperparameters leading to low accuracy (bright colors in \autoref{fig:heatmap_mean}) typically show high sensitivity to hyperparameter changes (dark colors in \autoref{fig:heatmap_std}).
We do indeed observe this negative correlation for all case studies.
The values of point-biserial correlation, shown in the \emph{S-C} (Sensitivity-Correlation) columns of \autoref{tab:big_table}, are strongly negative in most cases, with $p$-value $< 0.05$ in 358 out of 390 cases.
In low accuracy cases, small changes to number of samples and number of epochs have a high effect on accuracy.
\begin{tcolorbox}
\textbf{Summary (RQ4)}: \textit{For a given supervisor and model, for what concerns the number of training epochs and samples used for quantification, the higher the supervised accuracy, the lower the model's sensitivity to small hyperparameter changes.}
\end{tcolorbox}
\subsection{Lessons Learned}
Based on our answers to the RQs, we distilled the following primary lessons learned, possibly useful for a practical usage of
uncertainty-quantifiers based uncertainty monitoring for DLS robustness:
\begin{description}
\item [Anything is better than nothing:]
While the selection of the ideal supervisor is problem dependent,
all the supervisors we tested showed some capability to increase the accuracy of the DNN when supervised.
Thus including any uncertainty monitoring supervisor in a DLS increases its fail-safety.
\item [Ensembles are powerful:]
Not only did Ensembles show the best average rank on the $S_1$ score (ignoring the pre-trained Imagenet studies), in many cases even the unsupervised accuracy increased.
Furthermore, the relatively low number of 20 atomic models was sufficient to achieve good results in our study.
Thus, provided sufficient system resources, we suggest software architects to use Ensembles instead of Point Predictors. On the other hand, the latter may represent a good compromise solution when computational resources are severely constrained.
\item [Number of samples affects choice of quantifier:]
For Ensembles and MC-Dropout, VR was the best quantifier on average, but it requires a large number of samples to allow for precise threshold selection.
Thus, if computational resources allow the calculation of a high number of samples, our experiments suggest to use VR as quantifier. Otherwise, MS showed good performance, despite its simplicity.
\item [In-production supervisor assessment is needed:]
Since there is no uncertainty quantifier which performs best in all cases,
we want to emphasize the importance of in-production assessment of the supervisors performance on the actual system to be supervised, by comparing different supervisors for optimal selection. The metrics that we propose in Section~\ref{sec:assessment} are specifically designed for such assessment.
\end{description}
\subsection{Threats to Validity}
\textbf{External Validity}: While we considered only four subjects, we diversified them as much as possible. In particular, besides the benchmark subjects often used in DNN testing (Mnist, Cifar10 and Imagenet), we included an additional subject, Traffic, which consists of unbalanced data, partially of low quality. It implements a functionality (traffic sign recognition) commonly integrated in autonomous vehicles.
\textbf{Internal Validity}:
The selection of hyperparameters for DNN training
might be critical and the selected values may not be representative of other contexts.
To address this threat, we refrained from selecting any hyperparameter for the case study models ourselves, wherever possible, and instead relied on architectures available from the literature.
For what concerns the internal hyperparameters of the supervisors, we evaluated the sensitivity of the results to their choice in a dedicated research question (RQ4).
Another threat to the internal validity of our study is that the OOD inputs used in our experiments might not be representative of the uncertainties that may be observed in practice.
Indeed, this is unavoidable and intrinsic to the problem of DNN supervision, as unexpected conditions occurring in practice cannot be by definition simulated ahead of time.
\section{Conclusion}
\label{sec:conclusion}
Despite their fundamental role to support fail-safe execution, uncertainty estimators are rarely used by developers when integrating DNNs into complex DLS.
This might be due to the high complexity of some of the approaches and the lack of a tool to facilitate the use of such techniques.
This paper closes such gap through the following contributions:
(1) We compared the most widely used uncertainty-aware DNN types, providing an easy start into the relevant literature;
(2) We released our tool \tool, which allows to build and evaluate uncertainty-aware DNNs obtained transparently from unchanged, traditional DNNs;
(3) We reported our empirical evaluation, summarized into practical guidelines on how to set up an uncertainty monitoring DNN supervisor for a production system.
The non-optimality of any approach under all conditions may allow a complementary use of multiple supervisors.
To this extent, we plan to investigate the combination of different supervisors as future work,
hopefully leading to an overall more stable and universally applicable supervision.
\section{Introduction}
\label{sec:introduction}
Deep neural networks (DNNs) are a powerful tool to identify patterns
in large amounts of data and to make predictions on new, previously unseen data.
Thanks to the increased hardware capabilities, DNNs can be run even on small, battery powered hardware
and can be trained in performance-optimized GPUs. Correspondingly, the use of DNNs has gained a lot of popularity in the last decade.
Moreover, the introduction of high level APIs such as \tfkeras (see \href{https://www.tensorflow.org/}{tensorflow.org})
allows even software engineers without previous experience in artificial intelligence to define, train and use custom DNNs.
DNNs are now used in many \emph{Deep Learning based Systems (DLS)}, like self driving cars, to interpret observed sensor measurements and control the car's actuators,
in medical systems, to support physicians to make their diagnosis, and in web services, for image processing and analysis.
Relying solely on the predictions made by a deep learning component might be dangerous, as there is always some \emph{uncertainty} about the correctness of the prediction. In fact, the \textit{contract} between the overall system and its DNN based components is necessarily a probabilistic one, and while the probability of an error can be low, it is never zero.
The uncertainty intrinsic with DNNs is either caused by entropy in the input
or by inadequate training.
While the first type of uncertainty is inherent to a problem and cannot be avoided by definition,
the latter cannot also be avoided for practical reasons:
in most applications, the input space consists of a huge number of input contexts (e.g., the different weather or light conditions in which a car is driven), and it is impossible to collect data which perfectly represents all of them.
Faced with a problem for which a prediction is subject to high uncertainty,
a human intelligence may consider to refuse to make a prediction and instead say `I do not know'.
DNNs on the other hand, will calculate a prediction for any given input, independently of the uncertainty of the prediction.
If such predictions are trusted by a DLS, the DLS may fail due to a wrong prediction, as is best illustrated by the following two examples: a self-driving car has recently crashed into an overturned truck.
A likely explanation for such a crash is that overturned trucks are not sufficiently represented in the cars training data.\cite{Templeton2020Forbes}
Second, an online photo storage service classified an image of a black person as a a picture of a gorilla,
leading to negative press, which deemed the service as racist.
Again, such error is likely caused by insufficient training data of the machine learning component which classified the image.\cite{Vincent2018GoogleFotos}
The fact that problems like these happen even in software from leading companies in the machine learning domain shows that preventing such errors is quite challenging.
Even more so, in the second example the solution put in place was a drastic workaround:
the label \emph{Gorilla} was removed from the set of possible predictions for any input.
We propose that DLS include a \emph{supervisor}, which monitors the DNN uncertainty for any given input at runtime,
such that the system can ignore predictions for high-uncertainty inputs and can run a safe fallback process instead, such as stopping the self-driving car at the side of the street or,
in the second example, delegating the classification of the image to a human.
The machine learning community has investigated \emph{uncertainty-aware} types of DNNs, which support the deployment of such a supervisor.
This paper aims at closing the gap between uncertainty-aware DNNs and the deployment of an effective supervisor in a DLS. Specifically, it makes the following contributions:
\begin{description}[noitemsep]
\item [Metrics Comparison] Description and comparison of the most investigated uncertainty metrics for DNNs, with a discussion of their advantages and disadvantages.
\item [\tool] Python library which allows zero-knowledge, transparent implementation of uncertainty-aware DNNs.
\item [Evaluation Framework] We present existing and propose novel metrics to evaluate DLS which include a supervisor.
\item [Lessons learned] We discuss key findings from our empirical evaluation of various uncertainty metrics applied to four different case studies.
\end{description}
\section{Related Work}
\label{sec:related}
\textbf{Empirical Studies of Uncertainty-Aware Deep Neural Networks:}
Oliveira \textit{et al.}\xspace \cite{Oliveira2016} compared, amongst others, MC-Dropout based uncertainty and a variational approximation of BNN.
