text
stringlengths 14
5.77M
| meta
dict | __index_level_0__
int64 0
9.97k
⌀ |
|---|---|---|
Obby & Jappari stays up to date with the right mix of analogue and digital
Alif Ibrahim
Glittering 3D graphics, dystopian metal-inspired typography, and futuristic streetwear feature heavily in the work of Obby & Jappari, a graphic design studio based in Frankfurt that was founded just last year. The studio's core team is made up of Obby, Jappari, and Pia Graf, with the first two preferring to go by their mononymic artist names. Much like hip hop darling boyband Brockhampton, they found each other through the internet while working jobs unrelated to the discipline, eventually collaborating on small personal projects. "We then started working together on personal projects for like 10 hours a day. For our first big project, a campaign for Under Armour, we worked with the third founder Pia for the first time. The project welded us together and we officially founded the studio," Jappari tells It's Nice That about their final member, who mostly focuses on animation and illustration for the studio.
The three founders' initial graphic design inspirations were quite different. Pia, for instance, was mainly inspired by anime and manga. "I was very much into Japanese culture and design," Pia says. Georgia-born Jappari, like many hailing from the country, started designing out of necessity. "In Georgia, I founded a death metal band called Discord. I was the frontman of the band, but I also made all the covers and merchandise," Jappari explains. Obby's origins are also music related, redesigning album covers on Leanified, a now-defunct image editing app that leaned heavily towards the vaporwave aesthetics of the peak Yung Lean era. "Once the app got removed from the app store, I bought myself a photoshop license," Obby adds.
Nowadays, the studio's creative practice sees it mixing fluid and dynamic 3D graphics with clean typography, creating a visual language that's very much in vogue today. The studio makes highly atmospheric work, often more than just the sum of its parts, for projects like Nike's Vapormax campaign or Tunica's ambitious publication published back in June. "It's more about a general look and vibe than specific objects," the studio says, "in general, we try to explore digital and analogue imagery and the contrast between these."
Obby & Jappari: Nike Vapormax
The studio's project with photographer Sivan Miller is a perfect example of this, for instance. In The Future is Here, the language of fashion photography was met with interlocking tubes and plastic forms, all in translucent pink to complement the jacket and glasses that the model wears. "Here we blended 2D into 3D design instead of editing the photograph in the usual way by just photoshopping it in. In this case, we tried to translate a flat photograph into a 3D world," the studio explains. Another project, this time personal, started out with analogue techniques with the studio dripping honey onto a spray painted material. "We scanned it and worked on it again in photoshop and Cinema 4D to create the unique look," the studio explains, resulting in the highly-textured, sticky image in sunset pink.
Citing mentor Eike König as one of their inspirations, the studio says that "his advice has influenced us in our growth as designers and humans." Obby & Jappari, as another group of designers that's looking for beauty in an increasingly fast-paced world, where stories disappear as soon as they surface, will hopefully find its answer in never becoming too comfortable with its chosen medium or process.
Obby & Jappari: Tunica Magazine
Obby & Jappari: The Future is Here
In collaboration with Sivan Miller
Obby & Jappari
Obby & Jappari: Nike Air
Obby & Jappari: Apparat
Obby & Jappari: 3Teeth
ThoughtMatter uses physical design to cleverly point audiences to Tacoma Arts Live
Ahmad Hammoud's vibrant and organic designs are achieved through "aimless sketching"
Vera Lucía Jiménez is a designer making striking work "without a big budget or fancy materials"
obbyxjappari.tumblr.com
Alif joined It's Nice That as an editorial assistant from September to December 2019 after completing an MA in Digital Media at Goldsmiths, University of London. His writing often looks at the impact of art and technology on society.
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 6,742
|
Henry Brydges may refer to:
Henry Brydges (died 1539), MP for Ludgershall
Henry Brydges, 2nd Duke of Chandos (1708–1771)
Henry James Brydges, 1st Duke of Chandos (1673–1744)
See also
Henry Bridges (disambiguation)
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 7,447
|
\section{Introduction}
Machine learning and deep learning have become the most popular data analysis techniques over these years due to its high accuracy, reliability and repeatable results.
Especially in the field of biomedical image processing~\cite{r8},
where these cutting-edge techniques can be used to assist medical professionals with their diagnosis.
In traditional medical imaging,
diagnosis mainly relays on human eyes and experiences,
which can sometime be inconsistent and problematic.
Neural network is proved to possess an extraordinary ability to recognize patterns and make classifications.
It is able to capture high-level features of signals and images,
which are very subtle or even impossible for human eyes.
As such,
neural network is a powerful tool for solving biomedical image processing problems and is widely used over the past few years.
However,
neural network,
a data-driven method,
relies heavily on the training dataset,
which will lead to unexpected performance if it does not have a decent size or proper label~\cite{r25}.
In many medical research works,
it is challenging to obtain a dataset with a huge size and labelling them correctly due to many factors(e.g. cost, domain knowledge).
Labelling such a large dataset tends to be erroneous due to human mistakes and medical misjudgment.
It will therefore need more quality control (QC) procedures such as reliability cross-check~\cite{r26}.
And these will not always be possible as QC can double the number of data that needs to be labelled.
In this work,
to overcome these issues and design a more robust training scheme that can be used for data classification
when the biomedical signal is collected under an imperfect condition,
we proposed a \textbf{Negative-ResNet} with 5 residual blocks (12-layer Convolutional Neural Network in total),
which can be trained and showed good performance even when the dataset contains a significant number of \textbf{noisy labels}.
Unlike usual Positive Learning (PL),
the \textbf{Negative Learning (NL)} training scheme we mention above provides the information that a given signal doesn't belong to a certain label,
which is more likely to be correct.
For example,
in Fig. \ref{fig_NLPL},
we plot 6 types of data in total in the dataset.
If we assume a label might be noisy:
For positive learning,
information "This is a Type 2" will be provided,
As such, there's only 1/6 chance that the label is correct.
However, for negative learning, information "This is not a Type 2" will be given.
The probability of correctness increases to 5/6.
Hence, NL will usually provide more precise information when the label is noisy.
\begin{figure}[h]
\centering
\includegraphics[width=8cm]{NL_vs_PL.png}
\caption{Negative Learning(NL) versus. Positive Learning(PL),
NL is more likely to provide the correct information}
\label{fig_NLPL}
\end{figure}
To verify the effeteness of the training scheme,
we designed a portable Electrocardiogram (ECG) patch
- iRealCare
to collect and construct our ambulatory ECG dataset.
There are 5 types of signals in this dataset,
which indicates 5 different status of the heartbeat signal.
And the size of the dataset is more than 4 million of heartbeat signals.
The main contributions of this work are listed as below:
\begin{itemize}
\item We \textbf{propose a novel semi-supervised training scheme}:
a confidence-level-based training method,
following by a negative training scheme using a deep residual block based neural network backbone.
\item We develop a \textbf{novel portable IoT ECG Patch} named iRealCare and constructed a \textbf{large real-patient dataset} using this IoT device with more than 4 million ECG signal waveforms.
\item We apply the proposed scheme on our dataset and achieve a satisfactory\textbf{ classification accuracy of 91.0\,\%},
which is \textbf{6.2\,\% higher than a normal training process} with ResNet.
\end{itemize}
The remainder of this manuscript is organized as below:
Section \ref{related work} mainly talks about the previous relevant study related to our work,
such as ECG signal and noisy label training.
Section \ref{s2} describes how our dataset is constructed and issues that comes with it.
It mainly introduces the ECG patch, the dataset and the data pre-processing procedure.
Section \ref{s3} demonstrates the methodology of the project,
including detailed explanations of the concept of negative learning and confidence-level-based learning; the novel structure of the residual block.
Section \ref{s4} introduces our proposed deep learning architecture and the experiments for performing and evaluating the proposed method.
Section \ref{s5} gives analytical results of the procedure.
In the end, Section \ref{s6} concludes the work.
\section{Related Work\label{related work}}
\subsection{ECG Signal}
Arrhythmia,
also known as irregular heartbeat,
is an important indicator for most of the heart diseases~\cite{r1}.
Many infrequent arrhythmias are harmless while some may cause serious damage or even death~\cite{r1}.
Electrocardiogram(ECG) is currently widely used for detecting arrhythmia by measuring the electrical activities generated by the heart as contracts~\cite{r1,r2,r3,r4}. An ECG signal can be characterized by two features:
peaks and intervals.
Each peak (P, Q, R, S, T) has its normal amplitude,
and each interval (P-R, R-R, Q-R-S, S-T) has its duration period.
Fig. \ref{fig_1} illustrates an ECG signal,
and Table \ref{TB:tb_1} shows the ECG features and their typical values\cite{r5,r6}.
\begin{figure}[h]
\centering
\includegraphics[width=6cm]{ecg.png}
\caption{Standard waveform of an Electrocardiogram}
\label{fig_1}
\end{figure}
\begin{table}[h]
\caption{Feature value of normal ECG}
\begin{center}
\begin{tabular}{|c|c|}
\hline
Feature&Value\\
\hline
PR interval& 0.12-0.20s\\
P-wave interval&0.00-0.12s\\
QRS&0.06-0.11S\\
RR interval&0.60-1.00s\\
R-wave amplitude& $>$ 0.50mV\\
QT interval& 0.33-0.43s\\
T-wave amplitude& $>$ 0.50mV\\
P-wave amplitude& $<$ 0.25mV\\
\hline
\end{tabular}
\end{center}
\label{TB:tb_1}
\end{table}
However, ECG signal data are usually of large size,
and it would be a huge workload for cardiologists to
review all of them manually.
Hence,
over the past few decades,
many researchers and engineers have devoted themselves to design robust and accurate equipment and algorithms for ECG based arrhythmia detection and classification.
Conventionally,
ECG decomposition and hand-crafted manual features are used for ECG classification.
In~\cite{r7},~\cite{r8} and~\cite{r9},
the wavelet transform was used to extract features of each QRS complex. In~\cite{r10} and~\cite{r11}, adaptive filtering was designed and used for arrhythmia detection.
Even though good results were reported,
these model-based feature extraction methods are not generic and robust.
Individual differences in ECG signals are significant and may lead to poor performance for different individuals.
Neural networks which are able to extract high-level features are proved to be a powerful tool to learn distributions of a dataset and generalize from training data~\cite{r12,r13}.
And the authors of~\cite{r14} and~\cite{r15} both presented a machine learning based method which consisted of a 1-D Convolutional Neural Network (CNN) for real-time ECG classification.~\cite{r15} used a nine-layer CNN to classify five classes of arrhythmia and reached an accuracy of 94\% on MIT-BIH dataset. Even though~\cite{r15} reached a very high accuracy, it failed to consider performance on individual patients.~\cite{r14} presented a better way of introducing patient-specific training using transfer learning. And it is also based on MIT-BIH database.
Besides CNN, other neural network backbone models are also commonly used in ECG classification. In~\cite{r16}, a 3-layer Recurrent Neural Network was used for 5-type classification and an 85\% of accuracy was performed. In~\cite{r17,r18,r19}, long short-term memory(LSTM) showed very promising result in arrhythmia detection and classification tasks.
Even though~\cite{r10},~\cite{r14},~\cite{r7},~\cite{r8} and~\cite{r12} showed very significant improvement in classification than using naked eyes,
they suffer from either patient-generation ability or adaptability to the real-world data.
In~\cite{r20} and~\cite{r21} ,
patient-specific ECG classifiers were built using transfer learning of 2-D convolutional neural networks.
A convolutional neural network was trained and used for feature extraction first.
Then an individual classifier was trained for each patient transferring the whole pre-trained model from the training set domain to a specific patient domain.
However,
these researches were all built on the MIT-BIH dataset,
which is a very classic public and wholesome dataset~\cite{r22}.
The data is clean and collected from a 24-hour monitoring Holter~\cite{r23,r24},
which has a significant difference from the data collected by an ambulatory IoT patch.
In many Smart Health applications involving ECG,
the collected data will not be as clean and high-quality as the MIT-BIH dataset and therefore requires extra data analysis.
\subsection{Noisy Label Training}
In the era of big data, obtaining large-scale dataset with high quality and precise labels is one of the most important procedures for most of data analysis related projects,
especially when data-driven methods such as machine learning and deep learning are applied.
However, annotating such a large dataset usually requires tremendous human resources and sometimes even requires domain knowledge.
As such, human mistakes occur very often due to the reasons mentioned above, which will lead to a dataset with a certain number of noisy labels.
To solve this problem, numerous studies are done in the field of noisy label training in the past several decades.
And here, we refer to this survey~\cite{01} as a comprehensive review.
Many recent studies have illustrate directly training the noisy labels will impair the training scheme and therefore negatively impact the classification accuracy~\cite{02}. To avoid this issue, there are 2 major types of noisy label training schemes: Noisy Model Based Methods and Noisy Model Free Methods.
\subsubsection{Noisy Model Based Methods}
Some approaches advocate to use noise model based methods which mainly aims to model the noise structure and extract the noise feature,
so that the information of the noise can be estimated and applied on the training scheme. The performance of this method usually depends on the accuracy of noisy estimation.
Hence, it's more suitable when prior knowledge about noise structure is known.
In~\cite{03, 04, 05, 07, 08}, noise channels were modelled at the training phase and the noise got removed at the evaluation phase. Even through good results were performed, the most significant problem in this method is the scale-ability. The calculation would become intractable when the number of class increases in classification task.
Some approaches attempted to identify and clean noisy labels by picking the suspicious samples and send them to human for further annotation~\cite{09, 10}. However, this could still be not feasible when the dataset is at a large size.
Other approaches including dataset pruning~\cite{11, 12,13}, sample choosing~\cite{14, 15,16,17}, sample importance weighting~\cite{18, 19,20,21} also show significant improvement on noisy data classification. However, all of these methods are highly dependent on the calculation and estimation of the noisy structure. As such, prior information tends to be essential in this case.
\subsubsection{Noisy Model Free Methods}
On the other hand, instead of building a noise structure model,
noisy model free methods tends to achieve label noise robustness.
These methods requires that the robustness of the classifier,
which cannot be too sensitive to noise.
Some methods aimed to design a robust loss function so that the noisy would not decrease the accuracy~\cite{22,23,24,25}. Many researchers proposed different type of loss function which showed good result, such as distance-based loss~\cite{26}, information-theoretic loss~\cite{27}, classification-calibrated loss functions~\cite{28}.
With the development of deep neural networks, researchers tend to transfer from hand-designed features to autonomously learned machine learning techniques. So does in noisy label training, a machine learning algorithm called Meta Learning is widely used to learn the high level features on their own~\cite{29,30,31}.
In addition, there are also some other techniques such as regularizer~\cite{32,33,34}, which was used to prevent overfitting from noisy labels. Besides, complementary labels~\cite{35} and prototype learning~\cite{36} were used for noisy label training as well.
\section{ECG Data Acquisition, Processing and Accuracy considerations}\label{s2}
\subsection{iRealCare ECG patch}
Our dataset was collected by a device named iRealCare~\cite{r27} as shown in Fig.\ref{fig_2}.
It's a portable ECG monitoring device with signals sampled at 250Hz.
The patch is lightweight with only 13 grams.
In addition,
the patch has the characteristic of being able to perform long-term continuous monitoring,
which can last 72 hours.
As such,
it has significant advantages over the traditional ECG Holter.
\begin{figure}[h]
\centering
\includegraphics[width=6cm]{patch.png}
\caption{iRealCare IoT patch}
\label{fig_2}
\end{figure}
\subsection{Dataset}
Our experiments were performed on an ECG signal dataset collected from 65 patients.
4,831,517 heartbeats were identified considering the position of QRS,
among which heartbeats were divided into 6 categories,
Normal Heartbeat(N); Ventricular premature beat (V); Supraventricular premature beat(S); Atrial fibrillation(A); Electromagnetic Interference(E) and Motion Interference(Q).
In order to construct a balanced dataset,
each category contains the same number of data points.
Therefore,
no bias will be introduced to the neural network during training because of the different numbers of training samples for each category.
In other words,
each category is equally represented by data samples,
which avoids the over-fitting problem for certain categories.
Table~\ref{tab_2} shows the number of six types of ECG signals.
\begin{table}[h]
\caption{Number for six types of ECG signals}
\begin{center}
\begin{tabular}{|c|c|c|}
\hline
Type&Meaning&number\\
\hline
N&Normal Heartbeat& 3,501,003\\
V&Ventricular premature beat& 194,469\\
S& Supraventricular premature beat&45,536\\
A& Atrial fibrillation&96,052\\
E& Electromagnetic Interference& 46,437\\
Q& Movement Interference& 148,020\\
\hline
\end{tabular}
\end{center}
\label{tab_2}
\end{table}
As mentioned before,
the dataset generated by the patch would be more noisier than the Holter due to the limitation of the IoT patch, which could generate a significant number of Electromagnetic Interference.
In addition,
a considerable amount of movement interference will be introduced when the patients wearing the patch every day.
Besides,
there might also be some errors in labelling as well.
A set of possible noisy labels are presented below:
For example,
Fig. \ref{fig_3} shows the standard waveform of type V premature ventricular contraction~\cite{r3}.
\begin{figure}[h]
\centering
\includegraphics[width=8cm]{typeV.png}
\caption{Standard waveform of type V}
\label{fig_3}
\end{figure}
In our dataset,
there are a few signals that have very similar features as shown in Fig. \ref{fig_4A}.
As such,
we would consider it as a "clean label".
However,
there are also a few "noisy labels" in this dataset,
such as the one shown in Fig.~\ref{fig_4B}.
It was labelled as type V,
but it's revised to be other type because it doesn't show the significant features of type V (Ventricular premature beat).
\begin{figure}[h]
\centering
\includegraphics[width=6cm]{clean_label.png}
\caption{Example of "Clean label" in the dataset}
\label{fig_4A}
\end{figure}
\begin{figure}[h]
\centering
\includegraphics[width=6cm]{noisy_label.png}
\caption{Example of "Noisy label" in the dataset}
\label{fig_4B}
\end{figure}
\subsection{Data Segmentation}
To identify the QRS position,
both low pass filtering and high pass filtering are used to mitigate the noise and locate the QRS.
Low pass filters can be used to cancel the high pass noise efficiently while the high pass filters can limit the interference of P and T wave. After filtering using low pass and high pass, we used hand-crafted features to identify the QRS wave respectively.
\subsection{Data Pre-processing}
Data normalization is an essential step before feeding data into neural network for training.
In deep learning,
data are usually normalized to a mean of close to zero,
which will usually speed up the learning process and result in a faster convergence.
The reasons are as follows.
We define the binary cross-entropy loss as,
\begin{equation}
L=Y\times log(W\cdot X )+(1-Y)\log(1-W\cdot X)
\end{equation}
where $Y$ is the training label vector, $X$ is the input vector and $W$ is the weight matrix.
If we take the partial derivative with respect to each weight, we have,
\begin{equation}
\frac{\partial L}{\partial w_i}=x_i\times (a-y)
\label{eq2}
\end{equation}
where $x_i$ is corresponding activation output of this weight, $a$ is the network output probability and $y$ is the training label.
Weights update according to the following,
\begin{equation}
w_i:= w_i - \alpha \times \frac{\partial L}{\partial w_i}
\label{eq3}
\end{equation}
where $\alpha$ is the learning rate.
Even though magnitudes are different,
weights will be updated with some dependence on each other.
And that usually indicates a slow convergence,
because independent weight updates ensure all possible combinations of weights,
which is crucial for approaching a local minimum.
Therefore,
in this work,
we normalized our data to the range of $-0.5$ to $0.5$ according to Eq. (\ref{eq4}),
which also makes sure the scale of our input data is consistent,
suppressing the outliers.
\begin{equation}
x_i=\frac{x_i-\frac{\left\{\max{\left(x_i\right)}+\min{\left(x_i\right)}\right\}}{2}}{\max{\left(x_i\right)}-\min{\left(x_i\right)}}
\label{eq4}
\end{equation}
\section{METHODOLOGY}\label{s3}
In this section,
we discuss the methodology of our work.
To implement this training scheme, there are four steps we need to do:\\
1. Find noisy signals by performing Confidence-level-based Learning;\\
2. Generate a complementary labels for noisy labels;\\
3. Perform negative learning by flipping the loss and gradient;\\
4. Change the outcome from "NOT Type X" into "Type X".
\subsection{Negative Learning and Confidence-level-based Learning}
\subsubsection{Positive Training}
Positive training is the traditional way of training a neural network,
where the neural network is trained to learn the distribution of a given dataset and corresponding labels~\cite{r28}.
For example,
for a neural network to learn what is an image of a cat,
we would "show" many pictures of cat and "tell" the neural network that "it's a cat" multiple times until the neural network can recognize cats on its own.
\subsubsection{Negative Training}
Negative training is the opposite of positive training.
Instead of telling the neural network "it is a cat",
we tell it "it is not a dog".
This idea was first proposed in~\cite{r29} and used for increasing the accuracy of the label and decreasing the noisy labels.
The labels used for negative training processes are called negative labels in this work.
\subsubsection{Example}
Usually,
negative labels are more likely to be correctly labelled compared to positive labels under the same circumstances.
This is because the accuracy of deciding whether an object belongs to a label is lower than deciding whether it doesn't belong to the label.
For example,
there are totally 6 types of animals in a given dataset.
A cat image is mislabelled as a dog image accidentally,
which might impair the training process.
However,
negative labelling process might give it a "not a monkey" label,
which is still more likely to be correct($P_{correct}$= 5/6).
\subsubsection{Generate Complementary Labels }
As such,
we can generate correct negative labels from positive labels even if it is not labelled correctly.
\begin{figure}[h]
\centering
\includegraphics[width=8cm]{PLvsNL.png}
\caption{Training process of Positive Learning and Negative Learning}`
\end{figure}
Therefore,
we can use a combination of positive training and negative training to improve the network performance when the training dataset is noisy.
The authors of~\cite{r28} showed the performance of this training system on an ImageNet dataset with randomly added Gaussian noise.
To implement negative learning,
\textbf{a complementary label $y'$ needs to be generated in the first place,
which means the input doesn't belong to the complementary label $y'$}.
For example, if the data was labelled as Type 1 and its confidence level is relatively low.
Then we will find its complementary label by randomly select a type from Type 2 - Type 6.
For instance, we select a Type 4 in this case.
And the probability of "this signal is NOT Type 4" is 5/6,
hence the probability of "this signal is Type 4" is 1/6.
The detailed algorithm of generating a complementary label and negative label is shown in Fig. \ref{fig_negtr}.
\begin{figure}[h]
\centering
\includegraphics[width=8cm]{algr1.png}
\caption{Algorithm of Negative Training}
\label{fig_negtr}
\end{figure}
\subsubsection{Negative Training implementation}
In the implementation,
\textbf{a negative version of the cross-entropy loss function is used, }
in which the output possibility is flipped.
For example,
for a specific type,
if the output possibility is $0.8$ in a regular cross-entropy loss measurement,
the flipped one will be $0.2$.
In implementation,
the "not monkey" training is realized by reverting the activation function at the output layer,
such that when the gradient is passed through layers,
the model will fit the opposite way as opposed to positive learning.
In more detail,
if we want to train a "not monkey" label,
we first calculate the output activation function for a label of "monkey",
which contains the information on how the weights affect the network output for a label of "monkey".
Then,
we take the negative of the output activation function,
which holds the information on how the weights affect the network output for a label of "not monkey".
Hence,
passing gradients calculated in this way will result in fitting the opposite way of "monkey",
i.e. "not monkey",
yield a negative learning method.
By doing this,
we can 'trick' the training process to negatively fit our negative labels.
Hence,
we can get our negative label $y\_$ here.
In the implementation,
we don't need to expand the labelling by considering other negative cases.
\textbf{The main difference between positive training and negative training is the loss function. }
Generally,
when training with positive learning,
the expression of cross entropy loss function is given by:
\begin{equation}
L(f,y) = - \sum_{k=1}^{n}{y_k\times \log\left(p_k\right)}
\end{equation}
where $L$ denotes the cross entropy loss,
$n$ is the number of class,
$y_k$ denotes current digit of the training label (which assumes to be one-hot encoding),
and $p_k$ denotes output probability of current class.
The equation measures the performance of a classification model whose output is a probability value between $0$ and $1$.
It calculates and accumulates the difference between predicted and actual labels.
Here,
we sum up the loss of each type by calculating the product of $y_k$ and $\log {(p_k)}$.
However,
negative training has exactly opposite loss function, where the output probability is reverted.
The expression of the cross entropy loss function is
\begin{equation}
L(f,y\_) = - \sum_{k=1}^{n}{(y\_)_k\times \log\left(1-p_k\right)}
\end{equation}
\subsection{ResNet and Residual block}
Residual block is the building block of our proposed model.
It contains a normal convolutional layer,
except for a residual connection from the input of the block to output of the block.
It was first proposed in~\cite{r30} as a solution for vanishing gradient when neural network goes deeper.
Similar to~\cite{r30},
let's consider $M(x)$ as the underlying mapping which the residual block is supposed to learn and $x$ denotes input to the residual block.
The idea of residual learning is that rather than learning the underlying mapping $M(x)$,
we learn the residual function $F(x)=M(x)-x$.
And the underlying mapping can be reconstructed by an identity connection $F(x)+x$,
which simply adds input to the output.
\begin{figure}[h]
\centering
\includegraphics[width=5cm]{residualNet.png}
\caption{Residual Block}
\label{fig_7}
\end{figure}
It has been proved that residual function F(x) is easier to learn and with extra information passed from input~\cite{r30},
the model can be trained deeper and deeper.
If the output dimension is different from the input dimension,
for example filter number changes,
a linear projection $W_s$ needs to perform at the identity connection as $F\left(x\right)+W_s\times x$ (where $W_s$ is a matrix with a dimension of a$\times$b), ensuring the identity connection's dimension matches the output dimension, such that we can perform an element-wise addition.
Fig. \ref{fig_7} shows the structure of our proposed residual block.
It consists of two 1-D convolution layers followed by Batch Normalization,
ReLu activation and Dropout which improve the convergence of the training process.
As can be seen,
an identity connection is established from the input to the output with a simple addition operation.
\section{Experimental Setup}\label{s4}
\subsection{Neural Network Backbone Model}
This backbone is a residual block based deep 1-D Convolution neural network which consists of 12 convolutional layers as shown in Fig. \ref{fig8}.
Residual blocks which consist of a novel residual connection from the input to the output propagating information to deeper layers are basic building blocks of this backbone model~\cite{r30}.
The number of filters in the residual block is double every four residual block and a max-pooling operation is performed on the residual connection such that the input is sub-sampled by half before adding to its output for every two residual blocks.
\begin{figure}[h]
\centering
\includegraphics[width=9cm]{network_structure.png}
\caption{Network Structure of Negative-ResNet}
\label{fig8}
\end{figure}
We have constructed a baseline model with the same structure but normal positive training scheme.
The result of baseline model is shown in \ref{fig9}. The highest accuracy achieved at convergence is $71\%$.
\begin{figure}[h]
\centering
\includegraphics[width=6cm]{baseline.png}
\caption{Baseline Model Training Accuracy and Loss}
\label{fig9}
\end{figure}
\subsection{Training Process}
The whole training process of the system is shown in Fig. \ref{fig10}.
The raw data is firstly feed into the CNN.
Later on,
the dataset is divided into clean data and noisy data based on the confidence level.
Here,
the confidence level refers to the output probability of the model.
If the output confidence level is higher than a certain threshold,
it is considered as clean data, vice versa.
The clean data (high confidence level) would be trained positively,
and the noisy data (low confidence level) would go through negative learning process.
Loss calculated from both positive and negative learning will be added together,
and then passed back for gradient descent.
Weights will be updated at the same time.
\begin{figure}[h]
\centering
\includegraphics[width=8cm]{training_process.png}
\caption{Training process of Negative-ResNet}
\label{fig10}
\end{figure}
Different confidence levels are tested in our experiments.
As can be seen from the results of Table \ref{tabIII},
the best result can be achieved when the confidence level is set to $80\%$,
which is around $10\%$ higher than the baseline model.
The confidence level controls how many data samples should be trained positively and how many data should be trained negatively.
Therefore,
it reflects the level of noisy labels contained in the dataset.
And the best confidence level varies depending on different datasets.
Note that originally there are six labels which consist of two interference labels: Type Q and Type E.
\begin{table}[h]
\caption{Accuracy of Negative-ResNet with different confidence level}
\begin{center}
\begin{tabular}{|c|c|}
\hline
Confidence Level&Accuracy \\
\hline
99\%&~83.0\%\\
90\%&~87.3\%\\
80\%&~90.7\%\\
70\%&~88.4\%\\
60\%&~87.2\%\\
50\%&~85.0\%\\
40\%&~83.1\%\\
30\%&~83.1\%\\
\hline
\end{tabular}
\end{center}
\label{tabIII}
\end{table}
As such,
the accuracy reaches $81\%$ when training with $80\%$ confidence level.
The accuracy and loss plot of training with $80\%$ of confidence level is shown \label{fig11}.
Note: The confidence level for Type A is set to be 50\% such that few Type A sample will go through the Negative Learning. It is because Type A in our dataset is confirmed by our medical expert to be of high labeling correctness.
\begin{figure}[h]
\centering
\includegraphics[width=6cm]{conf_level80.png}
\caption{Training Accuracy and Loss Plot of 80\% Confidence Level}
\label{fig11}
\end{figure}
\section{ASSESSMENT CRITERION AND RESULT ANALYSIS}
\label{s5}
In this work,
we use accuracy,
precision,
and recall to evaluate the performance of our classifier.
These are the commonly used and standardized parameters for machine learning and deep learning.
The performance of the model can be determined by these parameters since it can not only show how precise the model is,
but also whether the model is biased.
In the confusion matrix below,
Type Q and Type E are combined as Type Q.
The definitions of these parameters are listed below,
where TP stands for True Positive,
FP stands for False Positive,
and FN stands for False Negative:
\begin{equation}
precision=\frac{TP}{TP+FP};
\end{equation}
\begin{equation}
recall=\frac{TP}{TP+FN};
\end{equation}
\begin{equation}
F1 score=2 * \frac{precision \times recall}{precision + recall}
\end{equation}
Here,
we also used $K$-Fold cross-validation to validate our results.
$N$ is chosen to be two.
In more detail,
two patients will be selected randomly and the remaining patients will be used as training data.
\subsection{Baseline Model}
The confusion matrix for the baseline model is shown in Fig. \ref{fig12}.
The precision and recall of the baseline model are shown in Table \ref{tab4}.
\begin{figure}[h]
\centering
\includegraphics[width=7cm]{confusionBaseline.png}
\caption{Confusion Matrix of Baseline Model}
\label{fig12}
\end{figure}
\begin{table}[h]
\caption{Precision and Recall for Baseline Model}
\begin{center}
\begin{tabular}{|c|c|c|c|c|c|}
\hline
&N&V&S&Q&A\\
\hline
mean Precision&0.80&0.87&0.79&0.92& 0.81\\
mean Recall&0.81& 0.88& 0.91&0.81& 0.87\\
mean F1 score&0.80&0.88&0.85&0.86&0.84\\
\hline
\end{tabular}
\end{center}
\label{tab4}
\end{table}
\subsection{Negative ResNet}
The confusion matrix,
precision and recall of optimal Negative ResNet are shown in Fig. \ref{fig13} and Table \ref{tab5}, respectively.
\begin{figure}[h]
\centering
\includegraphics[width=8cm]{confusionNL.png}
\caption{Confusion Matrix of Negative ResNet}
\label{fig13}
\end{figure}
\begin{table}[h]
\caption{Precision and Recall for Negative ResNet}
\begin{center}
\begin{tabular}{|c|c|c|c|c|c|}
\hline
&N&V&S&Q&A\\
\hline
mean Precision&0.82 & 0.95& 0.92& 0.95& 0.88\\
mean Recall&0.86& 0.94 &0.95& 0.92& 0.87\\
mean F1 score&0.84&0.94&0.93&0.94&0.87\\
\hline
\end{tabular}
\end{center}
\label{tab5}
\end{table}
Generally speaking,
the result of negative training is $4-11\%$ better than the normal training process,
especially for interference,
where an increment of $11\%$ accuracy can be achieved.
However, for type A,
the accuracy remains the same when going through negative training.
This is mainly because for type A (Atrial fibrillation),
the original labelling process is relatively accurate.
As such, using negative learning in this case will not have a significant improvement in the accuracy of type A.
It proves our idea that negative learning performs better in a noisy dataset.
\section{CONCLUSION AND FUTURE DIRECTION }
\label{s6}
This work introduced a single-lead ambulatory ECG patch iRealCare,
which was used to construct a large real-patient dataset.
In addition, a residual block based neural network with a novel negative training scheme was proposed to train an ECG classifier from this dataset.
Even with a considerable amount of erroneous labels in the dataset,
our proposed training scheme showed a satisfactory and better result compared to the baseline model, a 6.2\% of increase in accuracy was observed and the overall accuracy reached 91.0\%.
Our proposed training scheme is useful in reducing the reliance on quality of the training dataset, such that the training performance can be improved and it can be extended to more applications.
Further down this research, we believe that neural network and our proposed training scheme can not only applied into the cardiology diagnosis process, it can improved other medical diagnosis processes in a similar manner as well. We will keep on exploring towards this direction and hopefully applying it on other biomedical signal processing field.
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 1,732
|
McDowell Incentives, Inc. - Incentives, Recognition, Promotions | What Does Every Most Admired Global Company Have In Common?
I have traveled almost 4 million miles working with organizations and their leaders, helping them create cultures that produce exceptional employee and customer experiences.
We have done implementations in over 50 countries and 15 languages.
The one thing each and every one of these most admired companies shares in common is they provide skills that make their associates more marketable.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 8,640
|
Adé releases debut EP 'Always Something'
Adé, formerly known as Phil Ade, is a DMV artist who started out releasing mixtapes and collaborating with rap stars like Wale, Mac Miller, Raekwon, Logic, and countless others. After a few quiet years of working on music, he says he has finally found his sound.
The Silver Spring, Maryland rapper just released his debut EP, Always Something. With each track's title containing the word "something," he said that the project is about there being Always Something to remain positive about. It's an excellently conceptual body of work that not only embraces positivity, but sounds great doing it. Alongside features from Wale, GoldLink, Fatman Scoop, and Lil Baby, Adé crafts six songs that each stand strongly on their own but come together to form one of the better debuts of the past few years. It's evident he's spent thousands of hours developing his sound; he sounds confident and comfortable.
He raps with a slick style and special rawness that make the GoldLink and Wale features make perfect sense. Still, his song with Lil Baby is without a doubt a highlight from Always Something. "Something New" sees Adé switch between choppy and smooth, melodic raps before Lil Baby comes in with a more laid back verse. The EP is stacked with honest, exciting rhymes that pair perfectly with the often head-nodding, quirky production. Check out Adé's first release with Epic Records, Always Something, up above.
Connect with Adé: Twitter | SoundCloud | Facebook | Instagram
Phil Ade
Send me the best music you heard today.
Ben Sterling brings the 'Energy' in with Hot Creations debut
Yoste delivers a beautifully nostalgic EP 'try to be okay'
Lindsey Oh
MOAN finds "Gratitude" in sobriety [Video]
Kae Kuzi shows us his life as the "Local Tyrant"
Saba taps Krayzie Bone of Bone Thugs-N-Harmony out for his new single, "Come My Way"
Atlanta rapper $ODAMAN returns with his high-powered new visual for "ALOT X3" [Video]
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 7,566
|
We are so privileged that Puffball Mushrooms are growing in our friend's garden across the road, and even more privileged that they are prepared to to share it with us for a MUSHROOM FEAST every year. So it was with great excitement that they called and announced that the puffballs are up. We went to take photographs today and will harvest them tomorrow for the feast in the evening. We shall report on recipes and results the day after tomorrow!!! How good can life be?
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 3,518
|
\section{A brief Reminder of Association}
We then may begin to introduce to the associated random variables concept
which goes back to Lehmann (1966) \cite{lehmann} in the bivariate case.
Notice that we will lessen the notation by putting $k(n)=k$ in the sequel
\newline
\noindent The concept of association for random variables generalizes that
of independence and seems to model a great variety of stochastic models
\newline
\noindent This property also arises in Physics, and is quoted under the name
of FKG property (Fortuin, Kastelyn et Ginibre (1971) \cite{fortuin}), in
percolation theory and even in Finance (see Pan Jiazhu \cite{jiazhu})
\newline
\noindent The definite definition is given by Esary, Proschan et Walkup
(1967) \cite{esary} as follows.
\begin{definition}
A finite sequence of rv's $(X_{1},...,X_{n})$ are associated when for any
couple of real and coordinate-wise non-decreasing functions $h$ and $g$
defined on $\mathbb{R}^{n}$, we have
\begin{equation}
Cov(h(X_{1},...,X_{n}),\ \ g(X_{1},...,X_{n}))\geq 0
\end{equation
An infinite sequence of rv's are associated whenever all its finite
subsequences are associated.
\end{definition}
We have a few number of interesting properties to be found in Rao (\cite{rao
) :
\bigskip
\noindent \textbf{(P1)} A sequence of independent rv's is associated.\newline
\noindent \textbf{(P2)} Partial sums of associated rv's are associated
\newline
\noindent \textbf{(P3)} Order statistics of independent rv's are associated
\newline
\noindent \textbf{(P4)} Non-decreasing functions and non-increasing
functions of associated variables are associated.\newline
\noindent \textbf{(P5)} Let the sequence $Z_{1},Z_{2},...,Z_{n}$ be
associated and let $(a_{i})_{1\leq i\leq n}$ be positive numbers and
(b_{i})_{1\leq i\leq n}$ real numbers. Then the \textit{rv}'s
a_{i}(Z_{i}-b_{i})$ are associated.\newline
\noindent As immediate other examples of associated sequences, we may cite
Gaussian random vectors with nonnegatively correlated components (see Pitt
\cite{pitt}) and a homogenuous Markov chain is also associated (Daley \cit
{daley}).\newline
\noindent Demimartingales are set from associated centered variables exactly
as martingales are derived from partial sums of centered independent random
variables. We have
\begin{definition}
A sequence of rv's $\{S_{n},n\geq 1\}$ \ in $L^{1}(\Omega ,\mathcal{A}
\mathbb{P})$ \ is a demimartingale when for any $j\geq 1$, for any
coordinatewise nondecreasing function $g$ \ defined on $\mathbb{R}^{j}$, we
have
\begin{equation}
\mathbb{E}\bigl((S_{j+1}-S_{j})\ g(S_{1},...,S_{j})\bigr)\geq 0,\ \ j\geq 1.
\label{defmarting}
\end{equation}
\end{definition}
\bigskip Two particular cases should be highlighted. First any martingale is
a demimartingale. Secondly, partial sums $S_{0}=0$, $S_{n}=X_{1}+...+X_{n}$,
$n\geq 1$, of associated and centered random variables $X_{1},X_{2},...$
form a demimartingale for, in this case, (\ref{defmarting}) becomes :
\begin{equation*}
\mathbb{E}\left\{ (S_{j+1}-S_{j})\ g(S_{1},...,S_{j})\right\} =\mathbb{E
\left\{ X_{j+1}\ g(S_{1},...,S_{j})\right\} =Cov\left\{
X_{j+1},g(S_{1},...,S_{j})\right\} ,
\end{equation*
since $\mathbb{E}X_{j+1}=0$. Since $(x_{1},...,x_{j+1})\longmapsto x_{j+1}$
et $(x_{1},...,x_{j+1})\longmapsto g(x_{1},...,x_{j})$ are coordinate-wise
nondecreasing functions and since the $X_{1},X_{2},..$ are associated, we ge
\begin{equation*}
\mathbb{E}\left\{ (S_{j+1}-S_{j})\ g(S_{1},...,S_{j})\right\} =Cov\left\{
X_{j+1}\ g(S_{1},...,S_{j})\right\} \geq 0.
\end{equation*}
\bigskip
\bigskip
\section{Key results for associated sequences}
\begin{lemma}
\label{lemg1} Let $(X,Y)$ be a bivariate random vector such that $\mathbb{E
(X^{2})<\infty $ and $\mathbb{E}(Y^{2})<\infty .$ If $\left(
X_{1},Y_{1}\right) $ and $\left( X_{2},Y_{2}\right) $ are two independent
copies of $(X,Y),$ then We have
\begin{equation*}
2Cov(X,Y)=\mathbb{E}(X_{1}-X_{2})(Y_{1}-Y_{2}).
\end{equation*
We also hav
\begin{equation*}
Cov(X,Y)=\int_{-\infty }^{+\infty }\int_{-\infty }^{+\infty }H(x,y)dxdy,
\end{equation*}
\end{lemma}
\noindent where
\begin{equation*}
H(x,y)=\mathbb{P}(X>x,Y>y)-\mathbb{P}(X>x)\mathbb{P}(Y>y).
\end{equation*}
\bigskip
\noindent Before the proof of the lemma, we observe that :
\begin{equation}
H(x,y)=\mathbb{P}(X>x,Y>y)-\mathbb{P}(X>x)\mathbb{P}(Y>y)=\mathbb{P}(X\leq
x,Y\leq y)-\mathbb{P}(X\leq x)\mathbb{P}(Y\leq y). \label{equiv}
\end{equation}
Indeed we have
\begin{equation*}
\mathbb{P}(X>x,Y>y)-\mathbb{P}(X>x)\mathbb{P}(Y>y)=\mathbb{E}(\mathbb{I
_{\left( X>x\right) }\mathbb{I}_{\left( Y>y\right) })-\mathbb{E}(\mathbb{I
_{\left( X>x\right) })\mathbb{E}(\mathbb{I}_{\left( Y>y\right) })
\end{equation*
\begin{equation*}
=Cov(\mathbb{I}_{\left( X>x\right) },\mathbb{I}_{\left( Y>y\right) })
\end{equation*
\begin{equation*}
=Cov(1-\mathbb{I}_{\left( X>x\right) },1-\mathbb{I}_{\left( Y>y\right) })
\end{equation*
\begin{equation*}
=Cov(\mathbb{I}_{\left( X\leq x\right) },\mathbb{I}_{\left( Y\leq y\right) })
\end{equation*
\begin{equation*}
=\mathbb{P}(X\leq x,Y\leq y)-\mathbb{P}(X\leq x)\mathbb{P}(Y\leq y).
\end{equation*}
\bigskip
\noindent \textbf{Proof. }We have
\begin{equation*}
\mathbb{E}(X_{1}-X_{2})(Y_{1}-Y_{2})=\mathbb{E}(X_{1}Y_{1})-\mathbb{E}(X_{1}
\mathbb{E}(Y_{2})-\mathbb{E}(X_{2})\mathbb{E}(Y_{1})+\mathbb{E}(X_{2}Y_{2})
\end{equation*
\begin{equation*}
=2\mathbb{E}(X_{1}Y_{1})-2\mathbb{E}(X_{1})\mathbb{E}(Y_{1})
\end{equation*
\begin{equation*}
=2Cov(X_{1},Y_{1}).
\end{equation*}
\bigskip \noindent Next, for $a\in \mathbb{R},$by Fubini's Theorem for
nonegative random variables,
\begin{equation*}
\int_{a}^{\infty }\int_{a}^{\infty }\mathbb{P}(X>x,Y>y)dxdy=\mathbb{E
\int_{a}^{\infty }\int_{a}^{\infty }\mathbb{I}_{\left( X>x\right) }\mathbb{I
_{\left( Y>y\right) }dxdy
\end{equation*
\begin{equation*}
=\mathbb{E}(\int_{a}^{X}dx\int_{a}^{Y}dy)
\end{equation*
\begin{equation*}
=\mathbb{E}[(X-a)(Y-a)].
\end{equation*}
\bigskip \noindent \noindent We have
\begin{equation*}
2Cov(X_{1},Y_{1})=\mathbb{E}(X_{1}-X_{2})(Y_{1}-Y_{2})
\end{equation*
\begin{equation*}
=\mathbb{E}(\left\{ (X_{1}-a)-(X_{2}-a)\right\} \left\{
(Y_{1}-a)-(Y_{2}-a)\right\} ))
\end{equation*
\begin{equation*}
=\mathbb{E}(X_{1}-a)(Y_{1}-a)-\mathbb{E}(X_{1}-a)(Y_{2}-a)
\end{equation*
\begin{equation*}
-\mathbb{E}(X_{2}-a)(Y_{1}-a)+\mathbb{E}(X_{2}-a)(Y_{2}-a)
\end{equation*
\begin{equation*}
=\int_{a}^{\infty }\int_{a}^{\infty }\mathbb{P}(X_{1}>x,Y_{1}>y)dxdy
\int_{a}^{\infty }\int_{a}^{\infty }\mathbb{P}(X_{1}>x,Y_{2}>y)dxdy
\end{equation*
\begin{equation*}
-\int_{a}^{\infty }\int_{a}^{\infty }\mathbb{P}(X_{2}>x,Y_{1}>y)dxdy
\int_{a}^{\infty }\int_{a}^{\infty }\mathbb{P}(X_{2}>x,Y_{2}>y)dxdy.
\end{equation*}
\bigskip \noindent By the independence of $\{X_{1},Y_{1}\}$ and
\{X_{2},Y_{2}\},$ $\mathbb{P}(X_{1}>x,Y_{2}>y)=\mathbb{P}(X_{1}>x)\times
\mathbb{P}(Y_{1}>y)$ and $\mathbb{P}(X_{2}>x,Y_{1}>y)=\mathbb{P
(X_{1}>x)\times \mathbb{P}(Y_{1}>y),
\begin{equation*}
2Cov(X,Y)=2\left( \int_{a}^{\infty }\int_{a}^{\infty }\left\{ \mathbb{P
(X_{1}>x,Y_{1}>y)-\mathbb{P}(X_{1}>x)\times \mathbb{P}(Y_{1}>y)\right\}
dxdy\right) .
\end{equation*}
\bigskip
\noindent We get the final result by letting $a\rightarrow -\infty .$
\begin{lemma}
\label{lemg2} Suppose that $X$, $Y$ are two random variables with finite
variance and, $f$ and $g$ are $C^{1}$ complex valued functions on $\mathbb{R
^{1}$ with bounded derivatives $f^{\prime }$ and $g^{\prime }.$ Then
\begin{equation*}
|Cov(f(X),h(Y))|\leq ||f^{\prime }||_{\infty }||g^{\prime }||_{\infty
}Cov(X,Y)
\end{equation*}
\end{lemma}
\bigskip
\noindent \textbf{Proof.} By Lemma \ref{lemg1}, we have
\begin{equation*}
2Cov(f(X),g(Y))=\mathbb{E}(f(X_{1})-f(X_{2}))(g(Y_{1})-g(Y_{2}))
\end{equation*
\begin{equation*}
=\mathbb{E}\left( \int_{X_{1}}^{X_{2}}f^{\prime
}(x)dx\int_{Y_{1}}^{Y_{2}}g^{\prime }(x)dx\right) .
\end{equation*
Bu
\begin{equation*}
\int_{X_{1}}^{X_{2}}f^{\prime }(x)dx=\int_{X_{1}}^{+\infty }f^{\prime
}(x)dx-\int_{X_{2}}^{+\infty }f^{\prime }(x)dx
\end{equation*
\begin{equation*}
=\int_{\mathbb{R}}f^{\prime }(x)\left\{ 1_{(X_{1}\leq x)}-1_{(X_{2}\leq
x)}\right\} dx
\end{equation*
Applying this to $\int_{Y_{1}}^{Y_{2}}g^{\prime }(x)dx$ and combining all
that, leads t
\begin{equation}
2Cov(f(X),g(Y))=\mathbb{E}\int_{\mathbb{R}^{2}}f^{\prime }(x)g^{\prime }(y)\left\{
1_{(X_{1}\leq x)}-1_{(X_{2}\leq x)}\right\} \left\{ 1_{(Y_{1}\leq
y)}-1_{(Y_{2}\leq y)}\right\} dxdy. \label{int1}
\end{equation
It is easy to see tha
\begin{equation*}
\mathbb{E}\left\{ 1_{(X_{1}\leq x)}-1_{(X_{2}\leq x)}\right\} \left\{
1_{(Y_{1}\leq y)}-1_{(Y_{2}\leq y)}\right\}
\end{equation*
\begin{equation*}
=2(\mathbb{P}(X\leq x,Y\leq y)-\mathbb{P}(X\leq x)\mathbb{P}(Y\leq y))
\end{equation*
and by (\ref{equiv}), this is equal to $2H(x,y).$ By applying Fubini's
theorem in (\ref{int1}), we ge
\begin{equation*}
2Cov(f(X),g(Y))=2\int_{\mathbb{R}^{2}}f^{\prime }(x)g^{\prime }(y)H(x,y)dxdy.
\end{equation*
This gives, since $H(x,y)\geq 0$ for associated $rv$'s,
\begin{equation*}
\left\vert Cov(f(X),g(Y))\right\vert \leq ||f^{\prime }||_{\infty
}||g^{\prime }||_{\infty }\int_{\mathbb{R}^{2}}H(x,y)dxdy.
\end{equation*
And we complete the proof by applying Lemma \ref{lemg1}.
\bigskip
\noindent \textbf{Remark }: We used the proof of Yu(1993) here.
\begin{theorem}
\label{theo1} \label{theog2} Let $X_{1},X_{2},...,X_{n}$ be associated, then
we have for all $t=(t_{1},...,t_{n})\in \mathbb{R}^{k},
\begin{equation}
\left\vert \psi _{_{(X_{1},X_{2},...,X_{n})}}(t)-\prod\limits_{i=1}^{n}\psi
_{_{X_{i}}}(t_{i})\right\vert \leq \frac{1}{2} \sum_{1\leq i\neq j\leq
n}\left\vert t_{i}t_{j}\right\vert \left\vert Cov(X_{i},X_{j})\right\vert .
\label{decomp}
\end{equation}
\end{theorem}
\bigskip
\bigskip \noindent \textbf{Proof :} First, we prove this for $n=2.$ Use the
Newman inequality in Lemma \ref{lemg2}. Let $X$ and $Y$ be two associated
random variables. For $(s,t)\in R^{2},$ put $U=f(X)=:e^{isX}$ and
V=g(Y)=:e^{itY}.$ We hav
\begin{equation*}
Cov(U,V)=E(e^{(isX+tY)})-E(e^{isX})E(e^{itY})=\psi _{(X,Y)}(s,t)-\psi
_{X}(s)\psi _{Y}(t).
\end{equation*}
\bigskip \noindent But Lemma \ref{lemg2} implies
\begin{equation*}
\left\vert Cov(U,V)\right\vert =\left\vert Cov(f(X),g(Y))\right\vert \leq
\left\vert st\right\vert \left\Vert f^{\prime }\right\Vert_{\infty}
\left\Vert g^{\prime }\right\Vert \left\vert_{\infty} Cov(X,Y)\right\vert
=\left\vert st\right\vert \left\vert Cov(X,Y)\right\vert .
\end{equation*}
\begin{equation*}
=\frac{1}{2} \left\vert st\right\vert \left\vert (Cov(X,Y)+cov(Y,X))
\right\vert .
\end{equation*}
\bigskip \noindent And ($\ref{decomp}$) is valid for $n=2.$ Now we proceed
by induction and suppose that \ref{decomp} is true up to $n.$ Consider
associated random variables $X_{1},X_{2},...,X_{n+1}$ and let
t=(t_{1},...,t_{n+1})\in R^{n+1}.$ If all the $t_{i}$ are nonnegative, we
have $U=t_{1}X_{1}+...+t_{n}X_{n}$ and $V=X_{n+1}$ are associated. We hav
\begin{equation*}
\psi _{_{(X_{1},X_{2},...,X_{n}+1)}}(t)=\psi _{_{(U,V)}}(1,t_{n+1})\text{
and \ }\psi _{_{U}}(1)=\psi _{_{(X_{1},X_{2},...,X_{n})}}(t_{1},...,t_{n}).
\end{equation*}
\bigskip \noindent By the induction hypothesis, we hav
\begin{equation}
\left\vert \psi _{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\psi
_{_{(X_{1},X_{2},...,X_{n})}}(t_{1},...,t_{n})\psi
_{X_{n+1}}(t_{n+1})\right\vert \label{d2}
\end{equation
\begin{equation*}
\leq \left\vert t_{n+1}\right\vert \left\vert
cov(X_{n+1},t_{1}X_{1}+...+t_{n}X_{n})\right\vert
\end{equation*
\begin{equation*}
\leq \frac{1}{2} \sum_{j=1}^{n}\left\vert t_{i}t_{n+1}\right\vert \left\vert
cov(X_{n+1},X_{i})\right\vert .
\end{equation*}
\bigskip \noindent Next
\begin{equation*}
\left\vert \psi
_{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\prod\limits_{i=1}^{n+1}\psi
_{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation*}
\leq \left\vert \psi _{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\psi
_{_{(X_{1},X_{2},...,X_{n})}}(t_{1},...,t_{n})\psi
_{X_{n+1}}(t_{n+1})\right\vert
\end{equation*
\begin{equation*}
+\left\vert \psi _{_{(X_{1},X_{2},...,X_{n})}}(t_{1},...,t_{n})\psi
_{X_{n+1}}(t_{n+1})-\prod\limits_{i=1}^{n+1}\psi
_{_{X_{i}}}(t_{i})\right\vert .