In their experiments, the performance of the two were comparable, but MC-Dropout was much faster.
They did not consider Ensemble models.
Similarly, but more extensively, Ovadia \textit{et al.}\xspace \cite{Ovadia2019} compared various uncertainty aware DNNs against each other, including MC-Dropout,
Deep Ensembles and a variational approximation of BNN. Consistently with our results, Ensembles performed the best in their experiments.
As opposed to our work, both of these studies consider fewer subjects, do not investigate the impact of different quantifiers,
and do not have the constraint of transparently introducing uncertainty estimators into DNNs without altering their inner architecture, as possible instead with variational BNN approximations.
Zhang \textit{et al.}\xspace \cite{Zhang2020} compare MC-Dropout against PCS to detect misclassifications caused by adversarial examples,
i.e., examples deliberately modified to trick the DNN into making prediction errors.
Their results show, similarly to ours, that there is no strict dominance between these two approaches of network supervision. Ours is the first large scale study where uncertainty estimators are injected transparently into existing DNNs (thanks to \tool). We are also the first to introduce practical metrics for in-production assessment of supervisors and to distill a list of lessons learned that can be used as guidelines for practical usage of supervisors.
\textbf{Other types of DNN supervisors:}
While our focus is on uncertainty measures based on the variability of the output for a given input,
there have been various proposals of supervisors that are not directly based on the output distribution of the supervised network.
Berend \textit{et al.}\xspace\cite{Berend2020} compared various such techniques, typically based on neuron activations, which re\-co\-gnize activation patterns that were not sufficiently represented in the training data. While such an approach may be powerful against epistemic uncertainty, it can not help against aleatoric uncertainty.
Stocco \textit{et al.}\xspace\cite{Stocco2020} and Henriksson \textit{et al.}\xspace\cite{Henriksson2019} proposed the use of autoencoders, i.e., anomaly detectors trained on the DLS training set as DNN supervisors. This approach does not consider the inner state of the DNN or its predictions (i.e., it is black-box). On the contrary, the supervisors supported by \tool take advantage of the predictions of the DNN being supervised.
\section{\tool}
\label{sec:wizard}
\input{figures/wizard_snippet}
With 148'742 stars and 82'784 forks on github.com, Tensorflow is presumably the most popular deep learning framework.\footnote{Its main alternative, \emph{PyTorch} has 42'621 stars and 11'102 forks.}~\cite{TfPytorchCompare}
In its recent versions, a large part of its API, \tfkeras, is based on the popular Keras API, a simple yet powerful high-level API to develop, train and deploy DNNs.
The simplicity of the \tfkeras API allows researchers and practitioners outside of the machine learning community to get started with deep learning easily.
Unfortunately, such simple API does not expose equally simple methods to quantify uncertainty.
Thus, we release \tool, an extension of \tfkeras which allows developers to easily create Stochastic and Ensemble DNNs
and apply all uncertainty quantifiers described in Section \ref{sec:approaches}.
The core features of \tool are:
\begin{description}[noitemsep]
\item [Sequential API] Sequential models are the most straightforward way to use \tfkeras
and are thus very popular.
However, the sequential API does not allow dropout at prediction time. Hence, it also does not allow the implementation of MC-Dropout.
\tool closes this gap by supporting the creation of stochastic models using the sequential, as well as the functional, API
in plain \tfkeras syntax. An example of this is given in \autoref{lst:stochastic_sequential_snippet}.
\item [Dynamic Randomness] In \tfkeras, the dropout behavior at prediction time is unchangeable for a given model:
Either it is disabled (as required in point predictors) or enabled (as required in stochastic models).
Thus, despite relying on the same architecture and weights, a \tfkeras model cannot be used both as stochastic model and as point predictor.
Converting them is nontrivial and includes the creation of a new model, which is memory and performance intensive.
\tool's sequential models dynamically enable and disable stochastic behavior based on whether the passed quantifier
expects a point prediction or sampled predictions.
This is shown in \autoref{lst:stochastic_sequential_snippet} as well.
\item [Conversion from Keras] Often, the user of a DNN is not the same person or group that trained the model.
To allow users to use such a model for MC-Dropout nonetheless,
\tool supports the import of any \tfkeras model (which has at least one dropout layer) as a stochastic model.
\item [Parallelized Ensembles] \tfkeras API does not expose simple functionality for parallel training of multiple models.
Especially with smaller models, which do not require the full use of the existing hardware to be loaded and executed,
sequential training of an ensembles atomic models has a large negative impact on training and prediction time.
Additionally, it can also lead to pollution of the global tensorflow runtime
due to memory leaks and eager processing.\footnote{See e.g. tensorflow issues \href{https://github.com/tensorflow/tensorflow/issues/33030}{33030}
and \href{https://github.com/tensorflow/tensorflow/issues/37505}{37505}.}
\tool treats ensembles lazily: every atomic model is stored on the file system,
and lazily loaded into its own tensorflow runtime during execution.
This allows faster, parallelized execution without runtime pollution.
\item [Dependency-Light \textit{pip install}] \tool is platform independent,
importable
through \textit{pip install uncertainty-wizard}
and has only one dependency: Tensorflow version 2.3.0 or later.
\end{description}
Due to space constraints, the description of \tool at this place is brief.
We provide a more extensive discussion in a technical tool paper \cite{Weiss2020Wizard}.
\tool and a comprehensive user guide can be found online:
\textbf{\href{https://github.com/testingautomated-usi/uncertainty-wizard}{github.com/testingautomated-usi/uncertainty-wizard}}
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 9,332 |
Pass extra arguments to your npm scripts through the command line (instead of creating an extra script for that).
### Syntax
Invoke your script using `npm run` for a regular run.
Append a `--` at the very end of the same command to pass extra parameters.
<sub>
Note: the first '--' does not replace or eliminate the need for the dashes in optional arguments.
In other words, you still need to preppend your optional arguments with the corresponding '-' or '--' (see example below).
</sub>
### Example
In your package.json
```json
{
"scripts": {
"build": "babel-node --presets es2015 script.js"
}
}
```
In the command line (terminal)
```bash
$ npm run build
Failed because X, Y or Z...
# Want to know what went wrong?
# Send the --debug optional argument to your 'build' script, this way babel-node will give you more details...
$ npm run build -- --debug
```
### References
NPM Run Script \([https://docs.npmjs.com/cli/run-script](https://docs.npmjs.com/cli/run-script)\)
### Tags
[#tip](../../tips.md)
[#package-managers](../package-managers.md)
[#npm](npm.md) | {
"redpajama_set_name": "RedPajamaGithub"
} | 673 |
function BxPaymentProviderChargebeeV3(oOptions) {
this.init(oOptions);
}
BxPaymentProviderChargebeeV3.prototype = new BxPaymentMain();
BxPaymentProviderChargebeeV3.prototype.init = function(oOptions) {
this._sProvider = oOptions.sProvider;
this._sActionsUrl = oOptions.sActionUrl;
this._sObjName = oOptions.sObjName == undefined ? 'oPaymentProviderChargebeeV3' : oOptions.sObjName;
this._sAnimationEffect = oOptions.sAnimationEffect == undefined ? 'fade' : oOptions.sAnimationEffect;
this._iAnimationSpeed = oOptions.iAnimationSpeed == undefined ? 'slow' : oOptions.iAnimationSpeed;
this._sObjNameCart = oOptions.sObjNameCart;
this._iClientId = oOptions.iClientId;
this._iSellerId = oOptions.iSellerId;
this._iModuleId = oOptions.iModuleId;
this._iItemId = oOptions.iItemId;
this._sItemName = oOptions.sItemName;
this._iItemCount = oOptions.iItemCount;
this._sRedirect = oOptions.sRedirect;
this._sCustom = oOptions.sCustom;
this._rHandler = Chargebee.init({
site: oOptions.sSite
});
this._rHandler = Chargebee.getInstance();
};
BxPaymentProviderChargebeeV3.prototype.subscribe = function(oLink) {
var $this = this;
var oDate = new Date();
oLink = jQuery(oLink);
if(oLink.hasClass('bx-btn-disabled'))
return;
oLink.addClass('bx-btn-disabled');
this._rHandler.openCheckout({
hostedPage: function() {
return $.post({
url: $this._sActionsUrl + 'call/' + $this._sProvider + '/get_hosted_page/' + $this._iClientId + '/' + $this._iSellerId + '/' + $this._sItemName + '/',
dataType: 'json'
});
},
success: function(sHostedPageId) {
$this.loadingInPopup(oLink, true);
var oParams = {
seller_id: $this._iSellerId,
seller_provider: $this._sProvider,
module_id: $this._iModuleId,
item_id: $this._iItemId,
item_count: $this._iItemCount,
redirect: $this._sRedirect,
custom: $this._sCustom,
page_id: sHostedPageId,
_t: oDate.getTime()
};
$.post(
$this._sActionsUrl + 'subscribe_json/',
oParams,
function(oData){
$this.loadingInPopup(oLink, true);
processJsonData(oData);
},
'json'
);
},
close: function() {
oLink.removeClass('bx-btn-disabled');
}
});
return false;
};
/** @} */
| {
"redpajama_set_name": "RedPajamaGithub"
} | 2,128 |
As Dunas do São Francisco são uma ecorregião brasileira, pertencente ao bioma da caatinga. Localiza-se no semi-árido baiano, no trecho do Médio São Francisco, abrangendo uma área de 36.170 km². Ocupa partes dos municípios de Barra (Bahia), Casa Nova, Pilão Arcado e Xique-Xique, a uma distância de cerca de 700 km de Salvador. Consistem de grandes depósitos eólicos, as dunas, que podem atingir 100 metros de altura, intercaladas por trechos de solos arenosos sem dunas.
Características
Os solos são compostos por areia quatzosa e de baixa fertilidade. Nas dunas, as altitudes variam de 450 a 500 metros, enquanto que no restante da ecorregião as altitudes variam de 150-700 metros. O potencial hídrico é baixo, Nas depressões entre as dunas, a potencialidade hídrica é mais favorável.
O clima é semi-árido quente, com 7 a 8 meses de seca. As chuvas ocorrem de outubro a março, com uma pluviosidade que varia de 400 a 800 mm.
A vegetação predominante é a caatinga hipoxerófila (arbustiva) e hiperxerófila (arbórea). O primeiro tipo ocorre às margens do rio São Francisco, enquanto que a última verifica-se sobre as dunas. A diversidade de espécies varia desde a favela, o pinhão bravo (euphorbia), araçá de boi (Myrtaceae), a macambira (bromelia sp), cacto quipá (opuntia) da caatinga arbustiva à bombacácea, celastrácea, o xiquexique e a coroa de frade da caatinga arbórea.
A mata ciliar do rio São Francisco apresenta carnaubeiras, umari, quixabeira. Entre as dunas verifica-se a ocorrência de buriti, pindaíba e taboa.
A fauna local, ainda pouco estudada, apresenta espécies endêmicas como lagartos (Tropidurus amathites, Calyptommatus leiolepis, Procellosaurinus sp., Psilophthalmus sp.), serpentes (Typhlops spn.), roedores como o rabo-de-facho (Proechimys yonenagae) e artrópodes, como Mummucia maury.
A unidade encontra-se em bom estado de conservação. A extração de lenha ameaça a estabilidade das dunas. Há planos de abertura de estradas pelas dunas a partir de Barra.
Ligações externas
Sao Francisco
Áreas protegidas da Bahia
Subdivisões da Bahia
Ecorregiões
Caatinga
Meio ambiente da Bahia | {
"redpajama_set_name": "RedPajamaWikipedia"
} | 7,993 |
\section{Introduction}
As per the report by the National Crime Record Bureau (NCRB), the number of farmers died in 2019-2020 is approximately 10,281 [1]. Despite the popular image of farmers, suicide in agriculture has become more common. Numerous workers still adopt the regular methods in farming which brings about low yield and efficiency. A study conducted by the Centre for Study of Developing Societies (CSDS) [2] found that 76\% of the farmers want to quit farming. It also reports that 74\% of farmers didn't get basic farming-related information such as fertilizer doses from officials of the agriculture department. Hence, the precision agriculture system assists and helps the farmers with robotizing and upgrading them to improve rural profitability and contribute to making farming systems smart [3]. The introduction of new technologies such as IoT-based devices would definitely have a positive impact. For the most part, large-scale setups are hard to execute in real-world scenarios. This paper involves using existing advancements like, for instance, sensor-based modules that are demanding and easy to carry out.
An IoT-based smart irrigation system is designed which maintains a balance in the water level of the field. It turns the water pump 'on' when the field is dry and turns it 'off' when it is wet. Also, considering one of the major problems in today's world is water scarcity [4], this system will help managing water levels and only irrigates fields when needed.
Crop mainly required three macronutrients: n (nitrogen), p (phosphorus), k (potassium). Shortage of nutrients can cause deficiency and affect crop's health. Fertilizers are used to provide nutrients to the soil; it is a natural or chemical substance. Overuse of fertilizers can have a negative impact not only on the crops but also on the groundwater. So, a fertilizers dose recommendation system can help to increase crop productivity and controls the overuse of fertilizers. Using firebase, data can be stored with help of an IoT-based fertilizers dose recommendation system. One can easily monitor the results from the previously collected data in the firebase database.
Deep learning is becoming popular and common since the introduction of ImageNet by Alex Krizhevsky et al [5]. Various Pre-trained CNN models are available such as VGG16, VGG19, ResNet50, InceptionV3 which can help in image-related classification tasks. This work also adopts the idea of transfer learning [6] which helps to detect the abnormalities in crop images. Supervised machine learning techniques are also employed to predict crop damage. These models can definitely assist farmers in the early prediction of diseases and damages in crops or plants automatically.
The idea behind presenting a paper like this was to come up with a study that gathers all state-of-the-art approaches on the multimodal systems for precision agriculture use. When looking at research papers and talking to researchers in the area of precision agriculture, we learned that there is a slight disconnect between what researchers believe state-of-the-art and what actually is state-of-the-art. Our biggest contribution for this paper is not an approach, but a study of approaches with an aim to help fellow researchers leverage our findings to push the state-of-the-art further. The following are the notable contributions of this study:
\begin{itemize}
\item To the best of our knowledge, this is one of the initial attempts to present beginner-friendly extensive work related to precision agriculture systems.
\item Well-defined figures, including flowcharts and circuit diagrams, are additionally included to better understand the proposed IoT systems.
\item We successfully leverage deep learning and machine learning methods that help oversee new researchers to follow the examination work from traditional and non-conventional approaches.
\item A bird's eye view on current techniques are provided that will assist readers further push the state-of-the-art systems.
\end{itemize}
The remaining sections are compiled as follows: Related works where we discuss the developments made in the smart precision-based farming system. The Methodology section contains the methods adopted in this project. The result segment provides the evaluation of machine and deep learning models using different metrics and last is the conclusion section.
\section{Related works}
In this section, an bird's eye view of studies and research work is provided that has been conducted in precision-based agriculture from all over the world.
Soil is a significant part in the field of agriculture because overall yield development is dependent on the soil so its sampling is needed to make further critical decisions. The primary goal of the soil examination is to check the strength of a land i.e. whether a field is nutrition deficient or not. Depending upon the soil conditions and weather, soil tests can be done accordingly [7]. The factors that can be determined by analyzing soil are soil types, irrigation, moisture levels, etc. These elements provide an overview regarding the synthetic, physical and natural status of the soil. Currently, there are various toolkits and sensors available to check the soil quality. These toolkits help to monitor the different soil behaviour, for example, water-holding limit, strength, and so forth. A testing kit is developed by Agrocares [8] that may test up to 100 soil samples per day. Additionally, distinguishing polluted soil by utilizing IoT innovations can further shields the field from overfertilization and yield loss [9].