\end{equation*}
\bigskip \noindent The first term in the right side member is bounded as in
\ref{d2}). The second term is bounded, due to the induction hypothesis, b
\begin{equation*}
\left\vert \psi _{X_{n+1}}(t_{n+1})\right\vert \left\vert \psi
_{_{(X_{1},X_{2},...,X_{n})}}(t_{1},...,t_{n})-\prod\limits_{i=1}^{n}\psi
_{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation*}
=\left\vert \psi
_{_{(X_{1},X_{2},...,X_{n})}}(t_{1},...,t_{n})-\prod\limits_{i=1}^{n}\psi
_{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation}
\leq \frac{1}{2} \sum_{1\leq i\neq j\leq n}\left\vert t_{i}t_{j}\right\vert
\left\vert cov(X_{i},X_{j})\right\vert . \label{d3}
\end{equation}
\bigskip \noindent By putting (\ref{d2}) and (\ref{d3}) together, we get
that (\ref{decomp}) is valid. By re-arranging the $t_{i},$ we observe that
we have proved( \ref{decomp}) for $n=3$. if at least $n$ of the $t_{i}$ are
nonnegative. Also, if at least $n$ of them are nonpositive, we consider the
sequence $-X_{1},...,-X_{n+1}$ that is also associated and get the same
conclusion. This means that (\ref{decomp}) is true. It remains the case
where exactely $p$ of the $t_{i}$ are nonnegative with $2\leq p\leq n-2.$ By
re-arranging the $t_{i}$ if necessary, we may consider that $t_{i}\geq 0$
for $1\leq i\leq p$ and $t_{i}<0$ for $i>p.$ Now, by putting
U=t_{1}X_{1}+...+t_{p}X_{p}$ and $U=t_{p+1}X_{p+1}+...+t_{n+1}X_{n+1}.$
Since $U$ et $-V$ are associated and sinc
\begin{equation*}
\psi _{_{(X_{1},X_{2},...,X_{n}+1)}}(t)=\psi _{(U,-V)}(1,-1),
\end{equation*}
\bigskip \noindent we have by the induction hypothesi
\begin{equation}
\left\vert \psi _{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\psi _{U}(1)\psi
_{-V}(-1)\right\vert \leq \frac{1}{2} \left\vert Cov(U,-V)\right\vert \leq
\frac{1}{2} \sum_{i=1}^{p}\sum_{j=p+1}^{n+1} \left\vert t_i t_j\right\vert
\left\vert cov(X_{i},X_{j})\right\vert \label{d3a}
\end{equation}
\bigskip \noindent Now use
\begin{equation}
\left\vert \psi
_{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\prod\limits_{i=1}^{n+1}\psi
_{_{X_{i}}}(t_{i})\right\vert \leq \left\vert \psi
_{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\psi _{U}(1)\psi _{-V}(-1)\right\vert
\label{d4}
\end{equation
\begin{equation*}
+\left\vert \psi _{U}(1)\psi _{-V}(-1)-\psi
_{U}(1)\prod\limits_{i=p+1}^{n+1}\psi _{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation*}
\leq \left\vert \psi _{_{(X_{1},X_{2},...,X_{n}+1)}}(t)-\psi _{U}(1)\psi
_{-V}(-1)\right\vert +\left\vert \psi _{U}(1)\psi
_{-V}(-1)-\prod\limits_{i=1}^{p}\psi _{_{X_{i}}}(t_{i})\psi
_{-V}(-1)(t_{i})\right\vert
\end{equation*
\begin{equation*}
+\left\vert \prod\limits_{i=1}^{p}\psi _{_{X_{i}}}(t_{i})\psi
_{-V}(-1)-\prod\limits_{i=1}^{n+1}\psi _{_{X_{i}}}(t_{i})\right\vert
\end{equation*}
\bigskip \noindent The first term already handled in (\ref{d4}). The second
term is bounded as follow
\begin{equation*}
\left\vert \psi _{U}(1)\psi _{-V}(-1)-\prod\limits_{i=1}^{p}\psi
_{_{X_{i}}}(t_{i})\psi _{-V}(-1)(t_{i})\right\vert =\left\vert \psi
_{-V}(-1)(t_{i})\right\vert \times \left\vert \psi
_{U}(1)-\prod\limits_{i=1}^{p}\psi _{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation*}
\leq \left\vert \psi _{U}(1)-\prod\limits_{i=1}^{p}\psi
_{_{X_{i}}}(t_{i})\right\vert =\left\vert \psi
_{_{(X_{1},X_{2},...,Xp)}}(t_{1},...,t_{p})-\prod\limits_{i=1}^{p}\psi
_{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation}
\leq \frac{1}{2} \sum_{1\leq i\neq j\leq p}^{p} \left\vert t_i t_j
\right\vert \left\vert cov(X_{i},X_{j})\right\vert. \label{d5}
\end{equation}
\bigskip \noindent where we used the induction hypothesis in the last
formula. The last term i
\begin{equation*}
\left\vert \prod\limits_{i=1}^{p}\psi _{_{X_{i}}}(t_{i})\psi
_{-V}(-1)-\prod\limits_{i=1}^{n+1}\psi _{_{X_{i}}}(t_{i})\right\vert
=\left\vert \prod\limits_{i=1}^{p}\psi _{_{X_{i}}}(t_{i})\psi
_{-V}(-1)-\prod\limits_{i=1}^{n+1}\psi _{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation*}
\leq \left\vert \prod\limits_{i=1}^{p}\psi _{_{X_{i}}}(t_{i})\right\vert
\times \left\vert \psi _{-V}(-1)-\prod\limits_{i=p+1}^{n+1}\psi
_{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation*}
\leq \left\vert \psi
_{_{(X_{p+1},...,X_{n+1})}}(t_{p+1},...,t_{n+1})-\prod\limits_{i=p+1}^{n+1
\psi _{_{X_{i}}}(t_{i})\right\vert
\end{equation*
\begin{equation}
\leq \frac{1}{2} \sum_{p+1\leq i\neq j\leq n+1} \left\vert t_i
t_j\right\vert \left\vert cov(X_{i},X_{j})\right\vert, \label{d6}
\end{equation}
\bigskip \noindent where we used again the induction hypothesis. We complete
the proof by putting (\ref{d3a}), (\ref{d4}), (\ref{d6}) and (\ref{d5})
together, we arrive at the result (\ref{decomp}).
\section{Central limit theorem for a stricly stationary and associated
sequence}
\noindent In this section, we provide all the details of the sharpest result
in this topic by Newman and Wright \cite{newmanwright}. This came as a
concluding paper for a series of papers by Newman.\newline
\noindent We present here all the materials used in the proof of Newman and
Wright in a detailed writing that makes it better understandable by a broad
public.\newline
\noindent First, we have this simple lemma.
\begin{lemma}
\label{lemg3} Let $X$ and $Y$ be finite variance random variables such that
\begin{equation}
E(X,Y1_{(Y\leq 0)})\geq 0. \label{c1}
\end{equation
Then, we have
\begin{equation}
\mathbb{E}[(\max (X,X+Y))^{2}]\leq E(X+Y)^{2}. \label{c2}
\end{equation}
\end{lemma}
\bigskip \noindent If $X$ and $Y$ are associated and $X$ is mean zero, then
\ref{c1}) holds and (\ref{c2}) is true.\newline
\noindent \textbf{Proof.} We hav
\begin{equation*}
\max (X,X+Y)^{2}=\left\{ X1_{(Y\leq 0)}+(X+Y)1_{(Y>0)}\right\} ^{2}
\end{equation*
\begin{equation*}
=X^{2}1_{(Y\leq 0)}+(X+Y)^{2}1_{(Y>0)}=X^{2}1_{(Y\leq
0)}+(X^{2}+Y^{2}+2XY)1_{(Y>0)}
\end{equation*
\begin{equation*}
=X^{2}+Y^{2}-Y^{2}1_{(Y\leq 0)}+2(XY)1_{(Y>0)}
\end{equation*}
\qquad
\begin{equation*}
=X^{2}+Y^{2}+2XY-2XY1_{(Y\leq 0)}-Y^{2}1_{(Y\leq 0)}
\end{equation*}
\begin{equation*}
=\left( X+Y\right) ^{2}-2XY1_{(Y\leq 0)}-Y^{2}1_{(Y\leq 0)}
\end{equation*}
\bigskip \noindent We get the desired result wheneve
\begin{equation*}
E(XY1_{(Y\leq 0)})=Cov(X,Y1_{(Y\leq 0)})\geq 0
\end{equation*
Now if $X$ \ and $Y$ are associated, we hav
\begin{equation*}
XY1_{(Y\leq 0)}=(-X)(-Y)1_{(-Y\geq 0)}.
\end{equation*}
\bigskip \noindent Since $(-X)$ and $(-Y)$ are associated too and
1_{(-Y\geq 0)}$ is a nondecreasing function of $(-Y)$, and reminding that $X$
is mean zero, we get that
\begin{equation*}
E(XY1_{(Y\leq 0)})=E((-X)(-Y)1_{(-Y\geq 0)})=Cov((-X),(-Y)1_{(-Y\geq
0)})\geq 0.
\end{equation*}
\begin{theorem}[Maximal inequality of Newman and Wright]
\label{theo2} Let $X_{1},X_{2},\cdots ,X_{n}$ be associated, mean zero,
finite variance, random variables and $M_{n}=\max (S_{1},S_{2},\cdots
,S_{n}) $ where $S_{n}=X_{1}+X_{2}+\cdots +X_{n}$, we have
\begin{equation}
\mathbb{E}(M_{n}^{2})\leq V(S_{n}). \label{nw}
\end{equation}
\end{theorem}
\noindent \textbf{Proof.} Let us prove (\ref{nw}) by induction. It is
obviously true \ for $n=1$ and for $n=2$ by Lemma \ref{lemg3}. Let us
suppose that it is true for $j,2\leq j<n.$ By putting
L_{j}=X_{2}+...+X_{j}, $ $j\geq 2,$we have
\begin{equation*}
M_{n}=\max (X_{1},X_{1}+L_{2},...,X_{1}+L_{n})=X_{1}+\max(0,L_{2},...,L_{n}).
\end{equation*
Bu
\begin{equation*}
\max (X_{1},X_{1}+\max (L_{2},...,L_{n}))=X_{1}+\max (0,\max
(L_{2},...,L_{n}))
\end{equation*
We obviously hav
\begin{equation*}
\max (0,\max (L_{2},...,L_{n}))=\max (0,L_{2},...,L_{n}).
\end{equation*}
\bigskip \noindent Then
\begin{equation*}
\mathbb{E}M_{n}^{2}=E\max (X_{1},X_{1}+\max (L_{2},...,L_{n}))^{2}
\end{equation*}
\bigskip \noindent Since $X_{1}$ and $\max (L_{2},...,L_{n})$ are associated
and $X_{1}$ is mean zero, then use Lemma \ref{lemg3} to ge
\begin{equation*}
EM_{n}^{2}=E\max (X_{1},X_{1}+\max (L_{2},...,L_{n}))^{2}\leq
EX_{1}^{2}+E\max (L_{2},...,L_{n})^{2}
\end{equation*}
\bigskip \noindent And then, apply (\ref{nw}) on $\mathbb{E}\max
(L_{2},...,L_{n})^{2}$ for $(n-1)$ mean zero associated rv's to hav
\begin{equation*}
\mathbb{E}\max (L_{2},...,L_{n})^{2}\leq EX_{2}^{2}+...+X_{n}^{2}.
\end{equation*
We conclude tha
\begin{equation*}
EM_{n}^{2}\leq \mathbb{E}X_{1}^{2}+\mathbb{E}X_{2}^{2}+...+\mathbb{E
X_{n}^{2}.
\end{equation*}
\bigskip
\begin{lemma}
Let $X_{1},X_{2},\cdots ,X_{n}$ be a second-order stationary sequence with
\sigma ^{2}=V(X_{1})+2\sum_{j=2}^{\infty }|Cov(X_{1},X_{j})|<\infty ,$ then
\begin{equation*}
V\left( \dfrac{S_{n}}{\sqrt{n}}\right) \rightarrow \sigma
^{2}=V(X_{1})+2\sum_{j=2}^{\infty }Cov(X_{1},X_{j}).
\end{equation*}
\end{lemma}
\noindent\textbf{Proof.} We have
\begin{equation*}
\alpha _{n}=V\left( \dfrac{S_{n}}{\sqrt{n}}\right) =\frac{1}{n}\left\{
\sum_{j=1}^{n}V(X_{i})+\sum_{1\leq i\neq j\leq n}Cov(X_{i},X_{j})\right\} .
\end{equation*}
\bigskip \noindent By stationarity, we hav
\begin{equation*}
V\left( \dfrac{S_{n}}{\sqrt{n}}\right) =V(X_{1})+\frac{2}{n}\sum_{1\leq
i<j\leq n}Cov(X_{i},X_{j})
\end{equation*
\begin{equation*}
=V(X_{1})+\frac{2}{n}\sum_{j=2}^{n}(n-j+1)Cov(X_{1},X_{j}).
\end{equation*
Let $\epsilon >0$. Since $\sum_{j=2}^{\infty }Cov(X_{1},X_{j}) < +\infty$,
there exists $K>0$ such that for any $k\geq K,$
\begin{equation*}
\sum_{j\geq k+1}Cov(X_{1},X_{j})<\epsilon .
\end{equation*}
\bigskip \noindent We fix that $k\geq K$ and write,
\begin{equation*}
\alpha _{n}=V(X_{1})+2\left[ \sum_{j=2}^{k}\left( 1-\dfrac{j-1}{n}\right)
Cov(X_{1},X_{j})+\sum_{j=k+1}^{n}\left( 1-\dfrac{j-1}{n}\right)
Cov(X_{1},X_{j})\right]
\end{equation*}
\bigskip \noindent and observe that
\begin{equation*}
\left\vert \alpha _{n}-V(X_{1})-2\sum_{j=2}^{k}\left( 1-\dfrac{j-1}{n
\right) Cov(X_{1},X_{j})\right\vert \leq 2\epsilon .
\end{equation*}
\bigskip \noindent Thus, we get
\begin{equation*}
\lim \inf V(X_{1})+2\sum_{j=2}^{k}\left( 1-\dfrac{j-1}{n}\right)
Cov(X_{1},X_{j})-2\epsilon \leq \lim \inf \alpha _{n}
\end{equation*
\begin{equation*}
\leq \lim \sup \alpha _{n}\leq \lim \sup V(X_{1})+2\sum_{j=2}^{k}\left( 1
\dfrac{j-1}{n}\right) Cov(X_{1},X_{j})+2\epsilon .
\end{equation*}
\bigskip \noindent Therefore, for any $k\geq K,$
\begin{equation*}
V(X_{1})+2\sum_{j=2}^{k}Cov(X_{1},X_{j})-2\epsilon \leq \lim \inf \alpha
_{n}\leq \lim \sup \alpha _{n}
\end{equation*
\begin{equation*}
\leq V(X_{1})+2\sum_{j=2}^{k}Cov(X_{1},X_{j})+2\epsilon .
\end{equation*
We finish the proof by letting $k\rightarrow \infty $ and next by \ letting
\epsilon \rightarrow 0.$
\begin{theorem}
Let $X_{1},X_{2},\cdots ,X_{m}$ be a strictly stationary, mean zero,
associated random variables such that
\begin{equation*}
\sigma ^{2}=V(X_{1})+2\sum_{j=2}^{+\infty}Cov(X_{1},X_{j})<\infty ,
\end{equation*
then
\begin{equation*}
\dfrac{S_{n}}{\sqrt{n}}=\dfrac{X_{1}+X_{2}+\cdots +X_{n}}{\sqrt{n}
\rightarrow N(0,\sigma ^{2})\ as\ n\rightarrow \infty
\end{equation*}
\end{theorem}
\noindent \textbf{Proof.} Let us fix $\ell >1$ an integer and let us set $m=
\frac{n}{\ell }]$, that is $m\ell \leq n\leq m\ell +\ell .$ Let us define
\Psi _{n}(r)=\mathbb{E}(e^{irS_{n}/\sqrt{n}}),$ $r\in \mathbb{R}$. \noindent
First, we have for $r\in \mathbb{R}$,
\begin{equation*}
|\Psi _{n}(r)-\Psi _{m\ell }(r)|=|\mathbb{E}(e^{irS_{n}/\sqrt{n}})-\mathbb{E
(e^{irS_{m\ell }/\sqrt{m\ell }})|
\end{equation*
\begin{equation*}
=\left\vert \mathbb{E}\left[ e^{irS_{m\ell }/\sqrt{m\ell }}\left( e^{ir\left[
(S_{n}/\sqrt{n})-(S_{m\ell }/\sqrt{m\ell })\right] }-1\right) \right]
\right\vert
\end{equation*
\begin{equation}
\leq \mathbb{E}\left\vert e^{ir\left( \frac{S_{n}}{\sqrt{n}}-\frac{S_{m\ell
}{\sqrt{m\ell }}\right) }-1\right\vert . \label{b}
\end{equation
But for any $x\in \mathbb{R}$,
\begin{equation*}
|e^{ix}-1|=|(\cos x-1)+i\sin x|=|2\sin \frac{x}{2}|\leq |x|.
\end{equation*}
\bigskip Thus the second member of $(\ref{b})$ is, by the Cauchy-Schwarz's
inequality, bounded by
\begin{equation*}
|r|\mathbb{E}\left\vert \frac{S_{n}}{\sqrt{n}}-\frac{S_{m\ell }}{\sqrt{m\ell
}}\right\vert \leq |r|V\left( \frac{S_{n}}{\sqrt{n}}-\frac{S_{m\ell }}{\sqrt
m\ell }}\right) ^{\frac{1}{2}}.
\end{equation*}
\bigskip \noindent Let us compute the quantity between brackets for fixed
\ell $ and $n\rightarrow \infty $ $(m\rightarrow \infty )$, we get
\begin{equation*}
\frac{S_{n}}{\sqrt{n}}-\frac{S_{m\ell }}{\sqrt{m\ell }}=\frac{S_{n}}{\sqrt{n
}-\frac{S_{m\ell }}{\sqrt{n}}+\frac{S_{m\ell }}{\sqrt{n}}-\frac{S_{m\ell }}
\sqrt{m\ell }}
\end{equation*
\begin{equation*}
=\frac{S_{n}-S_{m\ell }}{\sqrt{n}}-\frac{\sqrt{n}-\sqrt{m\ell }}{\sqrt
nm\ell }}S_{m\ell }
\end{equation*
and
\begin{equation*}
\delta _{m,\ell }=V\left( \frac{S_{n}}{\sqrt{n}}-\frac{S_{m\ell }}{\sqrt
m\ell }}\right) =V\left( \frac{S_{n}-S_{m\ell }}{\sqrt{n}}\right) +\left(
\frac{\sqrt{n}-\sqrt{m\ell }}{\sqrt{n}}\right) ^{2}V\left( \frac{S_{m\ell }}
\sqrt{m\ell }}\right)
\end{equation*
\begin{equation*}
-2\frac{\sqrt{n}-\sqrt{m\ell }}{\sqrt{nm\ell }}Cov(S_{n}-S_{m\ell },S_{m\ell
}).
\end{equation*
$Cov(S_{n}-S_{m\ell },S_{m\ell })\geq 0$ by association. Thus
\begin{equation*}
\delta _{m,\ell }\leq V\left( \frac{S_{n-m\ell }}{\sqrt{n}}\right) +\left(
\frac{\sqrt{n}-\sqrt{m\ell }}{\sqrt{n}}\right) ^{2}V\left( \frac{S_{m\ell }}
\sqrt{m\ell }}\right) .
\end{equation*
Since $0\leq n-m\ell \leq \ell $, and $Cov(X_{1},X_{j})\geq 0$ by
association,
\begin{equation*}
V(S_{n-m\ell })=\sum_{i=1}^{n-m\ell }V(X_{i})+\sum_{1\leq i\neq j\leq
n-m\ell }Cov(X_{i},X_{j})
\end{equation*
\begin{equation*}
\leq \sum_{i=1}^{\ell }V(X_{i})+\sum_{1\leq i\neq j\leq \ell
}Cov(X_{i},X_{j})=A(\ell ).
\end{equation*}
\bigskip \noindent Further, $m\ell \leq n\leq (m+1)\ell $ implies
\begin{equation*}
0\leq \frac{\sqrt{n}-\sqrt{m\ell }}{\sqrt{n}}\leq \left( 1-\sqrt{\frac{m\ell
}{n}}\right) \rightarrow 0\text{ as }n\rightarrow +\infty .
\end{equation*
Then when $m\rightarrow \infty $ $(n\rightarrow \infty )$
\begin{equation*}
V\left( \frac{S_{m\ell }}{\sqrt{m\ell }}\right) \rightarrow
V(X_{1})+2\sum_{j=2}^{\infty }Cov(X_{1},X_{j})<\infty
\end{equation*
and
\begin{equation*}
\delta _{m,\ell }\leq \frac{A(\ell )}{n}+\left( 1-\sqrt{\frac{m\ell }{n}
\right) ^{2}V\left( \frac{S_{m\ell }}{\sqrt{m\ell }}\right) \rightarrow 0
\end{equation*
for fixed $\ell $, $n\rightarrow \infty $, we get
\begin{equation*}
|\Psi _{n}(r)-\Psi _{m\ell }(r)|\rightarrow 0.
\end{equation*}
\bigskip \noindent Now, let us set $Y_{j}=(S_{j\ell }-S_{\ell (j-1)})/\sqrt
\ell }$, for a fixed $\ell $. By strict stationarity, the $Y_{j}^{\prime }s$
are associated and identically distributed. Let $\Psi_{\ell} $ be the common
characteristic function of $Y_{1},\cdots ,Y_{m}$. Furthermore
\begin{equation*}
\frac{S_{m\ell }}{\sqrt{m\ell }}=\frac{1}{\sqrt{m}\sqrt{\ell }
\sum_{j=1}^{m}(S_{j\ell }-S_{\ell (j-1)})=\frac{1}{\sqrt{m}
\sum_{j=1}^{m}Y_{j}.
\end{equation*}
\bigskip \noindent According to the Newman's Theorem (see Theorem \ref{theo1
)
\begin{equation*}
\left\vert \Psi _{m\ell }(r)-\left( \Psi _{\ell }\left( \frac{r}{\sqrt{m}
\right) \right) ^{m}\right\vert \leq \frac{r^{2}}{2m}\sum_{1\leq j\neq k\leq
m}Cov(Y_{j},Y_{k}),
\end{equation*
and we know that
\begin{equation*}
V\left( \sum_{j=1}^{m}Y_{j}\right) =\sum_{j=1}^{m}V(Y_{j})+\sum_{1\leq j\neq
k\leq m}Cov(Y_{j},Y_{k}).
\end{equation*}
\bigskip \noindent Thus, by using the stationarity again, we get
\begin{equation*}
\frac{1}{m}\sum_{1\leq j\neq k\leq m}Cov(Y_{j},Y_{k})=\frac{1}{m}V\left(
\sum_{j=1}^{m}Y_{j}\right) -\frac{1}{m}\sum_{j=1}^{m}V(Y_{j})
\end{equation*
\begin{equation*}
=V\left( \frac{1}{\sqrt{m}}\sum_{j=1}^{m}Y_{j}\right) -\frac{1}{m
\sum_{j=1}^{m}V\left( Y_{j}\right)
\end{equation*
\begin{equation*}
V\left( \frac{S_{m\ell }}{\sqrt{m\ell }}\right) -V\left( \frac{S_{\ell }}
\sqrt{\ell }}\right) =\sigma _{m\ell }^{2}-\sigma _{\ell }^{2},
\end{equation*
where for any $p\geq 2,$
\begin{equation*}
\sigma _{p}^{2}=\frac{1}{p}\sum_{i=1}^{p}V(Y_{i})+\frac{1}{p}\sum_{1\leq
i\neq j\leq p}Cov(Y_{i},Y_{j})
\end{equation*
Now, when $m\rightarrow \infty $, $\sigma _{m\ell }^{2}\rightarrow \sigma
^{2}$ and
\begin{equation*}
\left( \Psi _{\ell }\left( \frac{r}{\sqrt{m}}\right) \right) ^{m}\rightarrow
e^{-\sigma _{\ell }^{2}r^{2}/2},
\end{equation*}
\bigskip \noindent where $\sigma _{\ell }^{2}$ is the common variance of
Y_{j}^{\prime }s$,
\begin{equation*}
\sigma _{\ell }^{2}=\sum_{i=1}^{\ell }V(X_{i})+\frac{1}{\ell }\sum_{1\leq
i\neq j\leq m}Cov(X_{i},X_{j}).
\end{equation*
Then it comes out that
\begin{equation*}
\varlimsup \left\vert \Psi _{m\ell }(r)-e^{-\sigma _{\ell
}^{2}r^{2}/2}\right\vert \leq \frac{r^{2}}{2}(\sigma ^{2}-\sigma _{\ell
}^{2}).
\end{equation*
We complete the proof by letting $\ell \rightarrow \infty .$ Thus $\sigma
_{\ell }^{2}-\sigma ^{2}\rightarrow 0$ and we get
\begin{equation*}
\lim_{n\rightarrow \infty }\left\vert \Psi _{n}(r)-e^{-\sigma
^{2}r^{2}/2}\right\vert =0.
\end{equation*
Remark. We finish this exposition by these important facts. A number of
CLT's and invariance principles are available in the literature for strictly
stationary sequences of associated random variables and not stationary ones.
The most general CLT seems to be the one provided by Cox and Grimmet \cit
{cox} for arbitrary associated rv's fulfulling a number of moment
conditions. Dabrowski and co-authors (see \cite{burton} and \cite{dabro})
considered weakly associated random variables to establish principle
invariances in the lines of Newman and Wright \cite{newmanwright}, as well
as Berry-Essen-type results and functional LIL's. But almost all these
results use the original adaptation of the original method of Newman we have
described here.
\bigskip
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 6,072
|
The Toowoomba Branch of QJA is based in Toowoomba (nicknamed 'The Garden City') a city in the Darling Downs region located 125km west of Brisbane.
Meetings of the branch are held at the Dr Price Meeting Room, 6 Little Street, (Behind Mothercraft Childhood Centre), every 3rd Monday of the month commencing at 5.30pm. The Annual General Meeting (AGM) is held in July each year immediately after the July ordinary meeting.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 9,866
|
Q: How to turn a video n degrees I've got a video that needs to be rotated n degrees (not just 90 or 180 which I can do using ffmpeg but for instance 1 or 10). How can do this?
I've googled and no such tools seem to be freely available.
A: I've figured out a way to do this that's not very elegant, but it works for me so I wanted to share it.
First I've twisted the video based on code from learnopencv.com:
# source learnopencv.com
import numpy as np
import cv2
from tqdm import tqdm
def fixBorder(frame):
s = frame.shape
# Scale the image 4% without moving the center
T = cv2.getRotationMatrix2D((s[1] / 2, s[0] / 2), 0, 1.04)
frame = cv2.warpAffine(frame, T, (s[1], s[0]))
return frame
def straighten_vid(input_name, output_name, d_dx=0, d_dy=0, d_da=0):
d_da *= np.pi / 180.
# Read input video
cp = cv2.VideoCapture(input_name)
# To get number of frames
n_frames = int(cp.get(cv2.CAP_PROP_FRAME_COUNT))
# To check the number of frames in the video
width = int(cp.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cp.get(cv2.CAP_PROP_FRAME_HEIGHT))
# get the number of frames per second
fps = cp.get(cv2.CAP_PROP_FPS)
# Try doing 2*width
out = cv2.VideoWriter(output_name, 0x7634706d, fps, (width, height))
transforms_smooth = np.zeros((n_frames - 1, 3), np.float32)
for i in tqdm(range(n_frames - 2)):
transforms_smooth[i] = [d_dx, d_dy, d_da]
# Reset stream to first frame
cp.set(cv2.CAP_PROP_POS_FRAMES, 0)
# Write n_frames-1 transformed frames
for i in range(n_frames - 2):
# Read next frame
success, frame = cp.read()
if not success:
break
# Extract transformations from the new transformation array
dx = transforms_smooth[i, 0]
dy = transforms_smooth[i, 1]
da = transforms_smooth[i, 2]
# Reconstruct transformation matrix accordingly to new values
m = np.zeros((2, 3), np.float32)
m[0, 0] = np.cos(da)
m[0, 1] = -np.sin(da)
m[1, 0] = np.sin(da)
m[1, 1] = np.cos(da)
m[0, 2] = dx
m[1, 2] = dy
# Apply affine wrapping to the given frame
frame_stabilized = cv2.warpAffine(frame, m, (width, height))
# Fix border artifacts
frame_stabilized = fixBorder(frame_stabilized)
# Write the frame to the file
frame_out = frame_stabilized #cv2.hconcat([frame, frame_stabilized])
# If the image is too big, resize it.
if (frame_out.shape[1] > 1920):
frame_out = cv2.resize(frame_out, (frame_out.shape[1] // 2, frame_out.shape[0] // 2))
# cv2.imshow("Before and After", frame_out)
# cv2.waitKey(10)
out.write(frame_out)
# Release video
cp.release()
out.release()
# Close windows
cv2.destroyAllWindows()
The d_da is the angle turned and d_dx and d_dy is used to center the video. However, this leaves a file with no sounds. To add this back in I do this:
import moviepy.editor as mpe
input_folder = "my_folder\\my_folder\\"
file_name = "my_file_name.MOV"
input_file = input_folder + file_name
output_file = input_folder + file_name.replace(".MOV", "_straightened.MOV")
output_file_final = input_folder + file_name.replace(".MOV", "_straightened_with_sound.mp4")
straighten_vid(input_file, output_file, 0, -70, 0.1)
my_clip = mpe.VideoFileClip(output_file)
audio_background = mpe.VideoFileClip(input_file).audio
final_clip = my_clip.set_audio(audio_background)
final_clip.write_videofile(output_file_final)
A: You can use the rotate filter.
For example rotate 10 degree.
ffmpeg.exe -i input.mp4 -vf 'rotate=10/180*PI/2' output.mp4
Or you can use a dynamic degree.
*
*ffprobe.exe -v quiet -show_entries format=duration -print_format json -i input.mp4 to get the video lenght as T.
*Then rotate it with ffmpeg.exe -i input.mp4 -vf 'rotate=t/T*PI/2' output.mp4
|
{
"redpajama_set_name": "RedPajamaStackExchange"
}
| 5,140
|
package alphapoint
import (
"context"
"errors"
"strconv"
"time"
"github.com/thrasher-corp/gocryptotrader/common"
"github.com/thrasher-corp/gocryptotrader/config"
"github.com/thrasher-corp/gocryptotrader/currency"
exchange "github.com/thrasher-corp/gocryptotrader/exchanges"
"github.com/thrasher-corp/gocryptotrader/exchanges/account"
"github.com/thrasher-corp/gocryptotrader/exchanges/asset"
"github.com/thrasher-corp/gocryptotrader/exchanges/order"
"github.com/thrasher-corp/gocryptotrader/exchanges/orderbook"
"github.com/thrasher-corp/gocryptotrader/exchanges/protocol"
"github.com/thrasher-corp/gocryptotrader/exchanges/request"
"github.com/thrasher-corp/gocryptotrader/exchanges/ticker"
"github.com/thrasher-corp/gocryptotrader/exchanges/trade"
"github.com/thrasher-corp/gocryptotrader/log"
"github.com/thrasher-corp/gocryptotrader/portfolio/withdraw"
)
// GetDefaultConfig returns a default exchange config for Alphapoint
func (a *Alphapoint) GetDefaultConfig() (*config.Exchange, error) {
return nil, common.ErrFunctionNotSupported
}
// SetDefaults sets current default settings
func (a *Alphapoint) SetDefaults() {
a.Name = "Alphapoint"
a.Enabled = true
a.Verbose = true
a.API.Endpoints = a.NewEndpoints()
err := a.API.Endpoints.SetDefaultEndpoints(map[exchange.URL]string{
exchange.RestSpot: alphapointDefaultAPIURL,
exchange.WebsocketSpot: alphapointDefaultWebsocketURL,
})
if err != nil {
log.Errorln(log.ExchangeSys, err)
}
a.API.CredentialsValidator.RequiresKey = true
a.API.CredentialsValidator.RequiresSecret = true
a.Features = exchange.Features{
Supports: exchange.FeaturesSupported{
REST: true,
Websocket: true,
RESTCapabilities: protocol.Features{
AccountInfo: true,
TickerFetching: true,
TradeFetching: true,
OrderbookFetching: true,
GetOrders: true,
CancelOrder: true,
CancelOrders: true,
SubmitOrder: true,
ModifyOrder: true,
UserTradeHistory: true,
CryptoDeposit: true,
CryptoWithdrawal: true,
TradeFee: true,
},
WebsocketCapabilities: protocol.Features{
AccountInfo: true,
},
WithdrawPermissions: exchange.WithdrawCryptoWith2FA |
exchange.AutoWithdrawCryptoWithAPIPermission |
exchange.NoFiatWithdrawals,
},
}
a.Requester, err = request.New(a.Name,
common.NewHTTPClientWithTimeout(exchange.DefaultHTTPTimeout))
if err != nil {
log.Errorln(log.ExchangeSys, err)
}
}
// FetchTradablePairs returns a list of the exchanges tradable pairs
func (a *Alphapoint) FetchTradablePairs(ctx context.Context, asset asset.Item) ([]string, error) {
return nil, common.ErrFunctionNotSupported
}
// UpdateTradablePairs updates the exchanges available pairs and stores
// them in the exchanges config
func (a *Alphapoint) UpdateTradablePairs(ctx context.Context, forceUpdate bool) error {
return common.ErrFunctionNotSupported
}
// UpdateAccountInfo retrieves balances for all enabled currencies on the
// Alphapoint exchange
func (a *Alphapoint) UpdateAccountInfo(ctx context.Context, assetType asset.Item) (account.Holdings, error) {
var response account.Holdings
response.Exchange = a.Name
acc, err := a.GetAccountInformation(ctx)
if err != nil {
return response, err
}
balances := make([]account.Balance, len(acc.Currencies))
for i := range acc.Currencies {
balances[i] = account.Balance{
CurrencyName: currency.NewCode(acc.Currencies[i].Name),
Total: float64(acc.Currencies[i].Balance),
Hold: float64(acc.Currencies[i].Hold),
Free: float64(acc.Currencies[i].Balance) - float64(acc.Currencies[i].Hold),
}
}
response.Accounts = append(response.Accounts, account.SubAccount{
Currencies: balances,
AssetType: assetType,
})
err = account.Process(&response)
if err != nil {
return account.Holdings{}, err
}
return response, nil
}
// FetchAccountInfo retrieves balances for all enabled currencies on the
// Alphapoint exchange
func (a *Alphapoint) FetchAccountInfo(ctx context.Context, assetType asset.Item) (account.Holdings, error) {
acc, err := account.GetHoldings(a.Name, assetType)
if err != nil {
return a.UpdateAccountInfo(ctx, assetType)
}
return acc, nil
}
// UpdateTickers updates the ticker for all currency pairs of a given asset type
func (a *Alphapoint) UpdateTickers(assetType asset.Item) error {
return common.ErrFunctionNotSupported
}
// UpdateTicker updates and returns the ticker for a currency pair
func (a *Alphapoint) UpdateTicker(ctx context.Context, p currency.Pair, assetType asset.Item) (*ticker.Price, error) {
tick, err := a.GetTicker(ctx, p.String())
if err != nil {
return nil, err
}
err = ticker.ProcessTicker(&ticker.Price{
Pair: p,
Ask: tick.Ask,
Bid: tick.Bid,
Low: tick.Low,
High: tick.High,
Volume: tick.Volume,
Last: tick.Last,
ExchangeName: a.Name,
AssetType: assetType,
})
if err != nil {
return nil, err
}
return ticker.GetTicker(a.Name, p, assetType)
}
// FetchTicker returns the ticker for a currency pair
func (a *Alphapoint) FetchTicker(ctx context.Context, p currency.Pair, assetType asset.Item) (*ticker.Price, error) {
tick, err := ticker.GetTicker(a.Name, p, assetType)
if err != nil {
return a.UpdateTicker(ctx, p, assetType)
}
return tick, nil
}
// UpdateOrderbook updates and returns the orderbook for a currency pair
func (a *Alphapoint) UpdateOrderbook(ctx context.Context, p currency.Pair, assetType asset.Item) (*orderbook.Base, error) {
orderBook := new(orderbook.Base)
orderbookNew, err := a.GetOrderbook(ctx, p.String())
if err != nil {
return orderBook, err
}
orderBook.Bids = make(orderbook.Items, len(orderbookNew.Bids))
for x := range orderbookNew.Bids {
orderBook.Bids[x] = orderbook.Item{
Amount: orderbookNew.Bids[x].Quantity,
Price: orderbookNew.Bids[x].Price,
}
}
orderBook.Asks = make(orderbook.Items, len(orderbookNew.Asks))
for x := range orderbookNew.Asks {
orderBook.Asks[x] = orderbook.Item{
Amount: orderbookNew.Asks[x].Quantity,
Price: orderbookNew.Asks[x].Price,
}
}
orderBook.Pair = p
orderBook.Exchange = a.Name
orderBook.Asset = assetType
err = orderBook.Process()
if err != nil {
return orderBook, err
}
return orderbook.Get(a.Name, p, assetType)
}
// FetchOrderbook returns the orderbook for a currency pair
func (a *Alphapoint) FetchOrderbook(ctx context.Context, p currency.Pair, assetType asset.Item) (*orderbook.Base, error) {
ob, err := orderbook.Get(a.Name, p, assetType)
if err != nil {
return a.UpdateOrderbook(ctx, p, assetType)
}
return ob, nil
}
// GetFundingHistory returns funding history, deposits and
// withdrawals
func (a *Alphapoint) GetFundingHistory(ctx context.Context) ([]exchange.FundHistory, error) {
// https://alphapoint.github.io/slate/#generatetreasuryactivityreport
return nil, common.ErrNotYetImplemented
}
// GetWithdrawalsHistory returns previous withdrawals data
func (a *Alphapoint) GetWithdrawalsHistory(ctx context.Context, c currency.Code) (resp []exchange.WithdrawalHistory, err error) {
return nil, common.ErrNotYetImplemented
}
// GetRecentTrades returns the most recent trades for a currency and asset
func (a *Alphapoint) GetRecentTrades(_ currency.Pair, _ asset.Item) ([]trade.Data, error) {
return nil, common.ErrNotYetImplemented
}
// GetHistoricTrades returns historic trade data within the timeframe provided
func (a *Alphapoint) GetHistoricTrades(_ context.Context, _ currency.Pair, _ asset.Item, _, _ time.Time) ([]trade.Data, error) {
return nil, common.ErrNotYetImplemented
}
// SubmitOrder submits a new order and returns a true value when
// successfully submitted
func (a *Alphapoint) SubmitOrder(ctx context.Context, s *order.Submit) (*order.SubmitResponse, error) {
if err := s.Validate(); err != nil {
return nil, err
}
fPair, err := a.FormatExchangeCurrency(s.Pair, s.AssetType)
if err != nil {
return nil, err
}
response, err := a.CreateOrder(ctx,
fPair.String(),
s.Side.String(),
s.Type.String(),
s.Amount,
s.Price)
if err != nil {
return nil, err
}
return s.DeriveSubmitResponse(strconv.FormatInt(response, 10))
}
// ModifyOrder will allow of changing orderbook placement and limit to
// market conversion
func (a *Alphapoint) ModifyOrder(_ *order.Modify) (order.Modify, error) {
return order.Modify{}, common.ErrNotYetImplemented
}
// CancelOrder cancels an order by its corresponding ID number
func (a *Alphapoint) CancelOrder(ctx context.Context, o *order.Cancel) error {
if err := o.Validate(o.StandardCancel()); err != nil {
return err
}
orderIDInt, err := strconv.ParseInt(o.OrderID, 10, 64)
if err != nil {
return err
}
_, err = a.CancelExistingOrder(ctx, orderIDInt, o.AccountID)
return err
}
// CancelBatchOrders cancels an orders by their corresponding ID numbers
func (a *Alphapoint) CancelBatchOrders(ctx context.Context, o []order.Cancel) (order.CancelBatchResponse, error) {
return order.CancelBatchResponse{}, common.ErrNotYetImplemented
}
// CancelAllOrders cancels all orders for a given account
func (a *Alphapoint) CancelAllOrders(ctx context.Context, orderCancellation *order.Cancel) (order.CancelAllResponse, error) {
if err := orderCancellation.Validate(); err != nil {
return order.CancelAllResponse{}, err
}
return order.CancelAllResponse{},
a.CancelAllExistingOrders(ctx, orderCancellation.AccountID)
}
// GetOrderInfo returns order information based on order ID
func (a *Alphapoint) GetOrderInfo(ctx context.Context, orderID string, pair currency.Pair, assetType asset.Item) (float64, error) {
orders, err := a.GetOrders(ctx)
if err != nil {
return 0, err
}
for x := range orders {
for y := range orders[x].OpenOrders {
if strconv.Itoa(orders[x].OpenOrders[y].ServerOrderID) == orderID {
return orders[x].OpenOrders[y].QtyRemaining, nil
}
}
}
return 0, errors.New("order not found")
}
// GetDepositAddress returns a deposit address for a specified currency
func (a *Alphapoint) GetDepositAddress(ctx context.Context, cryptocurrency currency.Code, _, _ string) (string, error) {
addreses, err := a.GetDepositAddresses(ctx)
if err != nil {
return "", err
}
for x := range addreses {
if addreses[x].Name == cryptocurrency.String() {
return addreses[x].DepositAddress, nil
}
}
return "", errors.New("associated currency address not found")
}
// WithdrawCryptocurrencyFunds returns a withdrawal ID when a withdrawal is
// submitted
func (a *Alphapoint) WithdrawCryptocurrencyFunds(_ context.Context, _ *withdraw.Request) (*withdraw.ExchangeResponse, error) {
return nil, common.ErrNotYetImplemented
}
// WithdrawFiatFunds returns a withdrawal ID when a withdrawal is submitted
func (a *Alphapoint) WithdrawFiatFunds(_ context.Context, _ *withdraw.Request) (*withdraw.ExchangeResponse, error) {
return nil, common.ErrNotYetImplemented
}
// WithdrawFiatFundsToInternationalBank returns a withdrawal ID when a withdrawal is
// submitted
func (a *Alphapoint) WithdrawFiatFundsToInternationalBank(_ *withdraw.Request) (string, error) {
return "", common.ErrNotYetImplemented
}
// GetFeeByType returns an estimate of fee based on type of transaction
func (a *Alphapoint) GetFeeByType(_ *exchange.FeeBuilder) (float64, error) {
return 0, common.ErrFunctionNotSupported
}
// GetActiveOrders retrieves any orders that are active/open
// This function is not concurrency safe due to orderSide/orderType maps
func (a *Alphapoint) GetActiveOrders(ctx context.Context, req *order.GetOrdersRequest) ([]order.Detail, error) {
if err := req.Validate(); err != nil {
return nil, err
}
resp, err := a.GetOrders(ctx)
if err != nil {
return nil, err
}
var orders []order.Detail
for x := range resp {
for y := range resp[x].OpenOrders {
if resp[x].OpenOrders[y].State != 1 {
continue
}
orderDetail := order.Detail{
Amount: resp[x].OpenOrders[y].QtyTotal,
Exchange: a.Name,
ExecutedAmount: resp[x].OpenOrders[y].QtyTotal - resp[x].OpenOrders[y].QtyRemaining,
AccountID: strconv.FormatInt(int64(resp[x].OpenOrders[y].AccountID), 10),
OrderID: strconv.FormatInt(int64(resp[x].OpenOrders[y].ServerOrderID), 10),
Price: resp[x].OpenOrders[y].Price,
RemainingAmount: resp[x].OpenOrders[y].QtyRemaining,
}
orderDetail.Side = orderSideMap[resp[x].OpenOrders[y].Side]
orderDetail.Date = time.Unix(resp[x].OpenOrders[y].ReceiveTime, 0)
orderDetail.Type = orderTypeMap[resp[x].OpenOrders[y].OrderType]
orders = append(orders, orderDetail)
}
}
order.FilterOrdersByType(&orders, req.Type)
order.FilterOrdersBySide(&orders, req.Side)
err = order.FilterOrdersByTimeRange(&orders, req.StartTime, req.EndTime)
if err != nil {
log.Errorf(log.ExchangeSys, "%s %v", a.Name, err)
}
return orders, nil
}
// GetOrderHistory retrieves account order information
// Can Limit response to specific order status
// This function is not concurrency safe due to orderSide/orderType maps
func (a *Alphapoint) GetOrderHistory(ctx context.Context, req *order.GetOrdersRequest) ([]order.Detail, error) {
if err := req.Validate(); err != nil {
return nil, err
}
resp, err := a.GetOrders(ctx)
if err != nil {
return nil, err
}
var orders []order.Detail
for x := range resp {
for y := range resp[x].OpenOrders {
if resp[x].OpenOrders[y].State == 1 {
continue
}
orderDetail := order.Detail{
Amount: resp[x].OpenOrders[y].QtyTotal,
AccountID: strconv.FormatInt(int64(resp[x].OpenOrders[y].AccountID), 10),
Exchange: a.Name,
ExecutedAmount: resp[x].OpenOrders[y].QtyTotal - resp[x].OpenOrders[y].QtyRemaining,
OrderID: strconv.FormatInt(int64(resp[x].OpenOrders[y].ServerOrderID), 10),
Price: resp[x].OpenOrders[y].Price,
RemainingAmount: resp[x].OpenOrders[y].QtyRemaining,
}
orderDetail.Side = orderSideMap[resp[x].OpenOrders[y].Side]
orderDetail.Date = time.Unix(resp[x].OpenOrders[y].ReceiveTime, 0)
orderDetail.Type = orderTypeMap[resp[x].OpenOrders[y].OrderType]
orders = append(orders, orderDetail)
}
}
order.FilterOrdersByType(&orders, req.Type)
order.FilterOrdersBySide(&orders, req.Side)
err = order.FilterOrdersByTimeRange(&orders, req.StartTime, req.EndTime)
if err != nil {
log.Errorf(log.ExchangeSys, "%s %v", a.Name, err)
}
return orders, nil
}
// ValidateCredentials validates current credentials used for wrapper
// functionality
func (a *Alphapoint) ValidateCredentials(ctx context.Context, assetType asset.Item) error {
_, err := a.UpdateAccountInfo(ctx, assetType)
return a.CheckTransientError(err)
}
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 7,966
|
Mount Bierstadt ist ein 4286 m hoher Berg in der Front Range der Rocky Mountains. Er liegt im Bundesstaat Colorado und ist nach dem deutschstämmigen Landschaftsmaler Albert Bierstadt benannt.
Berg in Colorado
Berg in Nordamerika
Berg in den Rocky Mountains
Clear Creek County
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 8,595
|
{"url":"https:\/\/codegolf.stackexchange.com\/questions\/185545\/ambassadors-and-translators\/185557","text":"Two ambassadors at a UN conference want to speak to each other, but unfortunately each one only speaks one language- and they're not the same language. Fortunately, they have access to several translators, who each understand and speak a few languages. Your task is to determine the shortest chain of translators (since you want as little to be lost in translation as possible) that allows the two ambassadors to speak with each other.\n\nCoding\n\nInput: two languages as 2-letter lowercase strings (each ambassador's language) and a list of lists of languages (one list per available translator)\n\nYou may alternatively take in integers instead of 2-letter codes.\n\nOutput: A sequence of translators either by index or value that is any one of the shortest chains of translators that allows the two ambassadors to communicate. If there is no valid chain of translators, the behavior is undefined. (You may crash, output any arbitrary value, or indicate an error)\n\nA valid chain of translators is one where the first translator speaks one ambassador's language, the second and subsequent translators share at least one language with the previous translator, and the last translator speaks the other ambassador's language.\n\nExamples\n\nUsing zero-based indexing:\n\nes, en, [\n[es, en]\n] ==> [0]\n\nen, en, [] ==> []\n\nen, jp, [\n[en, zh, ko, de],\n[jp, ko]\n] ==> [0, 1]\n\nes, ru, [\n[gu, en, py],\n[po, py, ru],\n[po, es]\n] ==> [2, 1]\n\nfr, gu, [\n[it, fr, de, es, po, jp],\n[en, ru, zh, ko],\n[jp, th, en],\n[th, gu]\n] ==> [0, 2, 3]\n\nfr, ru, [\n[fr, en],\n[en, ko, jp],\n[en, ru]\n] ==> [0, 2]\n\nde, jp, [\n[en, fr],\n[ko, jp, zh],\n[fr, po],\n[es, ko, zh],\n[de, en, th],\n[en, es],\n[de, fr]\n] ==> [4, 5, 3, 1]\n\n\nRules and Assumptions\n\n\u2022 Standard IO rules (use any convenient I\/O format) and banned loopholes apply.\n\u2022 You may assume that speaking and understanding languages is perfectly symmetric and that all possible translations between languages are equally efficient.\n\u2022 There is no concept of \"close enough\" languages. It is not good enough to use Portuguese on one end where Spanish is required, for instance.\n\u2022 If there are multiple shortest translator chains, any one of them will do.\n\u2022 If the ambassadors happen to speak the same language, the translator list should be empty\n\u2022 Which of the ambassadors is the first one doesn't matter; the translator list can be forward or reverse.\n\u2022 Ambassadors only speak one language for the sake of this challenge\n\u2022 Translators speak at least two languages\n\u2022 The 2-letter language codes do not need to correspond with real languages\n\u2022 You may assume there is a valid sequence of translators\n\u2022 If outputting the sequence by value, include the full set of available languages, not just the relevant ones.\n\nHappy Golfing!\n\n\u2022 Why the I\/O restriction to two-character strings, wouldn't integers do just as well? \u2013\u00a0Jonathan Allan May 13 '19 at 18:47\n\u2022 can the list of list of translators be in csv form like: en,fr,sp;en,gr;gr,fr \u2013\u00a0Quinn May 13 '19 at 19:49\n\u2022 @Quinn standard IO rules say yes. \u2013\u00a0Beefster May 13 '19 at 20:13\n\u2022 Can the ambassadors be included in the output at the beginning and end? \u2013\u00a0Nick Kennedy May 13 '19 at 20:29\n\u2022 @NickKennedy I'm gonna say no on that one. \u2013\u00a0Beefster May 13 '19 at 20:44\n\nPython 2, 138126120117 113 bytes\n\nF=lambda a,b,T,*U:a!=b and min([[t]+F(l,b,T,t,*U)for t in T if(t in U)<(a in t)for l in t-{a}]+[2*T],key=len)or[]\n\n\nTry it online!\n\n3 bytes thx to ArBo\n\nReturns a minimal length list of the translators as sets of languages, i.e. 'by value', from T that allow a to talk to b.\n\n\u2022 if t not in U and a in t can be changed to if(a in t)>U.count(t) to save 4 bytes. \u2013\u00a0mypetlion May 13 '19 at 21:06\n\u2022 @mypetition - I had a similar thought and squeezed out another 2. \u2013\u00a0Chas Brown May 13 '19 at 21:08\n\u2022 117 by using *args notation \u2013\u00a0ArBo May 13 '19 at 21:40\n\u2022 @ArBo: Nice; thx for 3 bytes. \u2013\u00a0Chas Brown May 14 '19 at 0:46\n\nJelly, 19 17 bytes\n\n\u0152P\u0152!\u20ac\u1e8ej@\u20acf\u019d\u1ea0$\u0187\u1e22\u1e0a\u1e56 Try it online! A dyadic link taking the list of translators as the left argument and list of ambassadors (each double-wrapped in a list) as the right argument. Returns a list of translators, each of which is a list of the languages they speak. Thanks to @KevinCruijssen for saving 2 bytes! Explanation \u0152P\u0152!\u20ac\u1e8ej@\u20acf\u019d\u1ea0$\u0187\u1e22\u1e0a\u1e56 | A dyadic link taking a list of translators as left argument and a list of ambassadors (double-wrapped in lists) as right argument\n\n\u0152P | Power set of translators\n\u0152!\u20ac | Permutations of each\n\u1e8e | Tighten, i.e. create a single list of all permutations of any length\nj@\u20ac | Join the ambassadors with each set of translators\n\\$\u0187 | Filter those where:\n\u1ea0 | all\nf\u019d | the neighbouring pairs have at least one in common\n\u1e22 | Take the first\n\u1e0a | Drop the first ambassador from the start\n\u1e56 | Drop the second ambassador from the end\n\n\u2022 You can save 2 bytes by removing the sort by length L\u00de, since the powerset + permurations already results in a list sorted by length. \u2013\u00a0Kevin Cruijssen May 14 '19 at 13:58\n\u2022 @KevinCruijssen thanks, good point! \u2013\u00a0Nick Kennedy May 14 '19 at 17:45\n\n05AB1E, 18 17 bytes\n\n\u00e6\u20ac\u0153\u20ac\u0292\u00b2\u0161\u00b3\u00aa\u00fc\u00e5\u20ac\u00e0P}\u043d\n\n\nInspired by @NickKennedy's Jelly answer, so make sure to upvote him!\n\nOutputs the lists themselves instead of their indices.\n\nExplanation:\n\n\u00e6 # Get the powerset of the (implicit) input-list of translators\n# i.e. [[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]]\n# \u2192 [[],[[\"ef\",\"gh\",\"bc\"]],[[\"bc\",\"ab\"]],[[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"]],[[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"]],[[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]]]\n\u20ac\u0153 # Get the permutations of each\n# \u2192 [[[]],[[[\"ef\",\"gh\",\"bc\"]]],[[[\"bc\",\"ab\"]]],[[[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"]],[[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]]],[[[\"ef\",\"cd\",\"de\"]]],[[[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"cd\",\"de\"],[\"ef\",\"gh\",\"bc\"]]],[[[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"cd\",\"de\"],[\"bc\",\"ab\"]]],[[[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"],[\"bc\",\"ab\"]],[[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"]],[[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"],[\"ef\",\"gh\",\"bc\"]],[[\"ef\",\"cd\",\"de\"],[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"]],[[\"ef\",\"cd\",\"de\"],[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]]]]\n\u20ac # Flatten each one level down (4D list becomes 3D list)\n# \u2192 [[],[[\"ef\",\"gh\",\"bc\"]],[[\"bc\",\"ab\"]],[[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]],[[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"]],[[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"cd\",\"de\"],[\"ef\",\"gh\",\"bc\"]],[[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"cd\",\"de\"],[\"bc\",\"ab\"]],[[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"cd\",\"de\"],[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]],[[\"ef\",\"cd\",\"de\"],[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"]],[[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"],[\"ef\",\"gh\",\"bc\"]],[[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"]],[[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"],[\"bc\",\"ab\"]],[[\"ef\",\"gh\",\"bc\"],[\"bc\",\"ab\"],[\"ef\",\"cd\",\"de\"]]]\n\u0292 # Filter this 3D list by:\n\u00b2\u0161 # Prepend the second input ambassador\n# i.e. [[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]] and \"ab\"\n# \u2192 [\"ab\",[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]]\n\u00b3\u00aa # Append the third input ambassador\n# i.e. [\"ab\",[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]] and \"ef\"\n# \u2192 [\"ab\",[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"],\"ef\"]\n\u00fc # For each adjacent pair of translator-lists:\n\u00e5 # Check for each item in the second list, if it's in the first list\n# i.e. [\"bc\",\"ab\"] and [\"ef\",\"gh\",\"bc\"] \u2192 [0,0,1]\n\u20ac\u00e0 # Then check if any are truthy by leaving the maximum\n# \u2192 1\nP # And then take the product to check if it's truthy for all pairs\n# i.e. [\"ab\",[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"],\"ef\"] \u2192 [1,1,1] \u2192 1\n}\u043d # After the filter: only leave the first list of translator-lists\n# i.e. [[[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]],[[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"],[\"ef\",\"cd\",\"de\"]]]\n# \u2192 [[\"bc\",\"ab\"],[\"ef\",\"gh\",\"bc\"]]\n# (which is output implicitly as result)\n\n\nJavaScript (ES6), \u00a0123\u00a0 121 bytes\n\nExpects integers instead of 2-letter codes.\n\n(a,b,l)=>((B=g=(m,s,i)=>m>>b&1?B<i||(o=s,B=i):l.map(a=>a.map(M=c=>M|=1<<c)|M&m&&m^(M|=m)&&g(M,[...s,a],-~i)))(1<<a,[]),o)\n\n\nTry it online!","date":"2020-03-30 14:29:39","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\": 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\": 0.29833099246025085, \"perplexity\": 7969.165307202416}, \"config\": {\"markdown_headings\": false, \"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-2020-16\/segments\/1585370497042.33\/warc\/CC-MAIN-20200330120036-20200330150036-00405.warc.gz\"}"}
| null | null |
Home Browse Categories Sports Water Sports Swimming Swimming Schools Canterbury Region Christchurch Fendalton Fendalton (8014)
Contact Swimming Schools for Kids in Fendalton (8014)
Book Swimming Schools for Kids in Fendalton (8014)
Swimming Schools for Kids - Fendalton (8014)
Fendalton, Christchurch, the Canterbury Region
Best match results for swimming schools for kids in Fendalton (8014) + 30km.
Kings Swim School
Sports Swimming Schools
109 Main South Road, Sockburn (8042), Canterbury Region
2-5 km from Fendalton (8014)
459 Cashel St, Linwood (8011), Canterbury Region
Other related businesses in and around Fendalton (8014) (within 50km)
Wharenui Swimming Club and Sports Centre
73 Elizabeth Street, Riccarton (8011), Canterbury Region
2 km from Fendalton (8014)
Wharenui Swim club has been here for over 70 years and we teach water confidence and swimming lessons to all ages. We are one of the top Coaching Clubs in New Zealand and have an excellent record of achievement within our club members Our Learn to Swim programme runs after school each day and mornings in the weekends Squads train up to 10 times per week under our Coach Martin Harris Swimming in New Zealand is so important as most of our holidays are around water, so lets make the most of it... Read more
Canterbury Swim School
25 Charente Way, Hornby (8042), Canterbury Region
5-10 km from Fendalton (8014)
Private swim school offering learn to swim lessons for all ages and abilities. Open 7 days per week with 2 pools - a 25m 6 lane pool and a 10m 4 lane pool - both of which are open to the public for casual swimming, except for between 3.30-6.30pm weekdays. Read more
If you are a business that provides Swimming Schools for Kids in Fendalton (8014), please add your details to the site today.
If you have used one of the kids activity providers listed on the website that provides Swimming Schools for Kids in Fendalton (8014) then please add a review for the benefit of other users.
Swimming Schools in Bryndwr (8052)
Swimming Schools in Burnside (8052)
Swimming Schools in Fendalton (8052)
Swimming Schools in Mairehau (8052)
Swimming Schools in Northcote (8052)
Fendalton (8014) only 0
Fendalton (8014) + 5km 2
Fendalton (8014) + 10km 2
Canterbury Region
Fendalton (8014)
Swimming Schools in Fendalton
Swimming Schools in Christchurch
Swimming Schools in the Canterbury Region
Swimming Schools in New Zealand
Swimming Activities in Fendalton (8014)
Water Sport Activities in Fendalton (8014)
Sport Activities in Fendalton (8014)
All Kids Activities in Fendalton (8014)
ActiveActivities is building New Zealand's most complete online marketplace for kids activities for all occasions, including events for kids and school holiday activities. If you are a provider of Swimming Schools for Kids in Fendalton (8014), please join our online marketplace by adding your details to the site today.
Map of Swimming Schools for Kids in Fendalton (8014)
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 3,338
|
<Record>
<Term>Emakalim</Term>
<SemanticType>Pharmacologic Substance</SemanticType>
<ParentTerm>Antihypertensive Agent</ParentTerm>
<ClassificationPath>Chemicals_and_Drugs_Kind/Drug, Food, Chemical or Biomedical Material/Pharmacologic Substance/Adjuvant/Cardiovascular Agent/Antihypertensive Agent/Emakalim</ClassificationPath>
<BroaderTerm>Chemicals_and_Drugs_Kind</BroaderTerm>
<BroaderTerm>Cardiovascular Agent</BroaderTerm>
<BroaderTerm>Pharmacologic Substance</BroaderTerm>
<BroaderTerm>Adjuvant</BroaderTerm>
<BroaderTerm>Antihypertensive Agent</BroaderTerm>
<BroaderTerm>Emakalim</BroaderTerm>
<BroaderTerm>Drug, Food, Chemical or Biomedical Material</BroaderTerm>
<Synonym>EMAKALIM</Synonym>
<Synonym>Emakalim</Synonym>
<Source>NCI Thesaurus</Source>
</Record>
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 8,840
|
\section{Introduction}
\label{sec:intro}
The memristor is the first (and possibly the only) non-linear fundamental circuit element, and, as such, has a lot to offer to those interested in non-linear dynamics. By virtue of being a fundamental element, we can be reasonably certain that it's the simplest non-linear circuit element. It is a physical device, announced to the world in 2008~\cite{strukov08}, although its existence was predicted many years earlier in 1971~\cite{Chua1971}. The non-linear behaviour arises because the memristor relates the total history of the charge (the time integral of current), that has passed through the device, with the total history of the magnetic flux (the time integral of voltage)~\cite{Chua1971}.
Since the first physical realization of a memristor \cite{strukov08} a huge amount of effort has been devoted to the study of materials and manufacturing techniques for memristor devices, as well as theoretical models, (as is covered in detail in a recent review~\cite{RevMemReRAM}). In terms of materials, the first and archetypal memristor is a device made of titanium dioxide sandwiched between platinum electrodes. Studies on TiO$_2$ memristor~\cite{strukov08} have shown that the main cause of the memristive effect is the motion of oxygen vacancies from the high resistance layer, the TiO$_2$ layer, to the sub-oxide (TiO$_{2-x}$), or low resistance layer, under the application of an external bias: it is this interconversion that changes the resistance of the device. The realization of the active layer requires expensive techniques, such nano-imprint lithography and atomic layer deposition, followed by subsequent annealing at high temperature. Most memristors are fabricated by atomic layer deposition as perfect crystals and annealing is usually required to introduce vacancies, as these are errors in a perfect crystal. The fabrication techniques significantly influence the device performance and significantly impact the cost-effectiveness of the device. For this reason, new methods, typical of the printing industry, screen printing or ink-jet printing, have been also investigated for the realization of memristive devices \cite{duraisamy,memZno}. Despite the numerous realizations of memristors, a breakthrough in ease of manufacturing was the fabrication of a device by spin-coating a titanium isopropoxide solution on a flexible plastic substrate \cite{hackett}. Unlike devices created by atomic deposition, sol-gel and ink-jet printed devices already have vacancy errors, so they require no annealing step for the formation of the active layer.
In this paper, we make use of flexible TiO$_2$ memristors \cite{ella2011}.
Memristors belong to the class of resistive switching devices, which in recent years, attracted a lot of interest for memory, logic and neuromorphic applications. In particular, the memory effect is due to the possibility to switch the status of the device, through SET and RESET processes, between two states, the high resistance state (HRS), and the low resistance state (LRS). In bipolar RRAM, the SET and RESET processes occur through the formation/dissolution of conductive filaments \cite{ielmini2011}. This mechanism is controlled by a compliance current. Depending on the compliance current, during the switching of a metal-insulator-metal structure between the HRS and the LRS, random telegraph noise (RTN) may be observed. RTN can be reduced with proper SET and RESET processes, but, on the contrary, fluctuations of current at different resistance states can be exploited for non-destructive studies of switching phenomena at microscopic scale \cite{ielmini10}. RTN is also used as the principle for true random number generators based on a contact resistive RAM \cite{Huang}.
As resistive switching devices, the memristors are also of interest for implementation of Boolean logic gates \cite{borghetti2010,Rose2012,Ella2013,Kvatinsky2014,Vourkas2014}, for applications as nonvolatile memories \cite{nonvolatileMem2011}, and as synapses in neuromorphic circuits \cite{Jo2010}.
Due to its non-linearity, the use of the memristor as nonlinear component in chaotic circuits \cite{muthuswamy10} has been envisaged. To this aim the memristor has been used to substitute for Chua's diode in versions of Chua's circuits \cite{itoh08,muthuswamy10,253} and simulated chaos has been observed \cite{82,61,70,232}. Most of the memristor-based oscillators used in these chaotic circuit simulations assume ideal characteristics for the memristor, \emph{e.g.} cubic or piece-wise-linear (PWL) nonlinearities \cite{itoh08}. Since the Strukov memristor is a passive non-symmetrical element having a non-linearity different from the one most frequently investigated, its use for chaotic circuits design is not trivial. Specifically, the non-linearity assumed in literature is usually a continuously varying function or a piece-wise linear-continuously varying function for filamentary type devices.
We would like to demonstrate experimentally that the memristor can be used as the non-linear element in a chaotic circuit. The first step is to demonstrate that simulated circuits including a realistic model of memristor exhibit chaos, this has been done in ~\cite{buscarino12,Buscarino2013} where recently a gallery of chaotic circuits using the HP model has been proposed: these papers utilised a particular configuration of two memristors connected in anti-parallel, which allowed the authors obtain a symmetrical characteristic suitable for chaos generation. This sub-circuit has been experimentally realised in ~\cite{spike} where circuits made of 2 or 3 memristors were found to give rise to complex dynamics. What has not been done is to conclusively demonstrate chaotic dynamics in an experimental circuit containing a memristor and to do so with only one non-linearity in the circuit (i.e. only one memristor). In our experiments we do not apply specific SET and RESET processes, but we drive the memristor by a signal which is function of the actual status of the device, exploiting its switching properties to obtain deterministic chaos.
The aim of this work is to provide an experimental demonstration of chaos with memristive devices. For this purpose, a simple experimental set-up is proposed in Sec.~\ref{sec:modelA}, where chaos can be uniquely attributed to the presence of the memristor, as described in Sec.~\ref{sec:resultsA}. We then perform a numerical simulation of the set-up as described in Sec.~\ref{sec:modelB} using the simplest model available and explore the parameter space to demonstrate that the chaos is a robust phenomenon observed over a large parameter space in Sec.~\ref{sec:resultsB}. In Sec.~\ref{sec:conclusion} the conclusions of the work are drawn.
\section{Methodology}
\label{sec:model}
\subsection{Experimental}
\label{sec:modelA}
The idea underlying the experimental set-up originates from the way in which memristors are usually characterized. One of the main fingerprints of memristors is the pinched hysteresis loop in the $v-i$ plane measured under periodic excitation \cite{Chua2011,Biolek2013}. A commonly used technique to obtain the $v-i$ characteristics is to apply a sinusoidal voltage input to the memristor (alternatively, the time response is studied through the application of pulse functions) and to measure the current through the device. To do this, a (Keithley 2602) programmable sourcemeter is commonly used because it allows the tuning of the input parameter and the recording of the current. The core of this work is to dynamically change the parameters of the input waveforms as a function of the actual state of the memristor, to our knowledge, this is the first experiment using such a feedback loop.
The memristor used in this work has a sandwich structure of Al-TiO$_2$-Al: the aluminium electrodes were sputter-coated onto PET plastic substrate via a mask as in \cite{hackett,ella2014}, the sol-gel layer \cite{ella2014} was deposited via spin-coating (spun at 33r/s for 60s) and then left under vacuum to hydrolyse for an hour as in \cite{ella2011}. The electrodes are 4mm wide, crossed at 90$^{\circ}$, giving an active area of 16mm$^2$, the titanium gel layer is 40nm thick. In contrast to other more expensive techniques for the fabrication of the active layer, the spin-coated memristor was created with a simple procedure, and without the need for a forming step.
The experimental methodology consists of two separate steps. In the first, we test the memristive behavior by exciting it with a periodic input and observing the behavior in the $v-i$ plane. In the second step, we drive the memristor with pulses which are related to the actual state of the memristor through a law discussed in details below. The first step corresponds to operating the memristor in open loop conditions, while the second in feedback mode. In feedback conditions the whole system operates without external inputs, so the actual status of the system (which includes the memristor status and the voltage applied to it) is the result of an autonomous evolution.
The memristor characterisation analysis (first step of our methodology) was performed by setting the sourcemeter to run linear voltage sweeps between a range of $\pm 1V$, with a voltage step size of $0.02V$ and a settling time of $0.01s$ (settling time is the delay period after which the measurement is made). The signal may be viewed as the DC equivalent of an AC waveform with a frequency of $0.5Hz$. The measurements have been repeated several times to test the device performance over iterated cycles.
In the second step, we investigate the effect of establishing a relationship between the measured current and the next sample of the applied voltage signal, that is, we drive the memristor on the basis of the current flowing through it. This is particularly simple to realize since it only requires a memristor and a programmable sourcemeter.
The scheme adopted to test for non-linear dynamics is illustrated in Fig.~\ref{fig:schema2}, where $v(t)$ indicates the voltage applied to the memristor, i.e. a digital-to-analog converted signal. The memristor current, indicated as $i(t)$, is sampled as a sequence of measurements $i_h$. Each sample $i_h$ is used to generate the next voltage sample $v_{h+1}$ through the relation:
\begin{equation}
\label{eq:vh1}
v_{h+1}=r(1-k i_h)
\end{equation}
\noindent where $k$ and $r$ are constant tunable parameters. The sequence of samples $v_{h}$ is converted into the analog signal $v(t)$ through the zero-order hold (ZOH), so that:
\begin{equation}
\label{eq:vt}
\begin{array}{ll}
v(t)=v_h, & t_h \leq t < t_{h+1}
\end{array}
\end{equation}
\noindent with $t_h=h\Delta T$, where $\Delta T$ is the sampling time. The sampling of $i(t)$ occurs immediately before the sweep of the voltage from $v_h$ to $v_{h+1}$, that is, $i_h=i(t)|_{t=t_{h+1}^-}$. We note that Eq. (\ref{eq:vh1}) is linear, so that any non-linearity in the system comes from the memristor.
The sampler, the processor implementing Eq. (\ref{eq:vh1}) and the ZOH are all implemented in the sourcemeter. During operations in feedback conditions, in the sourcemeter Eq. (\ref{eq:vh1}) is iterated, so that the voltage levels are automatically generated for a given time window during which a number of samples of the waveforms is acquired. Specifically, we used a Keithley 2602 programmable sourcemeter to apply a step-wise sweep voltage (both the step amplitude and the duration can be programmed). We used steps with fixed duration so that the applied waveform is a continuous-time signal generated by the conversion of a discrete-time signal through a zero-order hold. The measurement is made at each step after a specified delay period (the sampling time $\Delta T$). Lyapunov exponents of the data were calculated using the TISEAN package \cite{tisean}, which is based on Rosenstein's algorithm \cite{Rosenstein}.
\begin{figure}
\centering {{\includegraphics[width=8.5cm]{schema2.eps}}}
\caption{\label{fig:schema2} The memristor based circuit.}
\end{figure}
\subsection{Simulational}
\label{sec:modelB}
Many models of memristors, ranging from very simple ones to ones incorporating a very detailed level of description of the phenomena occurring in the device, have been proposed in literature. The first electronic engineering model of a memristor was Chua's original model in 1971 \cite{Chua1971}, this model offered a way to relate the circuit measurables to the state of the system, but was too abstract to model specific systems. The first materials science model of a memristor was in the original Strukov paper \cite{strukov08}, this included a very simple approach based on modelling the memristor as a space-conserving variable resistor, where the boundary, $w$, between TiO$_2$ and TiO$_{2-x}$ (Fig.~\ref{fig:HPmem}) is used as a state variable which holds the memory of the device (this fits the chemistry of the system, as when the voltage is removed the oxygen vacancies do not dissipate like electrons or holes do in semiconductors, but remain `remembering' a state). The Strukov model assumes a linear dopant drift under a finite uniform field, which is widely thought to be unlikely in thin-film~\cite{86}. Several papers have attempted to improve on this, by introducing a window function to prevent the system from going outside of its bounds 0 and D, the thickness of the device, and slowing the boundary down near the edges \cite{2,46,306}. Nonetheless, the original model allows ease of modelling and simplicity to the interpretation of results as long as the simulation code takes care to avoid $w$ taking un-physical values, and has been widely adopted for simulations \cite{345,346}. More realistic models have also been proposed, an example is \cite{220} which gives good agreement with the experimental data, but requires 8 fitting parameters and contains difficult to simulate terms.
\begin{figure}
\centering {{\includegraphics[width=8.5cm]{HPmem.eps}}}
\caption{\label{fig:HPmem} Schematic representation of a TiO$_2$ memristor.}
\end{figure}
We show here that the presence of chaos is demonstrated even with a very simple model capturing only the main characteristics of the memristive behavior. Specifically, numerical results are obtained by substituting the memristor in the scheme of Fig.~\ref{fig:schema2} with its modelling equations as proposed by Strukov et al. \cite{strukov08}: the resulting simulational scheme is shown in Fig.~\ref{fig:schema1}. The relationship between current and voltage in the memristor is governed by:
\begin{equation} \label{eqn:hp model}
i(t)=v(t)/ \left( R_{ON}{w(t)\over{D}}+{R_{OFF}\left( {1-{w(t)\over{D}}} \right)} \right) \: .
\end{equation}
The variable $w(t)$ (the width of the doped region) represents an internal memory variable limited to values between zero and $D$, the thickness of the device, and is characterized by the following dynamics:
\begin{equation} \label{eq:w2}
\dot{w}(t)={\eta \frac{\mu_v R_{ON}}{D}} i(t)
\end{equation}
\begin{figure}
\centering {{\includegraphics[width=8.5cm]{schema1.eps}}}
\caption{\label{fig:schema1} Scheme used for numerical analysis of the memristor based circuit.}
\end{figure}
\noindent where $\eta$ characterizes the polarity of the memristor ($\eta=1$ if the doped region is expanding, $\eta=-1$ otherwise), $\mu_v$ is the oxygen vacancy ion mobility, and $R_{ON}$ ($R_{OFF}$) is the resistance of the device in its lowest (highest) resistance state.
Following \cite{strukov08} Eqs.~(\ref{eqn:hp model}) and (\ref{eq:w2}) may be rescaled to obtain:
\begin{equation} \label{eqn:hp model2}
i(t)=v(t)/ \left( \bar{w}(t)+{\beta\left({1-\bar{w}(t)}\right)} \right )
\end{equation}
\noindent and
\begin{equation} \label{eq:w2_2}
\dot{\bar{w}}(t)={\eta} i(t)
\end{equation}
\noindent where $\beta=R_{OFF}/R_{ON}$ and time is now expressed in the units of $t_0=D^2/\mu_v v_0$, with $v_0=1V$. The numerical simulations have been based on Eqs. (\ref{eqn:hp model2}) and (\ref{eq:w2_2}), which are dimensionless and contain only one parameter ($\beta$).
\section{Results}
\label{sec:results}
\subsection{Experimental results}
\label{sec:resultsA}
The experimental results are reported for two devices, indicated as memristor 1 and memristor 2. The two memristors were fabricated with the same procedure, however parametric tolerances due to the fabrication process do appear. We first describe the $v-i$ characterization under periodic excitation of the memristors analyzed in this work. The results, which refer to memristor 1, are reported in Fig.~\ref{fig:v-iMv47}, showing that the resistance change is not fully reversed over the range of the hysteresis loop. However, in view of the application as chaos generator, this does not represent a severe issue, since the non-linearity is not required to be time-invariant. Similar behavior has been obtained for memristor 2.
\begin{figure}
\centering {{\includegraphics[width=8.5cm]{v-iMv47.eps}}}
\caption{\label{fig:v-iMv47} Two $v-i$ curves for one of the memristors (memristor 1) used in the experiments (similar results were seen for memristor 2 and are not shown). The input signal is a linear voltage sweep between a range of $\pm 1V$, with a voltage step size of $0.02V$ and a settling time of $0.01s$, that is the DC equivalent of an AC waveform with a frequency of $0.5Hz$.}
\end{figure}
The spin-coated TiO$_2$ memristor was then controlled by a feedback signal generated by using the scheme of Fig.~\ref{fig:schema2}. In both cases chaos has been observed, but the two devices required a different parameter tuning of the control law (\ref{eq:vh1}). For memristor 1 we have chosen $r=0.25$ and $k=1750$ and obtained the waveforms shown in Fig.~\ref{fig:timeevolutioniandv2}. For memristor 2 we have chosen $k=850$ and $r=0.25$; the time evolution of $i(t)$ and $v(t)$ is reported in Fig.~\ref{fig:timeevolutioniandv}. Starting from the acquired data reported in Figs.~\ref{fig:timeevolutioniandv2} and~\ref{fig:timeevolutioniandv} we have estimated the largest Lyapunov exponents: for the data referring to memristor 1 (Fig.~\ref{fig:timeevolutioniandv2}) the largest Lyapunov exponent is $\lambda_{max}=0.7947$, and for the data from memristor 2 (Fig.~\ref{fig:timeevolutioniandv}) $\lambda_{max}=0.9230$. The existence of positive values for the largest Lyapunov exponent indicates that the behavior is chaotic. As the rest of the set-up was linear, this chaos arises from the memristor, and thus we have demonstrated chaotic dynamics from a single memristor device.
\begin{figure}
\centering
\subfigure[]{\includegraphics[width=8cm]{i.eps}}
\subfigure[]{\includegraphics[width=8cm]{v.eps}}\\
\caption{\label{fig:timeevolutioniandv2} Experimental results: time evolution of current and voltage in the memristor 1 based circuit for $r=0.25$ and $k=1750$.}
\end{figure}
\begin{figure}
\centering
\subfigure[]{\includegraphics[width=8cm]{iM49.eps}}
\subfigure[]{\includegraphics[width=8cm]{vM49.eps}}\\
\caption{\label{fig:timeevolutioniandv} Experimental results: time evolution of current and voltage in the memristor 2 based circuit for $r=0.25$ and $k=850$.}
\end{figure}
\subsection{Numerical results}
\label{sec:resultsB}
In order to demonstrate that the chaos observed with our devices was a real effect generally due to the memristor behaviour, we simulated the set-up and numerically analyzed it with respect to the parameters $k$ and $\beta$. Several regions in the parameter space for which chaotic behavior arises were found. An example of chaotic behavior is reported in Fig.~\ref{fig:trend} where the time evolutions of the variables $v(t)$ and $i(t)$ are shown for $k=10$ and $\beta=100$. In Fig.~\ref{fig:bifk} the bifurcation diagram with respect to $k$ (for $\beta=100$) is illustrated, it shows alternating windows of periodic behavior and chaos. The bifurcation diagram with respect to $\beta$ is reported in Fig.~\ref{fig:bifbeta} (for $k=10$), showing how chaos is preserved for a wide range of values of the constitutive parameter $\beta$. Since $\beta$ may vary in real memristors due to technology, fabrication process and device characteristics, the robustness with respect to this parameter is particularly important.
\begin{figure}
\centering
\subfigure[]{{\includegraphics[width=8cm]{iSim.eps}}}
\subfigure[]{{\includegraphics[width=8cm]{vSim.eps}}}
\caption{\label{fig:trend} Trend of the current (a) and voltage (b) of the memristor circuit for $k=10$ and $\beta=100$.}
\end{figure}
\begin{figure}
\centering {{\includegraphics[width=8cm]{bif_k.eps}}}
\caption{\label{fig:bifk} Bifurcation diagram of the memristor circuit with respect to $k$ ($\beta=100$, $r=1$).}
\end{figure}
\begin{figure}
\centering {{\includegraphics[width=8cm]{bif_beta_k10.eps}}}
\caption{\label{fig:bifbeta} Bifurcation diagram of the memristor circuit with respect to $\beta$ ($k=10$, $r=1$).}
\end{figure}
We observe that the numerical results have been obtained with a dimensionless set of equations; the parameters of the control law ($k$ and $r$) have been rescaled during the experimental tests. The numerical results show that chaos can be generated using this approach. However, the waveforms observed in the experiments and in the numerical simulations do not coincide, indicating that the simplified mathematical model of Eqs. (\ref{eqn:hp model})-(\ref{eq:w2}) is able to capture the general behavior of the system, but not the details of the waveforms observed in the experiments. This shows that the chaotic behaviour comes from an aspect of the memristor that is captured qualitatively by the simplest possible memristor model. However, we strongly suspect that further work involving using more complicated models will produce numerical results closer to experimental data, and this might provide memristor modellers a test case for the improvement of their models.
It is interesting to ask why the memristor allows chaos to arise in such a simple system, and the fact that such a simple model of memristance demonstrates chaotic dynamics provides a clue. As mentioned earlier, the memristor non-linearity is not time-invariant, and it is known that memristors possess a memory. We posit that it is specifically the interaction of this time-varying memory with the time-based feedback that gives rise to chaos in this system (as the voltage for each step is generated on the data from the previous step, there is a memory in the experimental system). The simulation equations are so simple that there is only one time-varying variable, $w(t)$ which is the internal state of the memristor and its memory. This variable depends on $i(t)$ (which is linearly updated through the feedback) but also on the past history of the memristor through a functional which, instead, is non-linear.
\section{Conclusions}
\label{sec:conclusion}
In this work experimental findings on the generation of chaos with a spin-coated memristor are presented. The experimental set-up consists of a single memristor driven by a linear control law relating the applied voltage to the actual value of the current flowing in it. This feedback law does not introduce any non-linearity, so that the observed behavior can be uniquely attributed to the memristor characteristics.
Chaotic waveforms have been experimentally observed in several samples of spin-coated memristors and confirmed by a numerical analysis. The largest Lyapunov exponent has been calculated in the experimental time-series, and the positive value obtained indicates chaotic dynamics. A simple model of memristive behavior has been shown to be sufficient to reproduce the onset of chaos, although capturing all the features of the waveforms observed in the experiments requires a more detailed model. The experiments showing the irregularity of the behavior of the circuit demonstrate the suitability of this simple approach to generate chaos with memristors and suggests that memristors might be useful components for hardware chaos-based encryption.
The analysis performed through the use of the Strukov model suggests that the main mechanism explaining chaos in the experimental and simulation behaviors is the presence of the pinched hysteretic behavior. However, as in the experiments specific SET and RESET processes are not applied, it cannot be excluded the presence of RTN, which on the contrary could be a possible explanation for the differences between experimental and simulation results.
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 8,696
|
Rapport is een hechte en harmonieuze relatie tussen twee mensen met empathische aandacht en communicatie. Tussen therapeut en patiënt wordt dit wel werkrelatie, therapeutische relatie of therapeutische alliantie genoemd.
Aanvankelijke begripsvorming
Franz Anton Mesmer introduceerde het begrip dierlijk magnetisme of mesmerisme in 1773 waarmee hij stelde genezingen uit te kunnen voeren. Tijdens de behandeling plaatste de magnetiseur zich in de en rapport-positie – met de voeten en knieën tegen die van de patiënt – en legde de hand op het pijnlijke gebied. Deze sessies konden uren duren en een zeer hechte band met de behandelde was noodzakelijk. Pierre Janet bouwde voort op het werk van Mesmer met hypnose, rapport magnétique. Janet hechtte groot belang aan een goede relatie tijdens therapie, maar zag ook de gevaren van een krachtige wederzijdse aantrekkingskracht die de relatie tussen arts en patiënt met zich mee kan brengen en voor hem moest dit geneutraliseerd worden. Bij Sigmund Freud werd rapport overdracht, wat voor hem juist onderwerp van analyse was.
Therapeutische relatie
Het belang van een goede relatie tussen therapeut en patiënt wordt al lang onderkend. Freud stelde dat een serieuze belangstelling en sympathiek begrip door de therapeut van de patiënt het gezonde deel van de patiënt toe laat staan om een positieve band aan te gaan met de therapeut. Voor Carl Rogers was een goede relatie zelfs de belangrijkste factor voor succes bij zijn cliëntgerichte psychotherapie. Een niet-geaccepteerd persoon kan zich zo binnen de beperkte relatie met de therapeut geaccepteerd voelen en zo zichzelf leren accepteren, waarna zelfactualisatie kan plaatsvinden. Oprechte empathie van de therapeut is hiervoor een voorwaarde.
Voor strikte behavioristen was echter vooral de techniek relevant en de relatie nauwelijks van belang. Daar ging Edward S. Bordin tegenin:
I propose that the working alliance between the person who seeks change and the one who offers to be a change agent is one of the keys, if not the key, to the change process.
Daarbij onderscheidde hij drie processen:
[...] we can speak of the working alliance as including three features: an agreement on goals, an assignment of task or a series of tasks, and the development of bonds.
Er moet dus overeenstemming zijn tussen patiënt en therapeut over de doelen, de taakverdeling en de op te bouwen band, wat goed past bij cognitieve gedragstherapie.
Vanuit de psychoanalyse onderscheidden Frans de Jonghe, Piet Rijnierse en Rob Janssen de overdrachtsrelatie, de werkrelatie, de realistische relatie en de primaire relatie. In de overdrachtsrelatie komen relaties en omgangsstijl uit het verleden naar voren, in de werkrelatie de samenwerking, de realistische relatie is de gezonde volwassen relatie voor zover die niet al in de werkrelatie naar voren komt en in de primaire relatie komen de gevoelens en behoeften uit de jeugd naar voren.
Een goede relatie blijkt inderdaad bij te dragen aan de effectiviteit van de therapie, vooral als de patiënt het zo ervaart, de ervaring van de therapeut blijkt minder relevant. Het type therapie en de duur leken niet relevant bij Adam Horvath en Dianne Symonds. John C. Norcross en Bruce Wampold zagen dat het belang van de relatie minstens zo groot was als de gebruikte behandelmethode. Ook zou uitval verminderen bij een goede relatie, al neemt het belang daarvan af bij een hoger opleidingsniveau. Bij langere behandelduur nam uitval ook af.
Literatuur
; (2010): 'Building the Working Alliance in Brief Psychotherapies' in Communication in Cognitive Behavioral Therapy, Springer
(2014): Therapeutic Communication. Developing Professional Skills, SAGE
Noten
Sociale psychologie
Psychotherapie
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 1,684
|
set -euo pipefail
# This script will import slides from the Hacker Slides instance that
# Asheesh has been using into a directory structure that can live in
# this repo.
mkdir -p student-handout
pushd student-handout
# First, nab the slides.
wget --header "Cookie: $2" "https://$1/slides.md"
# Then, nab the static directory, and all that it references.
mkdir -p static
pushd static
wget -nH --cut-dirs=1 --mirror --no-parent --header "Cookie: $2" "https://$1/static/slides.html" || true # OK if this fails.
# Grab these extra dependencies.
for thing in "plugin/markdown/markdown.js" "plugin/markdown/marked.js" "css/print/paper.css" "plugin/zoom-js/zoom.js" "plugin/notes/notes.js" "plugin/notes/notes.html"
do
wget -nH --cut-dirs=1 --mirror --no-parent --header "Cookie: $2" "https://$1/static/$thing" || true # OK if this fails.
done
# Prepare this for git commit, but let the user be the one to actually
# commit it.
popd; popd
git add .
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 2,523
|
James Lind Institute (JLI) is a global training and education provider in India with information and administrative centers in Singapore, US, and many other regions such as Africa, and the Middle East. JLI is accredited by various agencies and is also a member of the International Council of Open and Distance Education (ICDE), Norway. JLI works with many fortune 500 healthcare companies and has students and alumni spread across more than 80 countries.
Self-paced learning with 24 x 7 access to the study material. Online submission of projects and assignments.
Admission process for James Lind Institute's programs is very simple. Create an account on this website, enter some basic information about yourself in your profile, upload your documents, save the information and submit the application online.
JLI offers a wide range of accredited programs in areas such as clinical & translational research, healthcare management, public health, drug safety etc. that will enhance your profile, improve your skill set & make you shine at your workplace.
We are committed to providing you with all the support required to pursue your program smoothly. Our support systems such as the Student Success Team, online chat support, job portal & other resources make for an excellent student experience.
Welcome to James Lind Institute (JLI). JLI is a global training and education provider with programs and courses for professional development conducted in a self-paced learning environment. Programs cover everything from taking online public health courses, clinical research courses, to earning a Master's degree in Health Management. A post graduate diploma or Master's degree greatly enhances your profile and enables you to take advantage of abundant career opportunities in the biotechnology, pharmaceutical and healthcare industries.
Career Building Programs: Our unique industry specific programs are designed to help you move to the next level in your career. We offer a wide range of programs in categories that include Clinical Development Sciences, Public Health & Topical Medicine Course, Pharmaceutical Medicine, and Medical Journalism. Choose from classes that provide healthcare management training, pharmacovigilance courses, hospital management courses, medical writing training, and much more.
Although you will have access to study materials at all hours of the day or night and study is self-paced, you won't be left to struggle on your own. At JLI, our students are provided with support to ensure student success and a productive and enjoyable experience. Individuals are able to find help through our Student Success Team, access support via online support, a job portal, and additional resources.
We encourage you to browse our website to explore our programs, find answers to commonly asked questions, and read what our students have to say about training and programs from JLI. In addition, you can calculate fees, find admissions information, and complete and submit your application for admissions through our website.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 5,700
|
Q: Docker CMD cannot add new user, permission denied I have a docker image that at the end launch this script:
# Start the server
CMD [ "bash", "/app/start.sh"]
And then my script start.sh
# Create ssh user
useradd -m -d /home/${SSH_MASTER_USER} -G ssh ${SSH_MASTER_USER} -s /bin/bash
echo "${SSH_MASTER_USER}:${SSH_MASTER_PASS}" | chpasswd
echo 'PATH="/usr/local/bin:/usr/bin:/bin:/usr/sbin"' >> /home/${SSH_MASTER_USER}/.profile
And I keep getting this error:
useradd: Permission denied.
useradd: cannot lock /etc/passwd; try again later.
If run with sudo it says:
sudo: no tty present and no askpass program specified
Tried to fix it like this:
sudo -S useradd -m -d /home/${SSH_MASTER_USER} -G ssh ${SSH_MASTER_USER} -s /bin/bash
But now is asking me for a password, but there is not password...
A: my base image is using root.
so i am adding appuser and installing sudo like this:
RUN adduser --disabled-password --gecos "" appuser
RUN apt-get update
RUN apt-get install -y sudo
RUN echo "appuser ALL=(root) NOPASSWD:ALL" > /etc/sudoers.d/appuser && chmod 0440 /etc/sudoers.d/appuser
Later on when you ready to use new user for all commands:
USER appuser:appuser
that is it, now appuser which is going to be user by default can use sudo without password.
A: I have solved by adding my existing user to the sudo group:
RUN echo "myUser ALL=(root) NOPASSWD:ALL" > /etc/sudoers.d/docker && chmod 0440 /etc/sudoers.d/docker
then the:
sudo -S useradd -m -d /home/${SSH_MASTER_USER} -G ssh ${SSH_MASTER_USER} -s /bin/bash
worked without password
|
{
"redpajama_set_name": "RedPajamaStackExchange"
}
| 8,137
|
{"url":"https:\/\/idris.readthedocs.io\/en\/latest\/proofs\/definitional.html","text":"In order to understand how to write proofs in Idris I think its worth clarifying some fundamentals, such as,\n\n\u2022 Propositions and judgments\n\u2022 Boolean and constructive logic\n\u2022 Curry-Howard correspondence\n\u2022 Definitional and propositional equalities\n\u2022 Axiomatic and constructive approaches\n\n# Propositions and Judgments\u00b6\n\nPropositions are the subject of our proofs, before the proof then we can\u2019t formally say if they are true or not. If the proof is successful then the result is a \u2018judgment\u2019. For instance, if the proposition is,\n\n 1+1=2\n\nWhen we prove it, the judgment is,\n\n 1+1=2 true\n\nOr if the proposition is,\n\n 1+1=3\n\nObviously we can\u2019t prove it is true, but it is still a valid proposition and perhaps we can prove it is false so the judgment is,\n\n 1+1=3 false\n\nThis may seem a bit pedantic but it is important to be careful, in mathematics not every proposition is true or false for instance, a proposition may be unproven or even unprovable.\n\nSo the logic here is different from the logic that comes from boolean algebra. In that case what is not true is false and what is not false is true. The logic we are using here does not have this \u2018law of excluded middle\u2019 so we have to be careful not to use it.\n\nA false proposition is taken to be a contradiction and if we have a contradiction then we can prove anything, so we need to avoid this. Some languages, used in proof assistants, prevent contradictions but such languages cannot be Turing complete, so Idris does not prevent contradictions.\n\nThe logic we are using is called constructive (or sometimes intuitional) because we are constructing a \u2018database\u2019 of judgments.\n\nThere are also many other types of logic, another important type of logic for Idris programmers is \u2018linear logic\u2019 but that\u2019s not discussed on this page.\n\n# Curry-Howard correspondence\u00b6\n\nSo how to we relate these proofs to Idris programs? It turns out that there is a correspondence between constructive logic and type theory. They are the same structure and we can switch backward and forward between the two notations because they are the same thing.\n\nThe way that this works is that a proposition is a type so this,\n\nIdris> 1+1=2\n2 = 2 : Type\n\n\nis a proposition and it is also a type. This is built into Idris so when an \u2018=\u2019 equals sign appears in a function type an equality type is generated. The following will also produce an equality type:\n\nIdris> 1+1=3\n2 = 3 : Type\n\n\nBoth of these are valid propositions so both are valid equality types. But how do we represent true judgment, that is, how do we denote 1+1=2 is true but not 1+1=3. A type that is true is inhabited, that is, it can be constructed. An equality type has only on constructor \u2018Refl\u2019 so a proof of 1+1=2 is\n\nonePlusOne : 1+1=2\nonePlusOne = Refl\n\n\nSo how can Refl, which is a constructor without any parameters, construct and equality type? If we type it on its own then it can\u2019t:\n\nIdris> Refl\n(input):Can't infer argument A to Refl, Can't infer argument x to Refl\n\n\nSo it must pattern match on its return type:\n\nIdris> the (1=1) Refl\nRefl : 1 = 1\n\n\nSo now that we can represent propositions as types other aspects of propositional logic can also be translated to types as follows:\n\n propositions example of possible type A x=y B y=z and A \/B Pair(x=y,y=z) or A \/ B Either(x=y,y=z) implies A -> B (x=y) -> (y=x) for all y=z exists y=z\n\n## And (conjunction)\u00b6\n\nWe can have a type which corresponds to conjunction:\n\nAndIntro : a -> b -> A a b\n\n\nThere is a built in type called \u2018Pair\u2019.\n\n## Or (disjunction)\u00b6\n\nWe can have a type which corresponds to disjunction:\n\ndata Or : Type -> Type -> Type where\nOrIntroLeft : a -> A a b\nOrIntroRight : b -> A a b\n\n\nThere is a built in type called \u2018Either\u2019.\n\n# Definitional and Propositional Equalities\u00b6\n\nWe have seen that we can \u2018prove\u2019 a type by finding a way to construct a term. In the case of equality types there is only one constructor which is \u2018Refl\u2019. We have also seen that each side of the equation does not have to be identical like \u20182=2\u2019. It is enough that both sides are definitionaly equal like this:\n\nonePlusOne : 1+1=2\nonePlusOne = Refl\n\n\nSo both sides of this equation nomalise to 2 and so Refl will type match and the proposition is proved.\n\nWe don\u2019t have to stick to terms, can also use symbolic parameters so the following will compile:\n\nvarIdentity : m = m\nvarIdentity = Refl\n\n\nIf a proposition\/equality type is not definitionaly equal but is still true then it is propositionaly equal. In this case we may still be able to prove it but some steps in the proof may require us to add something into the terms or at least to take some sideways steps to get to a proof.\n\nEspecially when working with equalities containing variable terms (inside functions) it can be hard to know which equality types are definitially equal, in this example plusReducesL is \u2018definitially equal\u2019 but plusReducesR is not (although it is \u2018propositionaly equal\u2019). The only difference between them is the order of the operands.\n\nplusReducesL : (n:Nat) -> plus Z n = n\nplusReducesL n = Refl\n\nplusReducesR : (n:Nat) -> plus n Z = n\nplusReducesR n = Refl\n\n\nplusReducesR gives the following error:\n\n- + Errors (1)\n-- proof.idr line 6 col 17:\nWhen checking right hand side of plusReducesR with expected type\nplus n 0 = n\n\nType mismatch between\nn = n (Type of Refl)\nand\nplus n 0 = n (Expected type)\n\nSpecifically:\nType mismatch between\nn\nand\nplus n 0\n\n\nSo why is \u2018Refl\u2019 able to prove some equality types but not others?\n\nThe first answer is that \u2018plus\u2019 is defined in such a way that it splits on its first argument so it is simple to prove when 0 is the first argument but not the second. So what is the general way to know if Refl will work?\n\nIf an equality type can be proved\/constructed by using Refl alone it is known as a definitional equality. In order to be definitinally equal both sides of the equation must normalise to unique values. That is, each step in the proof must reduce the term so each step is effectively forced.\n\nSo when we type 1+1 in Idris it is immediately converted to 2 because definitional equality is built in.\n\nIdris> 1+1\n2 : Integer\n\n\nIn the following pages we discuss how to resolve propositionaly equalies.\n\n# Axiomatic and Constructive Approaches\u00b6\n\nHow should we define types so that we can do proofs on them? In the natural numbers with plus example we could have started by treating it as a group based on the plus operator. So we have axioms:\n\n\u2022 for all x,y : x+y=y+x\n\u2022 for all x: x + 0 = x = 0 + x\n\u2022 for all x,y,z: (x + y) + z = x + (x + z)\n\nThen we can implement \u2018+\u2019 so that it respects these axioms (presumably implemented in hardware).\n\nThese are axioms, that is a propositions\/types that are asserted to be true without proof. In Idris we can use the \u2018postulate\u2019 keyword\n\ncommutePlus postulate: x -> y -> plus x y = plus y x\n`\n\nAlternatively we could define the natural numbers based on Zero and Successor. The axioms above then become derived rules and we also gain the ability to do inductive proofs.\n\nAs we know, Idris uses both of these approaches with automatic coercion between them which gives the best of both worlds.\n\nSo what can we learn from this to implement out own types:\n\n\u2022 Should we try to implement both approaches?\n\u2022 Should we define our types by constructing up from primitive types?\n\nProof theory affects these design decisions.","date":"2019-08-18 18:55:57","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\": 0.7855381369590759, \"perplexity\": 1054.532475561555}, \"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-2019-35\/segments\/1566027313996.39\/warc\/CC-MAIN-20190818185421-20190818211421-00337.warc.gz\"}"}
| null | null |
Osteo on the move
Osteopath Paul Hermann used his studies at VU as a springboard to an exciting career as an international consultant and Australian clinic owner.
"I chose osteopathy at VU because I believed it was the best program with the best location – central CBD.
The best part about my course was the lecturers. They had a wide range of experiences and backgrounds which enhanced the spectrum of learning substantially. The VU student-run osteopathy clinic is also fantastic grounding for moving into private practice.
I now run two successful multidisciplinary clinics with nearly 40 staff. I have lectured throughout the world several times and Australia. I recently worked at the Beach Volleyball World Championships in Holland and will be doing more consulting work overseas next year. My goal is to continue to be challenged and do what I love for as long as possible.
VU gave me the knowledge, experience and motivation to continue my studies and to go to the next level in my career."
Courses studied
Bachelor of Clinical Science, now Bachelor of Science (Osteopathy)
Master of Health Science (Osteopathy)
Find out more about health and biomedicine at VU.
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 513
|
Бандхавгарх (англ. Bandhavgarh National Park, хінді बांधवगढ राष्ट्रीय उद्दान) - національний парк в Індії. Розташований на сході штату Мадхья-Прадеш, в окрузі Умарія. Територія парку - горбиста, є безліч печер, загальна площа становить 437 км².
Створений в 1968 році, спочатку парк займав лише 105 км², але був збільшений в 1982 році. У 1993 він був зарахований до тигрові заповідникам. Свою назву парк отримав від фортеці Бандхавгарх, розташованої неподалік
Національні парки Індії
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 4,704
|
<header>
header
</header>
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 1,655
|
Luther College was a private black school in New Orleans, Louisiana, United States. It was established by the Evangelical Lutheran Synodical Conference of North America in 1903 as part of the conference's missionary work among African Americans in the Southern United States following the American Civil War. The school was founded the same year as Immanuel Lutheran College in Concord, North Carolina, and both schools had the same three departments: a secondary school, a normal school, and a seminary.
In 1910, the school closed its seminary, having graduated only one student in that department. In 1925, the conference closed the college, citing the lack of its seminary and the high estimated cost required for new school buildings. Funds for the college were diverted to the Alabama Lutheran College in Selma, Alabama, which had been established in 1922.
History
Following the American Civil War, the Evangelical Lutheran Synodical Conference of North America began to support missionary work directed towards the African American community in the southern United States. As part of this outreach, the conference founded several schools and colleges throughout the region. In 1903, the conference founded Immanuel Lutheran College in Concord, North Carolina, and later that year, the conference founded Luther College in New Orleans. Luther College officially opened in September 1903 in the by vestry of St. Paul's Church, under the leadership of the Reverend F. J. Lankenau. While classes were initially held inside the church, a school building was erected in 1904. The building, located adjacent to St. Paul's, was dedicated in November of that year and cost $4,007. Lankenau was assisted in his instructions by two assistants, the Reverends K. Kretzschmar and J. Kossmann. The school was modeled after the North Carolina institute, and like that school, it operated three departments: a high school, a normal school, and a seminary. In 1904, the school had 25 students, with two students in the seminary. Additionally, St. Paul's operated a parochial school that had an enrollment of over 200 students.
A 1908 directory of colleges and private schools in the United States listed Luther College as a co-ed institution that offered preparatory and college classes. That year, the school had 14 male and 9 female students, ranging in age from 13 to 24 years old, being taught by three white male teachers, with six students studying for the Christian ministry. In September 1908, Lankenau departed from New Orleans, with his duties as both president of Luther College and minister of St. Paul's being taken up by Frederick Wenger. By 1910, the school's property was valued at $6,000 and it had annual expenses of $2,500, which were primarily covered by the conference, with additional funding secured by tuition. That same year, the seminary closed, having graduated only one student, Clavin Peter Thompson, during its existence.
In February 1914, the school was visited by members of the United States Office of Education as part of a nationwide study on African American education. At the time, the school was headed by President R. E. Schmidt and had a staff of two teachers, all white men. The school had an enrollment of 37 students and offered ten grades, with sewing classes offered to students above the seventh grade. The office reported that most of the students were in the elementary grades. For the 1913–1914 financial year, the school had an income of $1,400, of which $1,300 came from the conference and the remainder from tuition. The property value was estimated at $10,500, including the land lot, two-story building, and various pieces of equipment. According to the office's report, they recommended that the school should raise its attendance and offer industrial education for its male students. Additionally, they recommended that several Lutheran parochial schools in the city should divert their income to the college "to make it a first-class secondary school".
In 1922, the conference opened the Alabama Lutheran College in Selma, Alabama, which had been a mission field of focus for the Lutherans. Several years later, in 1925, the conference closed Luther College, with the funds that had been going towards that school redirected to the Alabama institute. A 1927 book published by the conference cited two main reasons for the closure of Luther College: the lack of the seminary and the estimated $18,500 that would have been required for the construction of new buildings for the school.
Notes
References
Sources
Further reading
1903 establishments in Louisiana
1925 disestablishments in Louisiana
African-American history in New Orleans
Christian schools in Louisiana
Defunct Christian schools in the United States
Defunct high schools in New Orleans
Educational institutions disestablished in 1925
Educational institutions established in 1903
Former theological colleges in the United States
Historically black universities and colleges in the United States
Lutheran schools in the United States
Private high schools in New Orleans
Seminaries and theological colleges in Louisiana
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 6,929
|
The driving force for the creation of a Comprehensive Plan comes from the Washington State Growth Management Act (GMA) passed in 1990. The GMA recognizes that uncoordinated and unplanned growth, along with a lack of common goals can and has impacted the environment, economic development, and the high quality of life traditionally enjoyed by Washington citizens.
The overriding goal of the GMA is to give each community the tools to access its goals for the future, to evaluate community assets, and to write and implement goals through innovative techniques and regulations by creating a comprehensive plan that allows the community to reach its unique vision for the future.
The City of Fife Comprehensive Plan is intended to provide the community's decision makers with guidelines regarding issues affecting the future shape, character and form of the City. The GMA establishes early and continual citizen participation as the basis for the community's comprehensive planning process. The 2005 update was created using community surveys, public notices, public meetings and hearings, workshops, newspaper articles and all required legal notices.
The nature of planning for growth inherently means the Comprehensive Plan will need to be changed. Amendments are allowed to be considered no more than once a year and go through a citizen-based planning process. The outcome is an evolving Comprehensive Plan that is a policy document supplemented by strategies for implementing community goals regarding change over the next 20 years.
What is a Comprehensive Plan Amendment (CPA)?
How do I submit an proposed amendment?
When do I submit a proposed amendment?
What happens to your proposed amendment?
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 2,673
|
Manappattu is a village in Bahour Commune of Bahour taluk in the Union Territory of Puducherry, India. It lies east of NH-45A at a distance of 2 km from it.
Geography
Manappattu is bordered by Bahour in the west, Pillaiyarkuppam in the north, Bay of Bengal in east and Krishnavaram, Pudukuppam in the south.
Villages
Following are the list of villages under Manappattu Village Panchayat.
Manappattu
Kanniakoil
Kattukuppam
Varakalodaipet
Road Network
Manappattu is connected by Kanniakoil - Manappattu road. Also Pannithittu - Pudukuppam road connects Manappattu.
Gallery
Politics
Manappattu is a part of Bahour (Union Territory Assembly constituency) which comes under Puducherry (Lok Sabha constituency)
References
External links
Official website of the Government of the Union Territory of Puducherry
Villages in Puducherry district
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 7,662
|
Ferem (, also Romanized as Faram) is a village in Barik Rud Rural District, in the Central District of Fereydunkenar County, Mazandaran Province, Iran. At the 2006 census, its population was 1,041, in 275 families.
References
Populated places in Fereydunkenar County
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 5,379
|
Select Home Breakdown Cover Members Areas Road Safety Charity About Us Motoring Advice & News
Road Safety Charity
Motoring Advice & News
Wheels, Tyres and Brakes
►Driving Tips and Advice
General Driving Tips
►General Motoring
Money and Environment
General Road Safety
Road Safety Insight
Road Safety Statistics
Home > Motoring Advice > Car Reviews
Road test: Toyota C-HR
Posted on March 2nd, 2018 by James Luckhurst
The C-HR looks great, inside and out, but space feels limited and it really needs more poke.
A rendezvous with the delivery driver is arranged for 9.30am in a Gerrards Cross supermarket car park. I'm 150 miles from home with a few appointments (both business and personal) before driving back to Mid Wales tomorrow. First impressions are positive. The C-HR looks really good – futuristic, definitely a head-turner. Not too sure about the plasticky spoiler that juts out from the rear end of the roof, but otherwise, I'm impressed. Once inside, I take a while to find exactly the right seating position.
Having not long ago spent a week driving a Prius, I immediately see similarities with the C-HR. A 1.8-litre automatic petrol engine with additional electric motor is found in both models. As I quickly discover, this combination has a lot more work to do propelling the C-HR along. However, it appears to rise to the challenge as I move through a series of local appointments in the Thames Valley.