Drought is also one of the serious issues that farmers face. Remote sensing is being used to deal with these problems which helps to analyze water content in the soil . Soil Moisture and Ocean Salinity (SMOS) satellites were launched in 2009. The author in [10] utilizes SMOS L2 to compute the soil water deficit index (SWDI). In [11], the researchers used a moderate resolution imaging spectroradiometer (MODIS) to gather information about the land degradation risk. In [12], sensor and a vision-based self-governing robot named Agribot has been proposed that can help in planting seeds. To determine the seed flow rates a sensor which is equipped with LEDs is used [13]. The sign data is utilized to gauge the stream rate.
About 3\% of freshwater is accessible on the earth of which around 66\% is frozen as icy masses and polar ice covers [14]. In brief, the whole world relies on 0.5\% of total water. Various traditional methods are currently employed by farmers like drip irrigation and sprinkler which are ineffective and crop production is badly affected by them. Crop efficiency stress index (CWSI) based smart irrigation is proposed in [15]. Every sensor is associated to gather the estimations and it further sends the information to the advanced calculation centre to analyze farm data through various intelligent software applications.
Various IoT-based fertilizers approaches are being used to estimate the nutrient requirements like Normalized Difference Vegetation Index (NDVI). It utilizes satellite pictures to check crop status [16] and is dependent on the impression of apparent and close infrared light from vegetation used to decide crop wellbeing. Advances like GPS exactness [17], variable rate innovation (VRT) [18], are added to the intelligent system. There are other sensors that also help to gather data regarding plant health and pest situations like IoT-based automated traps [19]. It is used to count and characterize insect types. An IoT-based automated robot can locate and deal with pest problems.
Machine and Deep learning methods have also been leveraged by researchers for the prediction tasks like yield prediction, object classification, multimodal frameworks [20], [32]. Arun Kumar et al. [21] utilized ANN for regression analysis to anticipate crop yields dependent on yield efficiency. Authors in [22] used time series forecasting methods to analyze and predict the weather patterns. Factors such as environmental, soil, weather, and abiotic features are adopted in [23] to classify and foresee the groundnut yield using Random Forest, SVM, and KNN. Authors in [24] accentuate the utilization of a minimal expense UAV framework with a vision-based arrangement for the isolation of fundamental harvests from weed. It targets lessening the spread of herbicides and pesticides on crops to safeguard their quality. Strategies like harvest location are also utilized where the picture of the yield is specifically masked out of the background that incorporates soil and different items.
\section{Methodology}
In this work, we implemented the smart irrigation system, smart fertilizer dose recommendation system, crop disease detection system and crop damage prediction system. This section is divided into its various subsections for detailed discussion.
\subsection{Irrigation System}
In smart irrigation method, soil moisture sensor gathers the information of the moisture content present in soil and sends it to the Node MCU. It checks the condition whether to turn on or off the water pump by a program designed in Arduino IDE. Soil moisture sensor gives the values in the range of ADC that varies from (0 to 1023) using,
\begin{equation}
Analog Output = ADC Value / 1023
\end {equation}
\begin{equation}
Moisture = 100 - (Analog output * 100)
\end {equation}
The proposed flow of events is given in figure 1. Generally, moisture below 50\% is considered to be dry conditions which suggests the need to irrigate the field and above 50\% to be wet. If the value of the moisture content received from soil moisture sensor is less than the threshold value of moisture content required for the crop, it means that there is a need to irrigate the field. Node MCU sends a signal to the relay module to turn on the water pump which results in an increase in the water levels in the soil until it reaches the threshold value of moisture required for the crop. Once it reaches the threshold value, the relay module will automatically turn off the water pump. Node MCU also sends data to the firebase database. Firebase provides an online database to store data that can be extracted by the web portal. It can easily monitor the information about the moisture content of the field. The web portal also shows the information whether the water pump is on or off. We can also manually turned on and off the water pump through the web portal as shown in figure 2.
\begin{figure}[]
\centerline{\includegraphics[width=8
cm, height=5cm]{13.png}}
\caption{Proposed flow of events smart irrigation system}
\label{fig}
\end{figure}
\begin{figure}[]
\centerline{\includegraphics[width=8
cm, height=4cm]{wpi.png}}
\caption{Web Portal for Irrigation system}
\label{fig}
\end{figure}
The circuit diagram for smart irrigation system is given in Figure 3. The SIG pin of moisture sensor provides the analog signals that is attached to A0 pin of the Node MCU and reads the analog values. VCC pin of the soil moisture sensor is connected to 3v3 pin of Node MCU for the power supply. The ground pins of both are connected with each other to use it as (0V) reference to all other electronic parts. Motor pump is attached to the NodeMCU with ground and a Vin pin is used
In the Relay module, the Node MCU connection with the ground pins of both the devices are connected with each other. Relay module power is connected with D1 pin of the Node MCU and provides the data from the relay module and its signal pin is connected to Vin of the Node MCU for the power supply.
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=5cm]{111.png}}
\caption{Circuit Diagram Smart Irrigation System}
\label{fig}
\end{figure}
\subsection{Fertilizer System}
In smart fertilizer dose recommendation system, as shown in figure 4, we first dug an n-p-k sensor into the soil to gather the data on the nitrogen, phosphorus, potassium content of the soil. The n-p-k sensor sends the n-p-k values to the NodeMCU with the help of max 485 modbus. NodeMCU sends the data to the firebase through a program design in Arduino IDE. The values can be easily accessed by a web portal given in figure 5 to perform basic calculations that helps suggesting the quantity of fertilizers for a specific crop type. For instance, the n-p-k value of the soil given by the n-p-k sensor is (10, 5, 10 kg/ha) and the requirement is to grow wheat on this land. The n-p-k requirement of wheat is (100,20,60 kg/ha). The target is to calculate the doses of fertilizers through urea, muriate of potash (MOP) and decomposed organic phosphorus (DOP). The difference in n-p-k values is (90, 15, 50) which implies that we need 90 kg of nitrogen, 15 kg of phosphorus and 50kg of potassium.
\begin{figure}[htbp]
\centerline{\includegraphics[width=9
cm, height=5cm]{222.png}}
\caption{Proposed flow of events smart fertilizer system}
\label{fig}
\end{figure}
\begin{equation}
MOP = 60\% potassium
\end{equation}
\begin{equation}
Urea = 46\% nitrogen
\end{equation}
\begin{equation}
DAP = (18\% nitrogen + 46\% phosphorus)
\end{equation}
From 3, 4 and 5 we get,
\begin{enumerate}
\item 100 kg MOP → 60kg Potassium,
\item 50kg potassium → (100/60) * 50 of MOP = 8.3 kg of MOP
\item 100 kg DOP → 18kg of nitrogen and 46kg of phosphorus
\item 15kg phosphorus → (100/46) * 15 of DOP = 32.6kg of DOP
\item DOP also has Nitrogen, 32.6kg of DOP → (18/100) * 32.6 nitrogen = 5.86kg of nitrogen
\item Now, we require (90 - 5.86) kg of nitrogen = 84kg of nitrogen
\item 100kg of urea → 46kg of Nitrogen
\item 84kg of nitrogen → (100/46) * 84kg of Urea = 182.6 kg of urea
\end{enumerate}
For better wheat productivity we need 8.3kg of MOP, 32.6kg of DOP and 182.6 kg of urea on this type of soil.
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=6cm]{wpf.png}}
\caption{Web Portal for Fertilizer Recommendation}
\label{fig}
\end{figure}
The circuit diagram is presented in Figure 6. R0 and d1 pins of Modbus are connected to d2 and d3 of node MCU which sends the data collected by the n-p-k sensor to node MCU via d2 and d3 pins. The brown wire is VCC which needs a 9v-24v power supply. The Ground pin of NodeMCU is connected to the Ground pin of Modbus (black Wire). Ground pin maintains a reference level to all the other IOT connections (i.e 0v). Blue wire which is the B pin is connected to the B pin of MAX485 and Yellow wire is pin A connected to A pin of MAX485 which sends data from n-p-k to modbus.
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=4.5cm]{14.png}}
\caption{Circuit Diagram smart fertilizer system}
\label{fig}
\end{figure}
\subsection{Crop Disease Detection System}
In crop disease detection, we employed plant village dataset [25] containing approximately 54,000 images distributed in 38 classes. In this work, images of only six crops namely, potatoes, tomatoes, corn, peach, apple, grapes are utilized for classification. The dataset distibution of crop types used in this work is given in Table I. The images originally are of size 256*256 and are cropped to 64*64. An image augmentation pipeline using Image Data Generator is used to automatically augment the original images incorporating methods like flip, rotate, zoom in, zoom out, blur, rotate, etc. The augmentation is only performed for the training set to increase the accuracy and to avoid data leakage problems. Prior augmentation, the images were divided using the train test split method. We performed a standard 60:20:20 split in which 60 percent of total images belong to the train set and 20 percent to the test set. The validation set is also created to help train the model effectively which constitutes 20 percent of the total images.
\begin{table}[htbp]
\caption{Distribution of images}
\begin{center}
\begin{tabular}{|c|c|c|c|c|c|c|}
\hline
\textbf{Crop Type} & \textbf{Classes}& \textbf{Images}& \textbf{Train}& \textbf{Validation}& \textbf{Test}\\
\hline
Tomato& 10&18160 &10896&3632&3632 \\
\hline
Potato& 3&2152 &1291&431&430 \\
\hline
Apple& 4&3171 &1902&635&634 \\
\hline
Peach& 2&2657 &1594&532&531 \\
\hline
Grapes& 4&4062 &2437&813&812 \\
\hline
Corn& 4&3852 &2371&771&770 \\
\hline
\end{tabular}
\end{center}
\end{table}
The concept of transfer learning is used in this project to transfer the weight of the already trained models i.e. Pre-Trained CNN (Convolutional Neural Network) models [6]. It helps in reducing the computational power and speeds up the performance which shows indications of quicker and improved outcomes. In this study, we utilize only three models namely ResNet50, VGG16, and DenseNet121 [26], [27], [28] for comparative analysis. Our proposed pipeline for preparing the model contains three areas:
\begin{itemize}
\item Feature Generation - Separating the main features by calibrating the models through preparing the three State of the Art (SOTA) Pre-trained CNN structures. The extracted features after this phase act as an input to the modification phase.
\item Modification - Incorporated a concat layer which is a concatenation of features of three layers namely, MaxPooling2D, AveragePooling2D, flatten layer. In addition, a dropout layer with a dropout rate of 0.5 has additionally been fused.
\item Output - The output generated from the modification phase goes to the dense layer. The activation function used is sigmoid. For compilation of model , adam optimizer is used with a learning rate of 0.0002. The batch size is 32 and the model is trained for 30 epochs.
\end{itemize}
\subsection{Crop Damage Prediction System}
In crop damage prediction, the task is to use and employ feature engineering concepts with machine learning algorithms to predict the category of crop damage. We used 'machine learning in agriculture' dataset from the Analytics Vidhya website [33]. This dataset has 88858 rows and 10 columns which is divided into 75:25 using train test split. The description of the features are given in Figure 7.
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=3.5cm]{Feature.png}}
\caption{Dataset Description}
\label{fig}
\end{figure}
It is important to clean and visualize data before implementing machine learning algorithms. This dataset initially contains null values replaced by -999 using fillna method. It also contains columns like Estimated insects counts as shown in Figure 8 with some time series related pattern that suggests to generate features containing different lag values using shift and rolling method. A window size of 5 is used to extract the mean of five observations with a lag of 1 and 2.
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=3cm]{EIC.png}}
\caption{Estimated Insects Count}
\label{fig}
\end{figure}
We used five machine learning algorithms, namely, Random Forest, LGBM, KNN, Decision Tree and XGBoost [29], [30], [31], [9] for classification. Decision Trees is a simple supervised learning algorithm built on the concepts of trees. The tree has two components namely leaves and nodes. The leaves are the choices or the ultimate results whereas the nodes represent the points where the data is divided according to a fixed parameter for further classification. Random Forest is an ensemble-based supervised machine learning algorithm. The bagging method is used to train the ensemble of decision trees which help to build the forest. In simple terms, Random Forest assembles numerous decision trees and combines them to get a more exact and stable expectation. K-nearest neighbors (KNN) depends on the possibility of the likeness of similarities. It predicts the values of new points dependent on how closely the new information is identified with the upsides of the given training set. The KNN doesn't make any assumptions identified with the preparation information and it for the most part inputs all the data values for training.
Extreme gradient boosting (XGBoost) and light gradient boosting (LGBM) are based on the concept of gradient boosting. This technique leverage weak models utilizing ensemble methods that can develop new models to decrease the loss function estimated by gradient descent. However, there is a subtle difference these two models in terms of split. For XGBoost, histogram based filters are used while for LGBM gradient-based one-side sampling (GSOS) method is used to distinguish the best split. It has been observed that histogram based filters technique is computationally costly compared with GSOS which recommends that LGBM is more effective than XGBoost.
\subsection{Metrics}
In this research, we primarily perform two prediction tasks to classify crop damages and disease detection. Since both fall under the category of classification, the models are evaluated by metrics namely, precision, recall, f1 score and accuracy.
\begin{equation}
Precision = \frac{T_{p}}{T_{p}+F_{p}}
\end{equation}
\begin{equation}
\\Recall = \frac{T_{p}}{T_{p}+F_{n}}
\end{equation}
\begin{equation}
Accuracy = \frac{T_{p}+T_{n}}{T_{p}+T_{n} + F_{p}+F_{n}}
\end{equation}
\begin{equation}
F1score = 2.\frac{Precision.Recall}{Precision+Recall}
\end{equation}
where $T_p$, $T_n$, $F_p$, $F_n$ represents True positive, True negative, False positive, False negative.
\section{Results}
Table II shows the evaluation of metrics for crop disease detection. We performed the disease classification on six crops, tomato, potato, apple, peach, grapes and xorn using three Pre-trained CNN models, VGG16, ResNet50, Densenet121. All models performed well and there is a slight difference between the accuracy of the models. For tomato, corn and peach, Densenet121 outperforms other models in terms of all evaluation metrics. VGG16 also performed state of the art results for potato, apple and grapes. However, it marginally underperformed in classification for peach and tomato. The results forecasted from ResNet50 manifests that aside from class apple it failed to outperform other models. For better interpretation and visualization, bar plot is pictured in figure 9 showing precision of models on given classes.
\begin{table}[htbp]
\caption{Evaluation of Metrics for crop disease detection}
\begin{center}
\begin{tabular}{|c|c|c|c|c|c|c|}
\hline
\textbf{Crop Type} & \textbf{Model}& \textbf{Prec}& \textbf{Rec}& \textbf{F1}
& \textbf{Acc.}\\
\hline
& VGG16& 0.93&0.94 &0.94&0.945 \\
Tomato& ResNet50& \textbf{0.96}&0.96 &0.96&0.965\\
& Densenet121& \textbf{0.96}&\textbf{0.97} &\textbf{0.97}&\textbf{0.973}\\
\hline
& VGG16& \textbf{0.94}&0.96 &\textbf{0.95}&\textbf{0.981} \\
Potato& ResNet50& 0.89&\textbf{0.97} &0.92&0.960\\
& Densenet121& 0.87&0.96 &0.90&0.955\\
\hline
& VGG16&\textbf{0.99}&\textbf{0.98} &\textbf{0.98}&\textbf{0.981} \\
Apple& ResNet50& 0.98&0.96 &0.97&0.970\\
& Densenet121& 0.97&0.95 &0.96&0.970\\
\hline
& VGG16& 0.97&0.98 &0.97&0.986 \\
Peach& ResNet50& \textbf{1.00}&\textbf{0.99} &\textbf{0.99}&\textbf{0.996}\\
& Densenet121& \textbf{1.00}&\textbf{0.99} &\textbf{0.99}&\textbf{0.996}\\
\hline
& VGG16& \textbf{0.97}&\textbf{0.97} &\textbf{0.97}&\textbf{0.969} \\
Grape& ResNet50& 0.96&0.96 &0.96&0.948\\
& Densenet121& \textbf{0.97}&0.96 &0.96&0.963\\
\hline
& VGG16& 0.93&0.93 &0.93&0.946 \\
Corn& ResNet50& 0.93&0.93 &0.93&0.943\\
& Densenet121& \textbf{0.95}&\textbf{0.95} &\textbf{0.95}&\textbf{0.965}\\
\hline
\end{tabular}
\end{center}
\end{table}
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=5cm]{Dlearn.png}}
\caption{Bar Plot showing precision for Crop Disease Detection}
\label{fig}
\end{figure}
Analysis of evaluation metrics are also discussed for crop damage prediction. As we can see from Table III, the highest accuracy achieved is 94\% by LGBM and it completely beats other algorithms. Random Forest and XGBoost have very subtle difference in their predictions. KNN is the worst performing algorithm forestalling less than 5 percent precision for minority classes. The differences in the results of the various classes suggest that the algorithms failed to manage the issue of class imbalance. In general, tree based classifiers outperformed the standard machine learning algorithm. A bar plot is pictured in Figure 10 shows the precision of machine learning algorithms used for crop damage prediction on various sub classes.