Once done with a morning commitment in London, I am reunited with the C-HR in Marlow, having left it in the care of photographer Richard Sowersby. He's bubbling over with enthusiasm as he talks through everything he loves about the car. The long motorway journey back to my home gives me plenty of time to get used to the performance and the interior, as well as to the noticeable noise from the engine.
The worrying thing is the rather optimistic range guide on the dashboard. Starting at 400 miles-plus as I join the M4 near Maidenhead, it's down to a little over 100 by the time I arrive home. Ah-ha, you think – I must have been 'giving it' some to push the range down so much. Actually, no. I was within the speed limit, grabbing the opportunity to make progress, but never heavy on the gas. Am I on the wrong roads? Or do I need to adapt? Hopefully things will become clearer in the week ahead.
My youngest son Dylan, aged seven, is playing rugby this morning. He loves the look of the C-HR as we make ready to leave for the ground, but he can only just reach the very high rear door handle. If it weren't for his booster seat, he would not see anything out of the small rear windows, either. As he and his team mates warm up, I take note of the CH-R's interior layout, which is smart, dark and futuristic. The floors are black, the ceiling's black, the seats are black and the door panels are black. The key can sit in one of the cup holders, and the start button is just to the right of the steering wheel.
A blue strip runs from the driver's door handle, up and over the air vent. It disappears by the two big dials (the left for power, the right for speed and fuel tank), only to pick up again around the large infotainment screen, along the top of the glove box to finish on the passenger door handle.
The automatic gearshift is small but perfectly formed, the heating is controlled by four small tabs just above it. The window and wing mirror controls sit on the driver's door.
On a quick trip to Brecon, I pay close attention to Toyota's guide on making the most of electric power. It tells me a full hybrid is engineered to run on electric power as often as possible. This saves fuel and reduces exhaust emissions. An 'EV' button means I can switch to electric power for times when I'm going really slowly. I drive round Morrison's car park in this 'stealth' mode, being careful to follow the instructions and to drive smoothly and gently, with minimum additional drain on the battery (no Iron Maiden at peak volume, no full blast of air-conditioning). Come to think of it, I couldn't name a single song by Iron Maiden, as I'm more of a Bach man myself. But you get my point.
Braking early and gently maximises the power that's recycled from the C-HR's regenerative braking system, and it would appear best not to accelerate… at all. Or at least to keep out of the 'power' band shown clearly on the Eco monitor. That causes a slight problem, especially on the hilly roads around us. But I do my best.
Off to Hinckley this morning to be a delegate at the Roads Policing conference. It's a three-and-a-half hour drive through a lot of city and motorway traffic, but I can't fault the comfort. Chief Constable Anthony Bangham causes a stir at the event by demanding an end to tolerance for speeding, and in particular saying that there should be fines and penalty points for drivers caught at 31mph in 30 limits. This sets off a furore among journalists, as tomorrow's papers are sure to confirm.
After glancing at the headlines made by the Chief Constable, I leave the Conference and head down the A5 for a short meeting in Stony Stratford, before striking for home. The roads around Buckingham and Banbury prove bone-shakingly rutted, with some seriously big holes that would challenge even the most sophisticated suspension system.
I have a day at home, so my wife piles Dylan and two other children into the C-HR for post-school swimming lessons in Hay-on-Wye. Feedback supports Dylan's gripe from earlier in the week that the rear windows are too small and too high, although one child says it feels like being in a spaceship (I think that's a compliment, but it comes from a girl aged eight, so I am not 100 per cent sure).
An early start to meet the Toyota driver at Leominster. Having been an observer of all the speeding fines debate earlier in the week, I purr through the village of Winforton at 29, as displayed by the C-HR's speedometer. Winforton has one of those signs that tell you your speed – red if above 30 or green if 30 or below. So the car's 29 was, according to the sign, 26. Just thought I'd mention it, without using this page as an opportunity to enter any speed-related debate. I manage a final slow march (all 22 seconds of it, according to the monitor) in EV mode to park at Leominster station and wait for my lift back home.
We tested the Toyota C-HR Dynamic Hybrid 1.8 CVT, which is powered by a 1.8-litre petrol engine and an electric motor.
On board safety features include the Toyota 'Safety Sense' programme, which features pedestrian detection, adaptive cruise control, lane departure alert and road sign assist.
The Toyota 8-inch touchscreen combines a DAB radio, satnav, rear-view camera, and online connectivity. There's also a USB port and Bluetooth.
The list of kit is impressive: dual-zone air conditioning; power lumbar adjustment on the driver's seat, heated front seats and leather trim to the gearstick and steering wheel.
WE SAY the C-HR looks great, inside and out, but space feels limited and it really needs more poke.
Price: £30,225 as tested
Performance: 0-62 in 11 seconds
Economy: 72.4 mpg combined
Insurance: 14E
Tax: £90/£130
Figures for the Dynamic Hybrid 1.8 CVT
Take the stress and risk out of your European road journey
Safety organisation issues pothole advice
BMW 2 Series: car review
Toyota Land Cruiser: road test
Motoring Advice and News
Recovery EXTRA
Recovery RECLAIM
Breakdown Cover T&Cs
Classic Car Breakdown Cover
What to do if you breakdown
My Cover
Good Motoring Magazine
Email preferences updated
GEM Road Safety Charity
GEM Membership
Road Sign Test 1 Answers
FREE mobile app for iPhone and Android devices
GEM Motoring Assist's free mobile app has been developed to give our members extra peace of mind. This has been developed for you, the member, to ensure that if you break down you can tell us exactly where you are.
About GEM Motoring Assist
NOTE: Prices shown are for the first year only. GEM reserves the right to withdraw the offer at any time. All breakdown policies include a membership fee (inclusive of VAT) and an insurance premium (inclusive of IPT).
GEM Motoring Assist is a trading name of The Guild of Experienced Motorists a road safety association founded in 1932 and is authorised and regulated by the Financial Conduct Authority (FCA).
03 numbers are charged at the standard national geographic rate and are included in inclusive minute plans from landlines and mobiles.
© Copyright GEM Motoring Assist 2019
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 8,773
|
\section{Introduction}\label{intro}
Dark Matter (DM), although there are overwhelming evidences of its existence from astrophysics and cosmology, is still mysterious from the particle physics point of view and is one of few indications of physics beyond the Standard Model (SM)~\cite{Bertone:2004pz}. Traditionally, one hopes to find some of its properties either by observing nuclear recoils in the underground direct detection experiments~\cite{Freese:2012xd,Jungman:1995df,Smith:1988kw} or by the indirect searches via the measurements of cosmic and gamma rays~\cite{Cirelli:2012tf}. The most recent possible signal comes from the discovery of the unidentified X-ray line at about 3.55 keV from the observation of the Andromeda galaxy and many galaxy clusters, including the Perseus galaxy cluster by XMM-Newton X-ray Space observatory~\cite{Bulbul:2014sua,Boyarsky:2014jta}, which have inspired particle physicists to propose many interesting DM candidates~\cite{Bulbul:2014sua,Boyarsky:2014jta,GeneralModel}.
In this paper, we use a simple two-component DM model~\cite{MultiComponent} to explain the 3.55 keV X-ray line. The dark sector contains two Majorana fermionic DM fields and a heavy singly-charged scalar which enables the DMs to couple to the right-handed $\mu$ and $\tau$ leptons. Both the DM and singly-charged scalar fields are charged under a $Z_2$ symmetry to stabilize the two DM particles. The decay of the heavier DM particle to its lighter sister can only proceed via the dark transition electric dipole moment (EDM), resulting in the observed X-ray signal and guaranteeing the stability of the heavy DM within the Universe age. Furthermore, because of the nearly degenerate DM masses and the particular choice of the Yukawa structure, the correct DM relic abundance is achieved via the thermal production by the coannihilation of the two DM particles into $\mu^\pm \tau^\mp$ pairs in the $s$-wave with the almost one half for each component.
The paper is organized as follows. In Sec.~\ref{model}, we show the relevant Lagrangian of the DM sector. We then study the X-ray line
in Sec.~\ref{Xray} and the DM relic abundance in Sec.~\ref{DMRelic}. In Sec.~\ref{constraint},
we consider constraints on the model from flavor physics, DM direct detections and collider searches. In Sec.~\ref{result}, we present our numerical results. Our conclusions and discussions are given in
Sec.~\ref{conclusion}.
\section{The Model}\label{model}
In our model, in addition to the SM particle content, we introduce two Majorana fermionic DM fields $\chi_{1,2}$, which are neutral under the SM gauge groups, and a heavy singly-charged scalar mediator $\eta$ to allow the two DM fields to interact with the right-handed leptons. For simplicity, we restrict ourselves to the case where the two DMs only couple to the two heavier families of leptons, $\mu$ and $\tau$. Both $\chi_{1,2}$ and $\eta$ are odd under the dark $Z_2$ symmetry which stabilizes the DM particles. The quantum numbers of the relevant fields are summarized in Table~\ref{charge}.
\begin{table}\caption{Quantum Numbers for Relevant Fields}
\begin{center}
\begin{tabular}{c|c c c}\label{charge}
& $SU(2)_L$ & $U(1)_Y$ & $Z_2$ \\
\hline
$\ell_R$ & ${\mathbf 1}$ & $-2$ & + \\
$\chi_{1,2}$ & ${\mathbf 1}$ & 0 & $-$ \\
$\eta$ & ${\mathbf 1}$ & $-2$ & $-$ \\
\end{tabular}
\end{center}
\end{table}
With the assigned quantum numbers for the new particles, we can write down the general renormalizable Lagrangian related to the dark sector as follows:
\begin{eqnarray}
{\cal L}_{\rm DM} &=& \sum_{i=1,2}\bar{\chi}_i \slashed{\partial} \chi_i - \frac{1}{2}\sum_{i,j=1,2} \bar{\chi}^c_i M_{ij} \chi_j + (D^\mu \eta)^\dagger (D_\mu \eta)-V(\eta) - \sum_{\substack{i=1,2\\ \ell=\mu,\tau}}(h_{\ell i}\bar{\ell}P_L \chi_i\eta +{\rm h.c.}),\nonumber\\
V(\eta) &=& m_\eta^2\,\eta^\dagger\eta+ \frac{\lambda}{2}(\eta^\dagger\eta)^2,\, \mbox{and}\,\, D_\mu \eta = (\partial_\mu - i e A_\mu)\eta ,
\end{eqnarray}
where $M$ denotes the Majorana mass matrix for the two DM fields, $P_L=({\mathbb I}-\gamma^5)/2$ the usual left-handed projection operator and $A_\mu$ the electromagnetic field. We further assume that the Yukawa couplings $h_{\ell i}$ are all real in the bases of the DM and lepton mass states.
Furthermore, the constraints from the lepton flavor violating process $\tau\to\mu\gamma$ and the DM relic density force us to take the hierarchical form of the Yukawa couplings $h_{\mu 1} \sim h_{\tau_2} \gg h_{\mu 2} \sim h_{\tau 1}$. For simplicity, we assume the equal value $h_L(h_S)$ of the (off-)diagonal entry:
\begin{equation}\label{YukMat}
h_{\ell i} = \left(\begin{array}{cc}
h_L & h_S \\
h_S & h_L
\end{array}
\right),\;\;h_L\gg h_S\;.
\end{equation}
{\color{red} }
\section{X-Ray Line}\label{Xray}
It is well-known that the dipole operator of a Majorana particle vanishes. However, for a model with multiple neutral Majorana particles there still exist the transition dipole moments. In this section, we will show that the decay of the heavy Majorana DM particle into the lighter one via the transition electric dipole operator can naturally explain the cluster X-ray line at 3.55 keV. We start with the effective dipole operator in our model
\begin{eqnarray}\label{dipole}
{\cal O}_{\rm dipole} = \frac{1}{2}\bar{\chi}_1\sigma^{\mu\nu}[D_\chi\gamma^5+i\mu_\chi\mathbb{I}]\chi_2 F_{\mu\nu},
\end{eqnarray}
where $D_\chi$ and $\mu_\chi$ denote the dark transition electric dipole moment (EDM) and transition magnetic dipole moment (MDM), respectively.
\begin{figure}[ht]
\centering
\includegraphics[width=0.47\textwidth]{dip_c.ps}
\includegraphics[width=0.47\textwidth]{dip_n.ps}
\caption{Feynman diagrams for the transition dipole operator between $\chi_{2,1}$.}
\label{Fig_Dip}
\end{figure}
From Fig.~\ref{Fig_Dip}, we obtain
\begin{eqnarray}\label{EDM}
D_\chi = \frac{e R}{16\pi^2} \frac{\Delta m}{2 m_\chi^2} \Big[1+ \frac{m_\eta^2}{m_\chi^2}\ln\big(1-\frac{m_\chi^2}{m_\eta^2} \big) \Big] \approx \frac{e R }{64\pi^2} \frac{\Delta m}{ m_\eta^2}, \quad \mu_\chi =0,
\end{eqnarray}
where $m_\chi$ ($\Delta m$) stands for the average mass (the mass difference) of the two DM particles and $R=h_{\mu 1}h_{\mu 2}+ h_{\tau 1} h_{\tau 2} = 2 h_L h_S$. The decay rate of the heavier DM is then given by
\begin{equation}\label{DecRate}
\Gamma = \frac{\Delta m^3}{\pi} D_\chi^2 \approx \frac{\Delta m^3}{\pi} \Big( \frac{eR}{64\pi^2} \frac{\Delta m}{m_\eta^2} \Big)^2.
\end{equation}
Note that the vanishing value of the transition MDM can be traced back to our assumption of the real Yukawa couplings. However, if this assumption is relaxed, then the MDM would have a nonzero value of $\mu_\chi$, given by
\begin{eqnarray}
\mu_\chi = \frac{ e I}{16\pi^2} \frac{1}{m_\chi} \left[1+ \frac{m_\eta^2}{m_\chi^2}\ln\big(1-\frac{m_\chi^2}{m_\eta^2} \big) \right] \approx \frac{e I }{32\pi^2} \frac{m_\chi}{m_\eta^2},
\end{eqnarray}
where $I= {\rm Im} (h^*_{\mu 1}h_{\mu 2}+ h^*_{\tau 1} h_{\tau 2})$. Clearly, if the imaginary part of the Yukawa couplings is of the same size as the real one, the MDM would be the main contribution since $\mu_\chi/D_\chi \propto \Delta m/m_\chi \sim {\cal O}(10^{-7})$.
In order to make numerical studies, we need to transform the lifetime presented in Ref.~\cite{Boyarsky:2014jta} into our setup. The prototype model in Refs.~\cite{Boyarsky:2014jta,Bulbul:2014sua} contains the sterile neutrino with the mass $m_0=7.1$ keV, which decays into an active neutrino plus a photon via the small transition magnetic moment with the lifetime in the range $\tau_0 = 0.2 \sim 2\times 10^{28}~{\rm s}$. In our present model, we rescale the above lifetime to match the observed X-ray flux to be
\begin{eqnarray}\label{lifetime}
\tau_{\chi} = \frac{\tau_0 m_0}{2 m_\chi},
\end{eqnarray}
where the factor 2 accounts for the fact that each DM component carries only half of the DM density in the whole Universe and the Galaxy. From Eq.~(\ref{lifetime}), we see that the predicted lifetime depends on the average DM mass $m_\chi$. With Eqs.~(\ref{DecRate}) and (\ref{lifetime}),
we will determine the parameter space in our model to explain the cluster X-ray line.
The measurement of the cluster X-ray line gives some constraint to the mass of the charged scalar $m_\eta$.
From Eqs.~(\ref{DecRate}) and (\ref{lifetime}), we can derive the following formula:
\begin{eqnarray}
m_\eta = \left[ \frac{\Delta m^5}{\pi} \left(\frac{eR}{64\pi^2}\right)^2 \left(\frac{\tau_0 m_0}{2 m_\chi} \right) \right]^{\frac{1}{4}},
\end{eqnarray}
If we fix the values of the dark matter mass and the coupling ratio, for example, $m_\chi=10$~GeV and $R_h=h_S/h_L=0.1$,
it is interesting to see that there is an upper limit on $m_\eta$ in the light of the cluster X-ray anomaly and the perturbativity of the Yukawa coupling.
If we further take the largest lifetime $(\tau_0)_{\mathrm max}=2 \times 10^{28}$s and the maximal value $h_L=5$ allowed by the perturbativity,
we obtain this upper limit to be $1.86$~TeV. Yet, this result comes solely from the consideration of the perturbativity and the measured X-ray flux. If we consider more constraints from other experiments, the upper bound on $m_\eta$ would further decreased,
which will be discussed in Sec.~V.
\section{Dark Matter Relic Abundance}\label{DMRelic}
It is clear that a successful prediction of the correct DM relic abundance is one of the most important tests for the viability of a DM model. Usually, for a cold DM, the generation of its relic abundance is assumed to proceed by the thermal freezing-out mechanism, which relies on the thermally averaged effective DM annihilation cross section at the time of the freeze-out. Since the two DMs have the highly degenerate masses by the construction, the coannihilations between $\chi_1$ and $\chi_2$ have to be considered~\cite{Griest:1990kh}.
For a fermionic DM with its freezing-out mainly through $t/u$-channel annihilations, it is well-known that the thermally averaged effective annihilation is dominated by the $p$-wave contribution, which usually requires large couplings and/or light mediators to enhance the annihilation so as not to over-close the Universe. Let us make this point more precise by looking into the leading tree-level Feynman diagrams in Fig.~\ref{Fig_Ann}.
\begin{figure}[ht]
\centering
\includegraphics[width=0.45\textwidth]{ann_t.ps}
\includegraphics[width=0.45\textwidth]{ann_u.ps}
\caption{Annihilation diagrams of the DM particles.}
\label{Fig_Ann}
\end{figure}
Since the annihilation with the definite incoming DMs and outgoing leptons involves $t$- and $u$-channels, if the Yukawa couplings in the two diagrams are the same, the destructive interference between the two channels would cause the leading $s$-wave contribution to be vanishingly small. However, in the presence of the two DM components with the Yukawa structure in Eq.~(\ref{YukMat}), there is a special process, $\chi_1\chi_2 \rightarrow \mu^\pm \tau^\mp$, which does not suffer such a suppression. The $t$-channel diagram is only proportional to $h_{\mu 1}h_{\tau 2}=h_L^2$, while the corresponding $u$-channel $h_{\tau 1}h_{\mu 2}=h_S^2$. Based on our construction of $h_L\gg h_S$, the $t$-channel can not be fully cancelled by the corresponding $u$-channel, leading to a large dominant $s$-wave effective cross section:
\begin{eqnarray}\label{swave}
\langle\sigma v_{\rm rel}\rangle = \sigma_0 = \frac{(4m_\chi^2-m_\tau^2)^2}{512\pi m_\chi^4 (2m_\chi^2+2m_\eta^2-m_\tau^2)^2} \left(h_L^4(4m_\chi^2+m_\tau^2)-8h_L^2 h_S^2 m_\chi^2+h_S^4(4m_\chi^2+m_\tau^2)\right).\nonumber\\
\end{eqnarray}
In the following numerical computation of the DM relic abundance, we shall use Eq.~(\ref{swave}) as our effective annihilation cross section. As a result, the analytic solution of the Boltzmann equation describing the evolution of the DM density is given by~\cite{Kolb94,ArkaniHamed:2006mb}:
\begin{equation}\label{RelicDens}
\Omega h^2 = \frac{688 \pi^{5/2} T^3_\gamma x_f }{99\sqrt{5g_\star}(H_0/h)^2 M^3_{\rm Pl} \sigma_0 } = \frac{8.7\times 10^{-11} {\rm GeV}^{-2} x_f }{\sqrt{g_\star} \sigma_0},
\end{equation}
where the freeze-out temperature $T_f$ is implicitly defined in $x_f=m_\chi/T_f$, given by
\begin{equation}
x_f = X-\frac{1}{2} \log X, \quad X=25+\log\Big[\frac{g}{\sqrt{g_\star(T_f)}} m_\chi \sigma_0 \times 6.4\times 10^6 {\rm GeV} \Big]\,,
\end{equation}
and $g_\star(T_f)$ denotes the number of relativistic degrees of freedom at $T_f$, taken from the tabulated functions in Refs.~\cite{Srednicki:1988ce,Olive:1980wz}. The final result is shown as the thick red curve in the $m_\chi$-$h_L$ plane for several selected benchmark mass and coupling ratios in Fig.~\ref{Fig_Result}, where we have used the central value of the most recent Planck result $\Omega_{\rm DM} h^2 = 0.1187\pm0.0017$~\cite{Planck}. Finally, we remark that the neglect of the other annihilation channels to the effective cross sections does not affect the final prediction of the DM relic abundance much, with the correction within 5\% at best, since they are $p$-wave suppressed as mentioned before.
\section{Complementary Constraints on the Model}\label{constraint}
In the previous sections, we have shown that our two-component DM model can not only explain the X-ray line observed from the distant galaxy clusters but also naturally give rise to the desired DM relic abundance in the Universe. However, it is clear that in order for this scenario to be viable, we have to investigate if it satisfies with the constraints from other aspects of particle physics, such as flavor physics, direct DM detections, and collider searches.
\subsection{Flavor Physics}
The most stringent flavor constraint on the model comes mainly from the lepton flavor violation (LFV) process $\tau\to \mu\gamma$, which proceeds via the dipole operator $\frac{1}{2} \bar{\mu} \sigma^{\mu\nu} (A_L P_L+A_RP_R)\tau F_{\mu\nu}$ with the coefficients $A_{L,R}$:
\begin{equation}
A_L = \frac{e}{16\pi^2}(2h_L h_S) F_2\Big(\frac{m_\chi^2}{m_\eta^2}\Big)\frac{m_\tau}{2m_\eta^2}, \quad
A_R = \frac{e}{16\pi^2}(2 h_L h_S) F_2\Big(\frac{m_\chi^2}{m_\eta^2}\Big)\frac{m_\mu}{2m_\eta^2},
\end{equation}
where the function $F_2(x)$ is given by
\begin{equation}
F_2(x) = \frac{1-6x+3x^2+2x^3-6x^2\log x}{6(1-x)^4}.
\end{equation}
The branching ratio for this LFV process is~\cite{Schmidt:2012yg}:
\begin{eqnarray}\label{TMG}
{\mathcal B}(\tau\to \mu\gamma) = \frac{3\alpha_{\rm em}}{64\pi G_F^2 m_\eta^4} \Big|(2 h_L h_S)F_2\left(\frac{m_\chi^2}{m_\eta^2}\right)\Big|^2 {\mathcal B}(\tau\to \mu\nu_\tau\bar{\nu}_\mu),
\end{eqnarray}
where $\alpha_{\rm em} = e^2/(4\pi)$ and $G_F$ are the fine structure and Fermi constants, respectively. In Eq.~(\ref{TMG}), we have neglected the terms proportional to the lepton masses.
Using the current experimental upper bound of ${\mathcal B}(\tau\to\mu\gamma)<4.4\times 10^{-8}$~\cite{PDG}, we present the constrained parameter space as the blue region with the dashed line as the boundary in Fig.~\ref{Fig_Result}. From the figure, we also see that a lot of parameter space is already excluded by this LFV channel, especially when the hierarchy between $h_S$ and $h_L$ is not very large, {\it i.e.}, $h_S/h_L \sim 0.2$.
\subsection{Direct Dark Matter Detections}
Currently, the most stringent cross section bound for the dark matter mass range of interest is from the LUX experiment~\cite{LUX}, in which for the 33 GeV dark matter the spin-independent cross section bound is of ${\cal O} (10^{-46})$~${\rm cm}^2$. For the present leptophilic DM model, the relevant scattering of two DM components with the target nucleus is proceeded by the mediation of a virtual photon. As discussed in the literature~\cite{Bai:2014osa,Agrawal:2011ze,Agrawal:2014ufa,Chang:2014tea,Fitzpatrick:2010br,Ho:2012bg,DelNobile:2014eta}, we first need to identify the relevant effective operators with the DMs coupled to a photon, and then calculate the scattering probability of the DM particles with the target nucleus. Each DM can be viewed as a Majorana particle in the mass eigenstate basis $\chi_i$, so that if the scattering does not change the DM component, for the so-called component-conserving scattering, one has to only consider the following two kinds of dimension-six effective operators
\begin{eqnarray}
{\cal O}^c_{1 i} = -\bar{\chi_i} \gamma^\mu \gamma^5 \partial^\nu \chi_i F_{\mu\nu}+ {\rm h.c.}, \quad\quad {\cal O}^c_{2 i} = i \bar{\chi_i}\gamma^\mu \partial^\nu \chi_i F^{\alpha\beta}\epsilon_{\mu\nu\alpha\beta} + {\rm h.c.},
\end{eqnarray}
where $i=1$ and 2. However, as pointed in Ref.~\cite{Bai:2014osa}, it can be proved that ${\cal O}^c_{2i} = -2 {\cal O}^c_{1i}$ by using the Chisholm identity, which results in a single kind of the dimension-six operator for the Majorana fermions. This kind of operators can be matched to the electromagnetic anapole moment of each DM component, which further couples to the current from the target nucleus in the non-relativistic limit and realizes the anapole dark matter~\cite{Bai:2014osa,Fitzpatrick:2010br,Ho:2012bg,DelNobile:2014eta}. However, besides the above component-conserving interactions, there are also transition interactions induced by the real Yukawa couplings,
\begin{eqnarray}
{\cal O}^t_1 = -\bar{\chi_2} \gamma^\mu \gamma^5 \partial^\nu \chi_1 F_{\mu\nu}+ {\rm h.c.}, \quad\quad {\cal O}^t_2 = i \bar{\chi_2}\gamma^\mu \partial^\nu \chi_1 F^{\alpha\beta}\epsilon_{\mu\nu\alpha\beta} + {\rm h.c.},
\end{eqnarray}
The explicit calculation of the one-loop diagrams in Fig.~\ref{Fig_Dip} leads to the concrete expressions for the Wilson coefficients for these effective operators in the present model~\cite{Bai:2014osa,Fitzpatrick:2010br,DelNobile:2014eta}, given by
\begin{itemize}
\item Component-Conserving Couplings:
\begin{eqnarray}\label{CC}
c_{1i}^c = \frac{-e}{64\pi^2 m_\eta^2} \sum_{\ell}R^c_{\ell i}\Big[ \frac{1}{2} + \frac{2}{3}\ln\Big(\frac{m_\ell^2}{m_\eta^2}\Big) \Big],\quad\quad c_{2i}^c = \frac{-e}{64\pi^2 m^2_\eta}\frac{1}{4}\sum_\ell R^c_{\ell i},
\end{eqnarray}
\end{itemize}
where $R^c_{\ell i} = {\rm Re}(h_{\ell i}h^*_{\ell i})$ with $R^c_{\mu 1(2)} = R^c_{\tau 2(1)} = h_{L(S)}^2$, and
\begin{itemize}
\item Transition Couplings:
\begin{eqnarray}\label{CPCT}
c_{1}^t = \frac{-2e}{64\pi^2 m_\eta^2} \sum_{\ell}R^t_{\ell}\Big[ \frac{1}{2} + \frac{2}{3}\ln\Big(\frac{m_\ell^2}{m_\eta^2}\Big) \Big],\quad\quad c_{2}^t = \frac{-2e}{64\pi^2 m^2_\eta}\frac{1}{4}\sum_\ell R^t_{\ell},
\end{eqnarray}
\end{itemize}
where $R^t_{\ell} = {\rm Re}(h_{\ell 1}h^*_{\ell 2})=h_L h_S $. Note that the formulas in Eqs.~(\ref{CC}) and (\ref{CPCT}) only differs a factor 2. Nevertheless, at the amplitude level, such a difference would be compensated when differentiating the two identical Majorana particle fields in the component-conserving effective operators ${\cal O}^c_{1i}$ and ${\cal O}^c_{2 i}$.
With the above effective operators, the two DM particles couple to the charge and magnetic dipole moment of the nucleus, which give the spin-independent and spin-dependent differential cross sections
\begin{eqnarray}\label{XSI}
\frac{d\sigma^E_T}{d E_R} = \frac{1}{2}\Big[(c^c_{11} - 2c^c_{21})^2+ (c^c_{12} - 2c^c_{22})^2 + \frac{2}{4}(c^t_1-2c^t_2)^2\Big] e^2 Z^2 \frac{m_T}{4\pi} \Big( 2-\frac{m_T E_R}{\mu_{NT}^2 v^2} \Big) F_E^2(q^2),~~
\end{eqnarray}
and
\begin{eqnarray}\label{XSD}
\frac{d\sigma^M_T}{d E_R} = \frac{1}{2}\Big[(c^c_{11} - 2c^c_{21})^2+ (c^c_{12} - 2c^c_{22})^2 + \frac{2}{4}(c^t_1-2c^t_2)^2\Big] e^2 \frac{1}{2\pi} \frac{E_R}{v^2} \frac{m_T^2 \lambda^2_T}{m_N^2 \lambda_N^2} \frac{J_T+1}{3 J_T} F_M^2(q^2),~~
\end{eqnarray}
where $\lambda_{N(T)} = e/(2 m_{N(T)})$ is the magneton of the nucleon (target nucleus), $m_{N(T)}$ is the nucleon (target nucleus) mass, $J_T$ ($Z$) is the spin (charge) of the target nucleus, and $F_{E(M)}(q^2)$ is the form factor of the nucleus charge (magnetic dipole moment), respectively. Note that in Eqs.~(\ref{XSI}) and (\ref{XSD}), the factor 1/2 accounts for the half density carried by each DM component in the Universe, while 2/4 the right normalization of the transition scattering in accord with the previous two component-conserving ones.
From Eqs.~(\ref{XSI}) and (\ref{XSD}), it is obvious that the spin-independent part is the dominant one due to the large charge $Z^2$ enhancement for the heavy target nuclei, such as Xenon used in the LUX experiments. However, the differential cross section in the recoil energy $E_R$ is suppressed by an additional power of $v^2$ or $E_R$ compared with the usual Dirac fermionic DM case, resulting in a weak direct detection constraint. An estimate made in Ref.~\cite{Bai:2014osa} for $^{129}_{54}{\rm Xe}$ with the typical reference recoiled energy and the similar masses of $\chi_{1,2}$ and $\eta$ gives the DM-nucleon scattering cross section of ${\cal O}(10^{-49}){\rm cm}^2$, which is evidently much smaller than the current LUX bound for the spin-independent DM-nucleon cross section~\cite{LUX}. Therefore, we do not show the direct DM detection constraints on our model.
Finally, we point out that if the restriction on the real Yukawa couplings is relaxed, the same one-loop Feynman diagrams in Fig.~\ref{Fig_Dip} would also induce the following transition operators:
\begin{eqnarray}\label{CPoddO}
{\cal O}^t_3 = \bar{\chi_2} \gamma^\mu \partial^\nu \chi_1 F_{\mu\nu}+ {\rm h.c.}, \quad\quad {\cal O}^t_4 = -i \bar{\chi_2}\gamma^\mu\gamma^5 \partial^\nu \chi_1 F^{\alpha\beta}\epsilon_{\mu\nu\alpha\beta} + {\rm h.c.},
\end{eqnarray}
with the corresponding Wilson coefficients
\begin{eqnarray}\label{CPVT}
c_{3}^t = \frac{-2e}{64\pi^2 m_\eta^2} \sum_{\ell}I^t_{\ell}\Big[ \frac{1}{2} + \frac{2}{3}\ln\Big(\frac{m_\ell^2}{m_\eta^2}\Big) \Big],\quad\quad c_{4}^t = \frac{-2e}{64\pi^2 m^2_\eta}\frac{1}{4}\sum_\ell I^t_{\ell},
\end{eqnarray}
where $I^t_{\ell} = {\rm Im}(h_{\ell 1}h^*_{\ell 2})$. If the imaginary part of $h_{\ell i}$ is the same order as the real one, the dominant contribution to the direct detection experiments is given by the spin-independent cross section induced by the operator ${\cal O}^{\,t}_3$. In this case, there is no suppression caused by the small recoiled energy $E_R$ or the DM velocity $v^2$, and in turn the LUX experiments have already given strong constraints to this model, effectively ruling out all the parameter space trying to explain the X-ray excess.
\subsection{Collider Searches}\label{collider}
A large portion of the parameter space of the present model has already been probed by the LEP and LHC experiments. Since our two DM components do not directly couple to the electron at tree-level by the construction, they cannot be directly pair produced at the LEP so that the relevant monophoton constraints~\cite{Fox:2011fx} do not apply.
Another collider signal in our model is related to the production of the singly-charged scalar $\eta$ mediated by a virtual photon or $Z$-boson. When the mass of $\eta$ is lighter than about 100 GeV, it can be pair produced at the $e^+ e^-$ colliders, which has been already ruled out by the null results at the LEP. Since the maximum energy of the LEP is about 200 GeV, we take 100 GeV as the LEP bound for the charged mediator mass as an illustration in our study. At the LHC, the signal of this leptophilic model is the creation of the charged mediator pair via the Drell-Yan process, each of which decays further into a lepton plus one of the DM particles~\cite{Bai:2014osa,Agrawal:2014ufa,Chang:2014tea}. The signature at hadron colliders is a pair of opposite-sign leptons with some missing transverse energy. Such a signature is very similar to that of searching for the right-handed sleptons with mixings in the MSSM at the colliders, for which the current LHC bound of the mass of the charged scalar is about 240 GeV as presented in Fig.~8a in Ref.~\cite{ATLAS}, which is more stringent than the LEP direct search bound.
In this work, we use both LEP and LHC bounds in our numerical studies, which can be transformed into the mass bounds of the two nearly degenerate DM components for a given mass ratio $R_M= m_\eta/m_\chi$. In Fig.~\ref{Fig_Result}, the excluded regions for both experimental bounds are plotted as the purple and yellow regions bounded by the dot-dashed and dotted lines. Note that the latter LHC constraint can be relaxed a lot for the present model. The LHC bound of 240 GeV for the charged scalar $\eta$ is the most stringent one in Fig.~8a in Ref.~\cite{ATLAS}. The actual restricted region in the $m_\eta$-$m_\chi$ plane is much irregular, and the constraint becomes more loose when the DM mass approaches the mediator one due to the decreasing of the missing energy. In Fig.~\ref{Fig_ATLAS}, we show the actual LHC bound on the $m_\eta$-$m_\chi$ plane given in Ref.~\cite{ATLAS}, together with the allowed X-ray signal region and other constraints with the Yukawa couplings $(h_S, h_L)=(0.113,1.13)$. Another important issue is the assumption made in the analysis of Ref.~\cite{ATLAS} that the decay of the charged scalar is only to the first two generations of the charged leptons plus a large missing energy. Nevertheless, in our model, $\eta$ can only have about one fourth probability to $\mu^+ + \mu^-$, whereas the other 3/4 of the decay branching ratio related to the $\tau$ lepton cannot be probed. As the detailed analysis of the LHC bound on our model is beyond the scope of the present paper, we are content with the most stringent bound in Ref.~\cite{ATLAS} in our exploration of the model parameter space.
\section{Final Results}\label{result}
Our results on the parameter space are shown in Fig.~\ref{Fig_Result} for different choices of the mass ratios $R_M=m_\eta/m_\chi$ and the Yukawa coupling ratios $R_h=h_S/h_L$ . The orange region with the thin black boundaries represents the parameter space which can be used to explain the X-ray line fluxes, while the thick red curve gives the correct DM relic abundance. The blue, yellow and purple regions bounded by the dashed, dotted and dot-dashed lines are excluded by the constraints from the $\tau\to\mu\gamma$ process, LEP and LHC, respectively. A further constraint is originated from the perturbativity requirement of the large Yukawa coupling $h_L$, which is assumed to be less than 3.
\begin{figure}
\centering
\includegraphics[width=1\textwidth, natwidth=1170,natheight=700]{T9.ps}
\caption{Parameter spaces in the $m_\chi$-$h_L$ plane for $R_M=m_\eta/m_N=(5,\,10,\,15)$ and $R_h=h_S/h_L= (0.05,\,0.1,\,0.2)$, where the thick red curve gives the correct DM relic abundance, the orange region bounded by the thin black curves is accord with the X-ray line observation, and the blue, yellow and purple regions bounded by the dashed, dotted and dot-dashed lines are excluded by the FLV process $\tau\to\mu\gamma$, LEP and LHC, respectively.
}\label{Fig_Result}
\end{figure}
In Fig.~\ref{Fig_Result}, we arrange the diagrams in rows and columns in terms of the mass ratios $R_M$ and the coupling ratios $R_h$ in order to show the trend of how the allowed parameter space changes when these two parameters vary. It is seen that there are still parameter spaces in which the present two-component DM model can explain the cluster X-ray line and DM relic abundance in the Universe, simultaneously, despite of the very strong constraints from flavor and collider physics. As an illustration, we present a prototypical benchmark point: $h_S=0.113$, $h_L=1.13$, $m_\chi=30~$GeV and $m_\eta = 300~$GeV. Fig.~\ref{Fig_ATLAS} illustrates the parameter space in the $m_\eta$-$m_\chi$ plane with the fixed Yukawa couplings ($h_L,~h_S$)=(1.13,0.113), in which the current ATLAS bounds on $m_\eta$ and $m_\chi$ are shown more clearly.
\begin{figure}[th]
\centering
\includegraphics[width=0.4\textwidth,natwidth=300,natheight=300]{figureD.ps}
\caption{Label is the same as Fig.~\ref{Fig_Result} but in the $m_\eta$-$m_\chi$ plane for ($h_L$,$h_S$)=(1.13,0.113). The black point represents the benchmark point given in the context.}
\label{Fig_ATLAS}
\end{figure}
\section{Conclusions and Discussions}\label{conclusion}
We have proposed a two-component DM model to understand the recent unidentified X-ray line from the observation of the Andromeda galaxy and the distant galaxy clusters. In order for the two DM particles $\chi_{1,2}$ to only couple to the right-handed $\mu$ and $\tau$ via the Yukawa couplings, a new singly-charged scalar mediator $\eta$ is introduced. Both $\chi_{1,2}$ and $\eta$ are odd under the dark $Z_2$ symmetry. In spite of the very stringent constraints from flavor physics, direct DM searches and collider bounds from the LEP and LHC, we have found some parameter space which can explain the X-ray line with the correct DM relic abundance.
As seen from Eq.~(5), the successful explanation of the cluster X-ray flux depends on the 3.5 keV mass splitting $\Delta m$ between the two DM components, which is naturally motivated by the measured energy of the X-ray line. The next question is to investigate the stability of such a small splitting against the radiative corrections, especially when we notice the DM Majorana masses are of ${\cal O}$(10 GeV). However, such a worry can be avoided for the present DM model. Since the leading order radiative correction to the DM Majorana masses appears at three-loop level, we find that the correction to the mass splitting is always of order sub-keV, which is achieved by the suppressions from the loop factors and the various couplings. Therefore, we expect that the keV order of the DM mass splitting is naturally guaranteed against the radiative corrections in our model.
In our study, we have used the most stringent bound from the ATLAS~\cite{ATLAS} to constrain the charged scalar mediator mass $m_\eta$. However, this constraint can be relaxed if we take into account the difference of the present model from the assumptions made in Ref.~\cite{ATLAS}. This analysis is helpful because this constraint excludes all or part of the signal region in some parameter spaces, such as those shown in the second and third graphs in the first row of Fig.~\ref{Fig_Result}. A detailed investigation of this collider constraint may reopen these parameter windows, which will be worth of a future work.
\section*{Acknowledgments}
The work was supported in part by National Center for Theoretical Sciences, National Science
Council (NSC-101-2112-M-007-006-MY3) and National Tsing Hua
University (Grant Nos. 103N1087E1 and 103N2724E1).
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 5,362
|
namespace Microsoft.Spectrum.MeasurementStation.Contract
{
using System;
using System.ServiceModel;
[ServiceContract(SessionMode = SessionMode.NotAllowed)]
public interface IMeasurementStationService
{
[OperationContract]
void GetUpdatedSettings(string measurementStationAccessId, out string storageAccountName, out string storageAccessKey, out byte[] measurementStationConfiguration);
[OperationContract]
void ScanFileUploaded(string measurementStationAccessId, string blobUri, bool success);
[OperationContract]
void ReportHealthStatus(string measurementStationAccessId, int healthStatus, int messagePriority, string title, string description, DateTime occurredAt);
[OperationContract]
int GetHealthStatusCheckInterval(string meaurementStationAccessId);
[OperationContract]
int GetStationAvailability(string measurementStationAccessId);
}
}
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 3,411
|
{"url":"https:\/\/xfido.com\/html\/indexEnglishNote.html","text":"Lunar rover landing in the moon\nThu Jan 3 14:19:30 2019 China is the first country that landed a lunar rover in the far side of the noon in human history in January 3. 2019.\nConsisting of sth\nEvery open subset $U \\in M$ has an altas consisting of single chart, namely $(\\phi, U) = (id_{U}, U)$ where $id_{U}$ denotes the identity map of $U$\nA collection of sth\nAn Altas is a collection of Charts\nHow to use where clause\nThe pair $(\\phi, U)$ where $U$ is an open set and $\\phi$ is bijective.\nAvailable: Here is the word that I always forget the spelling.\nThe word is used quite frequently, e.g. in email, I will be available ..\nBase, Bases, Basic and Basics, Basis, Bases\nWed Jul 18 09:14:45 PDT 2018\nI just learned something new today.\nThe word \"Base\" can be adjustive in English.\nI thought it could be noun only.\nWe use \"Base case\" all the time in Mathmatic Induction so it makes scene\nThe word \"Basic\" is adjective is obvious but it can be used as noun too.\ne.g. we need to learn the basics of English\nNe Phrase:\nhundreds of thousands of people in the Paris Eiffel Tower\nscrub, scrutiny, scrutinise, screw\n18-18-2016 @ 10:34, new words today\nscrub - rub sth or someone hard so as to clean them, typically with a brush and water:\nHe had to scrub the floor\nShe was scrubbing herself down at the sink, she scrubbed furiously at the plates\nscrutiny [noun]\nscrutinies [verb]\nPara words\nParallel\nParallelogram\nParachute\nClose to something or someone\nBuild an array strip[] that contains all points close or closer than d to a line passing the middle of points. Tom doesn't even close to Nancy when it comes to artistic ability.\nCompared to, compare with\nThe store's price are high compared to what some other stores charge\nInstance coffee doesn't compare with freshly ground coffee.\nAs: simultaneous changes\nCompare\nWhen you get older, moving house gets harder.[one thing happens first and as the result the second thing is true]\nAs you get older, moving house gets harder.[The two things happen as the same time]\nAs as a preposition\nWe use it with an noun to refer to the role or purpose of a person or thing.\nI worked as waiter when I was a student.\nAs as a conjunction\nThey arrive as they were leaving. [time conjunction meaning 'while' or 'when']\nBut and However in sentense\nI have worked in the company for two years but I hate the job.\nI have worked in the company for two years. However, I hate the job\nEnglish Tenses: Past perfect [had + participle]\nI had worked on the project for two years before you moved to Orlando Florida\nHow to use AS IF in the sentense\nClauses the start with as if describle the unreal or unsure or improbable situation if they are followed by unreal tense\n[or past subjunctive or the past perfect subjunctive]\nHe seems as if he knew the answer. [No one know whether he know the answer or not]\nHe looks as if he knew the answer\nHe seems as if he hadn't slept for days\nNight club shooting in Orlando Florida\n50 people are killed and 53 people are injured in a nightclub mass shooting in Orlando Florida\nThis is the worst mass shooting in U.S. history, I'm not sure about that\nBelieve there are more than one shooter with assault rifle in the nightclub at 2am on Sunday.\nObama will come out and repeat the same thing. Hey, we need to solve the Guns issues..\nAmerican people said. Hey, own a gun is first amendment. We dont care how many get killed with illegal guns every year\nWhat we know about the suspect?\nThe killed is identified as Omar Mateen. a U.S citizen of Afghan descent\nU.S. President Barack Obama called it 'an act of terror and an act of hate.' Islamic State claims responsibility\nEnglish Sentence\nMon Aug 15 11:34:08 PDT 2016\nI just miss the phone technical interview today, and I need to lie about being stuck in traffic.\nI was stuck in traffic this morning, please reschedule it, LOL. I'm pretty sure this is typical excuse for being late\nHere are the top five excuses for being late\n1. My car broke down on the highway or on the road\n2. There was something came up suddenly this morning and I can't be on the phone.\nEnglish has too many Crazy Forms, Crazy English\nSat May 7 16:22:48 PDT 2016\n1. Continua, Continuous, Continuously, Continual, Continuity, continuation, Continguous\n11. Consecurive, Consecutively, Consecutiveness\n12. Sequence, Sequent, Sequently\n13. Dangerous, Dangerously, Dangle, Dangling\n2. Techie, Technical, Technician, Technology, Technologies, Technological, Technologically\n3. Practice[Practise British] Practical, Practicable, Practicing[a practicing architech], Practician, Practicably\n4. Traver[se], Traver[sal], Traver[sable]\n5. Success[noun], Succeed[verb], Successful, Successfully, Successfulness\n6. Ancestor, Ancestral, Ancestry, Sibling, descend, scend, scene\n7. Even [adverb]- Used in comparision for emphasis. he knows even less about it then I do\n8. Emphasis[noun], Emphasize[verb] - give special important or prominence in speaking or writing\n9. Extra, Extract, Extreme, Extremely\n10. Cash, Cache, Caching\n11. Temporal, Temporary, Temporarily\n12. Explosive, Explode, detonate, refuse, torch\n13. Wound, Wounded [past tense], Wounded [past participle]\n-- Flying class wounds his face and neck\n14. Wind, Wound[past tense], wound[past participle]\nSun Aug 14 13:57:39 PDT 2016\n1. Here are the keywords that I learn from the bombing article\n2. Explosive, Explode, Detonate, Defuse, Torch\n3. Torch - [set fire to]\n4. Here are the sentences that I learn from the article.\n5. The bomb is exploded near a market\n6. Five explosive devices were hidden in the plant pot near the upscale resort in Hua Hui.\n7. Open-font door bar.\n8. The attack has killed four locals and wounded dozens\n9. Whoever carried out these attacks, they surely send a message, shaking public confidence in the military's ability to maintain peace and order.\n10. The southern insurgent has continued for 12 years and killed more than 6000 people,\n11. but rarely spreaded outside the three Maylay-Muslim provinces, and almost never targeded Thailand's many tourist hotspots.\n12. Another fire bomb was found in the island of Phuket and defused, local police said.\n13. The quality of the bombs are bad. One works and other other two didn't\nI'm wondering how many people use their real names in Starbucks\nThu Apr 14 18:37:51 PDT 2016\nI went to Starbucks quite often this day.\nMost of employees knew my name so far. I never use my real name actually.\nI have nothing to hide but I usually use random name when they ask my name.\nI'm wondering how many people use their real name when they order their Ceramel Macchiato\nI'm also wondering what kind of reaction they have if I ask their cell number when they ask me name in Starbucks\nWhy Starbucks should create some English names for their drink\nI love Camerel Macchiato but I never can remember how to spell and how to pronounce it.\nWhat is wrong with English\nThu Apr 14 00:52:46 PDT 2016\nI always don't remember how to spell Cash and Cache since they have the same pronunciation.\nCash - [k\u00e6\u0283]\nCache - [k\u00e6\u0283]\nMon Apr 11 21:49:14 PDT 2016\n[Recurit a new Soccer player]\nWhen I went to Richmond soccer field for a walk today, I saw a guy who is familliar to me.\nI asked him whether he plays soccer in Burnaby every Saturday.\nHe told me he is the guy and his name is Lee something. We chatted a while.\nhe told me he moved here when he was 14 and he teach English here.\nI think he love to play soccer so I told him we play in the Richmond Soccer field at 3:30 every Saturday.\nHe seems pretty excited. Obviously he seems very excited about being part of the team.\nTue Apr 12 11:36:09 PDT 2016\nExplicit - state clearly, leaving no room for confusion or doubt\nImplicit - implied though no plainly expressed\nToday, I drove to the Richmond Library at around 11am. When I tried to park my car in the parking lots, I found out I forgot my bottle of water at home. I just drove the car back home and got the water. It took me ten minutes. I'm just looking at the EmojiOne website and they have some pretty interested Emoji in SVG file format. SVG file format can be scaled to any size without losing the detail of image.\nWed Feb 24 10:31:53 PST 2016\nIt takes only 30 seconds to form an impression. Write with confidence, not arrogance.\nWho are you?\nWhat challenges have you faced?\nPick one topic, describe it in detail, and use that to introduct yourself.\nI'm a fan of backetball, tennis and soccer.\nWhen I was growing up, I would watch soccer with my dad and brother every Saturday.\nDetail your talents and skills specifically.\nStructure the cover letter as a cause and effect. A cover letter should explain the employer or admission\nwhy you are the best candidate for the position, or what you should be admitted to the University or program to which you are applying\nI expored a number of options including Preview and Adobe PDF reader.\nI expored a number of options including bank robbery and hacking\nI expored a number of options to improve my English skills including reading one article and writing one article a day.\nI expored serval options to brush up on my English skills including reading and wrting an article a day\nI expored a few optinos to brush up on my English skills including reading and writing an article daily\nI expored all the possibilities.\nI investigate all the possibilities to come up solution.\nExpore - examine, evaluate,\n- travel in or through an unfamililar country or area in order to familiarize oneself with it\nI continue to expore new ways to generate income.\nCommunication skills include listening, speaking and writing skill.\nThu Jan 28 17:32:10 PST 2016\nIf you want someone rely on you, it is better to understimate your abilities and overestimate your risks than to go in a direction that\nactually involves more uncertainty than you can justfy.\n\nShe pushed the anxious uncertainties out of her mind.\nMon Sep 14 02:38:43 PDT 2015\nI just learn a new vocabulary:\nCircumvent [s\u0259rk\u0259m \u02c8vent] find a way around an obstacle\nCircumference\nCircumcise cut off foreskin of\nFri Sep 18 22:01:50 PDT 2015\ndescend [d\u0268\u02c8send] move or fall downward\ndescendant [d\u0268\u02c8send\u0259nt] nonu, a person or animal that is descended from a particular ancestor\nbe descended from, be a direct blood relative from\nancestor [\u02c8\u00e6n\u02ccsest\u0259r]\nSun Sep 20 17:59:45 PDT 2015\n\u00e6[x00E6]\n\u0283[x0283]\n\nprefix [ ] add thing in font of sth or at the beginning of a word, add sth at the beginning of\npostfix [ ] append thing at the end of sth, append as a suffix\nappend [ ] add sth as an attachment or supplement\nelaborate [\u0268, l\u0251b\u0103r ] involving carefully arranged parts or detailed or complicated in design\nthe key idea of the book is [expressed] in the title and [elaborated] in the text\nFri Sep 25 20:41:49 PDT 2015\nNew Words:\nrecord [noun] [verb]\nexecute [noun] [verb]\n\ndisguise [] a different appearance in order to conceal one's identity\ndisguise [] a different appearance in order to conceal one's identity\ndisguise [] a different appearance in order to conceal one's identity\ngust [] noun - a brief or rush wind\ndisgust [] profound disapproved aroused by sth unpleasant or offensive\nelaborate [] involving carefully arranged part or details, detailed and complicated in design and planning\nelaborate [] involking carefully arranged parts or details, detailed and complicated in design and planning.\nelaborate [] involking carefully arranged parts or details, detailed and complicated in design and planning.\n\ngrandiose [ ] impressive and magnificent in appearance or style\n\nliberal - adjective, open to new behavious or opinion and willing to discard traditional values bind [] [past and past participle bound]\nvisible [] seeable, obversable[obverse] notify, note, notice, noticeable, detectable[detect]\ne.g. the action putStrLn \"Hello World\" with previous result bound to sth.\nHave + Object + past participle\ne.g. I have my hair cut.[I have cut my hair]\n\nThe structure being + adjective is used to talk about actions and behaviour\ne.g. why are you being so silly\ne.g. you are being cruel when you hurt others with your words or actions.\n\ne.g. this involes being liberal with your SpaceBar and Enter\/Return Key Being + past participle\nBeing can be followed by a past participle. This structure is used in the passive forms of present and past continuous tenses.\ne.g. mother is cooking the dinner\ne.g. the dinner is being cooked\ne.g. I'm quite sure someone is following me[Active]\ne.g. I'm quite sure I'm being followed[Passive]\n\nBeing in participle clauses\ninstead of because\/since\/as, we sometimes use an abverbial participle clause with being.\ne.g. Being late, he could't watch the movie[Becase he is late, he could't watch the movie]\n\n[Therefore, Consequently, For this reason, As a result, Thus] thing have changed recently-which led to people asking the questions\nlay out the current setup\nlines starting with '#'\nlines ending with certain strings\nsentence starting with capital letter\nprint all the words separated by white spaces\nuse pattern matching and regular expression\nhow to split words separated by one or more whitespace\nplease [explain] why the merge is necessary\nconnecting and linking words in English\nit is [essiential] to understand linking words, as part of speech, can be used\nto combine ideas in writing, and thus ensure that ideas within sentences and\nparagraphs are elegantly connected - for the benefit of the reader.\nThis will help to improve your writing[e.g. essay, comment, summary]\nI just learn some interested words:\nthe words have different pronunciation but same spelling is called heteronym.\nshow you how to 3d touch in iphone6\npop up little options, quick aciton, message will show you some recent message,\nright now quick action, message, will show you show recent msg, photo show you recent camera\nwhen i click on weather, wallet let you add a card, note let you cratch, 3d touch make the screen press sensitive\nok, back to quick aciton, news have quick aciton, let you jump to different category,\nlet pee and pop, and i hold down a email message, if i keep holding down, perk and pop, if i hold down,\nif i keep holding down, here is perk and pop, i'm going to athelet, this is perk and pop, let see how 3d toouch affect the keyboard,\nthe keyboard become track pad, swip through, let me see how 3d touch work in Safari, if you hold down on the link,\nthis is really neat, this is PC maC, apple called quick acitons,\nmessage will show you some recent message, photo will show your recent photos,\nselfies, if i click on, map let you pin current location, note let you\nNevertheless,\nIn conclusion,\nIn summary, there is not clear case for one tax system compared to another\nIn Summary, there is not clear cases for one tax system compared to another\nNevertheless,\nto some extend,\nit is essential to understand connecting and linking words, as parts of speech, can be used to combine sentences and paragraphs in writing,\nand thus ensure the idea within the sentences and paragraphs are elegrantly connected\nit is essential to understand connecting and linking words, as parts of speech, can be used to combine sentences and paragraph in writing,\nand thus ensure the idea within the sentences and paragraphs are elegrantly connected\nIn summary, there is not clear tax system compared to another\nprint all the words separated with whitespace\nsplit all the words separated with whitespace\nit is essential to understand the connecting and linking words, as parts of speech, can be used to combine sentences and paragraphs in writing,\nand thus ensure the idea within the sentences and paragraphs are elegrantly connected\nIn summary, there is not clear tax system compared to another.\nIf no input-file is specified, file is read from stdin. Otherwise, ...\n[If sth..., sth.... Otherwise...]\nIf input-file is empty, the method returns false. Otherwise, return true ...\nlanguage attribute, language name, file type,\nDisable syntax highlighting for code block and inlines, even though when a language attribute is given.\n[disable ... for ..., even though something]\nsomthing, even though when somthing\nRemove the whitespace, even though the string is empty.\nReturn false, even though the input-file is empty.\n[something ... even though ...]\nsplit the string separated by whitespace, even though input-file is empty\n[something ... even though ...]\nnumerical entities instead of UTF-8 when the option is selected\n[something ... instead of something when ....]\nSan Francisco Disrupe News\nWhere do delimiter go?\nThe null delimiter, whitespace, empty string, null pointer, null reference, null value, zero value, empty value, empty string\nsplit the string with delimiter[default is whitespace] preserving the null delimiter\nsplit's return value\nfortuniately, fortuniate, future, fortran, fortune, furniture\nThe star quantifier[zero or more]\nif the split patter is unspecified, the method returns false.\n\nif a file is written by the first action and the file is read by the second action, then the change to the file is visible to the read\n[if sth.. and sth.., then the change to the file is visible..]\nSun Sep 27 13:31:14 PDT 2015\nLast night, I try to write a simple script to read a text file and surround each line with Html tags.\nThe script should be fairly to write, but I decide to write it in Haskell.\nI knew a bit of Haskll before so I still remember the basic syntax.\nIn Haskell, there is interactive command line tool called GHCi.\nI primarily use the GHCi to get help on many unfamilar functions\n\n:info concat - show the definition of concat function\n:e helloword.hs - edit the helloword.hs[editor has to be set before that, e.g. vim]\n:load helloword.hs following by main to run the code\n\nWed Sep 30 11:50:27 PDT 2015\nI went to WallMart this morning to buy a money order.\nI ask for 100 cash back and the WallMart employee just give me 100 bill.\nI ask him can you give me to 20 bills, he said, sorry, 'we don't have any 20 bills any more'\nThis is pretty ridiculous behavious.\n\nHere is Why WallMart sucks:\nThe groceries are crappy. For example, it is extremely hard to find a good apple in WallMart.\nThe services are poor. For example, they don't have 20 bills for my 100 cash back.\nThe lines are extremely slow. For example, they don't hire enough cashiers\n\nNew words of the day:\ncash [] - money in coins or notes\ncash in [] - take advantage or exploit [a situation]\ncashier [] - a person who collects payments in stores - a person handling payments and receipts in store, bank, or other business.\ncache [] - store away in hiding for future use\n\nWed Jan 31 09:27:31 2018\nIt's easier and more intuitive to declare... If you take a look at this article where students have shown FP is harder than imperative programming\nIt illustrate FP is easier than imperative programming.\nAs today is Chinese New Year, there are many New Year decoratoins on the street\nWed Jan 31 20:58:36 2018\nMy manager asked me to send the performance today, obviously the guy try to nail me at some point\nI will switch to other team next week(if everthing are going well hopefully), but the guy still try to nail me.\nI really don't understand what is his intention? but I can feel he has some shit on his mind\nAir conditioned car.\nThe senior manager asks me the same shit to send him the email about my performance\nI think this is not a good sign for me in next week, they might block me to transfer to other team\nWell, finger cross and see what will happen, it would be excited and depressive\nThu Feb 1 07:38:21 2018\nIt would be excited and depressive today and I will see how the guy pull the shit of me.\nPersonally, I would like to leave the company right away. However, I always think what other people is thinking about me\nlike what company do you work for?\ndo you have house? do you have car? how much money do you make? do you work for top company?... this is all shit.. and we need to get rid of that in our mind. Thu Feb 1 19:34:46 2018\nMy new manager told me I won't be able to move to new team next Monday and I got a new message from him and said I might be able to move to new team next Monday.\nfinger cross, I will see what will happen?\nAir conditioned car? what is conditioned here anyway?\nDo you have any update from HR?.. I told me he will get me posted.\nNath guy send me the info how to connect to Dynamoc DB and how to connect to Oracle. Well, he did not send it to me at the same day when he assigned me the task?\nSend me the task and ask me here is the task, and you need to fix the shit? what the fuck that I need to do actually?\nCreate Test DB account and I still can't make the connection to DB.. not what is going with those BS..\nMy manager told me you need to finsh the task without anyone help?? seriously?? he guy obviously does not know what he's talking about???..\nWithout any help from anyone?... what the fuck the guy is talking about?\nUse Sqlplus to connect, but get error to make the connection\nYou should best it on Beta environment?\nJus try to install java8 in my MacOX, you can install Java8 with homebrew:\nbrew update\nbrew install java8(not java)\nThere is crazy guy ask me where I buy my jacket in Skytrain today, I told the crazy guy I bought it in Walmark\nMon Feb 19 21:48:41 2018\nSomething is very unsually happening today. When I was in the washroom today and washing my hand, a guy came to my shoulder and told me that DK left the company. I asked him what was happening? he said \"I have no idea\". Dk left the company this morning suddenly. The first question that I ask myself \"Why it is today, it was Monday\" The senior guy who came to our office to have a meeting with the team from Seattle and I specular there is something to do with the senior guy. I never hear anyone said Dk will leave the company and it is in such unusual time. Tue Feb 20 09:06:18 2018\nAll food will be disposed and all containers will be emptied You have been added to a prototype with Amzon Private Cloud. Hannah Grossman has added you as collaborator on a R4S prototype - Mobile. There is some interesting thing happening this moring in the Skytain. A girl who's wearing a red jacket and standing beside of me, she's constantly looking at me when I was making sound in my mouth.","date":"2022-05-26 17:17:48","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\": 0.23818320035934448, \"perplexity\": 5257.427257579197}, \"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-2022-21\/segments\/1652662619221.81\/warc\/CC-MAIN-20220526162749-20220526192749-00299.warc.gz\"}"}
| null | null |
\section{Introduction}
Camera calibration is of immense importance in the extraction of information from video surveillance data. It could either be used to deal with the perspective distortion of the object in the image plane or it can be used for photogrammetric measurements like distances, velocities, trajectories, etc. It is also fundamental for performing the multiview 3D reconstruction. Besides, with the aid of 3D information, it could also be used for vehicle tracking or object detection, robust to occlusion.