\begin{table}[htbp]
\caption{Evaluation of Metrics for crop damage prediction}
\begin{center}
\begin{tabular}{|c|c|c|c|c|c|}
\hline
\textbf{Model } & \textbf{Class}& \textbf{Prec}& \textbf{Rec}& \textbf{F1} & \textbf{Acc.}\\
\hline
& 0& 0.95&\textbf{0.99} &0.97&\\
RF& 1& 0.78&0.72 &0.75&0.93\\
& 2& 0.57&0.10 &0.16&\\
\hline
& 0& \textbf{0.97}&\textbf{0.99} &\textbf{0.98}& \\
LGBM& 1& \textbf{0.82}&\textbf{0.78} &\textbf{0.80}&\textbf{0.94}\\
& 2& 0.44&0.20 &\textbf{0.28}&\\
\hline
& 0& 0.96&0.91 &0.93& \\
DT& 1& 0.50&0.59 &0.54&0.85\\
& 2& 0.19&\textbf{0.32} &0.24&\\
\hline
& 0& 0.95&0.98 &0.96& \\
XGB& 1& 0.72&0.70 &0.72&0.92\\
& 2& \textbf{0.60}&0.06 &0.11&\\
\hline
& 0& 0.85&0.97 &0.90& \\
KNN& 1& 0.23&0.06 &0.09&0.84\\
& 2& 0.05&0.01 &0.01&\\
\hline
\end{tabular}
\end{center}
\end{table}
\begin{figure}[htbp]
\centerline{\includegraphics[width=8
cm, height=5cm]{Mlearn.png}}
\caption{Bar Plot showing Precision for Crop Damage Prediction}
\label{fig}
\end{figure}
\section{Conclusion}
A multimodal precision farming system is implemented in this project which consists of intelligent fertilizer, irrigation, crop disease and damage prediction which help reduce the efforts and labors in the agriculture sector. In this work, we intend to provide the methodology with circuit diagram, flowchart and theoretical aspects for IoT systems. Our work is well-organized, easy to read, contains well-defined figures and diagrams. This would definitely help new researchers to get a better understanding of the problem statement and further push the state of the art systems. Subsequently, we leverage multimodal approach to train an image classification and machine learning models related to crop disease detection and damages. In multiclass image classification, we employed three Pre-Trained CNN architectures namely, ResNet50, VGG16, DenseNet121. In crop damage prediction, LightGBM beats XGBoost and Random Forest by a slight margin. However, the aformentioned models aren't able to produce quality results for minority classes. Oversampling techniques like SMOTE and ADASYN are required to curb the dominance of majority class. Hyperparameter optimization techniques such as evolutionary algorithms are missed in this research due to time constraints and less computational power. It will be a part of the future work.
| {
"redpajama_set_name": "RedPajamaArXiv"
} | 4,030 |
{"url":"http:\/\/tex.stackexchange.com\/questions\/68465\/clean-import-of-latex-into-lyx-environment-instead-of-ert","text":"# Clean import of LaTeX into LyX: environment instead of ERT [closed]\n\nI am importing a LaTeX code into LyX. This code has multiple uses of amsmath environments (lemma, theorem etc.). After importing, all theese environments turn into ERT boxes in LyX. I was wondering if there is any way to import LaTeX into LyX preserving the amsmath environments? I think there should be a way to do this as AMS modules exist in LyX.\n\nTo experiment a little bit, I tried to check if LyX can import its own LaTeX code properly. I first created a LyX document using Theorems (AMS) module. Then I exported the LyX document into LaTeX code (see below) and imported the code back into LyX. This produced a LyX document with ERT instead of the theorem environment. What can possibly go wrong?\n\n%% LyX 1.6.7 created this file. For more info, see http:\/\/www.lyx.org\/.\n%% Do not edit unless you really know what you are doing.\n\\documentclass[english]{article}\n\\usepackage[T1]{fontenc}\n\\usepackage[latin9]{inputenc}\n\\usepackage[letterpaper]{geometry}\n\\geometry{verbose,tmargin=1in,bmargin=1in,lmargin=1in,rmargin=1in}\n\\usepackage{amsthm}\n\\usepackage{amsmath}\n\\usepackage{amssymb}\n\\makeatletter\n%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Textclass specific LaTeX commands.\n\\theoremstyle{plain}\n\\newtheorem*{thm*}{Theorem}\n%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% User specified LaTeX commands.\n\\makeatother\n\\makeatother\n\\usepackage{babel}\n\\makeatother\n\\usepackage{babel}\n\\begin{document}\n\\begin{thm*}\nLet $\\theta$ be a real number, then$\\sin^{2}\\theta+\\cos^{2}\\theta=1.$\n\\end{thm*}\n\\end{document}\n\n-\n\n## migration rejected from stackoverflow.comAug 4 '13 at 10:23\n\nThis question came from our site for professional and enthusiast programmers. Votes, comments, and answers are locked due to the question being closed here, but it may be eligible for editing and reopening on the site where it originated.\n\n## closed as off-topic by Joseph Wright\u2666Aug 4 '13 at 10:23\n\n\u2022 This question does not fall within the scope of TeX, LaTeX or related typesetting systems as defined in the help center.\nIf this question can be reworded to fit the rules in the help center, please edit the question.\n\nThis issue has also been reported here. I am sure many people stumble upon this while collaborating with LaTeX users. It would be great to know if this problem persists in LyX 2.0. \u2013\u00a0 MHT Aug 23 '12 at 14:28\nThis issue still exists in LyX 2.0.4. \u2013\u00a0 Werner Aug 23 '12 at 16:35\nWelcome to TeX.sx! Your question was migrated here from Stack Overflow. Please register on this site, too, and make sure that both accounts are associated with each other (by using the same OpenID), otherwise you won't be able to comment on or accept answers or edit your question. \u2013\u00a0 Werner Aug 23 '12 at 16:36\nThis question appears to be off-topic because it is about a feature request or bug report for Lyx (depending on your point of view). \u2013\u00a0 Joseph Wright Aug 4 '13 at 10:23","date":"2015-10-06 05:35:59","metadata":"{\"extraction_info\": {\"found_math\": true, \"script_math_tex\": 0, \"script_math_asciimath\": 0, \"math_annotations\": 0, \"math_alttext\": 0, \"mathml\": 0, \"mathjax_tag\": 0, \"mathjax_inline_tex\": 2, \"mathjax_display_tex\": 0, \"mathjax_asciimath\": 1, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 0, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.8582621812820435, \"perplexity\": 2163.7763312915827}, \"config\": {\"markdown_headings\": true, \"markdown_code\": true, \"boilerplate_config\": {\"ratio_threshold\": 0.18, \"absolute_threshold\": 10, \"end_threshold\": 15, \"enable\": true}, \"remove_buttons\": true, \"remove_image_figures\": true, \"remove_link_clusters\": true, \"table_config\": {\"min_rows\": 2, \"min_cols\": 3, \"format\": \"plain\"}, \"remove_chinese\": true, \"remove_edit_buttons\": true, \"extract_latex\": true}, \"warc_path\": \"s3:\/\/commoncrawl\/crawl-data\/CC-MAIN-2015-40\/segments\/1443736678520.45\/warc\/CC-MAIN-20151001215758-00185-ip-10-137-6-227.ec2.internal.warc.gz\"}"} | null | null |