Owing to its importance, a significant portion of literature in computer vision addresses the problem of camera calibration. Almost all the methods of calibration could be categorized into two major approaches:-
\begin{itemize}
\item Finding image-world correspondences\cite{MITpress, Hartley:2003:MVG:861369,Tsai, Zhang}.
\item Vanishing Point-based methods\cite{Caprile1990, 10.1007/3-540-47979-1_12, Zisserman, 1661552, 4587780}.
\end{itemize} The first method tries to exploits the properties of the 3D scene structure to find correspondences between the real-world and its 2D image captured by the camera. With the aid of these correspondences, intrinsic as well as extrinsic estimation could be made. Further, the accuracy can be improved with increasing such correspondences. The vanishing point methods are majorly based on estimating the orthogonal vanishing points in the scene and requires no a priori knowledge for recovering the extrinsic and intrinsics matrices.
If the camera is already pre-calibrated using 3D rigs or checker-board, this data could be used directly in the second stage. However, in most cases, this data is not always available and hence there is a need for automatic intrinsic calibration also. Most of the literature in traffic surveillance considers an ideal camera model which assumes that the pixels are perfectly square with zero skew and the optical center of the camera coincides exactly with the image center. The last assumption is not necessarily true and in such cases, the optical center is calculated in a slightly different fashion.
For extrinsic calibration in the context of traffic surveillance, the correspondences approach may require annotation of lane marking with its lane-width\cite{article}\cite{doi:10.1139/cjce-2011-0456} or presence of ground control points, or using regional heuristics such as average vehicle dimensions\cite{bhardwaj2017autocalib} or speed. However, the majority of the above methods involve a human intervention or are location-dependent and are unsuitable for generalized auto-calibration. Thus for such applications, we make use of vanishing points. The vanishing points may be generated from the static scene structures or lane markings, or motion of vehicles and pedestrians\cite{4020751}. For robust auto camera calibration, it is always advisable that the calibration process is independent of the scene in general and hence our approach will use only the moving objects in the scene.
This paper especially deals with the wide-angle lenses cameras, that are ideal for most of the photogrammetry applications. However, they are always accompanied by various distortion effects, among which fisheye effects are dominating. It is essential to remove such effects before vanishing points estimation. The remaining paper is organized into the following sections: Section II defines the Camera Model, Section III and IV describes the process of Intrinsic as well as Extrinsic Calibration, Section V presents the results of our approach and Section VI explains the Conclusion and Future Scope of the algorithm.
\begin{figure*}
\begin{tabular}{cc}
\centering
\subfloat[]{\includegraphics[width=8cm]{snap1.jpg}}&
\subfloat[]{\includegraphics[width=8cm]{Selection_086.png}}\\
\subfloat[]{\includegraphics[width=8cm]{Selection_087.png}}&
\subfloat[]{\includegraphics[width=8cm]{Selection_088.png}}
\end{tabular}
\caption{(a) Original distorted image of the traffic Scene, (b) Extracting trajectories using KLT-tracker, (c) Filtering top 10 trajectories, (d) Undistorted image of the same scene}%
\label{intrinsic}%
\end{figure*}
\section{Camera Model}
We have assumed our camera to obey the pin-hole camera model. In this model, perspective projection of a 3D point into the 2D image point can be represented as follows :
\begin{equation}{\lambda \begin{bmatrix}u \\ v \\ 1\end{bmatrix} =
\begin{bmatrix}
f_x & s & cx\\
0 & f_y & cy\\
0 & 0 & 1
\end{bmatrix}
\begin{bmatrix}
r_{11} & r_{12} & r_{13} & t_x\\
r_{21} & r_{22} & r_{23} & t_y\\
r_{31} & r_{32} & r_{33} & t_z
\end{bmatrix}
\begin{bmatrix}
X\\
Y\\
Z\\
1
\end{bmatrix}}
\end{equation}
\\The first matrix on R.H.S is referred to as the intrinsic matrix,
as it is only dependent on the internal properties of the camera.
$(f_x, f_y)$ represents the focal length in x and y directions, s
represents the skew present in the pixels. $(cx, cy)$ represents
the optical center. In our approach we have assumed that focal
length in x and y direction are identical, the skew present is zero
and the optical center is also the image center. On simplifying
the equation of a camera which captures an image of size $(H x W)$ becomes:
\begin{equation}
{\lambda \begin{bmatrix}u \\ v \\ 1\end{bmatrix} =
\begin{bmatrix}
f & 0 & W / 2\\
0 & f & H / 2\\
0 & 0 & 1
\end{bmatrix}
\begin{bmatrix}
r_{11} & r_{12} & r_{13} & t_x\\
r_{21} & r_{22} & r_{23} & t_y\\
r_{31} & r_{32} & r_{33} & t_z
\end{bmatrix}
\begin{bmatrix}
X\\
Y\\
Z\\
1
\end{bmatrix}}
\end{equation}
The second matrix in equation 1 is the extrinsic matrix and
is composed of 3x3 rotation matrix, augmented with a 3x1
translation matrix. In our approach, it would be possible to
estimate the rotation matrices and translation vector (with scale
ambiguity), if no input from the user is provided. However, if
the height of the camera from the ground is also provided, scale
ambiguity could be resolved.
To account for the deviation of the camera from the pin-hole model,
we consider a distortion matrix separately. Here we assume only radial effects, fitting a polynomial model\cite{Ronda2019} as :
\begin{equation}
x_u = x_d * (1 + k_1 * r^2 + k_2 * r^4 + k_3 * r^6)
\end{equation}
\begin{equation}
y_u = y_d * (1 + k_1 * r^2 + k_2 * r^4 + k_3 * r^6)
\end{equation}
Here, $(x_u, y_u), (x_d, y_d)$ denotes the undistorted and distorted normalized coordinates of the image and [k1, k2, k3] are known as distortion coefficients. r is defined as follows:
\begin{equation}
r_d^2 \xrightarrow{} r^2 = x_d^2 + y_d^2
\end{equation}
In this model, the distortion center coincides with the image center. Also, the fish-eye effect of distortion is assumed to be radial in nature.
\section{Intrinsic Calibration}
\subsection{Fish-eye Effect}\label{AA}
A pin-hole type world to image mapping is possible only by a rectilinear lens which satisfies:
\begin{equation}
r_u = f * tan(\theta)
\end{equation}
$\theta$ is the angle in in radians between a point in the real world and the optical axis, which goes from the center of the image through the center of the lens. f is the focal length of the lens and $r_u$ is radial position of a point on the image film or sensor. Since it abides by pin-hole model, it is also considered undistorted.
After observation across the datasets available, it was realized, to account for distortion, a global distortion model is to be used. The most common and simple among which was equidistant fish-eye. It is defined as follows :
\begin{equation}
r_d = f * \theta
\end{equation}
$r_d$ is radial position of a point on the image film or sensor. Since it do not abides by pin-hole model, it is also considered distorted. The equations (6) and (7) can be used for computing forward mapping i.e. from distorted cordinates to undisorted cordinates and inverse mapping i.e. from undistorted cordinates to distorted cordinates respectively as follows :
\begin{equation}
(x_u, y_u) \xrightarrow{} (x_d * \frac{tan(\frac{r_d}{f})}{\frac{r_d}{f}}, y_d * \frac{tan(\frac{r_d}{f})}{\frac{r_d}{f}} )
\end{equation}
\begin{equation}
(x_d, y_d) \xrightarrow{} (x_u * \frac{atan(\frac{r_u}{f})}{\frac{r_d}{f}} , y_d * \frac{atan(\frac{r_u}{f})}{\frac{r_d}{f}} )
\end{equation}
\begin{equation}
r_u^2 = x_u^2 + y_u^2
\end{equation}
Thus if we have enough correspondences between $(x_d, y_d)$ and $(x_u, y_u)$, it would be possible to estimate the focal length as well as distortion coefficients using a least square approach.
\subsection{Calibration}
To make the calibration independent of the scene, only moving
objects were used. It is done because it is not always possible
to have a similar geometric arrangement of static structures
in every scene on which calibration is performed. However,
moving objects could comprise of pedestrians and vehicles,
either of which is very easily present in every traffic scene.
There is only one unknown in the intrinsic matrix and
equations 8 and 9, i.e. the focal length f. Thus if we can determine f, the intrinsic matrix, and distortion coefficients
could be computed.
In general, on a road, it is assumed that vehicles move along
a straight line. However, due to distortion(fish-eye effect), these
straight lines become curved in the image. The relation between
the undistorted image and distorted image is only dependent
on the focal length. Thus, by adjusting the value of f to undistort the straight line, focal length can be computed.
The calibration process starts with extraction of moving
object trajectories from video footage as shown in the image.
In a video it is assumed that, there are sufficient number
of vehicles moving in orthogonal directions. To extract the
distorted straight lines, moving pedestrians or vehicles are
tracked in a video footage for a few seconds initially. Tracking
is achieved by optical flow with sift keypoints. Multiple
tracks are extracted and filtered as follows:
\begin{itemize}
\item
If the key point is not tracked for more than 80 \% of the
video interval, it is rejected.
\item
If the total distance in pixels of a keypoints is greater than
1.2 times of its displacement, the keypoints is rejected.
\end{itemize}
Once the trajectories are filtered, top ten longest trajectories are selected. These trajectories are fitted with straight lines
using least square method. The sum of least square errors of all
the trajectories gives us the estimate of the straightness of the
lines. Now the distorted points on trajectories are undistorted by varying the value of focal length from zero to diagonal of
image i.e. the maximum possible focal length in pixels. The
most appropriate focal length would be one with the minimum
least square error. Once the focal length is found, intrinsic
matrix is computed.
However, for estimating the distortion coefficients, the sift keypoints are to be normalized with intrinsic matrices as :
\begin{equation}
\lambda \begin{bmatrix}x_d \\ y_d \\ 1\end{bmatrix} =
\begin{bmatrix}
f & 0 & W / 2\\
0 & f & H / 2\\
0 & 0 & 1
\end{bmatrix}^{-1}
\begin{bmatrix}
u\\
v\\
1
\end{bmatrix}
\end{equation}
The undistorted cordinates for the same are calculated with the mapping equation 8. Now the cordinates of distorted and undistorted keypoints could be used to compute the Distortion coefficients using equation 11.
\begin{equation}
\frac{x_u}{x_d} - 1 = \begin{bmatrix}
r^2 & r^4 & r^6
\end{bmatrix}\begin{bmatrix}
k1\\
k2\\
k3
\end{bmatrix}
\end{equation}
The steps of intrinsic calibration are shown in Fig \ref{intrinsic}
\begin{figure}%
\centering
\includegraphics[width=8cm]{Selection_089.png}
\caption{Bimodal distribution of gradients of line segments}%
\label{histogram}%
\end{figure}
\begin{figure*}%
\centering
\subfloat[]{{\includegraphics[width=8cm]{Selection_083.png} }}%
\subfloat[]{{\includegraphics[width=8cm]{Selection_084.png} }}%
\caption{(a) Vanishing Point in x-direction, (b) Vanishing Point in y-direction}%
\label{votes}%
\end{figure*}
\section{Extrinsic Calibration}
\subsection{Vanishing Point Method}
Given the intrinsics, the extrinsics are computed using vanishing points. Vanishing points are defined as points on image plane where parallel lines in 3D intersect. For calibration, minimum of two orthogonal vanishing points are needed. Due to undistortion, the focal length of the new image is different from the original image. Thus, we will have to first compute the new focal length. If we have two vanishing points say $(u_x, v_x)$ and $(u_y, v_y)$ the new focal length is computed using the relation :
\begin{equation}
(u_x - c_x)(u_y - c_y) + (v_x - c_y)(v_y - c_y) + f_{new}^2 = 0
\end{equation}
Once the new focal length is known, the elements of rotation matrices computed from equations 14, 15, 16 :
\begin{equation}
\frac{u_x - c_x}{f_{new}} = \frac{r_{11}}{r_{31}};
\frac{v_x - c_y}{f_{new}} = \frac{r_{21}}{r_{31}}; r_{11}^2 + r_{21}^2 + r_{31}^2 = 1
\end{equation}
\begin{equation}
\frac{u_y - c_x}{f_{new}} = \frac{r_{12}}{r_{32}};
\frac{v_y - c_y}{f_{new}} = \frac{r_{22}}{r_{32}}; r_{12}^2 + r_{22}^2 + r_{32}^2 = 1
\end{equation}
\begin{equation}
\begin{bmatrix}
r_{13} \\
r_{23} \\
r_{33}
\end{bmatrix} = \begin{bmatrix}
r_{11} \\
r_{21} \\
r_{31}
\end{bmatrix} \times \begin{bmatrix}
r_{12} \\
r_{22} \\
r_{32}
\end{bmatrix}
\end{equation}
Assuming the road surface to be planar and constitutes a Z = 0 plane, the equation 2 is transformed to :
\begin{equation}
{\lambda \begin{bmatrix}u \\ v \\ 1\end{bmatrix} =
\begin{bmatrix}
f & 0 & W / 2\\
0 & f & H / 2\\
0 & 0 & 1
\end{bmatrix}
\begin{bmatrix}
r_{11} & r_{12} & t_x\\
r_{21} & r_{22} & t_y\\
r_{31} & r_{32} & t_z
\end{bmatrix}
\begin{bmatrix}
X\\
Y\\
1
\end{bmatrix}}
\end{equation}
If we assume the origin of World Cordinate System corresponds with the center of image and height of the camera from ground is $H_e$ then:
\begin{equation}
t_x = 0; t_y = 0; t_z = -H_e / r_{33}
\end{equation}
\subsection{Calibration}
As mentioned above, extrinsic calibration is also done using moving objects only. The motion of pedestrians is highly erratic thus, only the motion of vehicles is used for vanishing point estimation. Vehicles in most of the cases follow each other along a straight line. If there are multiple key-points on a single vehicle, it can be seen to originate or converge to a point in image plane, as vehicle moves closer to or farther away from the camera. If there are multiple vehicles moving along two orthogonal directions, orthogonal vanishing points can easily be detected.
The algorithm mentioned below is computationally expensive, and requires significant movement of vehicles in pixels, thus it is performed in one out of every six frames. The first step of calibration is to undistort the image with the help of distortion coefficients, as it is not easy to detect vanishing points from curves. It is followed by YOLO-v3\cite{DBLP:journals/corr/abs-1804-02767} based detection of vehicles. This generates the regions that consists of vehicles with high probability. A mask of such regions is computed for every image on which processing is done. SIFT\cite{Lowe:2004:DIF:993451.996342} key points are generated in the masked images and is matched with the keypoints in the consecutive masked images. The positions of these matched keypoints for any two frames produces multiple line segments. All the line-segments are stored for future processing.
Once, the whole short video-footage is processed, a list of line segments are generated. Every line segment is converted into polar form i.e (orientation, magnitude, and distance from origin). A histogram of orientation is computed from all the segments as shown in Fig \ref{histogram}. It is observed that it shows a bimodal distribution. The bimodal distribution implies that there are two major directions in which vehicles move. This bimodal distribution helps in generating clusters of line segments in each orthogonal direction. Most appropriate line segments are selectively picked using the distribution peaks with a threshold of 5 degree in either direction.
The detection of vanishing point is achieved by voting based system. For each cluster, every line segment is extended to a line. The size of accumulator is equal to one pixel and its value is initially set to zero. Through whichever pixels any line passes, is incremented by one. Thus the pixels with maximum votes are considered most likely positions of vanishing points. The votes in the two directions are shown in the Fig \ref{votes}.
However, the point with maximum votes is not necessarily the vanishing point, and there can be more than one vanishing point because, we do not achieve a perfect undistortion. If more than one vanishing point exists, it may imply that the point with maximum votes lie some where in between the two or more vanishing points. In order, to account for this, we take the top 20 \% of the pixels with maximum votes and compute the mean and standard deviation. The estimates of vanishing points is made equal to the sum of mean and two or three times of standard deviation. The columns of rotation matrices does not vary much with the value of constant multiplied with standard deviation.
Once the vanishing points are computed, the equations 13, 14, 15, 16 and 18, could be used for extrinsic calibration.
To improve the rotation matrix and enforce constraints on it, SVD is performed on the matrix. The new matrix is defined as:
\begin{equation}
R = USV^T
\end{equation}
where S is an identity matrix. With the height of camera from the ground, translation vector is computed from equation 18.
\section{Results}
\subsection{Accuracy of Estimated Focal Length}
To estimate the accuracy of the computed focal length from the proposed algorithm, camera is calibrated from checkerboard pattern. When using checkerboard pattern the focal length in x and y direction may not be same, to get the estimate its geometric mean is considered.
In the table \ref{Focal}, Proposed column is one which uses the proposed algorithm for estimation of focal length and Checkerboard column uses checkerboard calibration. The results from 5 very different video-footages are used to account for generality of results. The accuracy is around 6.35\% of truth value (assumming checkerboard to be ground trruth).
\\
\begin{figure}%
\centering
\includegraphics[width=8cm]{Selection_085.png}
\caption{Top-view transformed Image}%
\label{top-view}%
\end{figure}
\begin{table}
\begin{tabular}{|c|c|c|}
\hline
Datasets & Proposed & Checker Board \\
\hline
1. & 1064.20 & 1012.21\\
\hline
2. & 619.66 & 584.95\\
\hline
3. & 638.52 & 584.95\\
\hline
4. & 937.41 & 1012.21\\
\hline
5. & 933.43 & 1012.21\\
\hline
\end{tabular}
\captionof{table}{Comparison of focal length with the proposed method and checkerboard method}\label{Focal}
\end{table}
\subsection{Accuracy of Estimated Distortion Coefficients}
To estimate the accuracy of distortion coefficients, mean least square error in trajectories is computed with the predicted distortion coefficients and checkerboard calibrated coefficients. The results are tabulated in Table \ref{Distortion}
In Table \ref{Distortion}, the second column represents the mean error of distortion in trajectory in original image. The third column represents the mean error of distortion in rectified image with proposed algorithm. the last column represents the mean error in distortion in rectified image with checker-board calibration coefficients. It can be seen, in most cases proposed method performs better undistortion then the checkerboard calibrations coefficients.
\\
\begin{table}[t]
\begin{tabular}{|c|c|c|c|}
\hline
Datasets & \pbox{10cm}{Mean LSE \\ before calibration \\for a trajectory} & \pbox{10cm}{Mean LSE\\ after calibration \\for a trajectory} &\pbox{10cm}{Mean LSE after \\checkerboard\\ calibration\\ for a trajectory}\\
\hline
1. & 0.572 & 0.103 & 0.149\\
\hline
2. & 0.923 & 0.251 & 0.456\\
\hline
3. & 0.528 & 0.069 & 0.111\\
\hline
4. & 0.166 & 0.004 & 0.061\\
\hline
5. & 0.052 & 0.006 & 0.007\\
\hline
\end{tabular}
\captionof{table}{Comparison of undistortion of trajectories before and after the Intrinsic Calibration with the proposed method and Checkerboard Method}\label{Distortion}
\end{table}
\subsection{Accuracy of Rotation Matrices}
It was not possible to assess the accuracy of rotation matrices because, the rotation data was not computed when camera was originally set-up, only its video-footage was accessible. However, it is possible to visually estimate its accuracy. It is done by converting the image of a scene to a top-view plane transformed image as shown in Fig \ref{top-view}. In such images, the crosswalks will appear rectangular and not trapezium-like, man-holes will appear circular rather than ellipse, road markings will appear parallel instead of intersecting. There are enough visual cues in such images to test the accuracy of the data visually.
\section {Conclusion and Future Scope}
The proposed algorithm can perform calibration automatically from a video-footage. The intrinsic calibration works well also for the thermal cameras, since calibration procedure is dependent on only moving objects. However, there are certain limitations associated with the proposed method :
\begin{itemize}
\item Tracking done using KLT tracker is not robust against occlusion.
\item Assumptions of vehicles move along a straight line may be violated in certain scenarios.
\item There should be sufficient number of vehicles moving in either directions to estimate vanishing points robust to noise.
Once, the calibration is performed it could be used in multiple applications concerning to photogrammetry, speed-monitoring, 3D-reconstruction, etc. This algorithm in future could also be made independent of the requirement of perpendicular intersections.
\end{itemize}
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 2,147
|
Home page>PRESS RELEASE>Companies News>Alstom recognised Top Em...
Alstom recognised Top Employer in Middle East
Recognition by the Top Employers Institute for the first time in Saudi Arabia, the United Arab Emirates and Egypt
Awarded Global Top Employer for 2023 certification in 22 countries
Alstom has been recognised as a "Top Employer" for the 2022 calendar year in Saudi Arabia, UAE and Egypt for the first time, alongside certifications in 22 other geographies the company operates in, including Australia, Belgium, Canada, China, Hong Kong, France, India, Italy, Mexico, Poland, Singapore, Spain, United Kingdom, United States of America, Chile, Panama, Romania, Sweden, and Thailand. Being certified as a Top Employer showcases an organisation's dedication to a better world of work and exhibits a people centric culture.
Alstom continues to expand its employee engagement efforts across the region, focusing on building a strong corporate culture that creates a sense of belonging and purpose among its people. The Top Employers Institute also commended the company's career and talent development programs that provide continuous learning opportunities. Additionally, employee wellness has come into sharper focus at Alstom, with the company offering health and well-being support programs to help them excel in their personal and professional lives.
"We are honoured to be recognised as a top employer in Saudi Arabia and very proud of the team of our talented and diverse individuals serving the Kingdom's mobility needs. As we focus to support the fulfilment of Saudi Arabia's Vision 2030, our commitment to people and diversity will be key to healthy, balanced and sustainable growth," said Mohamed Khalil, Managing Director of Alstom Saudi Arabia.
"Providing our people with a greater purpose at work and fostering a culture of transparency and respect are essential in our success in the Emirates. Our employees know the work they do has an impact, and in making public transport smarter, more accessible, and most importantly sustainable," said Tamer Salama, Managing Director of Alstom UAE & Qatar.
"This certification is a recognition of the work Alstom does for the well-being of its employees. We are delighted to be able to provide our employees a healthy and constructive work environment, allowing each of them to bring their best selves to support our customers bring the best solutions to address critical mobility challenges" said Ramy Salah El-Din, Managing Director of Alstom Egypt.
Through its collaboration with the Top Employers Institute, Alstom will now benefit from being able to benchmark its internal efforts against industry best practice, determining further areas of improvement. The certifications will also serve to attract new talent, as the company continues to expand its presence across the Middle East.
About Top Employer certification
Top Employer certification recognises a company's commitment to creating a better working environment for its employees and excellence in human resources' policies and practices. The Top Employers Institute certifies organisations on the basis of their results in the HR Best Practices Survey. The survey covers six HR areas, divided into 20 themes such as people strategy, work environment, talent acquisition, training, well-being, diversity and inclusion.
GC Labs to present full portfolio of diagnostic testing at Medlab Middle East 2023
Average take-home pay at a low point: SA
Kuwait Petroleum Corporation to recruit '600 Kuwaitis'
Canada expected to buck trend of big investment banking layoffs
Elevator-maker Kone to slash 1,000 jobs, beats Q4 profit expectations
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 5,377
|
Why the U.S. Mental Health Care System Needs More Places for People to Just Chill Out
By Theresa Fisher
On Jan. 18, 2011, Misha Kessler opened the window of his dorm room at George Washington University, where he was a sophomore, climbed onto the window ledge and prepared to jump into the winter night. Kessler had been dealing with anxiety and other mental health issues since the summer. Earlier that day, after a week without sleep, he'd had his first-ever panic attack.
Kessler's lack of control over his own mental state scared him, as did preparing for a six-story free-fall. But the scariest part of that night came later, when he voluntarily went to the emergency room. The prospect of going to the psych ward, to join the real crazy people, Kessler recalls, terrified him more than deliberating suicide.
"I still had the basic idea of a psych ward from TV," he said. "But when I got there, the other people were a lot like me, just experiencing too much stress without the right coping mechanisms."
In the end, he was glad he went. While the process of handing over his personal belongings and donning a hospital gown was by no means pleasant, Kessler knew he needed to get out of his dorm room and put his fate in someone else's hands. In fact, he credits his two-night stay in the psychiatric ward with saving his life. Today, as a suicide prevention specialist, Kessler urges teenagers and young adults who find themselves in a dark place to go get help, even if that means checking in somewhere so they don't check out.
Note: Kessler talks about his suicide attempt in the video below.
But Kessler ended up at the hospital because he didn't see another option. Emergency rooms are designed to revive trauma victims and pump stomachs, not address mental health crises. Patients who need immediate psychiatric care endure a drawn-out intake process that often involves repeatedly explaining why they're at the ER and waiting hours for beds to free up in a psych unit. It's re-traumatizing, ineffective and expensive. That's why Illinois created Living Rooms, state-funded, alternative ER centers for mental health crises.
The U.S. has radically shifted its approach to psychiatric care a number of times. During the second half of the 20th century, a policy of deinstitutionalization resulted in the discharge of thousands of mentally ill and developmentally disabled Americans from psychiatric hospitals, many of which subsequently shut down. The policy, though well-intentioned, received criticism for leaving former in-patients to flounder.
Similar criticism resurfaced recently, when studies revealed just how many incarcerated Americans suffer from severe mental illness. At least 15% of state prison inmates have a psychotic disorder. Some argue that the U.S. just shifted the burden — jails are the new mental hospitals. We haven't quite figured out a system of comprehensive mental health care that doesn't infringe on people's basic civil rights.
More Living Rooms may be part of the solution.
Illinois opened the first of its five Living Rooms in 2011 as a non-clinical crisis center for people in the throes of a mental health meltdown. People can drop in to get immediate help and access to resources for longer-term care. A mental health crisis, to use the same definition as The Living Room does, is "a state in which an individual becomes overwhelmed and their usual coping mechanisms are not adequate, which leave them with disorganized thoughts and life processes ... if a crisis state is not properly treated, the condition can quickly escalate, leading to a mental health emergency."
As the name suggests, the centers are supposed to feel more like homes than hospitals. When guests walk into the Living Room, according to a 2014 study published in the journal Issues in Mental Health Nursing, staff members greet them "with open arms" and offer beverages — anything to make them feel at home, supported and in control of their own treatment.
It's really a disarmingly simple idea: Normalize the practice of seeking help and offer people a place to chill out when they feel bad.
Within the mental health field, the Living Room is thought to be a creative approach, according to Ken Duckworth, medical director at the National Alliance on Mental Illness.
In addition to keeping doctors and psychiatric nurses on staff, the centers employ peer counselors who have been through, and emerged from, mental health crises themselves. Their presence has proven to be a strength of the Living Room model, according to Living Room coordinator Peter Robichaux. The stubborn stigma surrounding mental illness often makes people hesitant to seek treatment and uncomfortable when they do. Peer counselors get it; they've been there too. As one peer counselor said in the Issues in Mental Health Nursing study, "We're not going to judge you because we are you."
Kessler, who works at the Campbell Center, a D.C.-based nonprofit, says that peer support is a resource in hospitals too. As important as high-quality doctors are, young adults benefit when relatable liaisons enter the picture.
The Living Room model of care jibes with psychiatrist Patrick Corrigan's theory of destigmatization. Corrigan, a researcher at the Illinois Institute of Technology, believes that destigmatizing mental illness hinges on more people coming out as having a mentall illness. We can learn about the causes and rates of various mood disorders, but living and working alongside people who've openly battled schizophrenia or bipolar disease is ultimately the way to change public perception. Corrigan looks to the gay rights movement for comparison. The movement really took off, he pointed out, when non-straight people became more visible in everyday life.
The Living Room model also helps stretch the state's health care budget. During its first year, the original Living Room had 228 visits from 87 guests, most of them diverted from the ER, which saved upwards of $500,000.
Jon Bradley via Getty/Getty
It's really a disarmingly simple idea: Normalize the practice of seeking help and offer people a place to chill out when they feel bad. And while Illinois' Living Rooms stand out in the U.S., non-clinical mental health care centers are the norm in Australia. In 2006, the Australian government launched Headspace, a national network of fully funded health care centers where Australians between the ages of 12 and 25 go when they're having a tough time. People can stop in whether they're battling ongoing mental illness or just feeling stressed and lonely.
The Living Room approach may be catching on stateside too. Over the past few months, mental health urgent care centers have popped up on both coasts. Los Angeles County officials opened one such center in late 2014 to relieve pressure from county hospitals and jails running low on psychiatric beds. In January, Rhode Island opened its first walk-in mental health urgent care center for similar reasons, according to an AP story. Both facilities seem to skew more clinical than the Living Room, but they fill the same broad purpose: to be there for people who need help and provide services specifically tailored to mental illness. One county supervisor who led the effort in LA, Kaiser Health News reported, described the center as a more humane approach than traditional hospitalization.
It's hard to say that any single service will fix our mental health care system, but based on history, we're willing to make big changes. The Living Room model shows that small changes — a place to go, people to talk to — may be just as important.
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 8,532
|
The 1987 Arizona State Sun Devils baseball team represented Arizona State University in the 1987 NCAA Division I baseball season. The Sun Devils played their home games at Packard Stadium, and played as part of the Pacific-10 Conference. The team was coached by Jim Brock in his sixteenth season as head coach at Arizona State.
The Sun Devils reached the College World Series, their fourteenth appearance in Omaha, where they finished tied for seventh place after losing to eventual runner-up Oklahoma State and fifth place Florida State.
Personnel
Roster
Coaches
Schedule and results
References
Arizona State Sun Devils baseball seasons
Arizona State Sun Devils
College World Series seasons
Arizona State Sun Devils baseball
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 8,789
|
Randan es una población y comuna francesa, situada en la región de Auvernia, departamento de Puy-de-Dôme, en el distrito de Riom y cantón de Randan.
Demografía
Monumentos y lugares turísticos
El Castillo de Randan- .
El Monumento a los Muertos firmado por el escultor Eugène Benet y la fundición Durenne
Referencias
Enlaces externos
INSEE
Localidades de Puy-de-Dôme
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 970
|
If you have information about doping in sport we want to hear from you. Any information passed onto Drug Free Sport will be dealt with in the strictest confidence.
are concerned about the behaviour or practices of an athlete, coach or support personnel.
It doesn't matter how small your concerns are - we want to hear from you.
Call Drug Free Sport NZ in confidence on 0800 DRUGFREE (378 437), or email us confidentially.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 7,492
|
<?php
// change this with the path where you put the library
require_once("../baobab.php");
/* either create the conf_database.php or create here the following variable
$DB_CONFIG=array(
"host" => ...
"username" => ...
"password" => ...
"dbname" => ...
"port" => ...
"charset" => "utf8" // remember mysql charsets have no dash
// [http://dev.mysql.com/doc/refman/5.1/en/charset-charsets.html]
*/
require_once("../test/conf_database.php");
if (!isset($DB_CONFIG)) throw new Exception("Missing or misconfigured conf_database.php");
$conn=mysqli_connect(
$DB_CONFIG["host"],
$DB_CONFIG["username"],
$DB_CONFIG["password"],
$DB_CONFIG["dbname"],
$DB_CONFIG["port"]
);
try {
mysqli_set_charset($conn,$DB_CONFIG["charset"]);
if (count($argv)!=2 || FALSE===array_search($argv[1],
array('animals','forum'))) {
print_r("Usage: launcher.php [animals|forum]\n");
} else {
include($argv[1].".php");
main($conn);
}
$conn->close();
}
catch (Exception $e) {
$conn->close();
throw $e;
}
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 9,000
|
import { defineComponent, mount, createElement as htm } from 'js-surface';
const CounterLabel = defineComponent({
name: 'CounterLabel',
properties: {
value: {
type: Number
}
},
render({ props }) {
return htm('label', null, htm('b', null, props.value));
}
});
// --------------------------------------------------------------------
// Counter intents
const
// State transitions
INCREASE_COUNTER = Symbol('increaseCounter'),
RESET_COUNTER = Symbol('resetCounter'),
// Interactions (aka side effects)
LOG = Symbol('log');
const Counter = defineComponent({
name: 'Counter',
properties: {
initValue: {
type: Number,
defaultValue: 0
},
style: {
type: Object,
defaultValue: null
}
},
initState({ props }) {
return { counterValue: props.initValue };
},
stateReducer: {
[INCREASE_COUNTER](delta) {
return state => ({ counterValue: state.counterValue + delta });
},
[RESET_COUNTER](value) {
return state => ({ counterValue: value });
}
},
initMiddleware({ send }) {
return {
[LOG](msg, params) {
console.log(msg, JSON.stringify(params));
// alert(msg + JSON.stringify(params));
}
}
},
methods: {
resetCounter(n) {
return ({ props, state, send }) => {
send(RESET_COUNTER, n);
};
}
},
needsUpdate({ send }) {
send(LOG, 'check whether update needed', arguments[0]);
return true;
},
onNextProps({ send }) {
send(LOG, 'next props for Counter - params:', arguments[0]);
},
onWillMount({ send }) {
send(LOG, 'will mount Counter - params:', arguments[0]);
},
onDidMount({ send }) {
send(LOG, 'did mount Counter - params:', arguments[0]);
},
onWillUpdate({ send }) {
send(LOG, 'will update Counter - params:', arguments[0]);
},
onDidUpdate({ send }) {
send(LOG, 'did update Counter - params:', arguments[0]);
},
onWillUnmount({ send }) {
send(LOG, 'will unmount Counter - params:', arguments[0]);
},
render({ props, state, send }) {
return (
htm('span',
{style: props.style},
htm('button',
{ onClick: () => send(INCREASE_COUNTER, -1) },
'-'),
htm('div',
{ style: {width: '30px', display: 'inline-block', textAlign: 'center' }},
CounterLabel({value: state.counterValue})),
htm('button',
{ onClick: () => send(INCREASE_COUNTER, 1) } ,
'+'))
);
}
});
// --------------------------------------------------------------------
const CounterCtrl = defineComponent({
name: 'CounterCtrl',
render({ send }) {
let counterInstance = null;
return (
htm("div",
null,
htm('button', { onClick: () => send(() => counterInstance.resetCounter(0)) }, 'Reset to 0'),
' ',
Counter({ref: it => counterInstance = it, style: {margin: '0 20px'}}),
' ',
htm('button', { onClick: () => send(() => counterInstance.resetCounter(100)) }, 'Reset to 100')));
}
});
mount(CounterCtrl(), 'main-content');
/*
setTimeout(() => {
mount(htm('div', null, 'done'), 'main-content');
}, 4000);
*/
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 7,108
|
\section{Introduction}
The macroscopic response of the spin-glass phase to a magnetic field change is linear at low enough fields \cite{orm}. Experiments on the time and temperature dependence of the spin-glass magnetization can be used to learn about the underlying spin configuration and its spontaneous re-organization towards more favorable states (aging) \cite{ghost}. The response function, $R(t,ta)$ reflects the spin re-organization and appears stationary at observation times log($t$) $<$ log ($t_a$) and non-stationary when log($t$) $\sim$ log ($t_a$), where $t_a$ is the effective age of the spin glass and $t$ the observation time of the experimental probe. The apparent age of the system is governed by the time spent at constant temperature (wait time, $t_w$) or in a temperature dependent experiment by the cooling/heating rate.\\
In this study we have considered three model spin glasses of spatial dimensionality three, but with different spin dimensionality 1, 2 and 3: Fe$_{0.5}$Mn$_{0.5}$TiO$_3$ ($Ising$) \cite{Ising}, Eu$_{0.5}$Sr$_{1.5}$MnO$_4$ ($XY$) \cite{xy} and Cu(Mn) ($Heisenberg$) \cite{cumn}. For all systems, we have studied the evolution and memory of the spin configuration that has been imprinted by a halt at constant temperature, $T_h$, during cooling. The probe we have used in this study is the isothermal remanent magnetization (IRM) that is attained by intermittently applying a magnetic field at the halt temperature. In the $Heisenberg$ systems, a local and random Dzyaloshinsky-Moriya (DM) interaction\cite{DM} is inherently at play \cite{fert}. For comparison, we have also investigated a Au(Fe) system, another three dimensional $Heisenberg$ spin glass, however, with stronger apparent anisotropy \cite{dorothee,eric}.
\section{Experimental}
The probe materials are those listed in the introduction: an $Ising$ Fe$_{0.5}$Mn$_{0.5}$TiO$_3$ single crystal, $T_g$ $\approx$ 21 K), an $XY$ (single crystal, $T_g$ $\approx$ 18 K) and two $Heisenberg$ (Cu(Mn) polycrystalline, $T_g$ $\approx$ 57 K and Au(Fe) polycrystalline, $T_g$ $\approx$ 24 K ) spin glasses, all of spatial dimensionality three \cite{Ising,xy,cumn}.\\
The experiments were performed in an MPMS Superconducting Quantum Interference Device (SQUID) magnetometer with the low magnetic field option, allowing a weak, $<$ 0.05 mT, residual field.\\
The temperature dependence of the zero-field-cooled (ZFC), field-cooled (FC) and isothermal remanent magnetization (IRM) of the samples was recorded using a probe field of 1 mT. The isothermal remanent magnetization (IRM) was acquired by intermittently applying a 1 mT field during $t_{\Delta H}$=3000 s at constant temperature $T_h$, ($T_h/T_g$ $\approx$ 0.7) after two different wait times $t_w$ (3 and 3000 s), and measured vs. temperature in two different thermal protocols illustrated in Fig.~\ref{fig-sketch}. In the first procedure (1), the sample is immediately cooled (10 K/min) to a lowest temperature and the IRM is measured on heating (1 K/min). In the second procedure (2), the sample is cooled immediately after the IRM has been acquired and measured on both cooling (1 K/min) and heating (1 K/min).\\
The magnetic field employed in these experiments (1 mT) is weak enough to yield linear response; this implies that the global evolution (aging) of the spin configuration at constant temperature evolves ``undisturbed'' by this weak perturbation and that the system recovers a closely similar spin state at $T_h$, after the cooling protocol (memory).
\section{Results}
Figure~\ref{fig-zfcfc} shows the temperature dependence of the ZFC, FC and IRM magnetization of the four systems in reduced units. The $Ising$ sample is measured along the direction of the spins (the $c$-axis of the Ilmenite structure) and perpendicular to this axis, in which case the susceptibility is much lower, reversible and only weakly temperature dependent (c.f. Fig.~\ref{fig-zfcfc}(a)). The $XY$ sample is measured in the $ab$-plane of the layered tetragonal structure and perpendicular to this plane, in which case, the susceptibility is much lower, reversible and only weakly temperature dependent (c.f. Fig.~\ref{fig-zfcfc}(b)). In the following we consider and perform measurements on the $Ising$ and $XY$ systems only in the spin directions, i.e. along the $c$-axis for the $Ising$ system, and perpendicular to it for the $XY$ system. The two $Heisenberg$ samples have isotropic magnetic properties; the samples have cylindrical shape and were measured along the cylinder axis, and perpendicular to it for comparison. Similar results were obtained in the two cases; only the data collected along the cylinder axes are shown in Fig.~\ref{fig-zfcfc}(c-d). The overall behavior of the susceptibility curves of the four samples show the characteristic spin-glass features: a cusp in the ZFC curve, irreversibility between the FC and ZFC curves below the cusp, and a weakly temperature dependent FC curve below the irreversibility onset.\\
Figure~\ref{fig-irm} illustrates the temperature dependence of the isothermal remanent magnetization for the different samples. A superposed ``FC'' magnetization associated with the weak residual field in the magnet has been subtracted from the data in all cases; the magnitude of this is of the order of the IRM, see inset of Fig.~\ref{fig-irm}(d), where the measured ``raw'' data of the magnetization of the Cu(Mn) sample (IRM and background) is plotted. The fact that the nominally zero field in the magnetometer in reality is a weak finite field does not affect the IRM results of Fig.~\ref{fig-irm}; since the principle of superposition is valid for the recorded magnetisation response to a field change in the linear response regime \cite{orm}.\\
There are common features in the IRM for all four samples. The IRM is frozen in on cooling and shows only a weak temperature dependence on cooling and heating at low temperatures, and fades away rapidly as the temperature is increased above the temperature where it was acquired. A rather similar behavior of the IRM would be observed also for a system of non-interacting (or interacting) magnetic nanoparticles with a distribution of particle sizes \cite{sasaki}. Also in common is a wait time dependence of the magnitude of the IRM; a larger magnitude is obtained for 3 s wait time than for 3000 s wait time at $T_h$ before the 1 mT field is applied for 3000 s. This difference is explained by the wait time dependence of the response function; the increase of magnetization gets slower with wait time at constant temperature in spin glasses (aging) \cite{aging}. Also in common, is that independently of the wait time, the IRM measured on heating after rapid cooling has larger amplitude than after slower cooling.\\
There are however significant differences between the behavior of the IRM for the different samples, possibly emanating from the spin dimensionality. The wait time at $T_h$ (3 s and 3000 s) influences the magnitude of the IRM in the sequence $Heisenberg$ - $XY$ - $Ising$, going from strongest to weakest influence. This is in accord with the observation that the influence of aging on the response function follows the same sequence \cite{ghost,xy}. The cooling heating curves for the $XY$ and $Ising$ systems coalesce more rapidly below the halt temperature than the corresponding curves for the $Heisenberg$ systems, irrespective of wait time at $T_h$.
The most striking difference is the pronounced maximum in the IRM when returning to $T_h$ in Cu(Mn) and Au(Fe). This feature does not appear in the other two samples, where the IRM smoothly decays with temperature through and above $T_h$. Also, significantly, the heating curve in the protocol (2) where the sample is measured both on cooling and heating approaches closer to the curve measured only on heating (protocol (1)) in the neighborhood of $T_h$. This behavior of the $Heisenberg$ system implies that there is an additional memory component in the spin structure that is imprinted during the halt at $T_h$ \cite{epl} (see also \cite{dcagmn}).
\section{Discussion and Conclusions}
The spin structure of a spin glass is continuously re-organized at low temperature to adjust toward a structure that is most favorable at each temperature. This re-organization occurs always, and irrespective of whether a weak field is applied or a weak field is applied or removed. This implies that the magnetization that is recorded in an experiment may be composed of several independent superposed contributions. The pure IRM (as recorded during procedure (1)) mirrors the excess magnetization that is acquired from an intermittent field application at constant temperature. This magnetization is frozen in on cooling to lower temperatures. During the time it takes to cool to low temperature and again heat the sample back towards $T_h$, the spin structure is spontaneously reorganized on short time-length scales, however, a memory of the structure attained during the stop at $T_h$ remains \cite{ghost,dcagmn}. During this process also parts of the IRM are lost due to relaxation and this is reflected in a lower value of the magnetization when $T_h$ is recovered. For the $XY$ and $Ising$ systems this picture accords with the observations in Fig.~\ref{fig-irm}, and one needs to make an ordinary memory experiment to realize that the spin configuration at $T_h$ does carry a memory of the equilibration that occurred during the original cooling \cite{ghost,xy}.\\
The $Heisenberg$ systems on the other hand, show an additional directional memory of the equilibrated spin structure imprinted at $T_h$. Although a part of the IRM apparently (Fig.~\ref{fig-irm}(a) and (b)) is lost due to the relaxation and restructuring that occurs during cooling and heating (the relaxation is larger for Cu(Mn) than Au(Fe)), a large part of the original IRM magnetization is recovered when $T_h$ is approached as the sample is reheated. This pronounced increase of the IRM is surprising since it is not observed in the other two samples, where the part of the magnetization that has decayed during cooling heating remains lost also at $T_h$. In the $Heisenberg$ systems on the other hand both the equilibrated spin structure and the direction of the excess moment are recovered at $T_h$. This property of the Cu(Mn) and Au(Fe) spin glass possibly reflects a reminiscence of the directional character of the Dzyaloshinsky-Moriya interaction inherent to the $Heisenberg$ systems \cite{fert}.\\
Interestingly, it was shown earlier that the critical exponents associated with the $XY$ spin-glass phase transition had values between those of the $Heisenberg$ and $Ising$ systems \cite{xy}, and that ac or dc memory experiments suggested dynamical properties closer to those of $Ising$ systems. It is interesting to note that here the dynamical behavior of the $XY$ system probed by the IRM experiments is qualitatively similar to that of the $Ising$ system.
\section{Conclusion}
We have investigated the dynamical magnetic properties of spatially three dimensional spin glasses by performing measurements of the isothermal remanent magnetization (IRM). Qualitative differences in the IRM curves of the different systems, which have different spin dimensionality (1: $Ising$, 2: $XY$, or 3: $Heisenberg$), were observed. Our results suggest that in $Heisenberg$ systems, the equilibrated spin structure and the direction of the excess moment of the IRM are kept in memory, reflecting the directional character of the (random and local) Dzyaloshinsky-Moriya interaction inherently present in $Heisenberg$ systems.
\subsection*{Acknowledgements}
We thank the Swedish Research Council (VR) and the G\"oran Gustafsson Foundation for financial support.
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 7,306
|
package edacc.parameterspace.domain;
import java.util.LinkedList;
import java.util.List;
import java.util.Random;
public class RealDomain extends Domain {
protected Double low, high;
public static final String name = "Real";
private RealDomain() {
}
public RealDomain(Double low, Double high) {
this.low = low;
this.high = high;
}
public RealDomain(Integer low, Integer high) {
this.low = Double.valueOf(low);
this.high = Double.valueOf(high);
}
@Override
public boolean contains(Object value) {
if (!(value instanceof Number)) return false;
double d = ((Number)value).doubleValue();
return d >= this.low && d <= this.high;
}
@Override
public Object randomValue(Random rng) {
return rng.nextDouble() * (this.high - this.low) + this.low;
}
@Override
public String toString() {
return "[" + this.low + "," + this.high + "]";
}
public Double getLow() {
return low;
}
public void setLow(Double low) {
this.low = low;
}
public Double getHigh() {
return high;
}
public void setHigh(Double high) {
this.high = high;
}
@Override
public Object mutatedValue(Random rng, Object value) {
if (!contains(value)) return value;
double r = rng.nextGaussian() * ((high - low) * 0.2);
return Math.min(Math.max(this.low, ((Number)value).doubleValue() + r), this.high);
}
@Override
public List<Object> getDiscreteValues() {
List<Object> values = new LinkedList<Object>();
for (double d = low; d <= high; d += (high - low) / 100.0f) {
values.add(d);
}
return values;
}
@Override
public String getName() {
return name;
}
}
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 3,964
|
Мер () — муніципалітет у Франції, у регіоні Центр-Долина Луари, департамент Луар і Шер. Населення — осіб (2011).