<!DOCTYPE html>
<html>
<head>
<title>Snowflake | The Pink Fairy Book | Andrew Lang | Lit2Go ETC</title>
<link rel="stylesheet" href="http://etc.usf.edu/lit2go/css/screenless.css" type="text/css" media="screen" title="no title" charset="utf-8">
<link rel="stylesheet" href="http://etc.usf.edu/lit2go/css/printless.css" type="text/css" media="print" title="no title" charset="utf-8">
<!--[if lt IE 9]>
<script src="//html5shiv.googlecode.com/svn/trunk/html5.js"></script>
<![endif]-->
<script type="text/javascript" src="//ajax.googleapis.com/ajax/libs/jquery/1.10.2/jquery.min.js"></script>
<script type="text/javascript" defer src="http://etc.usf.edu/lit2go/js/js.min.js"></script>
<script type="text/javascript">
var _gaq = _gaq || [];
_gaq.push(['_setAccount', 'UA-5574891-1']);
_gaq.push(['_trackPageview']);
(function() {
var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true;
ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js';
var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s);
})();
$(document).ready(function() {
$('img').unveil();
$('#contactable').contactable({
url: 'http://etc.usf.edu/lit2go/welcome/feedback/',
subject: 'Lit2Go Feedback — Snowflake | The Pink Fairy Book | Andrew Lang — http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4841/snowflake/'
});
});
</script>
</head>
<body>
<div class="page"> <header>
<h1>
<a href="http://etc.usf.edu/lit2go/">Lit<span class="blue">2</span>Go</a>
</h1>
<ul>
<li id="search"><form action="http://etc.usf.edu/lit2go/search/"><input type="text" name="q" placeholder="Search" value=""></form></li>
</ul>
</header>
<nav id="shell">
<ul>
<li><a href="http://etc.usf.edu/lit2go/authors/" class="">Authors</a></li>
<li><a href="http://etc.usf.edu/lit2go/books/" class="selected">Books</a></li>
<li><a href="http://etc.usf.edu/lit2go/genres/" class="">Genres</a></li>
<li><a href="http://etc.usf.edu/lit2go/collections/" class="">Collections</a></li>
<li><a href="http://etc.usf.edu/lit2go/readability/" class="">Readability</a></li>
</ul>
</nav>
<section id="content">
<div id="contactable"><!-- contactable html placeholder --></div>
<div id="page_content">
<header>
<h2>
<a href="http://etc.usf.edu/lit2go/145/the-pink-fairy-book/">The Pink Fairy Book</a>
</h2>
<h3>
by <a href="http://etc.usf.edu/lit2go/authors/198/andrew-lang/">Andrew Lang</a>
</h3>
</header>
<h4>
Snowflake </h4>
<h5>
by <a href="http://etc.usf.edu/lit2go/authors/198/andrew-lang/">Andrew Lang</a>
</h5>
<div id="default">
<details open>
<summary>
Additional Information
</summary>
<div id="columns">
<ul>
<li>
<strong>Year Published:</strong>
1897 </li>
<li>
<strong>Language:</strong>
English </li>
<li>
<strong>Country of Origin:</strong>
England </li>
<li>
<strong>Source:</strong>
Lang, A. (Ed.). (1897). <em>The Pink Fairy Book</em>. London: Longmans, Green & Co. </li>
</ul>
</div>
<div id="columns">
<ul>
<li>
<strong>Readability:</strong>
<ul>
<li>
Flesch–Kincaid Level:
<a href="http://etc.usf.edu/lit2go/readability/flesch_kincaid_grade_level/4/" title="Flesch–Kincaid Grade Level 4.0">4.0</a>
</li>
</ul>
</li>
<li>
<strong>Word Count:</strong>
1,339 </li>
</ul>
</div>
<div id="columns">
<ul>
<li>
<strong>Genre:</strong>
<a href="http://etc.usf.edu/lit2go/genres/13/fairy-talefolk-tale/">Fairy Tale/Folk Tale</a>
</li>
<li>
<strong>Keywords:</strong>
family, hope, love </li>
<li>
<a class="btn" data-toggle="modal" href="#cite_this" >✎ Cite This</a>
</li>
<li>
<!-- AddThis Button BEGIN -->
<div class="addthis_toolbox addthis_default_style ">
<a addthis:ui_delay="500" href="http://www.addthis.com/bookmark.php?v=250&pub=roywinkelman" class="addthis_button_compact">Share</a>
<span class="addthis_separator">|</span>
<a class="addthis_button_preferred_1"></a>
<a class="addthis_button_preferred_2"></a>
<a class="addthis_button_preferred_3"></a>
<a class="addthis_button_preferred_4"></a>
</div>
<script type="text/javascript">$($.getScript("http://s7.addthis.com/js/250/addthis_widget.js?pub=roywinkelman"))</script>
<!-- <script type="text/javascript" src="http://s7.addthis.com/js/250/addthis_widget.js?pub=roywinkelman"></script> -->
<!-- AddThis Button END -->
</li>
</ul>
</div>
<h4>Downloads</h4>
<ul id="downloads">
<li>
<a href="http://etc.usf.edu/lit2go/audio/mp3/the-pink-fairy-book-014-snowflake.4841.mp3">Audio</a>
</li>
<li><a href="http://etc.usf.edu/lit2go/pdf/student_activity/4841/4841-1.pdf">Student Activity</a></li>
</ul>
<hr>
</details>
<div class="modal hide" id="cite_this">
<script>
$($('#myTab a').click(function (e) {e.preventDefault();$('#myTab a[href="#apa"]').tab('show');}));
</script>
<nav>
<ul id="myTab">
<li class="active">
<a href="#apa" data-toggle="tab">APA</a>
</li>
<li>
<a href="#mla" data-toggle="tab">MLA</a>
</li>
<li>
<a href="#chicago" data-toggle="tab">Chicago</a>
</li>
</ul>
</nav>
<div class="tab-content">
<div class="content tab-pane hide active" id="apa">
<p class="citation">
Lang, A. (1897). Snowflake. <em>The Pink Fairy Book</em> (Lit2Go Edition). Retrieved February 15, 2016, from <span class="faux_link">http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4841/snowflake/</span>
</p>
</div>
<div class="content tab-pane" id="mla">
<p class="citation">
Lang, Andrew. "Snowflake." <em>The Pink Fairy Book</em>. Lit2Go Edition. 1897. Web. <<span class="faux_link">http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4841/snowflake/</span>>. February 15, 2016.
</p>
</div>
<div class="content tab-pane" id="chicago">
<p class="citation">
Andrew Lang, "Snowflake," <em>The Pink Fairy Book</em>, Lit2Go Edition, (1897), accessed February 15, 2016, <span class="faux_link">http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4841/snowflake/</span>.
</p>
</div>
</div>
</div>
<span class="top"> <nav class="passage">
<ul>
<li><a href="http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4839/the-snow-man/" title="The Snow-Man" class="back">Back</a></li>
<li><a href="http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4843/the-water-of-life/" title="The Water of Life" class="next">Next</a></li>
</ul>
</nav>
</span>
<div id="shrink_wrap">
<div id="i_apologize_for_the_soup">
<audio controls style="width:99%;">
<source src="http://etc.usf.edu/lit2go/audio/mp3/the-pink-fairy-book-014-snowflake.4841.mp3" type="audio/mpeg" />
<source src="http://etc.usf.edu/lit2go/audio/ogg/the-pink-fairy-book-014-snowflake.4841.ogg" type="audio/ogg" />
The embedded audio player requires a modern internet browser. You should visit <a href="http://browsehappy.com/">Browse Happy</a> and update your internet browser today!