Муніципалітет розташований на відстані близько 145 км на південний захід від Парижа, 38 км на південний захід від Орлеана, 18 км на північний схід від Блуа.
Демографія
Розподіл населення за віком та статтю (2006):
Економіка
У 2010 році в муніципалітеті числилось 2598 оподаткованих домогосподарств, у яких проживали 5990,0 особи, медіана доходів виносила євро на одного особоспоживача
Галерея зображень
Посилання
[ Розташування муніципалітету Мер на мапі Франції та сусідні муніципалітети]
Див. також
Список муніципалітетів департаменту Луар і Шер
Примітки
Муніципалітети департаменту Луар і Шер
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 6,920
|
{"url":"https:\/\/www.gradesaver.com\/textbooks\/math\/algebra\/algebra-2-1st-edition\/chapter-10-counting-methods-and-probability-10-1-apply-the-counting-principles-and-permutations-10-1-exercises-skill-practice-page-686\/10","text":"## Algebra 2 (1st Edition)\n\nWe can choose 3 ways for Event A. For each of the ways we can choose 6 ways for event B, 5 for C, and 2 for D. We multiply to get $2*5*6*3=180$","date":"2019-10-19 20:34:55","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\": 0.7604995965957642, \"perplexity\": 365.80022567056204}, \"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-2019-43\/segments\/1570986697760.44\/warc\/CC-MAIN-20191019191828-20191019215328-00335.warc.gz\"}"}
| null | null |
{"url":"https:\/\/gamedev.stackexchange.com\/questions\/156847\/libgdx-animation-for-html5","text":"# Libgdx animation for HTML5\n\nI'm working on my first html5 game with libgdx, it's a platform android game that I made a while back that I now want to host on my website. Everything has been going okay till I got to animating the character. The desktop and android versions of the game run just fine but with the html version the game doesn't run and I get the following error:\n\nGwtApplication: exception: Couldn't find type for class 'com.badlogic.gdx.graphics.g2d.TextureAtlas\\$AtlasRegion'\n\nWhen I comment out the code sections for the animation the game runs fine (with the character not animated). What's causing the error and how can I fix it?\n\n\u2022 Probably reflection related, see stackoverflow.com\/questions\/44719343\/\u2026 \u2013\u00a0Niall Quinlan Mar 26 '18 at 8:51\n\u2022 I had looked at reflection and I've even used it in my current project for scene2D. Turns out I was using it incorrectly in this case though I was not specifying the full name of the class; I had <extend-configuration-property name=\"gdx.reflect.include\" value=\"com.badlogic.gdx.graphics.g2d.TextureAtlas\"\/> while I need <extend-configuration-property name=\"gdx.reflect.include\" value=\"com.badlogic.gdx.graphics.g2d.TextureAtlas.AtlasRegion\"\/>. It's now working. \u2013\u00a0Macharia Mar 27 '18 at 7:29","date":"2019-08-22 07:13:21","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\": 0.20184089243412018, \"perplexity\": 2024.8717383751966}, \"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-2019-35\/segments\/1566027316785.68\/warc\/CC-MAIN-20190822064205-20190822090205-00446.warc.gz\"}"}
| null | null |
// Licensed to the .NET Foundation under one or more agreements.
// The .NET Foundation licenses this file to you under the MIT license.
// See the LICENSE file in the project root for more information.
#nullable enable
using System;
using System.Collections.Generic;
using System.Collections.Immutable;
using System.Diagnostics;
using System.Linq;
using System.Threading;
using System.Threading.Tasks;
using Microsoft.CodeAnalysis.CodeActions;
using Microsoft.CodeAnalysis.CodeGeneration;
using Microsoft.CodeAnalysis.CodeStyle;
using Microsoft.CodeAnalysis.Diagnostics.Analyzers.NamingStyles;
using Microsoft.CodeAnalysis.Editing;
using Microsoft.CodeAnalysis.Operations;
using Microsoft.CodeAnalysis.Options;
using Microsoft.CodeAnalysis.PooledObjects;
using Microsoft.CodeAnalysis.Shared.Extensions;
using Microsoft.CodeAnalysis.Shared.Naming;
using Microsoft.CodeAnalysis.Shared.Utilities;
using Microsoft.CodeAnalysis.Text;
using Roslyn.Utilities;
namespace Microsoft.CodeAnalysis.InitializeParameter
{
internal abstract partial class AbstractInitializeMemberFromParameterCodeRefactoringProvider<
TTypeDeclarationSyntax,
TParameterSyntax,
TStatementSyntax,
TExpressionSyntax> : AbstractInitializeParameterCodeRefactoringProvider<
TTypeDeclarationSyntax,
TParameterSyntax,
TStatementSyntax,
TExpressionSyntax>
where TTypeDeclarationSyntax : SyntaxNode
where TParameterSyntax : SyntaxNode
where TStatementSyntax : SyntaxNode
where TExpressionSyntax : SyntaxNode
{
protected abstract SyntaxNode TryGetLastStatement(IBlockOperation? blockStatementOpt);
protected abstract Accessibility DetermineDefaultFieldAccessibility(INamedTypeSymbol containingType);
protected abstract Accessibility DetermineDefaultPropertyAccessibility();
protected override Task<ImmutableArray<CodeAction>> GetRefactoringsForAllParametersAsync(
Document document, SyntaxNode functionDeclaration, IMethodSymbol method, IBlockOperation? blockStatementOpt,
ImmutableArray<SyntaxNode> listOfParameterNodes, TextSpan parameterSpan, CancellationToken cancellationToken)
{
return SpecializedTasks.EmptyImmutableArray<CodeAction>();
}
protected override async Task<ImmutableArray<CodeAction>> GetRefactoringsForSingleParameterAsync(
Document document, IParameterSymbol parameter, SyntaxNode constructorDeclaration, IMethodSymbol method,
IBlockOperation? blockStatementOpt, CancellationToken cancellationToken)
{
// Only supported for constructor parameters.
if (method.MethodKind != MethodKind.Constructor)
return ImmutableArray<CodeAction>.Empty;
var typeDeclaration = constructorDeclaration.GetAncestor<TTypeDeclarationSyntax>();
if (typeDeclaration == null)
return ImmutableArray<CodeAction>.Empty;
// See if we're already assigning this parameter to a field/property in this type. If so, there's nothing
// more for us to do.
var assignmentStatement = TryFindFieldOrPropertyAssignmentStatement(parameter, blockStatementOpt);
if (assignmentStatement != null)
return ImmutableArray<CodeAction>.Empty;
// Haven't initialized any fields/properties with this parameter. Offer to assign
// to an existing matching field/prop if we can find one, or add a new field/prop
// if we can't.
var rules = await document.GetNamingRulesAsync(cancellationToken).ConfigureAwait(false);
var parameterNameParts = IdentifierNameParts.CreateIdentifierNameParts(parameter, rules);
if (parameterNameParts.BaseName == "")
return ImmutableArray<CodeAction>.Empty;
var fieldOrProperty = await TryFindMatchingUninitializedFieldOrPropertySymbolAsync(
document, parameter, blockStatementOpt, rules, parameterNameParts.BaseNameParts, cancellationToken).ConfigureAwait(false);
if (fieldOrProperty != null)
{
return HandleExistingFieldOrProperty(
document, parameter, constructorDeclaration,
blockStatementOpt, fieldOrProperty);
}
return await HandleNoExistingFieldOrPropertyAsync(
document, parameter, constructorDeclaration,
method, blockStatementOpt, rules, cancellationToken).ConfigureAwait(false);
}
private async Task<ImmutableArray<CodeAction>> HandleNoExistingFieldOrPropertyAsync(
Document document,
IParameterSymbol parameter,
SyntaxNode constructorDeclaration,
IMethodSymbol method,
IBlockOperation? blockStatementOpt,
ImmutableArray<NamingRule> rules,
CancellationToken cancellationToken)
{
// Didn't find a field/prop that this parameter could be assigned to.
// Offer to create new one and assign to that.
using var _ = ArrayBuilder<CodeAction>.GetInstance(out var allActions);
var options = await document.GetOptionsAsync(cancellationToken).ConfigureAwait(false);
var (fieldAction, propertyAction) = AddSpecificParameterInitializationActions(
document, parameter, constructorDeclaration, blockStatementOpt, rules, options);
// Check if the surrounding parameters are assigned to another field in this class. If so, offer to
// make this parameter into a field as well. Otherwise, default to generating a property
var siblingFieldOrProperty = TryFindSiblingFieldOrProperty(parameter, blockStatementOpt);
if (siblingFieldOrProperty is IFieldSymbol)
{
allActions.Add(fieldAction);
allActions.Add(propertyAction);
}
else
{
allActions.Add(propertyAction);
allActions.Add(fieldAction);
}
var (allFieldsAction, allPropertiesAction) = AddAllParameterInitializationActions(
document, constructorDeclaration, method, blockStatementOpt, rules, options);
if (allFieldsAction != null && allPropertiesAction != null)
{
if (siblingFieldOrProperty is IFieldSymbol)
{
allActions.Add(allFieldsAction);
allActions.Add(allPropertiesAction);
}
else
{
allActions.Add(allPropertiesAction);
allActions.Add(allFieldsAction);
}
}
return allActions.ToImmutable();
}
private (CodeAction? fieldAction, CodeAction? propertyAction) AddAllParameterInitializationActions(
Document document,
SyntaxNode constructorDeclaration,
IMethodSymbol method,
IBlockOperation? blockStatementOpt,
ImmutableArray<NamingRule> rules,
DocumentOptionSet options)
{
if (blockStatementOpt == null)
return default;
var parameters = GetParametersWithoutAssociatedMembers(blockStatementOpt, rules, method);
if (parameters.Length < 2)
return default;
var fields = parameters.SelectAsArray(p => (ISymbol)CreateField(p, options, rules));
var properties = parameters.SelectAsArray(p => (ISymbol)CreateProperty(p, options, rules));
var allFieldsAction = new MyCodeAction(
FeaturesResources.Create_and_assign_remaining_as_fields,
c => AddAllSymbolInitializationsAsync(
document, constructorDeclaration, blockStatementOpt, parameters, fields, c));
var allPropertiesAction = new MyCodeAction(
FeaturesResources.Create_and_assign_remaining_as_properties,
c => AddAllSymbolInitializationsAsync(
document, constructorDeclaration, blockStatementOpt, parameters, properties, c));
return (allFieldsAction, allPropertiesAction);
}
private (CodeAction fieldAction, CodeAction propertyAction) AddSpecificParameterInitializationActions(
Document document,
IParameterSymbol parameter,
SyntaxNode constructorDeclaration,
IBlockOperation? blockStatementOpt,
ImmutableArray<NamingRule> rules,
DocumentOptionSet options)
{
var field = CreateField(parameter, options, rules);
var property = CreateProperty(parameter, options, rules);
var fieldAction = new MyCodeAction(
string.Format(FeaturesResources.Create_and_assign_field_0, field.Name),
c => AddSingleSymbolInitializationAsync(document, constructorDeclaration, blockStatementOpt, parameter, field, c));
var propertyAction = new MyCodeAction(
string.Format(FeaturesResources.Create_and_assign_property_0, property.Name),
c => AddSingleSymbolInitializationAsync(document, constructorDeclaration, blockStatementOpt, parameter, property, c));
return (fieldAction, propertyAction);
}
private static ImmutableArray<IParameterSymbol> GetParametersWithoutAssociatedMembers(
IBlockOperation? blockStatementOpt,
ImmutableArray<NamingRule> rules,
IMethodSymbol method)
{
using var _ = ArrayBuilder<IParameterSymbol>.GetInstance(out var result);
foreach (var parameter in method.Parameters)
{
var parameterNameParts = IdentifierNameParts.CreateIdentifierNameParts(parameter, rules);
if (parameterNameParts.BaseName == "")
continue;
var assignmentOp = TryFindFieldOrPropertyAssignmentStatement(parameter, blockStatementOpt);
if (assignmentOp != null)
continue;
result.Add(parameter);
}
return result.ToImmutable();
}
private ImmutableArray<CodeAction> HandleExistingFieldOrProperty(Document document, IParameterSymbol parameter, SyntaxNode functionDeclaration, IBlockOperation? blockStatementOpt, ISymbol fieldOrProperty)
{
// Found a field/property that this parameter should be assigned to.
// Just offer the simple assignment to it.
var resource = fieldOrProperty.Kind == SymbolKind.Field
? FeaturesResources.Initialize_field_0
: FeaturesResources.Initialize_property_0;
var title = string.Format(resource, fieldOrProperty.Name);
return ImmutableArray.Create<CodeAction>(new MyCodeAction(
title,
c => AddSingleSymbolInitializationAsync(
document, functionDeclaration, blockStatementOpt, parameter, fieldOrProperty, c)));
}
private static ISymbol? TryFindSiblingFieldOrProperty(IParameterSymbol parameter, IBlockOperation? blockStatementOpt)
{
foreach (var (siblingParam, _) in GetSiblingParameters(parameter))
{
TryFindFieldOrPropertyAssignmentStatement(siblingParam, blockStatementOpt, out var sibling);
if (sibling != null)
return sibling;
}
return null;
}
private IFieldSymbol CreateField(
IParameterSymbol parameter,
DocumentOptionSet options,
ImmutableArray<NamingRule> rules)
{
var requireAccessibilityModifiers = options.GetOption(CodeStyleOptions2.RequireAccessibilityModifiers);
var parameterNameParts = IdentifierNameParts.CreateIdentifierNameParts(parameter, rules).BaseNameParts;
foreach (var rule in rules)
{
if (rule.SymbolSpecification.AppliesTo(SymbolKind.Field, Accessibility.Private))
{
var uniqueName = GenerateUniqueName(parameter, parameterNameParts, rule);
var accessibilityLevel = Accessibility.Private;
if (requireAccessibilityModifiers.Value == AccessibilityModifiersRequired.Never || requireAccessibilityModifiers.Value == AccessibilityModifiersRequired.OmitIfDefault)
{
var defaultAccessibility = DetermineDefaultFieldAccessibility(parameter.ContainingType);
if (defaultAccessibility == Accessibility.Private)
{
accessibilityLevel = Accessibility.NotApplicable;
}
}
return CodeGenerationSymbolFactory.CreateFieldSymbol(
default,
accessibilityLevel,
DeclarationModifiers.ReadOnly,
parameter.Type, uniqueName);
}
}
// We place a special rule in s_builtInRules that matches all fields. So we should
// always find a matching rule.
throw ExceptionUtilities.Unreachable;
}
private static string GenerateUniqueName(IParameterSymbol parameter, ImmutableArray<string> parameterNameParts, NamingRule rule)
{
// Determine an appropriate name to call the new field.
var containingType = parameter.ContainingType;
var baseName = rule.NamingStyle.CreateName(parameterNameParts);
// Ensure that the name is unique in the containing type so we
// don't stomp on an existing member.
var uniqueName = NameGenerator.GenerateUniqueName(
baseName, n => containingType.GetMembers(n).IsEmpty);
return uniqueName;
}
private IPropertySymbol CreateProperty(
IParameterSymbol parameter,
DocumentOptionSet options,
ImmutableArray<NamingRule> rules)
{
var requireAccessibilityModifiers = options.GetOption(CodeStyleOptions2.RequireAccessibilityModifiers);
var parameterNameParts = IdentifierNameParts.CreateIdentifierNameParts(parameter, rules).BaseNameParts;
foreach (var rule in rules)
{
if (rule.SymbolSpecification.AppliesTo(SymbolKind.Property, Accessibility.Public))
{
var uniqueName = GenerateUniqueName(parameter, parameterNameParts, rule);
var accessibilityLevel = Accessibility.Public;
if (requireAccessibilityModifiers.Value == AccessibilityModifiersRequired.Never || requireAccessibilityModifiers.Value == AccessibilityModifiersRequired.OmitIfDefault)
{
var defaultAccessibility = DetermineDefaultPropertyAccessibility();
if (defaultAccessibility == Accessibility.Public)
{
accessibilityLevel = Accessibility.NotApplicable;
}
}
var getMethod = CodeGenerationSymbolFactory.CreateAccessorSymbol(
default,
Accessibility.Public,
default);
return CodeGenerationSymbolFactory.CreatePropertySymbol(
default,
accessibilityLevel,
new DeclarationModifiers(),
parameter.Type,
RefKind.None,
explicitInterfaceImplementations: default,
name: uniqueName,
parameters: default,
getMethod: getMethod,
setMethod: null);
}
}
// We place a special rule in s_builtInRules that matches all properties. So we should
// always find a matching rule.
throw ExceptionUtilities.Unreachable;
}
private async Task<Document> AddAllSymbolInitializationsAsync(
Document document,
SyntaxNode constructorDeclaration,
IBlockOperation? blockStatementOpt,
ImmutableArray<IParameterSymbol> parameters,
ImmutableArray<ISymbol> fieldsOrProperties,
CancellationToken cancellationToken)
{
Debug.Assert(parameters.Length >= 2);
Debug.Assert(fieldsOrProperties.Length > 0);
Debug.Assert(parameters.Length == fieldsOrProperties.Length);
// Process each param+field/prop in order. Apply the pair to the document getting the updated document.
// Then find all the current data in that updated document and move onto the next pair.
var root = await document.GetRequiredSyntaxRootAsync(cancellationToken).ConfigureAwait(false);
var nodesToTrack = new List<SyntaxNode> { constructorDeclaration };
if (blockStatementOpt != null)
nodesToTrack.Add(blockStatementOpt.Syntax);
var trackedRoot = root.TrackNodes(nodesToTrack);
var currentDocument = document.WithSyntaxRoot(trackedRoot);
for (var i = 0; i < parameters.Length; i++)
{
var parameter = parameters[i];
var fieldOrProperty = fieldsOrProperties[i];
var currentSemanticModel = await currentDocument.GetRequiredSemanticModelAsync(cancellationToken).ConfigureAwait(false);
var currentCompilation = currentSemanticModel.Compilation;
var currentRoot = await currentDocument.GetRequiredSyntaxRootAsync(cancellationToken).ConfigureAwait(false);
var currentConstructorDeclaration = currentRoot.GetCurrentNode(constructorDeclaration);
if (currentConstructorDeclaration == null)
continue;
IBlockOperation? currentBlockStatementOpt = null;
if (blockStatementOpt != null)
{
currentBlockStatementOpt = (IBlockOperation?)currentSemanticModel.GetOperation(currentRoot.GetCurrentNode(blockStatementOpt.Syntax), cancellationToken);
if (currentBlockStatementOpt == null)
continue;
}
var currentParameter = (IParameterSymbol?)parameter.GetSymbolKey(cancellationToken).Resolve(currentCompilation, cancellationToken: cancellationToken).GetAnySymbol();
if (currentParameter == null)
continue;
// fieldOrProperty is a new member. So we don't have to track it to this edit we're making.
currentDocument = await AddSingleSymbolInitializationAsync(
currentDocument,
currentConstructorDeclaration,
currentBlockStatementOpt,
currentParameter,
fieldOrProperty,
cancellationToken).ConfigureAwait(false);
}
return currentDocument;
}
private async Task<Document> AddSingleSymbolInitializationAsync(
Document document,
SyntaxNode constructorDeclaration,
IBlockOperation? blockStatementOpt,
IParameterSymbol parameter,
ISymbol fieldOrProperty,
CancellationToken cancellationToken)
{
var workspace = document.Project.Solution.Workspace;
var root = await document.GetRequiredSyntaxRootAsync(cancellationToken).ConfigureAwait(false);
var editor = new SyntaxEditor(root, workspace);
var generator = editor.Generator;
var options = await document.GetOptionsAsync(cancellationToken).ConfigureAwait(false);
if (fieldOrProperty.ContainingType == null)
{
// We're generating a new field/property. Place into the containing type,
// ideally before/after a relevant existing member.
// First, look for the right containing type (As a type may be partial).
// We want the type-block that this constructor is contained within.
var typeDeclaration = constructorDeclaration.GetAncestor<TTypeDeclarationSyntax>()!;
// Now add the field/property to this type. Use the 'ReplaceNode+callback' form
// so that nodes will be appropriate tracked and so we can then update the constructor
// below even after we've replaced the whole type with a new type.
//
// Note: We'll pass the appropriate options so that the new field/property
// is appropriate placed before/after an existing field/property. We'll try
// to preserve the same order for fields/properties that we have for the constructor
// parameters.
editor.ReplaceNode(
typeDeclaration,
(currentTypeDecl, _) =>
{
if (fieldOrProperty is IPropertySymbol property)
{
return CodeGenerator.AddPropertyDeclaration(
currentTypeDecl, property, workspace,
GetAddOptions<IPropertySymbol>(parameter, blockStatementOpt, typeDeclaration, options, cancellationToken));
}
else if (fieldOrProperty is IFieldSymbol field)
{
return CodeGenerator.AddFieldDeclaration(
currentTypeDecl, field, workspace,
GetAddOptions<IFieldSymbol>(parameter, blockStatementOpt, typeDeclaration, options, cancellationToken));
}
else
{
throw ExceptionUtilities.Unreachable;
}
});
}
// Now that we've added any potential members, create an assignment between it
// and the parameter.
var initializationStatement = (TStatementSyntax)generator.ExpressionStatement(
generator.AssignmentStatement(
generator.MemberAccessExpression(
generator.ThisExpression(),
generator.IdentifierName(fieldOrProperty.Name)),
generator.IdentifierName(parameter.Name)));
// Attempt to place the initialization in a good location in the constructor
// We'll want to keep initialization statements in the same order as we see
// parameters for the constructor.
var statementToAddAfterOpt = TryGetStatementToAddInitializationAfter(parameter, blockStatementOpt);
InsertStatement(editor, constructorDeclaration, returnsVoid: true, statementToAddAfterOpt, initializationStatement);
return document.WithSyntaxRoot(editor.GetChangedRoot());
}
private static CodeGenerationOptions GetAddOptions<TSymbol>(
IParameterSymbol parameter, IBlockOperation? blockStatementOpt,
SyntaxNode typeDeclaration, OptionSet options, CancellationToken cancellationToken)
where TSymbol : ISymbol
{
foreach (var (sibling, before) in GetSiblingParameters(parameter))
{
var statement = TryFindFieldOrPropertyAssignmentStatement(
sibling, blockStatementOpt, out var fieldOrProperty);
if (statement != null &&
fieldOrProperty is TSymbol symbol)
{
var symbolSyntax = symbol.DeclaringSyntaxReferences[0].GetSyntax(cancellationToken);
if (symbolSyntax.Ancestors().Contains(typeDeclaration))
{
if (before)
{
// Found an existing field/property that corresponds to a preceding parameter.
// Place ourselves directly after it.
return new CodeGenerationOptions(afterThisLocation: symbolSyntax.GetLocation(), options: options);
}
else
{
// Found an existing field/property that corresponds to a following parameter.
// Place ourselves directly before it.
return new CodeGenerationOptions(beforeThisLocation: symbolSyntax.GetLocation(), options: options);
}
}
}
}
return new CodeGenerationOptions(options: options);
}
private static ImmutableArray<(IParameterSymbol parameter, bool before)> GetSiblingParameters(IParameterSymbol parameter)
{
using var _ = ArrayBuilder<(IParameterSymbol, bool before)>.GetInstance(out var siblings);
if (parameter.ContainingSymbol is IMethodSymbol method)
{
var parameterIndex = method.Parameters.IndexOf(parameter);
// look for an existing assignment for a parameter that comes before us.
// If we find one, we'll add ourselves after that parameter check.
for (var i = parameterIndex - 1; i >= 0; i--)
siblings.Add((method.Parameters[i], before: true));
// look for an existing check for a parameter that comes before us.
// If we find one, we'll add ourselves after that parameter check.
for (var i = parameterIndex + 1; i < method.Parameters.Length; i++)
siblings.Add((method.Parameters[i], before: false));
}
return siblings.ToImmutable();
}
private SyntaxNode? TryGetStatementToAddInitializationAfter(
IParameterSymbol parameter, IBlockOperation? blockStatementOpt)
{
// look for an existing assignment for a parameter that comes before/after us.
// If we find one, we'll add ourselves before/after that parameter check.
foreach (var (sibling, before) in GetSiblingParameters(parameter))
{
var statement = TryFindFieldOrPropertyAssignmentStatement(sibling, blockStatementOpt);
if (statement != null)
{
if (before)
{
return statement.Syntax;
}
else
{
var statementIndex = blockStatementOpt!.Operations.IndexOf(statement);
return statementIndex > 0 ? blockStatementOpt.Operations[statementIndex - 1].Syntax : null;
}
}
}
// We couldn't find a reasonable location for the new initialization statement.
// Just place ourselves after the last statement in the constructor.
return TryGetLastStatement(blockStatementOpt);
}
private static IOperation? TryFindFieldOrPropertyAssignmentStatement(IParameterSymbol parameter, IBlockOperation? blockStatementOpt)
=> TryFindFieldOrPropertyAssignmentStatement(parameter, blockStatementOpt, out _);
private static IOperation? TryFindFieldOrPropertyAssignmentStatement(
IParameterSymbol parameter, IBlockOperation? blockStatementOpt, out ISymbol? fieldOrProperty)
{
if (blockStatementOpt != null)
{
var containingType = parameter.ContainingType;
foreach (var statement in blockStatementOpt.Operations)
{
// look for something of the form: "this.s = s" or "this.s = s ?? ..."
if (IsFieldOrPropertyAssignment(statement, containingType, out var assignmentExpression, out fieldOrProperty) &&
IsParameterReferenceOrCoalesceOfParameterReference(assignmentExpression, parameter))
{
return statement;
}
}
}
fieldOrProperty = null;
return null;
}
private static bool IsParameterReferenceOrCoalesceOfParameterReference(
IAssignmentOperation assignmentExpression, IParameterSymbol parameter)
{
if (IsParameterReference(assignmentExpression.Value, parameter))
{
// We already have a member initialized with this parameter like:
// this.field = parameter
return true;
}
if (UnwrapImplicitConversion(assignmentExpression.Value) is ICoalesceOperation coalesceExpression &&
IsParameterReference(coalesceExpression.Value, parameter))
{
// We already have a member initialized with this parameter like:
// this.field = parameter ?? ...
return true;
}
return false;
}
private async Task<ISymbol?> TryFindMatchingUninitializedFieldOrPropertySymbolAsync(
Document document, IParameterSymbol parameter, IBlockOperation? blockStatementOpt, ImmutableArray<NamingRule> rules, ImmutableArray<string> parameterWords, CancellationToken cancellationToken)
{
// Look for a field/property that really looks like it corresponds to this parameter.
// Use a variety of heuristics around the name/type to see if this is a match.
var containingType = parameter.ContainingType;
var compilation = await document.Project.GetRequiredCompilationAsync(cancellationToken).ConfigureAwait(false);
// Walk through the naming rules against this parameter's name to see what
// name the user would like for it as a member in this type. Note that we
// have some fallback rules that use the standard conventions around
// properties /fields so that can still find things even if the user has no
// naming preferences set.
foreach (var rule in rules)
{
var memberName = rule.NamingStyle.CreateName(parameterWords);
foreach (var memberWithName in containingType.GetMembers(memberName))
{
// We found members in our type with that name. If it's a writable
// field that we could assign this parameter to, and it's not already
// been assigned to, then this field is a good candidate for us to
// hook up to.
if (memberWithName is IFieldSymbol field &&
!field.IsConst &&
IsImplicitConversion(compilation, source: parameter.Type, destination: field.Type) &&
!ContainsMemberAssignment(blockStatementOpt, field))
{
return field;
}
// If it's a writable property that we could assign this parameter to, and it's
// not already been assigned to, then this property is a good candidate for us to
// hook up to.
if (memberWithName is IPropertySymbol property &&
property.IsWritableInConstructor() &&
IsImplicitConversion(compilation, source: parameter.Type, destination: property.Type) &&
!ContainsMemberAssignment(blockStatementOpt, property))
{
return property;
}
}
}
// Couldn't find any existing member. Just return nothing so we can offer to
// create a member for them.
return null;
}
private static bool ContainsMemberAssignment(
IBlockOperation? blockStatementOpt, ISymbol member)
{
if (blockStatementOpt != null)
{
foreach (var statement in blockStatementOpt.Operations)
{
if (IsFieldOrPropertyAssignment(statement, member.ContainingType, out var assignmentExpression) &&
UnwrapImplicitConversion(assignmentExpression.Target) is IMemberReferenceOperation memberReference &&
member.Equals(memberReference.Member))
{
return true;
}
}
}
return false;
}
}
}
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 2,119
|
<?xml version="1.0" encoding="utf-8" standalone="no"?>
<resources xmlns:xliff="urn:oasis:names:tc:xliff:document:1.2">
<string name="talkback_on">TalkBack imewashwa</string>
<string name="talkback_suspended">TalkBack imesimamishwa</string>
<string name="talkback_resumed">TalkBack inaendelea</string>
<string name="talkback_disabled">Kipengele cha TalkBack kimezimwa</string>
<string name="spokenfeedback_on">Kipengele cha Maelezo Yanayotamkwa kimewashwa</string>
<string name="spokenfeedback_suspended">Kipengele cha Maelezo Yanayotamkwa kimesimamishwa</string>
<string name="spokenfeedback_resumed">Kipengele cha Maelezo Yanayotamkwa kimerejeshwa</string>
<string name="spokenfeedback_disabled">Kipengele cha Maelezo Yanayotamka kimezimwa</string>
<string name="notification_type_beginning_of_field">Mwanzo wa sehemu</string>
<string name="notification_type_end_of_field">Mwisho wa sehemu</string>
<string name="notification_type_selection_mode_on">Hali ya kuchagua imewashwa</string>
<string name="notification_type_selection_mode_off">Hali ya kuchagua imezimwa</string>
<string name="notification_type_selection_cleared">Uteuzi umefutwa</string>
<string name="notification_category_call">Simu</string>
<string name="notification_category_msg">Ujumbe</string>
<string name="notification_category_email">Barua pepe</string>
<string name="notification_category_event">Tukio</string>
<string name="notification_category_promo">Tangazo</string>
<string name="notification_category_alarm">Kengele</string>
<string name="notification_category_progress">Maendeleo</string>
<string name="notification_category_social">Mitandao Jamii</string>
<string name="notification_category_err">Hitilafu</string>
<string name="notification_category_transport">Kudhibiti maudhui</string>
<string name="notification_category_sys">Hali ya mfumo</string>
<string name="notification_category_service">Huduma</string>
<string name="template_drawer_opened">Menyu ya %1$s</string>
<string name="template_text_replaced">%1$s yamechukua nafasi ya %2$s</string>
<string name="template_text_removed">%1$s yamefutwa</string>
<string name="template_text_cut">%1$s, yamekatwa</string>
<string name="template_text_pasted">%1$s, yamebandikwa</string>
<string name="template_text_selected">%1$s, yamechaguliwa</string>
<string name="template_announce_selected_text">Maandishi yaliyochaguliwa: %1$s</string>
<string name="template_text_unselected">%1$s, yameondolewa</string>
<string name="template_text_error">Ingizo lisilosahihi %1$s</string>
<string name="template_replaced_characters">Herufi %2$s zilichukua nafasi ya herufi %1$s</string>
<string name="template_labeled_item">%1$s cha %2$s</string>
<string name="template_percent">Asilimia %1$s.</string>
<plurals name="template_containers">
<item quantity="one">%1$s inaonyesha kipengee %2$d.</item>
<item quantity="other">%1$s inaonyesha vipengee %2$d.</item>
</plurals>
<string name="template_node_error_with_error_message">Error: %1$s</string>
<string name="value_gridview">Gridi</string>
<string name="value_listview">Orodha</string>
<string name="value_tabwidget">Upao wa kichupo</string>
<string name="value_spinner">Orodha kunjuzi</string>
<string name="value_checkbox">Kisanduku cha kuteua</string>
<string name="value_radio_button">Kitufe cha mviringo</string>
<string name="value_button">Kitufe</string>
<string name="value_seek_bar">Kitelezi</string>
<string name="value_switch">Swichi</string>
<string name="value_edit_box">Kisanduku cha kubadilisha</string>
<string name="value_edit_box_editing">Inabadilisha</string>
<string name="value_image">Picha</string>
<string name="value_unlabelled">Kisicho na lebo</string>
<string name="value_pager">Mwonekano wa kurasa nyingi</string>
<string name="value_pager_page">Ukurasa</string>
<string name="value_progress_bar">Upau wa kuonyesha hatua</string>
<string name="value_webview">Mwonekano wa Wavuti</string>
<string name="value_password">nenosiri</string>
<string name="value_sliding_drawer_opened">Droo ya kutelezesha imefunguliwa</string>
<string name="value_options_menu_open">Chaguo.</string>
<string name="value_text_cleared">Maandishi yaliyofutwa</string>
<string name="value_text_max_length">Urefu hauwezi kuzidi hapa</string>
<string name="value_selected">Iliyochaguliwa</string>
<string name="value_checked">imeteuliwa</string>
<string name="value_not_checked">haijachaguliwa</string>
<string name="value_on">Kimewashwa</string>
<string name="value_off">Kimezimwa</string>
<string name="value_disabled">imezimwa</string>
<string name="value_expanded">imepanuliwa</string>
<string name="value_collapsed">imekunjwa</string>
<string name="value_double_tap">Gusa mara mbili</string>
<string name="value_single_tap">Gusa</string>
<string name="value_switch_press">Bonyeza \'Chagua Swichi\'</string>
<string name="value_press_select">Bonyeza kitufe cha chagua</string>
<string name="template_hint_clickable">%1$s ili uwashe.</string>
<string name="template_hint_clickable_keyboard">Bonyeza %1$s ili uwashe.</string>
<string name="template_hint_long_clickable">%1$s na ushikile ili kubonyeza kwa muda mrefu.</string>
<string name="template_hint_long_clickable_keyboard">Bofya %1$s ili ubonyeze kwa muda mrefu.</string>
<string name="template_hint_checkable">%1$s ili ugeuze.</string>
<string name="template_hint_checkable_keyboard">Bonyeza %1$s ili ugeuze.</string>
<string name="template_hint_edit_text">%1$s ili ubadilishe matini.</string>
<string name="template_hint_edit_text_keyboard">Bonyeza %1$s ili ubadilishe matini.</string>
<string name="template_hint_spinner">%1$s ili ubadilishe.</string>
<string name="template_hint_spinner_keyboard">Bonyeza %1$s ili ubadilishe.</string>
<string name="template_hint_pager">Telezesha kwa vidole viwili ili ubadilishe kurasa.</string>
<string name="template_hint_pager_single_page">Hakuna kurasa zaidi.</string>
<string name="template_hint_seek_control">Tumia vitufe vya sauti ili kurekebisha</string>
<string name="template_hint_seek_control_tv">%1$s ili ubadilishe.</string>
<string name="template_hint_custom_actions">Vitendo:</string>
<string name="template_custom_hint_for_actions">%1$s ili %2$s.</string>
<string name="template_custom_hint_for_actions_keyboard">Bonyeza %1$s ili %2$s.</string>
<string name="template_custom_hint_for_long_clickable_actions">%1$s na ushikile ili %2$s.</string>
<string name="template_sa_custom_hint_scrollable">%1$s ili usogeze.</string>
<string name="template_sa_custom_hint_for_multiple_actions">%1$s ili uone vitendo vinavyopatikana.</string>
<string name="template_sa_hint_seek_control">Bonyeza \'Chagua Swichi\' ili ufanye marekebisho</string>
<string name="value_caps_lock_on">Caps Lock imewashwa.</string>
<string name="value_caps_lock_off">Caps lock imezimwa.</string>
<string name="value_num_lock_on">Num Lock imewashwa.</string>
<string name="value_num_lock_off">Num Lock imezimwa.</string>
<string name="value_scroll_lock_on">Scroll Lock imewashwa.</string>
<string name="value_scroll_lock_off">Scroll Lock imezimwa.</string>
<string name="orientation_portrait">Wima</string>
<string name="orientation_landscape">Mlalo</string>
<string name="title_action_dismiss">Ondoa</string>
<string name="title_action_expand">Panua</string>
<string name="title_action_collapse">Kunja</string>
<string name="row_index_template">Safu mlalo %1$d</string>
<string name="column_index_template">Safu wima %1$d</string>
<string name="list_index_template">%1$d kati ya %2$d</string>
<string name="heading_template">Kichwa</string>
<string name="row_heading_template">Kichwa cha safu mlalo</string>
<string name="column_heading_template">Kichwa cha safu wima</string>
<string name="in_collection_role_description_with_name">Uko katika %1$s %2$s</string>
<string name="in_collection_role_description">Uko katika %1$s</string>
<string name="in_list_with_name">Umeorodheshwa katika %1$s</string>
<string name="in_list">Umeorodheshwa</string>
<string name="out_of_role_description_with_name">Hauko katika %1$s %2$s</string>
<string name="out_of_role_description">Hauko katika %1$s</string>
<string name="out_of_list_with_name">Haujaorodheshwa katika %1$s</string>
<string name="out_of_list">Haujaorodheshwa</string>
<string name="in_grid_with_name">Uko katika gridi ya %1$s</string>
<string name="in_grid">Uko katika gridi</string>
<string name="out_of_grid_with_name">Hauko katika gridi ya %1$s</string>
<string name="out_of_grid">Hauko katika gridi</string>
<string name="in_pager_with_name">Katika peja %1$s</string>
<string name="in_pager">Katika peja</string>
<string name="out_of_pager_with_name">Nje ya peja %1$s</string>
<string name="out_of_pager">Nje ya peja</string>
<string name="in_vertical_pager_with_name">Katika peja wima %1$s</string>
<string name="in_vertical_pager">Katika peja wima</string>
<string name="out_of_vertical_pager_with_name">Nje ya peja wima %1$s</string>
<string name="out_of_vertical_pager">Nje ya peja wima</string>
<string name="in_horizontal_pager_with_name">Katika peja mlalo %1$s</string>
<string name="in_horizontal_pager">Katika peja mlalo</string>
<string name="out_of_horizontal_pager">Nje ya peja mlalo</string>
<string name="in_grid_pager_with_name">Katika peja ya gridi %1$s</string>
<string name="in_grid_pager">Katika peja ya gridi</string>
<string name="out_of_grid_pager_with_name">Nje ya peja ya gridi %1$s</string>
<string name="out_of_grid_pager">Nje ya peja ya gridi</string>
<string name="template_collection_level">Kiwango cha %1$d</string>
<plurals name="template_list_total_count">
<item quantity="one">Kipengee %1$d</item>
<item quantity="other">Vipengee %1$d</item>
</plurals>
<plurals name="template_list_row_count">
<item quantity="one">Safu mlalo %1$d</item>
<item quantity="other">Safu mlalo %1$d</item>
</plurals>
<plurals name="template_list_column_count">
<item quantity="one">Safu wima %1$d</item>
<item quantity="other">Safu wima %1$d</item>
</plurals>
<plurals name="template_password_character_count">
<item quantity="one">Herufi %1$d</item>
<item quantity="other">Herufi %1$d</item>
</plurals>
<string name="template_screen_magnification_on">Imewasha kipengele cha kukuza skrini kwa %1$d%%</string>
<string name="template_screen_magnification_scale_changed">Imekuza skrini kwa %1$d%%</string>
<string name="screen_magnification_off">Imezima kipengele cha kukuza skrini</string>
<string name="template_speech_rate_change">Kasi ya matamshi %1$d%%</string>
<string name="template_volume_change_increase">Ongeza</string>
<string name="template_volume_change_decrease">Punguza</string>
<string name="template_volume_change_maximum">Kiwango cha Juu</string>
<string name="template_volume_change_minimum">Kiwango cha chini zaidi</string>
<string name="template_seekbar_range">%1$d%%</string>
<string name="summaryOff_pref_speak_passwords_without_headphones">Manenosiri hutamkwa wakati unatumia tu vipokea sauti vinavyobanwa kichwani</string>
<string name="template_hint_menu_type_high_verbosity">%1$s inapatikana, tumia %2$s ili uangalie.</string>
<string name="hint_suggestion">Mapendekezo ya tahajia yanapatikana.</string>
<string name="in_window_with_name">Dirisha la %1$s</string>
<string name="detected_icon_description">%1$s, imetambua aikoni</string>
<string name="ocr_text_description">%1$s, tuma SMS</string>
</resources>
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 36
|
\section{Introduction and results}
Since the late nineties, nonlocal integral operators like
\begin{equation}
\label{slaplacian}
(-\Delta)^s u (x) = c_s\lim_{\varepsilon\searrow 0}\int_{B^c_\varepsilon(x)}\frac{u(x)-u(y)}{|x-y|^{3+2s}} dy=\mathcal{F}^{-1}(|\xi|^{2s}\mathcal{F}(u)(\xi))(x), \qquad u\in C^\infty_c(\mathbb{R}^3),
\end{equation}
where $s\in(0,1)$ and
$$
c_{s} = s 2^{2s} \frac{\Gamma\big(\frac{3+2s}{2}\big)}{\pi^{3/2}\Gamma(1-s)},
$$
being $\Gamma$ the Gamma function, have been widely used
in the theory of L\'evy processes. Indeed, in view of
the L\'evy-Khintchine formula, the
generator ${\mathscr H}$ of the semigroup on $C^\infty_c(\mathbb{R}^3)$ associated to a general L\'evy process is given by
\begin{equation}
\label{LK}
{\mathscr H}u(x)= -a_{ij}\partial_{x_i x_j}^2 u(x)-b_i\partial_{x_i}u(x)-\lim_{\varepsilon\searrow 0}\int_{B^c_\varepsilon(0)}\Big(u(x+y)-u(x)-1_{\{|y|<1\}}(y)y\cdot\nabla u(x)\Big)d\mu,
\end{equation}
with summation on repeated indexes and where $\mu$ is a L\'evy nonnegative measure, namely
$$
\int_{\mathbb{R}^3}\frac{|y|^2}{1+|y|^2}d\mu<\infty.
$$
The last contribution in \eqref{LK}
represents the purely jump part of the {\em L\'evy process}, while the first two terms represent a {\em Brownian motion} with drift.
It is now well established that L\'evy processes with jumps are more appropriate for some mathematical models in finance.
Among L\'evy processes, the only stochastically stable ones having jump part are those corresponding to radial measures as
$$
d\mu=\frac{c_s}{|y|^{3+2s}}dy,
$$
hence the importance of the definition \eqref{slaplacian}. Moreover, the fractional Laplacian \eqref{slaplacian} allows to develop a generalization of quantum mechanics and also to describe the motion of a chain or array of particles that are connected by elastic springs and unusual diffusion processes
in turbulent fluid motions and material transports in fractured media (for more details see e.g. \cite{A,CT,Lask3,MK} and the references therein).
Due to the results of Bourgain-Br\'ezis-Mironescu \cite{bourg,bourg2}, up to correcting the operator \eqref{slaplacian} with the
factor $(1-s)$ it follows that $(-\Delta)^s u$ converges to $-\Delta u$ in the limit $s\nearrow 1$.\ Thus, up to normalization, we may
think the nonlocal case as an approximation of the local case.\\
A {\em pseudorelativistic extension} of the Laplacian is the well known pseudodifferential operator $\sqrt{-\Delta + m^2} - m$ where $m$ is a nonnegative number.
This operator appears in the study of free relativistic particles of mass $m$ and $\sqrt{-\Delta + m^2}$ is defined by
$\mathcal{F}^{-1}(\sqrt{|\xi|^2+ m^2}\mathcal{F}(u)(\xi))$ (see \cite{LS} for more details).
We observe that for $m=0$ we have the operator in \eqref{slaplacian} with $s=1/2$.\\
An important role in the study of particles which interact, e.g. using the Weyl covariant derivative, with a magnetic field $B=\nabla\times A$, $A:\mathbb{R}^3\to\mathbb{R}^3$, is assumed by another {\em extension} of the Laplacian, namely the {\em magnetic Laplacian} $(\nabla-{\rm i} A)^2$ (see \cite{AHS,ReedSimon}).
Nonlinear magnetic Schr\"odinger equations like
\[
- (\nabla-{\rm i} A)^2 u + u = f(u)
\]
have been extensively studied (see e.g.\ \cite{arioliSz,CS,DCVS,EL,K,S}).\\
In \cite{IT}, Ichinose and Tamura, through oscillatory integrals, introduce the so-called Weyl pseudodifferential operator defined with {\em mid-point prescription}
\begin{align*}
{\mathscr H}_A u (x)
&=
\frac{1}{(2\pi)^3}\int_{\mathbb{R}^6} e^{\i (x-y)\cdot \xi } \sqrt{\Big|\xi - A\big(\frac{x+y}{2}\big)\Big|^2 + m^2} u(y) dyd\xi\\
&=
\frac{1}{(2\pi)^3}\int_{\mathbb{R}^6} e^{\i (x-y)\cdot \left(\xi+ A\big(\frac{x+y}{2}\big)\right) } \sqrt{|\xi|^2 + m^2} u(y) dyd\xi
\end{align*}
as a {\em fractional relativistic} generalization of the magnetic Laplacian (see also \cite{I89}, the review article \cite{I10} and the references therein).\
The operator ${\mathscr H}_A$ takes the place of $\sqrt{-\Delta + m^2}$ and it is possible to show that for all $u\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$,
\begin{equation*
\begin{split}
{\mathscr H}_A u (x)
&=
m u(x)
- \lim_{\varepsilon\searrow 0}\int_{B^c_\varepsilon(0)} \left[e^{-\i y \cdot A \big( x+ \frac{y}{2}\big)} u(x+y) - u(x) - 1_{\{|y|<1\}}(y) y\cdot (\nabla - \i A(x))u(x)\right]d\mu\\
&=
m u(x)
+ \lim_{\varepsilon\searrow 0}\int_{B^c_\varepsilon(x)} \left[u(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(y)\right] \mu(y-x)dy ,
\end{split}
\end{equation*}
where
\[
d\mu=
\mu(y)dy =
\begin{cases}
2\left(\frac{m}{2\pi}\right)^2 \frac{K_2(m|y|)}{|y|^2}dy,
& m>0,\\
\frac{1}{\pi^2 |y|^4}dy, & m=0,
\end{cases}
\]
and $K_2$ is the modified Bessel function of the third kind of order $2$ (see e.g. \cite[Subsection 3.1]{I10}).\\
In this paper we are concerned with the operator
\begin{equation}
\label{operator}
(-\Delta)^s_Au(x)=c_s \lim_{\varepsilon\searrow 0}\int_{B^c_\varepsilon(x)}\frac{u(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(y)}{|x-y|^{3+2s}}dy,
\quad x\in\mathbb{R}^3,
\end{equation}
and, in particular, with ground state solutions of the equation
\begin{equation}
\label{prob}
\tag{$\mathscr{P}_{s,A}$}
(-\Delta)^s_A u+u=|u|^{p-2}u\quad \text{in $\mathbb{R}^3$.}
\end{equation}
The operator \eqref{operator} is consistent with the definition of fractional Laplacian given in \eqref{slaplacian} if $A=0$ and with
${\mathscr H}_A$ for $m=0$ and $s=1/2$. To our knowledge, this is the first mathematical contribution
to the study of nonlinear problems involving operator~\eqref{operator}.\
For the sake of completeness we mention that there exist other
different definitions of the {\em magnetic pseudorelativistic
operator} (see \cite{I10,IMP,LS}) and in \cite{franketal} a fractional
magnetic operator $(\nabla-{\rm i} A)^{2s}$ is defined through the
spectral theorem (see also discussion on the different definitions in
\cite[Proposition 2.6]{I10}).
\vskip4pt
\noindent
Throughout the paper we consider magnetic potentials $A$'s which have locally bounded gradient. We now state our results.
\\
Let $2<p<6/(3-2s)$ and
consider the minimization problem
\begin{equation}
\label{MA}
\tag{${\mathscr M}_A$}
{\mathscr M}_A=\inf_{u\in \mathscr{S}} \left(\int_{\mathbb{R}^3}|u|^2dx+\frac{c_s}{2}\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy\right),
\end{equation}
where
\begin{equation*}
\mathscr{S}=\Big\{u\in H^s_A(\mathbb{R}^3,\mathbb{C}):\int_{\mathbb{R}^3}|u|^p dx=1\Big\}
\end{equation*}
and $H^s_A(\mathbb{R}^3,\mathbb{C})$ is a suitable Hilbert space defined in Section~\ref{setting}.
Once a solution to ${\mathscr M}_A$ exists, due to the Lagrange Multiplier Theorem, we get a weak solution to \eqref{prob}, see Sections \ref{setting} and \ref{sect4}.\\
When $\mathscr{S}$ is restricted to radially symmetric functions, the problem is denoted by ${\mathscr M}_{A,r}$.
\vskip3pt
First we give the following
\begin{definition}\rm
We say that $A$ satisfies assumption $\mathscr{A}$, if
for any unbounded sequence $\Xi=\{\xi_n\}_{n\in{\mathbb N}}\subset\mathbb{R}^3$
there exist a sequence $\{H_n\}_{n\in{\mathbb N}}\subset\mathbb{R}^3$ and a function $A_\Xi:\mathbb{R}^3\to\mathbb{R}^3$ such that
\begin{equation}
\label{limA}
\lim_n A_n(x)=A_\Xi(x) \hbox{ for all } x\in\mathbb{R}^3 \hbox{ and }
\sup_n \|A_n\|_{L^\infty(K)}<\infty \hbox{ for all compact sets } K,
\end{equation}
where $A_n(x):=A(x+\xi_n)+H_n$ and $\{\xi_n\}$ is a subsequence of $\Xi$ such that $|\xi_n|\to\infty$.
\end{definition}
\noindent
We also set $\mathscr{X}:=\{\Xi=\{\xi_n\}_{n\in{\mathbb N}}\hbox{ unbounded}: \text{condition \eqref{limA} holds}\}$.
Observe that, if $A$ admits limit as $|x|\to\infty$,
then it satisfies assumption $\mathscr{A}$.\\
Our main result is
\begin{theorem}[Subcritical case]
\label{main}
The following facts hold:
\begin{enumerate}[label=(\roman*),ref=\roman*]
\item \label{i12} ${\mathscr M}_{A,r}$ has a solution;
\item \label{iii12}if $A$ is linear, then ${\mathscr M}_A$ has a solution;
\item \label{ii12}if $A$ satisfies $\mathscr{A}$ and
${\mathscr M}_A<\inf_{\Xi\in \mathscr{X}} {\mathscr M}_{A_\Xi}$,
then ${\mathscr M}_A$ has a solution.
\end{enumerate}
\end{theorem}
\noindent
We also consider the minimization problem
\begin{equation}
\label{MAC}
\tag{${\mathscr M}_A^c$}
{\mathscr M}_A^c:=\inf_{u\in \mathscr{S}^c}
\frac{c_s}{2}\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy,
\end{equation}
where
$$
\mathscr{S}^c=\Big\{u\in D^s_A(\mathbb{R}^3,\mathbb{C}):\int_{\mathbb{R}^3}|u|^{6/(3-2s)} dx=1\Big\}
$$
and $D^s_A(\mathbb{R}^3,\mathbb{C})$ is a suitable Hilbert space defined in Subection~\ref{subscrit}.
We are able to prove
\begin{theorem}[Critical case]
\label{main-2}
The following facts hold:
\begin{enumerate}[label=(\roman*),ref=\roman*]
\item \label{i13}if ${\mathscr M}_A^c$ has a solution $u$, there exist
$z\in\mathbb{R}^3$, $\varepsilon>0$ and $\vartheta_A:\mathbb{R}^3\to\mathbb{R}$ such that
$$
u(x)=d_{s}\left(\frac{\varepsilon}{\varepsilon^2+|x-z|^2}\right)^{\frac{3-2s}{2}}e^{\i \vartheta_A(x)};
$$
\item \label{ii13}if for some $k\in{\mathbb N}$ and $E\subset\mathbb{R}^6$ of positive measure
$$
(x-y)\cdot A\Big(\frac{x+y}{2}\Big)\not \equiv 2k\pi\quad \hbox{for all } (x,y)\in E,
$$
then ${\mathscr M}_A^c$ has no solution
$u$ of the form $e^{\i \vartheta} v(x)$ where $\theta\in\mathbb{R}$ and $v$ of fixed sign.
\end{enumerate}
\end{theorem}
\vskip3pt
\noindent
The local version of the above results can be found in the work \cite{EL} by Esteban and Lions.
In \cite{DipValPal}, for the case without magnetic field and with subcritical nonlinearities,
existence of ground states was obtained using different arguments,
namely without involving concentration compactness arguments, but instead symmetrizing the minimizing sequences, by using
$$
\int_{\mathbb{R}^6}\frac{||u(x)|^*-|u(y)|^*|^2}{|x-y|^{3+2s}}dxdy\leq \int_{\mathbb{R}^6}\frac{|u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy,
$$
for all $u\in H^s(\mathbb{R}^3)$, where $v^*$ denotes the Schwarz
symmetrization of $v:\mathbb{R}^3\to \mathbb{R}^+$. On the contrary, when $A\not\equiv 0$, the inequality
$$
\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}|u(x)|^*-|u(y)|^*|^2}{|x-y|^{3+2s}}dxdy\leq \int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy,
$$
does not seem to work and a different strategy for the proof has to be outlined.\
Dealing with the nonlocal case, it is natural to expect that, in the study of minimizing sequences, the hardest stage
is that of ruling out the dichotomy in the concentration compactness alternative. This is in fact the case, but thanks to a careful analysis
developed in Lemma~\ref{lemma-dic}, dichotomy can be ruled out allowing for tightness and hence the strong convergence of
minimizing sequences up translations and phase changes.
We organize the paper in the following way: in Section \ref{setting} we introduce the functional setting of the problem and we provide some basic properties about it; in Section \ref{sec3} we show further technical facts on the functional setting as well as some preliminary results about the Concentration-Compactness procedure; finally, in Section \ref{sec4}, we complete with the proofs of our results.
\vskip10pt
\noindent
{\bf Acknowledgments.} The research was partially supported by Gruppo Nazionale per l'Analisi Matematica,
la Probabilit\`a e le loro Applicazioni (INdAM).
\vskip10pt
\noindent
{\bf Notations.} We denote by $B_R(\xi)$ a ball in $\mathbb{R}^3$ of center $\xi$ and radius $R$.
For a measurable set $E\subset \mathbb{R}^3$ we denote by $E^c$ the complement
of $E$ in $\mathbb{R}^3$, namely $E^c=\mathbb{R}^3\setminus E$.\ We denote by $1_E$ the indicator function of $E$.\
The symbol ${\mathcal L}^n(\Omega)$ stands for the Lebesgue measure of a measurable subset $\Omega\subset\mathbb{R}^n$. For a
complex number $z\in\mathbb{C}$, the symbol $\Re z$ indicates its real part and $\Im z$ its imaginary part.
The modulus of $z$ is denoted by $|z|$. The standard norm of $L^p$ spaces is denoted by $\|\cdot\|_{L^p}.$
\section{Functional setting}
\label{setting}
\noindent
Let $L^2(\mathbb{R}^3,\mathbb{C})$ be the Lebesgue space of complex valued functions with
summable square endowed with the real scalar product
$$
\langle u,v \rangle_{L^2}:=\Re\int_{\mathbb{R}^3} u\bar vdx,\quad \text{for all $u,v\in L^2(\mathbb{R}^3,\mathbb{C})$},
$$
and $A:\mathbb{R}^3\to\mathbb{R}^3$ be a continuous function. We consider the magnetic
Gagliardo semi-norm defined by
$$
[u]^2_{s,A}:=\frac{c_s}{2}\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy,
$$
the scalar product defined by
$$
\langle u,v \rangle_{s,A}:=\langle u,v \rangle_{L^2}+\frac{c_s}{2}{\Re}\int_{\mathbb{R}^6}\frac{\left(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)\right)\overline{\left(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v(x)-v(y)\right)}}{|x-y|^{3+2s}}dxdy,
$$
and the corresponding norm denoted by
$$
\|u\|_{s,A}:=\big(\|u\|_{L^2}^2+[u]_{s,A}^2\big)^{1/2}.
$$
We consider the space ${\mathcal H}$ of measurable functions $u:\mathbb{R}^3\to\mathbb{C}$ such that $\|u\|_{s,A}<\infty.$
\begin{proposition}
\label{complete}
$({\mathcal H},\langle \cdot,\cdot \rangle_{s,A})$ is a real Hilbert space.
\end{proposition}
\begin{proof}
It is readily checked that $\langle u,v \rangle_{s,A}$ is a real scalar product. Let us prove that ${\mathcal H}$ with this scalar product is complete.
Let $\{u_n\}_{n\in{\mathbb N}}$ be a Cauchy sequence in ${\mathcal H}$, namely for every $\varepsilon >0$ there exists $\nu_\varepsilon\in\mathbb{N}$ such that for all $m,n>\nu_\epsilon$ we have $\|u_n - u_m\|_{s,A}<\varepsilon$. Thus $\{u_n\}_{n\in{\mathbb N}}$ is a Cauchy sequence on $L^2(\mathbb{R}^3,\mathbb{C})$ and then there exists $u\in L^2(\mathbb{R}^3,\mathbb{C})$ such that $u_n \to u$ in $L^2(\mathbb{R}^3,\mathbb{C})$ and a.e.\ in $\mathbb{R}^3$. Firstly, we prove that $u\in {\mathcal H}$. By Fatou Lemma we have
\[
[u]^2_{s,A}
\leq
\liminf_n
[u_n]^2_{s,A}
\leq
\liminf_n
([u_n - u_{\nu_1}]_{s,A} + [u_{\nu_1}]_{s,A})^2
\leq
(1+ [u_{\nu_1}]_{s,A})^2.
\]
Thus it remains to prove that $[u_n - u]_{s,A}\to 0$ as $n\to\infty$. Again, by Fatou Lemma
\[
[u_n - u]_{s,A} \leq \liminf_k [u_n - u_k]_{s,A} \leq \liminf_k \|u_n - u_k\|_{s,A} \leq \varepsilon,
\]
for all $\varepsilon >0$ and $n$ large.
\end{proof}
\noindent
For any function $w:\mathbb{R}^3\to\mathbb{C}$ and a.e.\ $x\in\mathbb{R}^3$, we set
\begin{equation}
\label{notation}
w_x(y):=e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}w(y),\quad \text{for $y\in\mathbb{R}^3$}.
\end{equation}
We have
\begin{proposition}
The space $C^\infty_c(\mathbb{R}^3,\mathbb{C})$ is a subspace of ${\mathcal H}$.
\end{proposition}
\begin{proof}
It is enough to prove that $[u]_{s,A}<\infty$, for any $u\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$.\ If $K$ is the compact support of $u$, we have
\begin{align*}
\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy\leq
2 \int_{K\times\mathbb{R}^3}\frac{|u_x(x)-u_x(y)|^2}{|x-y|^{3+2s}}dxdy.
\end{align*}
Observe that, since $\nabla A$ is locally bounded, the gradient of the function $(x,y)\mapsto u_x(y)$ is bounded on $K\times\mathbb{R}^3$.
Then we have $|u_x(x)-u_x(y)|\leq C|x-y|$ for any $(x,y)\in K\times\mathbb{R}^3$. Of course, we also have
$|u_x(x)-u_x(y)|\leq C$ for any $(x,y)\in K\times\mathbb{R}^3$. Hence, we get
\begin{align*}
\int_{K\times\mathbb{R}^3}\frac{|u_x(x)-u_x(y)|^2}{|x-y|^{3+2s}}dxdy&\leq C\int_{K\times\mathbb{R}^3}\frac{\min\{|x-y|^2,1\}}{|x-y|^{3+2s}}dxdy \\
&\leq C\int_{B_1(0)} \frac{1}{|z|^{1+2s}} dz+C\int_{B^c_1(0)} \frac{1}{|z|^{3+2s}}dz,
\end{align*}
which concludes the proof.
\end{proof}
\noindent
Thus we can give the following
\begin{definition}
We define $H^s_A(\mathbb{R}^3,\mathbb{C})$ as the closure of $C^\infty_c(\mathbb{R}^3,\mathbb{C})$ in ${\mathcal H}$.
\end{definition}
\noindent
Then, $H^s_A(\mathbb{R}^3,\mathbb{C})$ is a real Hilbert space by Proposition~\ref{complete}.\
For $A=0$ this space is consistent with the usual fractional space $H^s(\mathbb{R}^3,\mathbb{C})$ whose norm is denoted by $\|\cdot\|_{s}$.
For a given Lebesgue measurable set $E\subset\mathbb{R}^3$
the localized Gagliardo norms are defined by
\begin{align*}
\|u\|_{H^s_A(E)}&:=\left(\int_{E} |u(x)|^2dx+
\frac{c_s}{2}\int_{E\times E}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy\right)^{1/2}, \\
\|u\|_{H^s(E)}&:=\left(\int_{E} |u(x)|^2dx+
\frac{c_s}{2}\int_{E\times E}\frac{|u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy\right)^{1/2}.
\end{align*}
The operator $(-\Delta)^s_A : H^s_A(\mathbb{R}^3,\mathbb{C})\to H^{-s}_A(\mathbb{R}^3,\mathbb{C})$ is defined by duality as
\begin{align*}
\langle (-\Delta)^s_Au,v\rangle &:=\frac{c_s}{2} {\Re}\int_{\mathbb{R}^6}\frac{\left(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)\right)\overline{\left(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v(x)-v(y)\right)}}{|x-y|^{3+2s}}dxdy \\
& =\frac{c_s}{2} {\Re}\int_{\mathbb{R}^6}\frac{\left(u(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(y)\right)\overline{\left(v(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v(y)\right)}}{|x-y|^{3+2s}}dxdy.
\end{align*}
If $f\in H^{-s}_A(\mathbb{R}^3,\mathbb{C})$, we say that $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$ is a weak solution to
\begin{equation}
\label{problema}
(-\Delta)^s_Au+u=f,\quad\text{in $\mathbb{R}^3$,}
\end{equation}
if we have
\begin{align*}
&\frac{c_s}{2}{\Re}\int_{\mathbb{R}^6}\frac{\left(u(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(y)\right)\overline{\left(v(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v(y)\right)}}{|x-y|^{3+2s}}dxdy \\
&+{\Re}\int_{\mathbb{R}^3} u\overline{v} dx={\Re}\int_{\mathbb{R}^3} f\overline{v} dx,\quad \text{for all $v\in H^s_A(\mathbb{R}^3,\mathbb{C})$}.
\end{align*}
Of course, one can equivalently define the weak solution
by testing over functions $v\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$.\\
On smooth functions, the operator $(-\Delta)^s_A $ admits the point-wise representation \eqref{operator}. To show this we need the following preliminary results.
\begin{lemma}
\label{prelimll}
Let $K$ be a compact subset of $\mathbb{R}^3,$ $R>0$ and set $K'=\{x\in\mathbb{R}^3: d(x,K)\leq R\}$.
Assume that $f\in C^2(\mathbb{R}^6)$
and that $g\in C^{1,\gamma}(K')$
for some $\gamma\in [0,1]$.\
If $h(x,y)=f(x,y)g(y)$, then there exists a positive constant $C$ depending on $K,f,g,R$, such that
$$
|\nabla_y h(x,y_2)-\nabla_y h(x,y_1)|\leq C|y_2-y_1|^\gamma,
$$
for all $x\in K$ and every $y_2,y_1\in K'$.
\end{lemma}
\begin{proof}
The proof is omitted as it is straightforward.
\end{proof}
\begin{lemma}
\label{controllo}
Let $A\in C^2(\mathbb{R}^3)$ and $u\in C^{1,\gamma}_{\rm loc}(\mathbb{R}^3,\mathbb{C})$ for some $\gamma\in [0,1]$.\
Then, for any compact set $K\subset\mathbb{R}^3$ and $R>0$,
there exists a positive constant $C$ depending on $R,K,A,u$, such that
$$
|u_x(x+y)+u_x(x-y)-2u_x(x)|\leq C |y|^{1+\gamma},
$$
for every $x\in K$ and $y \in B_R(0)$.
\end{lemma}
\begin{proof}
Fix a compact set $K\subset \mathbb{R}^3$ and $R>0$.\ Consider $x\in K$ and $y\in B_R(0)$.\
Then, by the Mean Value Theorem,
there exist $\tau_1,\tau_2\in [0,1]$ such that
\begin{align*}
|u_x(x+y)+u_x(x-y)-2u_x(x)|&=|\nabla_y u_x(x+\tau_1 y)\cdot y- \nabla_y u_x(x-\tau_2 y)\cdot y|\\
& \leq |\nabla_y u_x(x+\tau_1 y)- \nabla_y u_x(x-\tau_2 y)||y|\leq C|y|^{1+\gamma},
\end{align*}
where in the last inequality we use Lemma~\ref{prelimll} with $f(x,y)=e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}$ and $g(y)=u(y).$
\end{proof}
\noindent
Thus in the case $u$ and $A$ are smooth enough, we have the following result
\begin{theorem}[Weak to strong solution]
\label{weakstrong}
Let $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$ be a weak solution to \eqref{problema}. Assume that
$A\in C^2(\mathbb{R}^3)$ and that
$$
u\in L^\infty(\mathbb{R}^3,\mathbb{C})\cap C^{1,\gamma}_{\rm loc}(\mathbb{R}^3,\mathbb{C}),\quad\text{for some
$\gamma\in (0,1]$ with $\gamma>2s-1$.}
$$
Then $u$ solves problem \eqref{problema} pointwise a.e.\ in $\mathbb{R}^3$.
\end{theorem}
\begin{proof}
With the notation introduced in \eqref{notation}, the definition of weak solution writes as
\begin{equation}
\label{varf-2}
\frac{c_s}{2}{\Re}\int_{\mathbb{R}^6}\frac{\left(u_x(x)-u_x(y)\right)\overline{\left(v_x(x)-v_x(y)\right)}}{|x-y|^{3+2s}}dxdy
+{\Re}\int_{\mathbb{R}^3} u\overline{v} dx={\Re}\int_{\mathbb{R}^3} f\overline{v} dx,
\end{equation}
for all $v\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$.\
Let us fix a $v\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$ and set $K:={\rm supp}(v)$.\
Now, for any $\varepsilon>0$, we introduce the auxiliary function
$g_\varepsilon:K\to\mathbb{R}$ defined by
$$
g_\varepsilon(x):=\frac{c_s}{2}\int_{\mathbb{R}^3}\frac{u_x(x)-u_x(y)}{|x-y|^{3+2s}}1_{B^c_\varepsilon(x)}(y)dy.
$$
Note that for all $x\in K$ we have that
\begin{equation}
\label{puntgeps}
g_\varepsilon (x)\to \frac{1}{2}(-\Delta)^s_A u (x),\quad\text{as $\varepsilon\to 0$ whenever the limit exists}.
\end{equation}
Simple changes of variables show that $g_\varepsilon$ can be equivalently written as
$$
g_\varepsilon(x)=-\frac{c_s}{4}\int_{\mathbb{R}^3}
\frac{u_x(x+y)+u_x(x-y)-2u_x(x)}{|y|^{3+2s}}1_{B^c_\varepsilon(0)}(y)dy.
$$
Furthermore, by Lemma~\ref{controllo}, there exist $C>0$ and $R>0$ such that
$$
|u_x(x+y)+u_x(x-y)-2u_x(x)|\leq C |y|^{1+\gamma},\quad \text{for $x\in K$
and $y \in B_R(0)$}.
$$
Therefore, taking into account that
$|u_x(y)|\leq \|u\|_{L^\infty}$ for all $y\in\mathbb{R}^3$,
we have the inequality
$$
\frac{|u_x(x+y)+u_x(x-y)-2u_x(x)|}{|y|^{3+2s}}\leq \frac{C}{|y|^{2+2s-\gamma}}1_{B_R(0)}(y)+
\frac{C}{|y|^{3+2s}}1_{B^c_R(0)}(y),
$$
for some constant $C$.\
Due to the assumption $\gamma>2s-1$, the right hand side belongs to $L^1(\mathbb{R}^3)$.
Then, by dominated convergence, the limit of $g_\varepsilon(x)$ as $\varepsilon\to 0$
exists a.e.\ in $K$ and it is thus equal to $\frac{1}{2}(-\Delta)^s_A u (x)$ by \eqref{puntgeps}. Since
also $|g_\varepsilon(x)|\leq C$ a.e.\ in $K$, again the dominated convergence yields
\begin{equation}
\label{loc-conv}
g_\varepsilon\to \frac{1}{2}(-\Delta)^s_A u,\quad\text{strongly in $L^1(K)$}.
\end{equation}
Now, the first term in formula \eqref{varf-2} can be treated as follows
\begin{align*}
& \frac{c_s}{2}\int_{\mathbb{R}^6}\frac{\left(u_x(x)-u_x(y)\right)
\overline{\left(v_x(x)-v_x(y)\right)}}{|x-y|^{3+2s}}dxdy \\
& =\lim_{\varepsilon\to 0}\frac{c_s}{2}\int_{\mathbb{R}^6} \frac{\left(u_x(x)-u_x(y)\right)
\overline{\left(v_x(x)-v_x(y)\right)}}{|x-y|^{3+2s}}1_{B^c_\varepsilon(x)}(y)dxdy \\
& =\lim_{\varepsilon\to 0}\Big(\int_{\mathbb{R}^3} g_\varepsilon(x)\overline{v(x)}dx
-\frac{c_s}{2}\int_{\mathbb{R}^6}\frac{\left(u_x(x)-u_x(y)\right)
\overline{v_x(y)}}{|x-y|^{3+2s}}1_{B^c_\varepsilon(x)}(y)dxdy\Big).
\end{align*}
By Fubini Theorem on the second term of the last equality, switching
the two variables and observing that
$$-\left(u_y(y)-u_y(x)\right)
\overline{v_y(x)}1_{B^c_\varepsilon(y)}(x)
=(u_x(x)-u_x(y))\overline{v(x)}1_{B^c_\varepsilon(x)}(y)$$
yields
\begin{align*}
\frac{c_s}{2}\int_{\mathbb{R}^6}\frac{\left(u_x(x)-u_x(y)\right)
\overline{\left(v_x(x)-v_x(y)\right)}}{|x-y|^{3+2s}}dxdy
=\lim_{\varepsilon\to 0}\int_{\mathbb{R}^3} 2g_\varepsilon(x)\overline{v(x)}dx=\int_{\mathbb{R}^3} (-\Delta)^s_A u(x)\overline{v(x)}dx,
\end{align*}
where we used \eqref{loc-conv} in the last equality. Then, from formula~\eqref{varf-2},
we conclude that
$$
\Re\Big(\int_{\mathbb{R}^3} \big((-\Delta)^s_A u+u-f\big)\overline{v} dx\Big)=0,
\quad\text{for all $v\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$,}
$$
yielding $(-\Delta)^s_A u+u=f$ a.e.\ in $\mathbb{R}^3$. The proof is complete.
\end{proof}
\noindent
We conclude the section with an observation about the formal consistency of the
spaces $H^s_A(\mathbb{R}^3,\mathbb{C})$, up to suitably correcting the norm, with the usual local Sobolev spaces without magnetic field
in the singular limit as $s\to 1$ and $A\to 0$ pointwise. Consider the modified norm
$$
|||u|||_{s,A}:=\big(\|u\|_{L^2}^2+(1-s)[u]_{s,A}^2\big)^{1/2}.
$$
By arguing as in the proof of Lemma~\ref{livelli}, it follows that
$$
\lim_{A\to 0}[u]_{s,A}^2=[u]_{s,0}^2, \quad \text{for all $u\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$}.
$$
Moreover, from the results of Brezis-Bourgain-Mironescu \cite{bourg,bourg2},
we know that
$$
\lim_{s\to 1}(1-s)[u]_{s,0}^2=\|\nabla u\|_{L^2}^2,
\quad \text{for all $u\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$}.
$$
In conclusion
$$
\lim_{s\to 1}
\lim_{A\to 0}
|||u|||_{s,A}=\|u\|_{H^1(\mathbb{R}^3)},
\quad \text{for all $u\in C^\infty_c(\mathbb{R}^3,\mathbb{C})$}.
$$
Hence $|||u|||_{s,A}$ approximates the $H^1$-norm for $s\sim 1$ and $A\sim 0$.
\section{Preliminary stuff}\label{sec3}
In this section we provide some technical facts about the functional setting of the problem as well as some preliminary results about the Concentration-Compactness procedure.
\begin{lemma}[Diamagnetic inequality]
\label{diamagnetic}
For every $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$ it holds $|u|\in H^s(\mathbb{R}^3)$. More precisely
$$
\| |u|\|_{s}\leq \|u\|_{s,A},\quad\text{for every $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$.}
$$
\end{lemma}
\begin{proof}
For a.e.\ $x,y\in \mathbb{R}^3$ we have
$$
\Re \big(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)} u(x)\overline{u(y)}\big)\leq |u(x)||u(y)|.
$$
Therefore, we have
\begin{align*}
|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2 &=|u(x)|^2+|u(y)|^2-2\Re \big(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)} u(x)\overline{u(y)}\big) \\
& \geq |u(x)|^2+|u(y)|^2-2|u(x)||u(y)|=||u(x)|-|u(y)||^2,
\end{align*}
which immediately yields the assertion.
\end{proof}
\begin{remark}[Pointwise Diamagnetic inequality]\label{diamremark}
\rm
There holds
$$
||u(x)|-|u(y)||\leq
|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|,\quad\text{for a.e.\ $x,y\in\mathbb{R}^3$.}
$$
\end{remark}
\noindent
We have the following local embedding of $H_A^s(\mathbb{R}^3,\mathbb{C})$.
\begin{lemma}[Local embedding in $H^s(\mathbb{R}^3,\mathbb{C})$]
\label{localemb}
For every compact set $K\subset \mathbb{R}^3,$ the space $H^s_A(\mathbb{R}^3,\mathbb{C})$
is continuously embedded into $H^s(K,\mathbb{C})$.
\end{lemma}
\begin{proof}
Fixed a compact $K\subset\mathbb{R}^3$, we have
\begin{align*}
\|u\|_{H^s(K)}^2 &=\int_{K} |u(x)|^2dx+
\frac{c_s}{2}\int_{K\times K}\frac{|u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy \\
& \leq \int_{\mathbb{R}^3} |u(x)|^2dx+
C\int_{K\times K}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\quad +C\int_{K\times K}\frac{|u(x)|^2|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}-1|^2}{|x-y|^{3+2s}}dxdy \\
& \leq C \|u\|_{s,A}^2+CJ,
\end{align*}
where we have set
$$
J:=\int_{K\times K}\frac{|u(x)|^2|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}-1|^2}{|x-y|^{3+2s}}dxdy.
$$
We now prove that $J\leq C\|u\|_{L^2}^2$, which ends the proof.
We have
\begin{align*}
J &=\int_{K}|u(x)|^2\int_{K\cap \{|x-y|\geq 1\}}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}-1|^2}{|x-y|^{3+2s}}dxdy \\
&\quad+\int_{K}|u(x)|^2\int_{K\cap \{|x-y|\leq 1\}}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}-1|^2}{|x-y|^{3+2s}}dxdy \\
&\leq C\int_{K}|u(x)|^2\int_{K\cap \{|x-y|\geq 1\}}\frac{1}{|x-y|^{3+2s}}dxdy \\
&\quad+C\int_{K}|u(x)|^2\int_{K\cap \{|x-y|\leq 1\}}\frac{1}{|x-y|^{1+2s}}dxdy,
\end{align*}
where in the last line we used that
$$
|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}-1|^2
\leq C|x-y|^2,\quad \text{for $|x-y|\leq 1$,\,\, $x,y\in K$,}
$$
since $A$ is locally bounded. The proof is now complete.
\end{proof}
\begin{lemma}
\label{gen-bound}
Let $\{A_n\}_{n\in{\mathbb N}}$ be a sequence of uniformly locally bounded functions
$A_n:\mathbb{R}^3\to\mathbb{R}^3$ with locally bounded gradient and, for any $n\in{\mathbb N}$, $u_n\in H^s_{A_n}(\mathbb{R}^3,\mathbb{C})$ be such that
$$
\sup_{n\in{\mathbb N}} \|u_n\|_{s,A_n}<\infty.
$$
Then, up to a subsequence, $\{u_n\}_{n\in{\mathbb N}}$ converges strongly to some function $u$ in $L^q(K,\mathbb{C})$ for
every compact set $K$ and any $q\in[1,6/(3-2s))$.
\end{lemma}
\begin{proof}
Arguing as in the proof of Lemma~\ref{localemb}, the assertion follows by
\cite[Corollary 7.2]{DPV}.
\end{proof}
\begin{lemma}[Magnetic Sobolev embeddings]
\label{embedd}
The injection
$$
H^s_A(\mathbb{R}^3,\mathbb{C})\hookrightarrow L^p(\mathbb{R}^3,\mathbb{C})
$$
is continuous for every $2\leq p\leq \frac{6}{3-2s}$. Furthermore, the injection
$$
H^s_A(\mathbb{R}^3,\mathbb{C})\hookrightarrow L^p(K,\mathbb{C})
$$
is compact for every $1\leq p<\frac{6}{3-2s}$ and any compact set $K\subset \mathbb{R}^3$.
\end{lemma}
\begin{proof}
By combining Remark~\ref{diamremark} with the continuous injection $H^s(\mathbb{R}^3)\hookrightarrow L^{6/(3-2s)}(\mathbb{R}^3)$
(see \cite[Theorem 6.5]{DPV}) yields
\begin{equation}
\label{sobol}
\|u\|_{L^{6/(3-2s)}(\mathbb{R}^3)}\leq C \Big(\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy\Big)^{1/2}
\quad\text{for all $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$}.
\end{equation}
Whence, by interpolation the first assertion immediately follows.\ For the compact embedding,
taking into account Lemma~\ref{localemb}, the assertion follows by \cite[Corollary 7.2]{DPV}.
\end{proof}
\begin{lemma}[Vanishing]
\label{lionslemma}
Let $\{u_{n}\}_{n\in{\mathbb N}}$ be a bounded sequence in $H^{s}(\mathbb R^{3})$ and
assume that, for some $R>0$ and $2\leq q < \frac{6}{3-2s}$, there holds
$$
\lim_n\sup_{\xi\in \mathbb{R}^3}\int_{B(\xi,R)}|u_{n}|^{q}dx=0.
$$
Then $u_{n}\to 0$ in $L^{p}(\mathbb R^{3})$ for $2< p <\frac{6}{3-2s}$.
\end{lemma}
\begin{proof}
See \cite[Lemma 2.3]{DSS}.
\end{proof}
\begin{lemma}[Localized Sobolev inequality]
\label{localized}
Let $\xi\in\mathbb{R}^3$ and $R>0$. Then, for $u\in H^s(B_R(\xi))$,
$$
\|u\|_{L^\frac{6}{3-2s}(B_R(\xi))}\leq C(s)\left(\frac{1}{R^{2s}}\int_{B_R(\xi)} |u(x)|^2dx+
\int_{B_R(\xi)\times B_R(\xi)}\frac{|u(x)-u(y)|^2}{|x-y|^{3+2s}}dydx\right)^{1/2}
$$
for some constant $C(s)>0$. In particular for every $1\leq p\leq \frac{6}{3-2s}$ there holds
$$
\|u\|_{L^p(B_R(\xi))}\leq C(s,R) \|u\|_{H^s(B_R(\xi))}
$$
for some constant $C(s,R)>0$ and all $u\in H^s(B_R(\xi))$.
\end{lemma}
\begin{proof}
See \cite[Proposition 2.5]{BP} for the first inequality. The second inequality immediately follows.
\end{proof}
\begin{lemma}[Cut-off estimates]
\label{cut}
Let $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$ and $\varphi\in C^{0,1}(\mathbb{R}^3)$ with $0\leq \varphi\leq 1$. Then,
for every pair of measurable sets $E_1,E_2\subset \mathbb{R}^3$, we have
\begin{align*}
\int_{E_1\times E_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}\varphi(x)u(x)-\varphi(y)u(y)|^2}{|x-y|^{3+2s}}dxdy
&\leq C\min\left\{\int_{E_1}|u|^2dx,\int_{E_2}|u|^2dx\right\} \\
&\quad+C\int_{E_1\times E_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy,
\end{align*}
where $C$ depends on $s$ and on the Lipschitz constant of $\varphi$.
\end{lemma}
\begin{proof}
The proof follows by arguing as in \cite[Lemma 5.3]{DPV}, where the case $A=0$ and $E_1=E_2$
is considered.\ For the sake of completeness, we show the details.
We have
\begin{align*}
& \int_{E_1\times E_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}\varphi(x)u(x)-\varphi(y)u(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\leq C\int_{E_1\times E_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy+
C\int_{E_1\times E_2}\frac{|u(y)|^2|\varphi(x)-\varphi(y)|^2}{|x-y|^{3+2s}}dxdy.
\end{align*}
On the other hand, the second integral splits as
$$
\int_{E_2}|u(y)|^2\int_{E_1\cap \{|x-y|\leq 1\}}\frac{1}{|x-y|^{1+2s}}dxdy+\int_{E_2}|u(y)|^2\int_{E_1\cap \{|x-y|\geq 1\}}\frac{1}{|x-y|^{3+2s}}dxdy
\leq C\int_{E_2}|u|^2dy.
$$
Analogously, we have
\begin{align*}
& \int_{E_1\times E_2}\frac{|\varphi(x)u(x)-e^{\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}\varphi(y)u(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\leq C\int_{E_1\times E_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy+
C\int_{E_1\times E_2}\frac{|u(x)|^2|\varphi(x)-\varphi(y)|^2}{|x-y|^{3+2s}}dxdy,
\end{align*}
and the second term can be estimated as before by $\int_{E_1}|u|^2dx$. The assertion follows.
\end{proof}
Thus we can prove
\begin{lemma}[Dicothomy]
\label{lemma-dic}
Let $\{u_n\}_{n\in {\mathbb N}}$ be a sequence in $H^s_A(\mathbb{R}^3,\mathbb{C})$ such that, for some $L>0$,
$$
\|u_n\|_{L^p(\mathbb{R}^3)}=1,\qquad \lim_n \|u_n\|_{s,A}^2=L,
$$
and let us set
$$
\mu_n(x)=|u_n(x)|^2+\int_{\mathbb{R}^3}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dy,\quad x\in\mathbb{R}^3,\,\, n\in{\mathbb N}.
$$
Assume that there exists $\beta\in (0,L)$ such
that for all $\varepsilon>0$ there exist $\bar R>0$, $\bar n\geq 1$,
a sequence of radii $R_n\to+\infty$ and
$\{\xi_n\}_{n\in{\mathbb N}}\subset \mathbb{R}^3$ such that for $n\geq \bar n$
\begin{align}
& \left|\int_{\mathbb{R}^3}\mu_n^1(x)dx-\beta\right|\leq \varepsilon,\qquad
\mu_n^1:= 1_{B_{\bar R}(\xi_n)}\mu_n, \nonumber\\%\label{uno}\\
& \left|\int_{\mathbb{R}^3}\mu_n^2(x)dx-(L-\beta)\right|\leq \varepsilon,\qquad
\mu_n^2:= 1_{B^c_{R_n}(\xi_n)}\mu_n, \nonumber\\%\label{due}\\
& \int_{\mathbb{R}^3}|\mu_n(x)-\mu_n^1(x)-\mu_n^2(x)|dx\leq \varepsilon. \label{tre}
\end{align}
%
%
Then there exist $\{u_n^1\}_{n\in{\mathbb N}},\{u_n^2\}_{n\in{\mathbb N}}\subset
H^s_A(\mathbb{R}^3,\mathbb{C})$ such that ${\rm dist}({\rm supp}(u_n^1),{\rm supp}(u_n^2))\to+\infty$ and
\begin{align}
& \left|\|u_n^1\|_{s,A}^2-\beta\right|\leq \varepsilon, \label{1-est}\\
& \left|\|u_n^2\|_{s,A}^2-(L-\beta)\right|\leq \varepsilon, \label{2-est}\\
& \|u_n-u_n^1-u_n^2\|_{s,A}\leq \varepsilon, \label{3-est} \\
& \Big|1-\|u_n^1\|_{L^p(\mathbb{R}^3)}^p-\|u_n^2\|_{L^p(\mathbb{R}^3)}^p\Big|\leq \varepsilon \label{normep}
\end{align}
for any $n\ge \bar n$.
\end{lemma}
\begin{proof}
Notice that we have
\begin{equation}\label{mu1}
\begin{split}
\int_{\mathbb{R}^3}\mu_n^1dx &=\int_{B_{\bar R}(\xi_n)}|u_n|^2 dx+\int_{B_{\bar R}(\xi_n)\times B_{{\bar R}}(\xi_n)}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\quad+\int_{B_{\bar R}(\xi_n)\times B_{\bar R}^c(\xi_n)}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy,
\end{split}
\end{equation}
as well as
\begin{align*}
\int_{\mathbb{R}^3}\mu_n^2dx
&=\int_{B^c_{R_n}(\xi_n)}|u_n|^2 dx+\int_{B^c_{R_n}(\xi_n)\times B^c_{R_n}(\xi_n)}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\quad +\int_{B^c_{R_n}(\xi_n)\times B_{R_n}(\xi_n)}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy,
\end{align*}
and, from inequality \eqref{tre}, we have,
for $n\geq \bar n$,
\begin{align}
\label{piccola}
& \int_{\{\bar R\leq |x-\xi_n|\leq R_n\}\times \mathbb{R}^3}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\leq \varepsilon, \\
& \int_{\mathbb{R}^3\times \{\bar R\leq |y-\xi_n|\leq R_n\}}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\leq \varepsilon,
\label{piccola2} \\
\label{piccola3}
&\int_{\{\bar R\leq |x-\xi_n|\leq R_n\}}|u_n|^2 dx\leq \varepsilon.
\end{align}
For every $r>0$, let $\varphi_r\in C^\infty(\mathbb{R}^3)$ be a radially symmetric function such that
$\varphi_r=1$ on $B_r(0)$ and $\varphi_r=0$ su $B^c_{2r}(0)$.\
In light of Lemma~\ref{cut} applied with $E_1=E_2=\mathbb{R}^3$,
for any $n\in{\mathbb N}$, we can consider the functions
$$
u_n^1:=\varphi_{\bar{R}}(\cdot-\xi_n)u_n\in H^s_A(\mathbb{R}^3,\mathbb{C}),\qquad u_n^2:=(1-\varphi_{R_n/2}(\cdot-\xi_n))u_n\in H^s_A(\mathbb{R}^3,\mathbb{C}).
$$
We observe for further usage that the functions $\varphi_{\bar{R}}(\cdot-\xi_n)$ and $1-\varphi_{R_n/2}(\cdot-\xi_n)$
have a Lipschitz constant which is uniformly bounded with respect to $n$.
Moreover, ${\rm dist}({\rm supp}(u_n^1),{\rm supp}(u_n^2))\to\infty$.
Let us consider $\{u_n^1\}_{n\in{\mathbb N}}$. We have $[u_n^1]_{s,A}^2=\sum_{i=1}^5 I^i_n$, where
\begin{align*}
I^1_n& :=\int_{B_{\bar R}(\xi_n)\times B_{\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy, \\
I^2_n&:=\int_{B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)\times B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy, \\
I^3_n&:=2\int_{B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)\times B_{\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy, \\
I^4_n&:= 2\int_{B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)\times B^c_{2\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy, \\
I^5_n& :=2\int_{B_{\bar R}(\xi_n)\times B^c_{2\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy.
\end{align*}
Concerning $I_n^i$ with $i=2,3,4$, since for suitable measurable sets $E_2^i\subset\mathbb{R}^3$ and $c_i>0$,
\[
I^i_n= c_i \int_{B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)\times E^i_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy,
\]
in light of Lemma~\ref{cut} and inequalities \eqref{piccola}-\eqref{piccola3}, we have
\begin{equation}\label{234}
\begin{split}
I_n^i
& \leq C\left[\int_{B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)}|u_n|^2dx+
\int_{B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)\times E^i_2}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\right]\\
& \leq C\varepsilon,
\end{split}
\end{equation}
being $B_{2\bar R}(\xi_n)\setminus B_{\bar R}(\xi_n)\subset \{\bar R\leq |x-\xi_n|\leq R_n\}$ for every $n$ large enough.\\
Concerning $I_n^5$, we have
\begin{align*}
I^5_n & =2\int_{B_{\bar R}(\xi_n)\times \{2\bar R\leq |y-\xi_n|\leq R_n\}}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\quad +2\int_{B_{\bar R}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy.
\end{align*}
Then, arguing as in \eqref{234} for $I_n^i$ ($i=2,3,4$)
we get
$$
\int_{B_{\bar R}(\xi_n)\times \{2\bar R\leq |y-\xi_n|\leq R_n\}}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy \leq C\varepsilon,
$$
for large $n$. On the other hand, as far as the second term in concerned, we get
\begin{align*}
& \int_{B_{\bar R}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^1(x)-u_n^1(y)|^2}{|x-y|^{3+2s}}dxdy
=\int_{B_{\bar R}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|u_n(x)|^2}{|x-y|^{3+2s}}dxdy,
\end{align*}
since $u^1_n(y)=0$ for all $y\in B^c_{R_n}(\xi_n)$ and $u^1_n(x)=u_n(x)$ for all $x\in B_{\bar R}(\xi_n)$.
Observe first that if $(x,y)\in B_{\bar R}(\xi_n)\times B^c_{R_n}(\xi_n)$,
then $|x-y|\geq R_n-\bar R\to\infty$, as $n\to\infty$. We thus have
\begin{equation}\label{deltapiccolo}
\begin{split}
& \int_{B_{\bar R}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|u_n(x)|^2}{|x-y|^{3+2s}}dxdy \\
& \leq \frac{1}{(R_n-\bar R)^\delta}\int_{B_{\bar R}(\xi_n)}|u_n(x)|^2\left(\int_{\{|x-y|\geq 1\}}\frac{1}{|x-y|^{3+2s-\delta}}dy\right)dx
\leq \frac{C}{(R_n-\bar R)^\delta}\leq C\varepsilon,
\end{split}
\end{equation}
where $0<\delta<2s$. Here we have used the boundedness of $\{u_n\}_{n\in{\mathbb N}}$ in $L^2(\mathbb{R}^3,\mathbb{C})$.
So we have that $[u_n^1]_{s,A}^2= I^1_n +\varsigma_{n,\varepsilon}$ with $\varsigma_{n,\varepsilon}\leq C\varepsilon$
for $n$ large, which implies on account of \eqref{piccola3}
\begin{equation}
\label{ecco1}
\|u_n^1\|_{s,A}^2=\int_{B_{\bar R}(\xi_n)}|u_n|^2 dx+ I^1_n +\varsigma_{n,\varepsilon},\quad \varsigma_{n,\varepsilon}\leq C\varepsilon.
\end{equation}
A similar argument involving $\{u_n\}_{n\in{\mathbb N}}$ in place of $\{u_n^1\}_{n\in{\mathbb N}}$ shows that formula \eqref{mu1} writes as
\begin{equation}
\label{ecco2}
\int_{\mathbb{R}^3}\mu_n^1dx =
\int_{B_{\bar R}(\xi_n)}|u_n|^2 dx+I^1_n+\hat\varsigma_{n,\varepsilon},\quad \hat\varsigma_{n,\varepsilon}\leq C\varepsilon.
\end{equation}
Indeed, since
\begin{align*}
&\int_{B_{\bar R}(\xi_n)\times B_{\bar R}^c(\xi_n)}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\\
&\quad \leq
\int_{B_{\bar R}(\xi_n)\times \{\bar R\leq |y-\xi_n|\leq R_n\}}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\\
&\qquad + C\left[
\int_{B_{\bar R}(\xi_n)\times B_{R_n}^c(\xi_n)}
\frac{|u_n(x)|^2}{|x-y|^{3+2s}}dxdy
+\int_{B_{\bar R}(\xi_n)\times B_{R_n}^c(\xi_n)}
\frac{|u_n(y)|^2}{|x-y|^{3+2s}}dxdy
\right],
\end{align*}
by \eqref{piccola2} and arguing as in \eqref{deltapiccolo} we can conclude.
By combining \eqref{ecco1} and \eqref{ecco2} we finally obtain the desired estimate \eqref{1-est}.\\
Now, concerning $\{u_n^2\}_{n\in{\mathbb N}}$, we have
$[u_n^2]_{s,A}^2=\sum_{i=1}^5 J^i_n,$
where we have set
\begin{align*}
J^1_n& :=\int_{B^c_{R_n}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy, \\
J^2_n& :=\int_{B_{R_n}(\xi_n)\setminus B_{R_n/2}(\xi_n)\times B_{R_n}(\xi_n)
\setminus B_{R_n/2}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^2(x)-u_n^2(y)|^2}{|x-y|^{3+2s}}dxdy, \\
J^3_n&:=2\int_{B_{R_n}(\xi_n)\setminus B_{R_n/2}(\xi_n)\times B_{R_n/2}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^2(x)-u_n^2(y)|^2}{|x-y|^{3+2s}}dxdy, \\
J^4_n&:= 2\int_{B_{R_n}(\xi_n)\setminus B_{R_n/2}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^2(x)-u_n^2(y)|^2}{|x-y|^{3+2s}}dxdy, \\
J^5_n& :=2\int_{B_{R_n/2}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n^2(x)-u_n^2(y)|^2}{|x-y|^{3+2s}}dxdy.
\end{align*}
Concerning $J_n^i$ with $i=2,3,4$, observe that the integration domains are $B_{R_n}(\xi_n)\setminus B_{R_n/2}(\xi_n)\times E_2^i$, for suitable measurable $E_2^i$'s, and they are subset of $\{\bar R\leq |x-\xi_n|\leq R_n\}\times \mathbb{R}^3$ for $n$ sufficiently large. Thus we can argue as in \eqref{234}.
Finally,
$J^5_n$ can be estimated with similar arguments to that used in \eqref{deltapiccolo} and, as for $\mu_n^1$, using also \eqref{piccola3}, we obtain
\begin{equation*}
\int_{\mathbb{R}^3}\mu_n^2(x)dx
=\int_{B^c_{R_n}(\xi_n)}|u_n|^2 dx+J^1_n+\bar\varsigma_{n,\varepsilon},\quad \bar\varsigma_{n,\varepsilon}\leq C\varepsilon.
\end{equation*}
By combining all these estimates we get \eqref{2-est}
for any $n$ large.\\
Conclusion \eqref{3-est} follows
by \eqref{piccola}-\eqref{piccola3}. In fact, setting
$$
v_n:=u_n-u_n^1-u_n^2=(\varphi_{R_n/2}(\cdot-\xi_n)-\varphi_{\bar{R}}(\cdot-\xi_n))u_n,
$$
for all $n$, inequality \eqref{piccola3} yields
$$
\int_{\mathbb{R}^3}|v_n|^2dx=\int_{\mathbb{R}^3}(\varphi_{R_n/2}(x-\xi_n)-\varphi_{\bar{R}}(x-\xi_n) )^2|u_n|^2 dx
\leq\int_{\{\bar R\leq |x-\xi_n|\leq R_n\}}|u_n|^2 dx\leq \varepsilon.
$$
Furthermore, $[v_n]_{s,A}^2=\sum_{i=1}^4 K_n^i$, where
\begin{align*}
K^1_n&:=\int_{B_{R_n}(\xi_n)\setminus B_{\bar R}(\xi_n)\times B_{R_n}(\xi_n)\setminus B_{\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v_n(x)-v_n(y)|^2}{|x-y|^{3+2s}}dxdy, \\
K^2_n&:=2\int_{B_{R_n}(\xi_n)\setminus B_{\bar R}(\xi_n)\times B_{\bar R}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v_n(x)-v_n(y)|^2}{|x-y|^{3+2s}}dxdy, \\
K^3_n&:= 2\int_{B_{R_n}(\xi_n)\setminus B_{\bar R}(\xi_n)\times B^c_{R_n}(\xi_n)}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v_n(x)-v_n(y)|^2}{|x-y|^{3+2s}}dxdy.
\end{align*}
Since $v_n=\tilde\varphi u_n$ with
$\tilde\varphi:=(\varphi_{R_n/2}(\cdot-\xi_n)-\varphi_{\bar{R}}(\cdot-\xi_n))$, we can repeat
the arguments performed in \eqref{234}.
Concerning the final assertion \eqref{normep}, we have for some $\vartheta>0$,
\begin{align*}
1-\|u_n^1\|_{L^p}^p-\|u_n^2\|_{L^p}^p
& =
\int_{\mathbb{R}^3} \left(1-\varphi_{\bar{R}}^p(x-\xi_n)-(1-\varphi_{R_n/2}(x-\xi_n))^p\right)|u_n|^pdx \\
& \leq \int_{\{\bar R\leq |x-\xi_n|\leq R_n\}}|u_n|^p dx\\
& \leq \Big(\int_{\{\bar R\leq |x-\xi_n|\leq R_n\}}\!\!\!|u_n|^2 dx\Big)^{\frac{\vartheta p}{2}}
\Big(\int_{\mathbb{R}^3} |u_n|^{\frac{6}{3-2s}} dx\Big)^{\frac{(1-\vartheta)p(3-2s)}{6}}\\
&\leq\varepsilon,
\end{align*}
in light of \eqref{piccola3} and Lemma~\ref{embedd}.\ This concludes the proof.
\end{proof}
\begin{lemma}[Partial Gauge invariance]
\label{gauge2}
Let $\xi\in\mathbb{R}^3$ and $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$. For $\eta\in\mathbb{R}^3$, let us set
$$
v(x)=e^{\i \eta\cdot x} u(x+\xi),\quad x\in\mathbb{R}^3.
$$
Then $v\in H^s_{A_\eta}(\mathbb{R}^3,\mathbb{C})$ and
$$
\|u\|_{s,A}= \|v\|_{s,A_\eta},\quad\text{where $A_\eta:=A(\cdot+\xi)+\eta$}.
$$
\end{lemma}
\begin{proof}
Of course $\|v\|_{L^2}=\|u\|_{L^2}$.\ Moreover, a change of variables yields
\begin{align*}
\int_{\mathbb{R}^6}
\frac{|e^{-\i (x-y)\cdot A_\eta\left(\frac{x+y}{2}\right)}v(x)-v(y)|^2}{|x-y|^{3+2s}}dxdy
&=
\int_{\mathbb{R}^6}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}+\xi\right)} u(x+\xi)- u(y+\xi)|^2}{|x-y|^{3+2s}}dxdy\\
&=
\int_{\mathbb{R}^6}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)} u(x)- u(y)|^2}{|x-y|^{3+2s}}dxdy,
\end{align*}
which yields the assertion.
\end{proof}
\noindent
If $A$ is linear, then, taking $\eta=-A(\xi)$ in Lemma~\ref{gauge2}, we get $A_\eta=A$ and hence
\begin{lemma}[Partial Gauge invariance]
\label{gauge}
Let $\xi\in\mathbb{R}^3$ and $u\in H^s_A(\mathbb{R}^3,\mathbb{C})$. Assume that $A$ is linear and let us set
$$
v(x)=e^{-\i A(\xi)\cdot x} u(x+\xi),\quad x\in\mathbb{R}^3.
$$
Then $v\in H^s_A(\mathbb{R}^3,\mathbb{C})$ and $\|u\|_{s,A}= \|v\|_{s,A}$.
\end{lemma}
\section{Existence of minimizers}\label{sect4}\label{sec4}
\noindent
Let $2<p<6/(3-2s)$ and consider the minimization problem \eqref{MA}. First of all observe that by Sobolev embedding, ${\mathscr M}_A>0$.
Once a solution to \eqref{MA} exists,
due to the Lagrange Multiplier Theorem, there is $\lambda\in\mathbb{R}$ such that
\begin{align*}
& \frac{c_s}{2}{\Re}\int_{\mathbb{R}^6}\frac{\big(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)\big)\overline{\big(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v(x)-v(y)\big)}}{|x-y|^{3+2s}}dxdy \\
&+\Re\int_{\mathbb{R}^3} u\bar v dx=\lambda \Re\int_{\mathbb{R}^3} |u|^{p-2}u\bar v dx,\quad \text{for all $v\in H^s_A(\mathbb{R}^3,\mathbb{C})$}.
\end{align*}
A multiple of $u$ removes the Lagrange multiplier $\lambda$ and provides a
weak solution to \eqref{prob}.
Moreover, if we set
$$
{\mathscr M}_A(\lambda):=\inf_{u\in \mathscr{S}(\lambda)} \|u\|_{s,A}^2,
$$
where
$$
\mathscr{S}(\lambda):=\left\{u\in H^s_A(\mathbb{R}^3,\mathbb{C}):\int_{\mathbb{R}^3}|u|^p dx=\lambda\right\},
$$
we have that for every $\lambda>0$
\begin{equation}
\label{scala}
{\mathscr M}_A(\lambda)=\lambda^{\frac{2}{p}}{\mathscr M}_A.
\end{equation}
\subsection{Subcritical symmetric case}
\noindent
Let $2<p<\frac{6}{3-2s}$ and consider the problem
$$
{\mathscr M}_{A,r}=\inf_{u\in \mathscr{S}_r} \|u\|_{s,A}^2,
$$
where
$$
\mathscr{S}_r=\left\{u\in H^s_{A,{\rm rad}}(\mathbb{R}^3,\mathbb{C}): \int_{\mathbb{R}^3}|u|^p dx=1\right\}.
$$
First we give the following preliminary result.
\begin{lemma}[Compact radial embedding]
\label{comp-emb}
For every $2<q<6/(3-2s),$ the mapping
$$
H^s_{A,{\rm rad}}(\mathbb{R}^3,\mathbb{C})\ni u\mapsto |u|\in L^q(\mathbb{R}^3),
$$
is compact.
\end{lemma}
\begin{proof}
By Lemma~\ref{diamagnetic}, namely the
Diamagnetic inequality, we know that the mapping
$$
H^s_A(\mathbb{R}^3,\mathbb{C})\ni u\mapsto |u|\in H^s(\mathbb{R}^3,\mathbb{R}),
$$
is continuous. Then, the assertion follows directly by \cite[Theorem II.1]{Lions-c}.
\end{proof}
\noindent
We are ready to prove (\ref{i12}) of Theorem \ref{main}
\begin{theorem}[Existence of radial minimizers]
\label{Ex1-rad}
For any $2<p<6/(3-2s)$, the minimization problem ${\mathscr M}_{A,r}$
admits a solution.\ In particular, there exists a nontrivial
radially symmetric weak solution $u\in H^s_{A,{\rm rad}}(\mathbb{R}^3,\mathbb{C})$ to the problem \eqref{prob}.
\end{theorem}
\begin{proof}
Let $\{u_n\}_{n\in{\mathbb N}}\subset \mathscr{S}_r$ be a minimizing sequence for ${\mathscr M}_{A,r}$, namely $\|u_n\|_{L^p(\mathbb{R}^3)}=1$ for all $n$ and
$\|u_n\|_{s,A}^2\to {\mathscr M}_{A,r}$, as $n\to\infty$. Then, up to a subsequence, it converges weakly to some radial function $u$.\ On account
of Lemma~\ref{embedd}, $u_n\to u$ a.e.\ up to a subsequence.
By Lemma~\ref{comp-emb}, up to a subsequence
$\{|u_n|\}_{n\in{\mathbb N}}$ converges strongly to some $v$ in $L^{q}(\mathbb{R}^3)$
for every $2<q<6/(3-2s)$.\ Of course, $v=|u|$ by pointwise convergence.
In particular we can pass to the limit into the constraint $\|u_n\|_{L^p(\mathbb{R}^3)}=1$ to get $\|u\|_{L^p(\mathbb{R}^3)}=1$. Then $u$ is a solution to
${\mathscr M}_{A,r}$, since by virtue of Fatou Lemma
\begin{align*}
{\mathscr M}_{A,r}& \leq
\int_{\mathbb{R}^3} |u(x)|^2dx+
\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy \\
& \leq
\liminf_n \left(\int_{\mathbb{R}^3} |u_n(x)|^2dx+
\int_{\mathbb{R}^6}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\right)
={\mathscr M}_{A,r}.
\end{align*}
This concludes the proof.
\end{proof}
\subsection{Subcritical case}
\noindent
In this subsection we study the minimization problem \eqref{MA} in the case $2<p<\frac{6}{3-2s}$.
\subsubsection{Constant magnetic field case}
Let us consider \eqref{MA} under the assumption that $A:\mathbb{R}^3\to\mathbb{R}^3$ is linear. The local case was extensively
studied in \cite{EL} for the magnetic potential
$$
A(x_1,x_2,x_3)=\frac{b}{2} (-x_2,x_1,0),\quad b\in\mathbb{R}\setminus\{0\}.
$$
\noindent
Hence we can prove (\ref{iii12}) of Theorem \ref{main}.
\begin{theorem}[Existence of minimizers, I]
\label{Ex1}
Assume that the potential $A:\mathbb{R}^3\to\mathbb{R}^3$ is linear. Then, for any $2<p<\frac{6}{3-2s}$ the minimization problem \eqref{MA}
admits a solution.
\end{theorem}
\begin{proof}
Let $\{u_n\}_{n\in{\mathbb N}}\subset \mathscr{S}$ be a minimizing sequence for ${\mathscr M}_A$, namely $\|u_n\|_{L^p}=1$ for all $n$ and
$\|u_n\|_{s,A}^2\to {\mathscr M}_A$, as $n\to\infty$.\ We want to develop a concentration compactness argument \cite{L1}
on the measure of density defined by
$$
\mu_n(x):=|u_n(x)|^2+\int_{\mathbb{R}^3}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dy,\quad x\in\mathbb{R}^3,\,\, n\in{\mathbb N}.
$$
Notice that $\{\mu_n\}_{n\in{\mathbb N}}\subset L^1(\mathbb{R}^3)$ and, since $\|u_n\|_{s,A}^2={\mathscr M}_A+o_n(1)$,
$$
\sup_{n\in{\mathbb N}}\int_{\mathbb{R}^3} \mu_n(x)dx<\infty.
$$
More precisely, we shall apply \cite[Lemma I.1]{L1} by taking $\rho_n=\mu_n$.
Only vanishing, dichotomy or tightness (yielding compactness) are possible.\ Vanishing can be ruled out.