</audio>
<p>
Once upon a time there lived a peasant called Ivan, and he had a wife whose name was Marie. They would have been quite happy except for one thing: they had no children to play with, and as they were now old people they did not find that watching the children of their neighbours at all made up to them for having one of their own.</p>
<p>
One winter, which nobody living will ever forget, the snow lay so deep that it came up to the knees of even the tallest man. When it had all fallen, and the sun was shining again, the children ran out into the street to play, and the old man and his wife sat at their window and gazed at them. The children first made a sort of little terrace, and stamped it hard and firm, and then they began to make a snow woman. Ivan and Marie watched them, the while thinking about many things.</p>
<p>
Suddenly Ivan's face brightened, and, looking at his wife, he said, 'Wife, why shouldn't we make a snow woman too?'</p>
<p>
'Why not?' replied Marie, who happened to be in a very good temper; 'it might amuse us a little. But there is no use making a woman. Let us make a little snow child, and pretend it is a living one.'</p>
<p>
'Yes, let us do that,' said Ivan, and he took down his cap and went into the garden with his old wife.</p>
<p>
Then the two set to work with all their might to make a doll out of the snow. They shaped a little body and two little hands and two little feet. On top of all they placed a ball of snow, out of which the head was to be.</p>
<p>
'What in the world are you doing?' asked a passer-by.</p>
<p>
'Can't you guess?' returned Ivan.</p>
<p>
'Making a snow-child,' replied Marie.</p>
<p>
They had finished the nose and the chin. Two holes were left for the eyes, and Ivan carefully shaped out the mouth. No sooner had he done so than he felt a warm breath upon his cheek. He started back in surprise and looked--and behold! the eyes of the child met his, and its lips, which were as red as raspberries, smiled at him!</p>
<p>
'What is it?' cried Ivan, crossing himself. 'Am I mad, or is the thing bewitched?'</p>
<p>
The snow-child bent its head as if it had been really alive. It moved its little arms and its little legs in the snow that lay about it just as the living children did theirs.</p>
<p>
'Ah! Ivan, Ivan,' exclaimed Marie, trembling with joy, 'heaven has sent us a child at last!' And she threw herself upon Snowflake (for that was the snow-child's name) and covered her with kisses. And the loose snow fell away from Snowflake as an egg shell does from an egg, and it was a little girl whom Marie held in her arms.</p>
<p>
'Oh! my darling Snowflake!' cried the old woman, and led her into the cottage.</p>
<p>
And Snowflake grew fast; each hour as well as each day made a difference, and every day she became more and more beautiful. The old couple hardly knew how to contain themselves for joy, and thought of nothing else. The cottage was always full of village children, for they amused Snowflake, and there was nothing in the world they would not have done to amuse her. She was their doll, and they were continually inventing new dresses for her, and teaching her songs or playing with her. Nobody knew how clever she was! She noticed everything, and could learn a lesson in a moment. Anyone would have taken her for thirteen at least! And, besides all that, she was so good and obedient; and so pretty, too! Her skin was as white as snow, her eyes as blue as forget-me-nots, and her hair was long and golden. Only her cheeks had no colour in them, but were as fair as her forehead.</p>
<p>
So the winter went on, till at last the spring sun mounted higher in the heavens and began to warm the earth. The grass grew green in the fields, and high in the air the larks were heard singing. The village girls met and danced in a ring, singing, 'Beautiful spring, how came you here? How came you here? Did you come on a plough, or was it a harrow?' Only Snowflake sat quite still by the window of the cottage.</p>
<p>
'What is the matter, dear child?' asked Marie. 'Why are you so sad? Are you ill? or have they treated you unkindly?'</p>
<p>
'No,' replied Snowflake, 'it is nothing, mother; no one has hurt me; I am well.'</p>
<p>
The spring sun had chased away the last snow from its hiding place under the hedges; the fields were full of flowers; nightingales sang in the trees, and all the world was gay. But the gayer grew the birds and the flowers the sadder became Snowflake. She hid herself from her playmates, and curled herself up where the shadows were deepest, like a lily amongst its leaves. Her only pleasure was to lie amid the green willows near some sparkling stream. At the dawn and at twilight only she seemed happy. When a great storm broke, and the earth was white with hail, she became bright and joyous as the Snowflake of old; but when the clouds passed, and the hail melted beneath the sun, Snowflake would burst into tears and weep as a sister would weep over her brother.</p>
<p>
The spring passed, and it was the eve of St. John, or Midsummer Day. This was the greatest holiday of the year, when the young girls met in the woods to dance and play. They went to fetch Snowflake, and said to Marie: 'Let her come and dance with us.'</p>
<p>
But Marie was afraid; she could not tell why, only she could not bear the child to go. Snowflake did not wish to go either, but they had no excuse ready. So Marie kissed the girl and said: 'Go, my Snowflake, and be happy with your friends, and you, dear children, be careful of her. You know she is the light of my eyes to me.'</p>
<p>
'Oh, we will take care of her,' cried the girls gaily, and they ran off to the woods. There they wore wreaths, gathered nosegays, and sang songs some sad, some merry. And whatever they did Snowflake did too.</p>
<p>
When the sun set they lit a fire of dry grass, and placed themselves in a row, Snowflake being the last of all. 'Now, watch us,' they said, 'and run just as we do.'</p>
<p>
And they all began to sing and to jump one after another across the fire.</p>
<p>
Suddenly, close behind them, they heard a sigh, then a groan. 'Ah!' They turned hastily and looked at each other. There was nothing. They looked again. Where was Snowflake? She has hidden herself for fun, they thought, and searched for her everywhere. 'Snowflake! Snowflake!' But there was no answer. 'Where can she be? Oh, she must have gone home.' They returned to the village, but there was no Snowflake.</p>
<p>
For days after that they sought her high and low. They examined every bush and every hedge, but there was no Snowflake. And long after everyone else had given up hope Ivan and Marie would wander through the woods crying 'Snowflake, my dove, come back, come back!' And sometimes they thought they heard a call, but it was never the voice of Snowflake.</p>
<p>
And what had become of her? Had a fierce wild beast seized her and dragged her into his lair in the forest? Had some bird carried her off across the wide blue sea?</p>
<p>
No, no beast had touched her, no bird had borne her away. With the first breath of flame that swept over her when she ran with her friends Snowflake had melted away, and a little soft haze floating upwards was all that remained of her.</p>
</div>
</div>
<span class="bottom"> <nav class="passage">
<ul>
<li><a href="http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4839/the-snow-man/" title="The Snow-Man" class="back">Back</a></li>
<li><a href="http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4843/the-water-of-life/" title="The Water of Life" class="next">Next</a></li>
</ul>
</nav>
</span>
</div>
</div>
</section>
<footer screen>
<div id="footer-text">
<p>
This collection of children's literature is a part of the <a href="http://etc.usf.edu/">Educational Technology Clearinghouse</a> and is funded by various <a href="http://etc.usf.edu/lit2go/welcome/funding/">grants</a>.
</p>
<p>
Copyright © 2006—2016 by the <a href="http://fcit.usf.edu/">Florida Center for Instructional Technology</a>, <a href="http://www.coedu.usf.edu/">College of Education</a>, <a href="http://www.usf.edu/">University of South Florida</a>.
</p>
</div>
<ul id="footer-links">
<li><a href="http://etc.usf.edu/lit2go/welcome/license/">License</a></li>
<li><a href="http://etc.usf.edu/lit2go/welcome/credits/">Credits</a></li>
<li><a href="http://etc.usf.edu/lit2go/welcome/faq/">FAQ</a></li>
<li><a href="http://etc.usf.edu/lit2go/giving/">Giving</a></li>
</ul>
</footer>
<footer print>
<div id="footer-text">
<p>This document was downloaded from <a href="http://etc.usf.edu/lit2go/">Lit2Go</a>, a free online collection of stories and poems in Mp3 (audiobook) format published by the <a href="http://fcit.usf.edu/">Florida Center for Instructional Technology</a>. For more information, including classroom activities, readability data, and original sources, please visit <a href="http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4841/snowflake/">http://etc.usf.edu/lit2go/145/the-pink-fairy-book/4841/snowflake/</a>.</p>
</div>
<div id="book-footer">
<p>Lit2Go: <em>The Pink Fairy Book</em></p>
<p>Snowflake</p>
</div>
</footer>
<script type="text/javascript" defer src="http://etc.usf.edu/lit2go/js/details.js"></script>
</div>
</body>
</html>
| {
"redpajama_set_name": "RedPajamaGithub"
} | 3,455 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.