In fact, assume by contradiction that, for all $R>0$ fixed, there holds
$$
\lim_{n}\sup_{\xi\in\mathbb{R}^N} \int_{B_R(\xi)} \mu_n(x)dx=0,
$$
namely
$$
\lim_{n}\sup_{\xi\in\mathbb{R}^N} \left(\int_{B_R(\xi)} |u_n(x)|^2dx+
\int_{B_R(\xi)\times\mathbb{R}^3}\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy\right)=0.
$$
By Remark~\ref{diamremark} it follows that
$$
\lim_{n}\sup_{\xi\in\mathbb{R}^N} \left(\int_{B_R(\xi)} |u_n(x)|^2dx+
\int_{B_R(\xi)\times\mathbb{R}^3}\frac{||u_n(x)|-|u_n(y)||^2}{|x-y|^{3+2s}}dxdy\right)=0.
$$
In particular, we get
$$
\lim_{n}\sup_{\xi\in\mathbb{R}^N} \||u_n|\|_{H^s(B_R(\xi))}^2
=0
$$
and this implies, by virtue of Lemma~\ref{localized}, that for any $R>0$
$$
\lim_{n}\sup_{\xi\in\mathbb{R}^N} \int_{B_R(\xi)} |u_n(x)|^pdx=0.
$$
Thus, in light of Lemma~\ref{lionslemma}, $u_n\to 0$ in $L^p$ which violates
the constraint $\|u_n\|_{L^p}=1$. Whence, vanishing cannot occur.\\
We now exclude the dicothomy.
According to \cite[Lemma I.1]{L1}, this, precisely, means
that there exists $\beta\in (0,{\mathscr M}_A)$ such
that for all $\varepsilon>0$ there are $\bar R>0$, $\bar n\geq 1$,
a sequence of radii $R_n\to+\infty$ and
$\{\xi_n\}_{n\in{\mathbb N}}\subset \mathbb{R}^3$ such that for $n\geq \bar n$
\begin{align*}
& \left|\int_{\mathbb{R}^3}\mu_n^1(x)dx-\beta\right|\leq \varepsilon,\qquad
\mu_n^1(x):= 1_{B_{\bar R}(\xi_n)}\mu_n, \\
& \left|\int_{\mathbb{R}^3}\mu_n^2(x)dx-({\mathscr M}_A-\beta)\right|\leq \varepsilon,\qquad
\mu_n^2(x):= 1_{B^c_{R_n}(\xi_n)}\mu_n, \\
& \int_{\mathbb{R}^3}|\mu_n(x)-\mu_n^1(x)-\mu_n^2(x)|dx\leq \varepsilon.
\end{align*}
Then, by virtue of Lemma~\ref{lemma-dic},
there exist two sequences $\{u_n^1\}_{n\in{\mathbb N}},\{u_n^2\}_{n\in{\mathbb N}}\subset
H^s_A(\mathbb{R}^3,\mathbb{C})$ such that ${\rm dist}({\rm supp}(u_n^1),{\rm supp}(u_n^2))\to+\infty$ and
\begin{align}
& \left|\|u_n^1\|_{s,A}^2-\beta\right|\leq \varepsilon, \label{vicine-a} \\
& \left|\|u_n^2\|_{s,A}^2-({\mathscr M}_A-\beta)\right|\leq \varepsilon, \label{vicine-b}\\
& \left|1-\|u_n^1\|_{L^p}^p-\|u_n^2\|_{L^p}^p\right|\leq \varepsilon, \label{vicine-p}
\end{align}
for any $n\ge \bar n$. Up to a subsequence, in view of \eqref{vicine-p},
there exist $\vartheta_\varepsilon,\omega_\varepsilon\in (0,1)$ such that
$$
\|u_n^1\|_{L^p}^p=:\vartheta_{n,\varepsilon}\to\vartheta_\varepsilon,
\qquad
\|u_n^2\|_{L^p}^p=:\omega_{n,\varepsilon}\to \omega_\varepsilon,
\qquad
|1-\vartheta_{\varepsilon}-\omega_\varepsilon|\leq \varepsilon,\quad\text{as $n\to\infty$.}
$$
Notice that $\vartheta_\varepsilon$ does not converge to $1$ as $\varepsilon\to 0$, otherwise by \eqref{scala} and \eqref{vicine-a},
for $\varepsilon$ small we get
\begin{equation*}
\beta+\varepsilon
\geq \limsup_{n}\|u_n^1\|_{s,A}^2
\geq \limsup_n {\mathscr M}_A(\vartheta_{n,\varepsilon})
={\mathscr M}_A \vartheta_\varepsilon^{2/p}>\beta+\varepsilon.
\end{equation*}
Of course $\vartheta_\varepsilon$ does not converge to $0$ either, as $\varepsilon\to 0$, otherwise $\omega_\varepsilon\to 1$
and a contradiction would again follow by arguing as above on $u_n^2$ and using \eqref{vicine-b}.\
Whence, by means of \eqref{scala}, \eqref{vicine-a}, \eqref{vicine-b}, and since $\lambda^{2/p}+(1-\lambda)^{2/p}>1$ for any $\lambda\in(0,1)$,
if $\varepsilon$ is small enough
\begin{align*}
{\mathscr M}_A+2\varepsilon
&\geq \limsup_{n}\left(\|u_n^1\|_{s,A}^2+\|u_n^2\|_{s,A}^2\right)
\geq \limsup_n \left({\mathscr M}_A(\vartheta_{n,\varepsilon})+{\mathscr M}_A(\omega_{n,\varepsilon})\right)\\
&={\mathscr M}_A \left(\vartheta_\varepsilon^{2/p}+\omega_\varepsilon^{2/p}\right)>{\mathscr M}_A+2\varepsilon,
\end{align*}
a contradiction.\ This means that
tightness needs to occur, namely there exists
a sequence $\{\xi_n\}_{n\in{\mathbb N}}$ such that for all $\varepsilon>0$ there exists $R>0$ with
$$
\int_{B^c_R(\xi_n)} |u_n(x)|^2dx+
\int_{B^c_R(\xi_n)\times\mathbb{R}^3}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy<\varepsilon
$$
for any $n$. In particular, setting $\bar u_n(x):=u_n(x+\xi_n)$, for all $\varepsilon>0$ there is $R>0$ such that
\begin{equation}
\label{ext-d}
\sup_{n\in{\mathbb N}}\int_{B^c_R(0)} |\bar u_n(x)|^2dx<\varepsilon.
\end{equation}
Let us consider
$$
v_n(x):=e^{-\i A(\xi_n)\cdot x} \bar u_n(x),\quad x\in\mathbb{R}^3.
$$
Since, by Lemma~\ref{gauge}, $\|v_n\|_{s,A}=\|u_n\|_{s,A}$, we have that $\{v_n\}_{n\in{\mathbb N}}$ is bounded in $H^s_A(\mathbb{R}^3,\mathbb{C})$.
Notice also that, since $|v_n(x)|=|\bar u_n(x)|$ for a.e.\ $x\in\mathbb{R}^3$ and any $n\in{\mathbb N}$, by \eqref{ext-d} we have
that for all $\varepsilon>0$ there is $R>0$ such that
\begin{equation}\label{vn}
\sup_{n\in{\mathbb N}}\int_{B^c_R(0)} |v_n(x)|^2dx<\varepsilon.
\end{equation}
Thus, in view of the compact injection
provided by Lemma~\ref{embedd}, up to a subsequence, $\{v_n\}_{n\in{\mathbb N}}$ converges weakly,
strongly in $L^2(B_R(0),\mathbb{C})$ and point-wisely to some function $v$.
Moreover, by \eqref{vn},
it follows that
$v_n\to v$ strongly in $L^2(\mathbb{R}^3,\mathbb{C})$ as well as
in $L^q(\mathbb{R}^3,\mathbb{C})$ for any $2<q<6/(3-2s)$, via interpolation. Hence $\|v\|_{L^p}=1$.
Hence, by Fatou's lemma, we have
$$
{\mathscr M}_A\leq \|v\|^2_{s,A}\leq \liminf_n \|v_n\|_{s,A}^2=\liminf_n \|u_n\|_{s,A}^2={\mathscr M}_A,
$$
which proves the existence of a minimizer.
\end{proof}
\subsubsection{Variable magnetic field case}
We now prove (\ref{ii12}) of Theorem \ref{main}.
\begin{theorem}[Existence of minimizers, II]
\label{Ex2}
Assume that the potential $A:\mathbb{R}^3\to\mathbb{R}^3$ satisfies assumption $\mathscr{A}$ and that
\begin{equation}
\label{stretta}
{\mathscr M}_A<\inf_{\Xi\in \mathscr{X}} {\mathscr M}_{A_\Xi}.
\end{equation}
Then, for any $2<p<\frac{6}{3-2s},$ the minimization problem \eqref{MA}
admits a solution.
\end{theorem}
\begin{proof}
By arguing as in the proof of Theorem~\ref{Ex1}, if $\{u_n\}_{n\in{\mathbb N}}$ is a minimizing sequence for ${\mathscr M}_A$, we can
find a sequence $\{\xi_n\}_{n\in{\mathbb N}}$
such that for all $\varepsilon>0$ there exists $R>0$ with
$$
\int_{B^c_R(\xi_n)} |u_n(x)|^2dx+
\int_{B^c_R(\xi_n)\times\mathbb{R}^3}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy<\varepsilon
$$
for any $n$.
In particular, setting again $\bar u_n(x):=u_n(x+\xi_n)$, for all $\varepsilon>0$ there is $R>0$ such that
\begin{equation*}
\sup_{n\in{\mathbb N}}\int_{B^c_R(0)} |\bar u_n(x)|^2dx<\varepsilon.
\end{equation*}
Assume by contradiction that the sequence $\{\xi_n\}_{n\in{\mathbb N}}$ is unbounded. Then, since $A$ satisfies condition
${\mathscr A}$, there exists a sequence $\{H_n\}_{n\in{\mathbb N}}\subset\mathbb{R}^3$ such that \eqref{limA} holds. We thus consider
the sequence
$$
v_n(x):=e^{\i H_n\cdot x} \bar u_n(x), \quad x\in\mathbb{R}^3.
$$
By virtue of Lemma~\ref{gauge2} it follows that
$$
\sup_{n\in{\mathbb N}} \|v_n\|_{s,A_n}=\sup_{n\in{\mathbb N}}\|u_n\|_{s,A}<\infty,\quad A_n(x)=A(x+\xi_n)+H_n.
$$
Then, by combining Lemma~\ref{gen-bound} with
\begin{equation*}
\sup_{n\in{\mathbb N}}\int_{B^c_R(0)} |v_n(x)|^2dx<\varepsilon,
\end{equation*}
up to a subsequence, $\{v_n\}_{n\in{\mathbb N}}$ is strongly convergent in $L^q(\mathbb{R}^3)$ for all $q\in[2,6/(3-2s))$ to some function $v$
which satisfies the constraint $\|v\|_{L^p}=1$. By combining Lemma~\ref{gauge2} with Fatou's Lemma
and \eqref{stretta}, we get
\begin{align*}
{\mathscr M}_{A_\Xi} \leq \|v\|^2_{s,A_{\Xi}}
&= \int_{\mathbb{R}^3} |v|^2 dx
+\int_{\mathbb{R}^6}
\frac{|e^{-\i (x-y)\cdot A_{\Xi}\left(\frac{x+y}{2}\right)}v(x)-v(y)|^2}{|x-y|^{3+2s}}dxdy \\
&\leq \lim_n \int_{\mathbb{R}^3} |v_n|^2 dx
+\liminf_n \int_{\mathbb{R}^6}
\frac{|e^{-\i (x-y)\cdot A_n\left(\frac{x+y}{2}\right)}v_n(x)-v_n(y)|^2}{|x-y|^{3+2s}}dxdy \\
&= \lim_n \int_{\mathbb{R}^3} |u_n|^2 dx
+\liminf_n \int_{\mathbb{R}^6}
\frac{|e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u_n(x)-u_n(y)|^2}{|x-y|^{3+2s}}dxdy \\
&={\mathscr M}_{A}<\inf_{\Xi\in \mathscr{X}} {\mathscr M}_{A_\Xi}\leq {\mathscr M}_{A_\Xi},
\end{align*}
a contradiction.\ Therefore, it follows that $\{\xi_n\}_{n\in{\mathbb N}}$ is bounded. The assertion then
immediately follows arguing on the original sequence $\{u_n\}_{n\in{\mathbb N}}$.
\end{proof}
\subsection{Critical case}\label{subscrit}
\noindent
Let $D^s_A(\mathbb{R}^3,\mathbb{C})$ be the completion of $C^\infty_c(\mathbb{R}^3,\mathbb{C})$ with respect to the semi-norm $[\cdot]_{s,A}$.\
The functions of $D^s_A(\mathbb{R}^3,\mathbb{C})$ satisfy the Sobolev inequality stated in formula~\eqref{sobol}.\
The space $D^s_A(\mathbb{R}^3,\mathbb{C})$ is a real Hilbert space with respect to the scalar product
$$
(u,v)_{s,A}:=\frac{c_s}{2}{\Re}\int_{\mathbb{R}^6}\frac{\left(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}u(x)-u(y)\right)\overline{\left(e^{-\i (x-y)\cdot A\left(\frac{x+y}{2}\right)}v(x)-v(y)\right)}}{|x-y|^{3+2s}}dxdy.
$$
\noindent
We consider the minimization problem \eqref{MAC}.
Of course, by density, we have
$$
{\mathscr M}_A^c=\inf_{u\in \mathscr{S}^c\cap C^\infty_c(\mathbb{R}^3,\mathbb{C})} [u]_{s,A}^2,
\quad \,\,\,
{\mathscr M}_0^c=\inf_{u\in \mathscr{S}^c_0\cap C^\infty_c(\mathbb{R}^3,\mathbb{C})} [u]_{s,0}^2.
$$
where $\mathscr{S}^c_0=\{u\in D^s(\mathbb{R}^3,\mathbb{C}):\|u\|_{L^{6/(3-2s)}}=1\}$.
Moreover, since $[|u|]_{s,0}\leq [u]_{s,0}$, we have
\begin{equation}
\label{real-rap}
{\mathscr M}_0^c=\inf_{u\in \mathscr{S}^c_0\cap C^\infty_c(\mathbb{R}^3,\mathbb{R})} [u]_{s,0}^2.
\end{equation}
\begin{remark}\rm
\label{classif}
It is known \cite{classif,costav} that all the real valued fixed sign solutions to ${\mathscr M}_0^c$ are given by
\begin{equation*}
\label{Talentiane}
{\mathscr U}_{z,\varepsilon} (x)=d_{s}\left(\frac{\varepsilon}{\varepsilon^2+|x-z|^2}\right)^{\frac{3-2s}{2}}
\end{equation*}
for arbitrary $\varepsilon>0$, $z\in \mathbb{R}^3$ and that these are also the unique fixed sign solutions to
\begin{equation*}
(-\Delta)^s u=u^{\frac{3+2s}{3-2s}}\quad \text{in $\mathbb{R}^3$.}
\end{equation*}
\end{remark}
\noindent
We now prove the following crucial lemma
\begin{lemma}
\label{livelli}
It holds ${\mathscr M}_A^c={\mathscr M}_0^c$.
\end{lemma}
\begin{proof}
Let $\varepsilon>0$ and $u\in C^\infty_c(\mathbb{R}^3,\mathbb{R})$ be such that
$$
\int_{\mathbb{R}^3}|u|^{6/(3-2s)} dx=1,\quad \,\,
[u]_{s,0}^2\leq {\mathscr M}_0^c+\varepsilon,
$$
in light of formula~\eqref{real-rap} for ${\mathscr M}_0^c$.\ Consider now the scaling
$$
u_\sigma(x)=\sigma^{-\frac{3-2s}{2}}u\Big(\frac{x}{\sigma}\Big),\quad \sigma>0,\,\,\, x\in\mathbb{R}^3.
$$
It is readily checked that
$$
\int_{\mathbb{R}^3}|u_\sigma|^{6/(3-2s)} dx=\int_{\mathbb{R}^3}|u|^{6/(3-2s)} dx=1,\quad
[u_\sigma]_{s,0}=[u]_{s,0},\quad \text{for all $\sigma>0$.}
$$
There holds
that
\[
[u_\sigma]^2_{s,A} = \int_{\mathbb{R}^6}\frac{|e^{-\i \sigma (x-y)\cdot A\left(\sigma\frac{x+y}{2}\right)}u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy.
\]
Then, we compute
\begin{align*}
[u_\sigma]^2_{s,A}-[u]_{s,0}^2 &=\int_{\mathbb{R}^6}\frac{|e^{-\i \sigma (x-y)\cdot A\left(\sigma\frac{x+y}{2}\right)}u(x)-u(y)|^2-|u(x)-u(y)|^2}{|x-y|^{3+2s}}dxdy \\
& =\int_{\mathbb{R}^6} \varTheta_\sigma(x,y)dxdy=\int_{K\times K} \varTheta_\sigma(x,y)dxdy,
\end{align*}
where $K$ is the compact support of $u$ and
\begin{align*}
\varTheta_\sigma(x,y)
&:=\frac{2\Re\left(\Big(1-e^{-\i \sigma (x-y)\cdot A\left(\sigma\frac{x+y}{2}
\right)}\Big)u(x) u(y)\right)}{|x-y|^{3+2s}}\\
&=\frac{2\big(1-\cos\big(\sigma (x-y)\cdot A\big(\sigma\frac{x+y}{2}\big)\big)\big)u(x)u (y)}{|x-y|^{3+2s}},
\end{align*}
a.e. in $\mathbb{R}^6$.
Of course $\varTheta_\sigma(x,y)\to 0$ for a.e.\ $(x,y)\in\mathbb{R}^6$ as $\sigma\to 0$.
Since $A$ is locally
bounded then
\[
1-\cos\Big(\sigma (x-y)\cdot A\Big(\sigma\frac{x+y}{2}\Big)\Big) \leq C |x-y|^2
\quad x,y\in K.
\]
Therefore, since $u$ is bounded, it follows that for some $C>0$
\begin{align*}
& |\varTheta_\sigma(x,y)|\leq \frac{C}{|x-y|^{1+2s}},\quad\text{for $x,y\in K$ with $|x-y|<1$,} \\
& |\varTheta_\sigma(x,y)|\leq \frac{C}{|x-y|^{3+2s}},\quad\text{for $x,y\in K$ with $|x-y|\geq 1$.}
\end{align*}
Then overall, we have
$$
|\varTheta_\sigma(x,y)|\leq w(x,y),\quad w(x,y)=C\min\left\{\frac{1}{|x-y|^{1+2s}},\frac{1}{|x-y|^{3+2s}}\right\}\quad\text{for $x,y\in K$,}
$$
for a suitable constant $C>0$.
Notice that $w\in L^1(K\times K)$, since
\begin{align*}
\int_{K\times K} w(x,y) dxdy&=\int_{(K\times K)\cap\{|x-y|<1\}} w(x,y) dxdy+\int_{(K\times K)\cap\{|x-y|\geq 1\}} w(x,y) dxdy \\
& \leq C\int_{\{|z|<1\}} \frac{1}{|z|^{1+2s}} dz+C\int_{\{|z|\geq 1\}} \frac{1}{|z|^{3+2s}}dz<\infty.
\end{align*}
Then, by the Dominated Convergence Theorem, we obtain
$$
{\mathscr M}_A^c\leq \lim_{\sigma\to 0}[u_\sigma]^2_{s,A}=[u]_{s,0}^2 \leq {\mathscr M}_0^c+\varepsilon,
$$
hence ${\mathscr M}_A^c\leq {\mathscr M}_0^c$ by the arbitrariness of $\varepsilon$. Since the opposite
inequality is trivial through the Diamagnetic inequality, the desired assertion follows.
\end{proof}
\noindent
Thus we can prove (\ref{i13}) of Theorem \ref{main-2}.
\begin{theorem}[Representation of solutions]
Assume that ${\mathscr M}_A^c$ admits a solutions $u\in D^s_A(\mathbb{R}^3,\mathbb{C})$. Then there exist
$z\in\mathbb{R}^3$, $\varepsilon>0$ and a function $\vartheta_A:\mathbb{R}^3\to\mathbb{R}$ such that
$$
u(x)=d_{s}\left(\frac{\varepsilon}{\varepsilon^2+|x-z|^2}\right)^{\frac{3-2s}{2}}e^{\i \vartheta_A(x)},\quad x\in\mathbb{R}^3.
$$
\end{theorem}
\begin{proof}
If $u\in D^s_A(\mathbb{R}^3,\mathbb{C})$ is a solution to ${\mathscr M}_A^c$, then by the Diamagnetic inequality and Lemma~\ref{livelli},
$$
{\mathscr M}_A^c={\mathscr M}_0^c\leq [|u|]_{s,0}^2\leq [u]_{s,A}^2={\mathscr M}_A^c.
$$
Then, it follows that ${\mathscr M}_0^c=[|u|]_{s,0}^2$, which implies the assertion by Remark~\ref{classif}.
\end{proof}
\noindent
For a function $u\in D^s_A(\mathbb{R}^3,\mathbb{C})$ we define $\Upsilon_u^A:\mathbb{R}^6\to\mathbb{R}$ by setting
$$
\Upsilon_u^A(x,y):=2\Re\Big(|u(x)||u(y)|-e^{-\i (x-y)\cdot A\big(\frac{x+y}{2}\big)}u(x)\bar u(y)\Big),\quad
\text{a.e.\ in $\mathbb{R}^6$}.
$$
Finally we have
\begin{theorem}[Nonexistence]
\label{nonex}
Assume that for a function $u\in D^s_A(\mathbb{R}^3,\mathbb{C})$ we have
\begin{equation}
\label{Aass-00}
\Upsilon^A_u(x,y)>0\text{ on $E\subset\mathbb{R}^6$ with ${\mathcal L}^6(E)>0$.}
\end{equation}
Then $u$ cannot be a solution to problem ${\mathscr M}_A^c$.
\end{theorem}
\begin{proof}
For every $u\in D^s_A(\mathbb{R}^3,\mathbb{C})$ we have $|u|\in D^s(\mathbb{R}^3)$ and there holds
$$
[u]_{s,A}^2-[|u|]_{s,0}^2=\int_{\mathbb{R}^6}\Upsilon^A_u(x,y)dxdy.
$$
Assume by contradiction that $u$ solves ${\mathscr M}_A^c$.
Then, since $\|u\|_{L^{6/(3-2s)}}=1$, by Lemma \ref{livelli} and assumption
\eqref{Aass-00}, we conclude that ${\mathscr M}_0^c={\mathscr M}_A^c=[u]_{s,A}^2>[|u|]_{s,0}^2\geq {\mathscr M}_0^c,$
a contradiction.
\end{proof}
\noindent
As a consequence we get (\ref{ii13}) of Theorem \ref{main-2}.
\begin{corollary}[Nonexistence of constant phase solutions]
Assume that
$$
(x-y)\cdot A(x+y)\not \equiv k\pi,\quad\text{for some $k\in{\mathbb N}$ and on some $E\subset\mathbb{R}^6$ with ${\mathcal L}^6(E)>0$.}
$$
Then ${\mathscr M}_A^c$ does not admit solutions $u\in D^s_A(\mathbb{R}^3,\mathbb{C})$ of the form $u(x)=e^{\i \vartheta} v(x)$
for some $\theta\in\mathbb{R}$ and $v\in D^s_A(\mathbb{R}^3,\mathbb{R})$ of fixed sign.
\end{corollary}
\begin{proof}
Assumption \eqref{Aass-00} is fulfilled, since
$$
\Upsilon_u^A(x,y)=2\big(1-\cos\big((x-y)\cdot A\big(\frac{x+y}{2}\big)\big)\big)v(x)v (y)>0,\quad
\text{for a.e. $(x,y) \in E$}.
$$
Hence, the assertion follows from Theorem~\ref{nonex}.
\end{proof}
\bigskip
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 4,299
|
title: Installation
permalink: /docs/installation/
---
Getting Jekyll installed and ready-to-go should only take a few minutes.
If it ever becomes a pain, please [file an issue]({{ site.repository }}/issues/new)
(or submit a pull request) describing the issue you
encountered and how we might make the process easier.
### Requirements
Installing Jekyll should be straight-forward if all requirements are met.
Before you start, make sure your system has the following:
- GNU/Linux, Unix, or macOS
- [Ruby](https://www.ruby-lang.org/en/downloads/) version 2.0 or above, including all development
headers
- [RubyGems](https://rubygems.org/pages/download)
- [GCC](https://gcc.gnu.org/install/) and [Make](https://www.gnu.org/software/make/) (in case your system doesn't have them installed, which you can check by running `gcc -v` and `make -v` in your system's command line interface)
#### Only required for Jekyll 2 and earlier
- [NodeJS](https://nodejs.org/), or another JavaScript runtime (for CoffeeScript support).
- [Python 2.7](https://www.python.org/downloads/)
<div class="note info">
<h5>Problems installing Jekyll?</h5>
<p>
Check out the <a href="../troubleshooting/">troubleshooting</a> page or
<a href="{{ site.repository }}/issues/new">report an issue</a> so the
Jekyll community can improve the experience for everyone.
</p>
</div>
<div class="note info">
<h5>Running Jekyll on Windows</h5>
<p>
While Windows is not officially supported, it is possible to get Jekyll running
on Windows. Special instructions can be found on our
<a href="../windows/#installation">Windows-specific docs page</a>.
</p>
</div>
## Install with RubyGems
The best way to install Jekyll is via
[RubyGems](https://rubygems.org/pages/download). At the terminal prompt,
simply run the following command to install Jekyll:
```sh
$ gem install jekyll
```
All of Jekyll's gem dependencies are automatically installed by the above
command, so you won't have to worry about them at all.
<div class="note info">
<h5>Installing Xcode Command-Line Tools</h5>
<p>
If you run into issues installing Jekyll's dependencies which make use of
native extensions and are using macOS, you will need to install Xcode
and the Command-Line Tools it ships with. Download them in
<code>Preferences → Downloads → Components</code>.
</p>
</div>
## Pre-releases
In order to install a pre-release, make sure you have all the requirements
installed properly and run:
```sh
gem install jekyll --pre
```
This will install the latest pre-release. If you want a particular pre-release,
use the `-v` switch to indicate the version you'd like to install:
```sh
gem install jekyll -v '2.0.0.alpha.1'
```
If you'd like to install a development version of Jekyll, the process is a bit
more involved. This gives you the advantage of having the latest and greatest,
but may be unstable.
```sh
$ git clone git://github.com/jekyll/jekyll.git
$ cd jekyll
$ script/bootstrap
$ bundle exec rake build
$ ls pkg/*.gem | head -n 1 | xargs gem install -l
```
## Optional Extras
There are a number of (optional) extra features that Jekyll supports that you
may want to install, depending on how you plan to use Jekyll. These extras
include LaTeX support, and the use of alternative content rendering engines.
Check out [the extras page](../extras/) for more information.
<div class="note">
<h5>ProTip™: Enable Syntax Highlighting</h5>
<p>
If you're the kind of person who is using Jekyll, then chances are you'll
want to enable syntax highlighting using <a href="http://pygments.org/">Pygments</a>
or <a href="https://github.com/jayferd/rouge">Rouge</a>. You should really
<a href="../templates/#code-snippet-highlighting">check out how to
do that</a> before you go any farther.
</p>
</div>
## Already Have Jekyll?
Before you start developing with Jekyll, you may want to check that you're up to date with the latest version. To find your version of Jekyll, run one of these commands:
```sh
$ jekyll --version
$ gem list jekyll
```
You can also use [RubyGems](https://rubygems.org/gems/jekyll) to find the current versioning of any gem. But you can also use the `gem` command line tool:
```sh
$ gem search jekyll --remote
```
and you'll search for just the name `jekyll`, and in brackets will be latest version. Another way to check if you have the latest version is to run the command `gem outdated`. This will provide a list of all the gems on your system that need to be updated. If you aren't running the latest version, run this command:
```sh
$ gem update jekyll
```
Now that you've got everything up-to-date and installed, let's get to work!
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 2,233
|
{"url":"https:\/\/labs.tib.eu\/arxiv\/?author=M.%20van%20den%20Akker","text":"\u2022 ### Measuring a Cherenkov ring in the radio emission from air showers at 110-190 MHz with LOFAR(1411.6865)\n\nNov. 25, 2014 astro-ph.IM, astro-ph.HE\nMeasuring radio emission from air showers offers a novel way to determine properties of the primary cosmic rays such as their mass and energy. Theory predicts that relativistic time compression effects lead to a ring of amplified emission which starts to dominate the emission pattern for frequencies above ~100 MHz. In this article we present the first detailed measurements of this structure. Ring structures in the radio emission of air showers are measured with the LOFAR radio telescope in the frequency range of 110 - 190 MHz. These data are well described by CoREAS simulations. They clearly confirm the importance of including the index of refraction of air as a function of height. Furthermore, the presence of the Cherenkov ring offers the possibility for a geometrical measurement of the depth of shower maximum, which in turn depends on the mass of the primary particle.\n\u2022 ### LORA: A scintillator array for LOFAR to measure extensive air showers(1408.4469)\n\nThe measurement of the radio emission from extensive air showers, induced by high-energy cosmic rays is one of the key science projects of the LOFAR radio telescope. The LOfar Radboud air shower Array (LORA) has been installed in the core of LOFAR in the Netherlands. The main purpose of LORA is to measure the properties of air showers and to trigger the read-out of the LOFAR radio antennas to register extensive air showers. The experimental set-up of the array of scintillation detectors and its performance are described.\n\u2022 ### The shape of the radio wavefront of extensive air showers as measured with LOFAR(1404.3907)\n\nJune 8, 2014 astro-ph.IM, astro-ph.HE\nExtensive air showers, induced by high energy cosmic rays impinging on the Earth's atmosphere, produce radio emission that is measured with the LOFAR radio telescope. As the emission comes from a finite distance of a few kilometers, the incident wavefront is non-planar. A spherical, conical or hyperbolic shape of the wavefront has been proposed, but measurements of individual air showers have been inconclusive so far. For a selected high-quality sample of 161 measured extensive air showers, we have reconstructed the wavefront by measuring pulse arrival times to sub-nanosecond precision in 200 to 350 individual antennas. For each measured air shower, we have fitted a conical, spherical, and hyperboloid shape to the arrival times. The fit quality and a likelihood analysis show that a hyperboloid is the best parametrization. Using a non-planar wavefront shape gives an improved angular resolution, when reconstructing the shower arrival direction. Furthermore, a dependence of the wavefront shape on the shower geometry can be seen. This suggests that it will be possible to use a wavefront shape analysis to get an additional handle on the atmospheric depth of the shower maximum, which is sensitive to the mass of the primary particle.\n\u2022 ### Detecting cosmic rays with the LOFAR radio telescope(1311.1399)\n\nNov. 6, 2013 astro-ph.IM, astro-ph.HE\nThe low frequency array (LOFAR), is the first radio telescope designed with the capability to measure radio emission from cosmic-ray induced air showers in parallel with interferometric observations. In the first $\\sim 2\\,\\mathrm{years}$ of observing, 405 cosmic-ray events in the energy range of $10^{16} - 10^{18}\\,\\mathrm{eV}$ have been detected in the band from $30 - 80\\,\\mathrm{MHz}$. Each of these air showers is registered with up to $\\sim1000$ independent antennas resulting in measurements of the radio emission with unprecedented detail. This article describes the dataset, as well as the analysis pipeline, and serves as a reference for future papers based on these data. All steps necessary to achieve a full reconstruction of the electric field at every antenna position are explained, including removal of radio frequency interference, correcting for the antenna response and identification of the pulsed signal.\n\u2022 LOFAR, the LOw-Frequency ARray, is a new-generation radio interferometer constructed in the north of the Netherlands and across europe. Utilizing a novel phased-array design, LOFAR covers the largely unexplored low-frequency range from 10-240 MHz and provides a number of unique observing capabilities. Spreading out from a core located near the village of Exloo in the northeast of the Netherlands, a total of 40 LOFAR stations are nearing completion. A further five stations have been deployed throughout Germany, and one station has been built in each of France, Sweden, and the UK. Digital beam-forming techniques make the LOFAR system agile and allow for rapid repointing of the telescope as well as the potential for multiple simultaneous observations. With its dense core array and long interferometric baselines, LOFAR achieves unparalleled sensitivity and angular resolution in the low-frequency radio regime. The LOFAR facilities are jointly operated by the International LOFAR Telescope (ILT) foundation, as an observatory open to the global astronomical community. LOFAR is one of the first radio observatories to feature automated processing pipelines to deliver fully calibrated science products to its user community. LOFAR's new capabilities, techniques and modus operandi make it an important pathfinder for the Square Kilometre Array (SKA). We give an overview of the LOFAR instrument, its major hardware and software components, and the core science objectives that have driven its design. In addition, we present a selection of new results from the commissioning phase of this new radio observatory.\n\u2022 ### LOFAR: Detecting Cosmic Rays with a Radio Telescope(1109.5805)\n\nSept. 27, 2011 astro-ph.IM, astro-ph.HE\nLOFAR (the Low Frequency Array), a distributed digital radio telescope with stations in the Netherlands, Germany, France, Sweden, and the United Kingdom, is designed to enable full-sky monitoring of transient radio sources. These capabilities are ideal for the detection of broadband radio pulses generated in cosmic ray air showers. The core of LOFAR consists of 24 stations within 4 square kilometers, and each station contains 96 low-band antennas and 48 high-band antennas. This dense instrumentation will allow detailed studies of the lateral distribution of the radio signal in a frequency range of 10-250 MHz. Such studies are key to understanding the various radio emission mechanisms within the air shower, as well as for determining the potential of the radio technique for primary particle identification. We present the status of the LOFAR cosmic ray program, including the station design and hardware, the triggering and filtering schemes, and our initial observations of cosmic-ray-induced radio pulses.","date":"2021-04-13 09:57:29","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\": 0.26461121439933777, \"perplexity\": 1713.2924988705915}, \"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-2021-17\/segments\/1618038072180.33\/warc\/CC-MAIN-20210413092418-20210413122418-00394.warc.gz\"}"}
| null | null |
Anna Massey Lea Merritt (1844-1930) est une artiste américaine connue pour ses peintures de portraits et de paysages. Née en 1844 à Philadelphie, elle accompagne sa famille pour vivre en Europe en 1867. Elle étudie brièvement à Dresde en Allemagne et à Florence en Italie, avant de s'installer durablement en Angleterre, où elle épouse un professeur de Londres qui la laisse bientôt veuve. Elle s'installe alors dans un petit village anglais. Son œuvre Eve Overcome with Remorse est le premier tableau réalisé par une femme dont le gouvernement britannique fait l'acquisition.
Liens externes
Peintre américain du XIXe siècle
Peintre américain du XXe siècle
Naissance en septembre 1844
Naissance à Philadelphie
Décès en avril 1930
Décès à 85 ans
Peintre américaine
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 5,601
|
Pachygone är ett släkte av tvåhjärtbladiga växter. Pachygone ingår i familjen Menispermaceae.
Kladogram enligt Catalogue of Life:
Bildgalleri
Källor
Externa länkar
Ranunkelordningen
Pachygone
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 3,383
|
{"url":"https:\/\/www.sys-ele.com\/CN\/abstract\/abstract1575.shtml","text":"\u2022 \u4f20\u611f\u5668\u4e0e\u4fe1\u53f7\u5904\u7406 \u2022\n\n\u9762\u5411SAR\u56fe\u50cf\u76ee\u6807\u8bc6\u522b\u7684\u9c81\u68d2\u5904\u7406\u7b97\u6cd5\n\n1. 1. \u5357\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66\u81ea\u52a8\u5316\u5b66\u9662\uff0c \u6c5f\u82cf \u5357\u4eac 210016\uff1b \ue004\n2. \u7a7a\u519b\u9a7b\u4eac\u660c\u5730\u533a\u519b\u4e8b\u4ee3\u8868\u5ba4\uff0c \u5317\u4eac 100041\uff1b\ue004\n3. \u5357\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66\u822a\u5929\u5b66\u9662\uff0c \u6c5f\u82cf \u5357\u4eac 210016\uff1b \ue004\n4. \u5357\u4eac\u822a\u7a7a\u822a\u5929\u5927\u5b66\u6c11\u822a\u5b66\u9662\uff0c \u6c5f\u82cf \u5357\u4eac 210016\n\u2022 \u51fa\u7248\u65e5\u671f:2013-12-24 \u53d1\u5e03\u65e5\u671f:2010-01-03\n\nRobust processing algorithm for SAR image target recognition\n\nLIU Zhong-jie1\uff0c2\uff0c CAO Yun-feng3\uff0c ZHUANG Li-kui3\uff0c DING Meng4\uff0c WANG Xi-chao1\n\n1. 1. College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China;\ue004\n2.Air Force Military Representative Office in Jingchang District, Beijing 100041, China;\ue004\n3. College of Astronautics, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China; \ue004\n4. College of Civil Aviation, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China\n\u2022 Online:2013-12-24 Published:2010-01-03\n\nAbstract:\n\nWith the existing synthetic aperture radar (SAR) image target recognition algorithm, image preprocessing has to be usually carried out. Preprocessing has a significant impact on the recognition rate. However, the adaptability of the preprocessing algorithm is difficult to be guaranteed. This paper proposes to apply the theory of kernel principal component analysis (KPCA) and sparse representation to the image to be recognited, thus achieving a target recognition result possessing a\u00a0 high recognition rate with only a few observation data and saving the data storage and computation. This paper describes the basic theory of compressed sensing first, and proposes an SAR image target recognition algorithm based on KPCA and sparse representation. An experiment is carried out with five kinds of SAR targets in the MSTAR database. The simulation results show that this proposed algorithm is still able to recognize the target effectively without prediction of the attitude angle. Compared with other recognition algorithms, it has a higher recognition rate to the image under the noise pollution on an equal basis.","date":"2023-03-20 16:34:03","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\": 0.4251096844673157, \"perplexity\": 2650.09743220302}, \"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-2023-14\/segments\/1679296943484.34\/warc\/CC-MAIN-20230320144934-20230320174934-00130.warc.gz\"}"}
| null | null |
Blighty: Jeet Kune Do Arrest and Catch
Release:Mar 05, 2012
Size:863.26 Mb
☆☆☆☆☆ Blighty: Jeet Kune Do Arrest and Catch ☆☆☆☆☆
★This application is currently in Chinese, with Chinese narration and Chinese Words.★
This app includes 23 paragraphs to explain that the basic steps' method of Jeet Kune Do. There have masters to demonstrate the gestures, and point out how to grasp the essentials. You can get the key to combat.
Jeet Kune Do is a hybrid martial arts system and life philosophy founded by martial artist Bruce Lee with direct, non-classical and straightforward movements. Due to the way his style works they believe in minimal movement with maximum effect and extreme speed.
The system works on the use of different 'tools' for different situations. These situations are broken down into ranges (Kicking, Punching, Trapping and Grappling), with techniques flowing smoothly between them. It is referred to as a "style without style".
Unlike more traditional martial arts, Jeet Kune Do is not fixed or patterned, and is a philosophy with guiding thoughts. It was named for the concept of interception, or attacking your opponent while he is about to attack. However, the name Jeet Kune Do was often said by Bruce Lee to be just a name. Through his studies Bruce came to believe that styles had become too rigid, and unrealistic. He called martial art competitions of the day "Dry land swimming". He believed that combat was spontaneous, and that a martial artist cannot predict it, only react to it, and that a good martial artist should "Be like water" and move fluidly without hesitation.
CONTACT US: blightystudio@gmail.com
VIST US: www.blightystudio.com
FOLLOW US: www.twitter.com/blightystudio
LIKE US: www.facebook.com/blightystudio
WATCH US: www.youtube.com/blightystudio
Your 5 Star Ratings keep us developing other great and fun apps.
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 8,682
|
Constantinos Daskalakis (* 29. April 1981 in Athen) ist ein griechischer theoretischer Informatiker.
Daskalakis studierte an der Nationalen Technischen Universität in Athen mit dem Diplom in Elektrotechnik und Informatik 2004 und an der University of California, Berkeley, an der er 2008 bei Christos Papadimitriou promoviert wurde (The complexity of Nash equilibria). Als Post-Doktorand war er bei Jennifer Chayes bei Microsoft Research. Die Dissertation erhielt den ACM Dissertation Award. Seit 2015 ist er Professor am Massachusetts Institute of Technology (MIT) und am dortigen CSAIL.
Daskalakis befasst sich mit Komplexitätstheorie im Schnittfeld mit Spieltheorie, Wirtschaftswissenschaften, Stochastik und Maschinenlernen. Er löste lange offene Probleme über die Komplexität von Nash-Gleichgewichten in der Spieltheorie, die Komplexität von Auktionen und das Verhalten von Methoden des Maschinenlernens wie den Expectation-Maximizaton Algorithmus. Er fand auch effiziente Methoden für das Testen statistischer Hypothesen, Maschinenlernen in hohen Dimensionen und erzielte Resultate zur Struktur und den Konzentrationseigenschaften hochdimensionaler Verteilungen.
2010 war er Sloan Research Fellow und 2012 erhielt er die Microsoft Research Faculty Fellowship. 2008 erhielt er zusammen mit Paul W. Goldberg und Christos Papadimitriou den Kalai-Preis der Game Theory Society. 2018 erhielt er den Nevanlinna-Preis, den Google Faculty Research Award und den Simons Foundation Investigator Award. Ebenfalls für 2018 wurde ihm der Grace Murray Hopper Award zugesprochen.
Schriften (Auswahl)
mit C. Papadimitriou: Three-player games are hard, Electronic colloquium on computational complexity, Band 139, 2005, S. 81–87
mit E. Mossel, S. Roch: Optimal phylogenetic reconstruction, Proceedings of the 38. Annual ACM Symposium on Theory of Computing (STOC), 2006 (und Probability Theory and Related Fields, Band 149, 2011, S. 149–189)
mit Paul Goldberg, Christos Papadimitriou: The Complexity of Computing a Nash Equilibrium, 38th ACM Symposium on Theory of Computing (STOC), 2006, sowie SIAM Journal on Computing, Band 39, 2009, S. 195–259
mit Goldberg, Papadimitriou: The complexity of computing a Nash equilibrium, Communications of the ACM, Band 52, Nr. 2, 2009, S. 89–97
mit A. Mehta, C. Papadimitriou: Progress in approximate Nash equilibria, Proceedings of the 8th ACM Conference on Electronic Commerce, 2007, S. 355–358
mit A. Mehta, C. Papadimitriou: A note on approximate Nash equilibria, Theoretical Computer Science, Band 410, 2009, S. 1581–1588
mit Y. Cai, S. M. Weinberg: Optimal multi-dimensional mechanism design: Reducing revenue to welfare maximization, 53. Annual Symposium on Foundations of Computer Science (FOCS), 2012
Weblinks
Homepage
Erica Klarreich, A Poet of Computation Who Uncovers Distant Truths, Quanta Magazine 1. August 2018
Einzelnachweise
Informatiker
Hochschullehrer (Massachusetts Institute of Technology)
Grieche
Geboren 1981
Mann
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 1,760
|
Alfonso de Ávalos i d'Aquino també anomenat Alonso de Ávalos (Ischia, 1502 - 31 de març de 1546) fou un militar i Condottiero del Regne de Nàpols, d'origen espanyol, al servei de l'Emperador Carles V. Va ser marquès de Pescara i del Vasto.
El 1523 es va casar amb Maria d'Aragó, filla de Ferran d'Aragó i Guardato, duc de Montalto, i de Castellana Folch de Cardona. El 1529 va ser nomenat senyor d'Ischia i Procida, després que la família Cossa prengués partit per la causa francesa. Entre 1538 i 1546 va ser Governador del Milanesat.
Nebot de Fernando de Ávalos, Alfons va succeir al seu oncle en el comandament de les tropes de l'Emperador Carles V. El 1532 combatre en Àustria contra les tropes de Solimà el Magnífic i va participar en les diferents expedicions de Carles V, com la jornada de Tunis i la batalla de Pavia i va comandar l'exèrcit imperial a la Guerra d'Itàlia de 1542–46, en la que el 1543 va aconseguir aixecar el setge de Niça de Khair ed-Din Barba-rossa i Francesc de Borbó, però a la batalla de Ceresole de 1544, va ser derrotat per les tropes franceses a les ordres del mateix Francesc de Borbó.
Referències
Bibliografia
Charles Oman, A History of the Art of War in the Sixteenth Century Londres: Methuen & Co(1937)
Militars del Regne de Nàpols
Governadors del Milanesat
Persones de la ciutat metropolitana de Nàpols
Morts a la Llombardia
Militars de la Campània
Morts el 1546
Naixements del 1502
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 9,866
|
Q: If $b\mid ca$, then $b\mid a$. Is this true? My proof: We want to show $b\mid a$ i.e. $a = bn$ for some integer $n$. Since $b\mid ca$, $ca = bm$ for some integer $m$. Substituting for $a$ gives us $c(bn) = bm \Rightarrow b(cn) = bm\dots$
After that I get stuck. I get the feeling that this may be false.
A: Hint:
$$4\mid 2\cdot 6\implies 4\mid 6...?$$
A: False. Take $b=2$, $a=3$, and $c=0$. Or $c=2$. Or $c$ any even number.
What is true is: if $(b,c)=1$ (i.e. $b$ and $c$ are relatively prime) and if $b\mid ac$ then $b\mid a$.
A: You are correct, in that this is false. In your proof, you are assuming what you want to prove:
You wrote "Substituting for $a$ gives us $c(bn)=bm$," where you used the equation $a=bn$ for your substitution. But $a=bn$ is exactly what you want to prove!
A: You are correct, the proposition is false:
To prove a claim is false, use proof by counterexample (for which you need to provide only one counterexample, and you are done!)
Consider the obvious counter examples where $b \mid c$ but $b \not\mid a$: in these cases we would still have $b\mid ca$, but only because $b \mid c$.
Consider, e.g. $b = 2, c = 4, a = 5$: $b\mid ca$ but $b\not\mid a$.
Simpler yet, the counterexample $b = 2, c = 2, a = 3$ suffices.
|
{
"redpajama_set_name": "RedPajamaStackExchange"
}
| 789
|
Huang Shixiang (born 1934) is a Chinese Beijing opera singer best known for his performances in "Jade River" Taijun Cichao (太君辭朝) and Hongzong Liema (紅鬃烈馬). His father, Pugong, was the son of Zaixun, a Manchu prince of the late Qing dynasty. Had the empire not been abolished, this would have made him fourth in line to the throne as of 2023, behind Jin Yuzhang and his brothers.
Huang has gone to some lengths to conceal his imperial ancestry. Rather than his father's family name Aisin Gioro or its Chinese form Jin, he goes by the maiden name of his mother Huang Yongni (stage name Xue Yanqin), who was also a famous Beijing opera actress. He is likewise registered as a member of his mother's Hui Chinese ethnic group rather than as a Manchu.
References
Living people
Manchu male actors
Manchu singers
1955 births
20th-century Chinese male actors
21st-century Chinese male actors
20th-century Chinese male singers
21st-century Chinese male singers
Male actors from Beijing
Singers from Beijing
Chinese male Peking opera actors
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 5,234
|
\section{Introduction}
\subsection{Background}
The Schr\"odinger operator on $\Rd$ is given by
\eqx{
\wt{\ll} f = -\Delta f + \vv\cdot f,
}
where $\Delta$ is the Laplace operator and $\vv$ is a function called a potential. If we assume that $\vv\in L^1_{loc}(\Rd)$ and $\vv\geq 0$ then one can find a densely defined, self-adjoint operator $\ll$ on $L^2(\Rd)$, that corresponds to $\wt{\ll}$. It is well known that $\ll$ generates the semigroup of contractions $\kk_t = \exp(-t\ll)$ and $\kk_t$ admits an integral kernel $\kk_t(x,y)$ such that
\alx{
&\kk_t f(x) = \int_\Rd \kk_t(x,y) f(y) dy,\\
0\leq &\kk_t(x,y) \leq (4\pi t)^{-d/2} \exp\eee{-\frac{|x-y|^2}{4t}}.
}
There have been wide studies on harmonic analysis related to Schr\"odinger operators and, more generally, operators with Gaussian bounds. We refer the reader to \cite{Auscher_Ali}, \cite{ADM}, \cite{Bernicot_Zhao}, \cite{Bui}, \cite{CzajaZienkiewicz_ProcAMS}, \cite{Deng}, \cite{Duong_Yan}, \cite{DP_Argentina}, \cite{DP_Arkiv}, \cite{DP_Potential}, \cite{DZ_Studia2}, \cite{DZ_Annali}, \cite{DZ_Revista2}, \cite{DZ_JFAA}, \cite{DZ_Potential_2014}, \cite{Grafakos_Hardy}, \cite{Hofmann_memoirs}, \cite{Shen}, and references therein. In particular, the Hardy spaces
\eq{\label{defff}
H^1(\ll) = \set{f \in L^1(\Rd) \ : \ \norm{f}_{H^1(\ll)} := \norm{\sup_{t>0} \abs{\kk_t f} }_{L^1(\Rd)}<\8}
}
related to $\ll$ were intensively studied. At this point let us mention that the classical Hardy space $H^1(-\Delta)$ has many equivalent definitions, e.g. in terms of: various maximal functions, singular integrals, square functions, etc. A particulary useful result is the atomic decomposition theorem (see \cite{CoifmanWeiss_BullAMS}, \cite{Latter_Studia}): a~function $f\in H^1(-\Delta)$ can be decomposed as $f(x) = \sum_k \la_k a_k(x)$, where $\sum_k |\la_k| \simeq \norm{f}_{H^1(-\Delta)}$ and $a_k$ are classical atoms, that is, there exist balls $B_k$ such that:
\eqx{\supp \, a_k \subseteq B_k,\quad \norm{a_k}_\8\leq |B_k|^{-1},\quad \int a(x) dx = 0.}
In other words atoms satisfy some localization, size, and cancellation conditions.
Let us mention that Hofmann et al. \cite{Hofmann_memoirs} have found general results (for $\vv$ satisfying $0\leq \vv\in L^1_{loc}(\Rd)$) saying that $H^1(\ll)$ given above is equal to the Hardy spaces via: square functions, atomic or molecular decompositions. However, atoms used in \cite{Hofmann_memoirs} are given in terms of $\ll$, to be more precise: a function $a\in L^2(\Rd)$ is an atom if there exist a ball $B$ and $b \in \mathrm{Dom}(\ll)$, such that: $a=\ll b$, $\supp \, b \subseteq B$ and $b, \ll b$ satisfy some size condition.
An another approach, started by Dziuba\'nski and Zienkiewicz in the 90's, was to find atomic spaces with simple geometric conditions that characterize $H^1(\ll)$. It appeared that this cannot be done in full generality, and~the properties of atoms depend strictly on the potential $\vv$ and the dimension~$d$. For example, if $d\geq 3$, $\vv\in C_c^\8(\Rd)$, and $\vv\not\equiv 0$, then atoms have modified cancellation condition $\int a(x) \omega(x) \, dx=0$, where $\omega$~is such that $0<C^{-1}\leq \omega(x) \leq C$. For this result and generalizations see \cite{DZ_Annali}, \cite{DZ_JFAA}, \cite{Preisner2}. Other results, see \cite{DZ_Studia2}, lead to Hardy spaces with local atoms. It was first observed by Goldberg \cite{Goldberg_Duke} that if we take supremum for $0<t\leq \tau^2$ in \eqref{defff}, then one obtains atomic space with classical atoms complemented with the atoms of~the form $|B|^{-1}\mathbbm{1}_B(x)$, where the ball $B$ has radius $\tau$. In \cite{DZ_Studia2} the authors assume that for $0\leq \vv \in L^1_{loc}(\Rd)$ there exists a family of cubes $\qq = \set{Q_k \ : \ k\in \NN}$ such that
\eqx{
\overline{\cup_k Q_k} = \Rd, \quad |Q_k\cap Q_j|=0 \text{ for }k\neq j, \quad d(Q_k)\simeq d(Q_j) \text{ if } Q_k^{***}\cap Q_j^{***} \neq \emptyset.
}
Here $d(Q)$ is the diameter of $Q$ and $Q^*$ is a cube that has the same center as $Q$ but with slightly enlarged diameter. The atomic space $H^1_{at}(\qq)$ is built on classical atoms and atoms of the form $|Q_k|^{-1} \mathbbm{1}_{Q_k}(x)$. The main result of \cite{DZ_Studia2} states that under two additional assumptions on $V,\qq,\kk_t$ (see \cite[p.41]{DZ_Studia2}, conditions: $(D)$, $(K)$) we have that $H^1(\ll) = H^1_{at}(\qq)$, see \cite[Thm. 2.2]{DZ_Studia2}. In other words the atoms for $H^1(\ll)$ are either classical atoms or local atoms related to some $Q_k \in \qq$.
Among examples of atoms for which one can find a family $\qq$ such that the assumptions of \cite[Thm. 2.2]{DZ_Studia2} are satisfied, there are potentials $\vv$ in the Reverse H\"older class in dimension $d\geq 3$. For more examples see \cite{DZ_Studia2}. Later, Czaja and Zienkiewicz \cite[Thm. 2.4]{CzajaZienkiewicz_ProcAMS} proved that in dimension one for any $0\leq \vv \in L^1_{loc}(\RR)$ there is a~family of intervals such that \cite[Thm. 2.2]{DZ_Studia2} gives local atomic decompositions for $H^1(\ll)$. Let us mention that results of this type in dimension $d=2$, are much unlike those in $d=1$ and $d\geq 3$, see \cite{DZ_Revista2}.
A question that we are concerned with is: what happens if we replace $-\Delta$ by the Bessel operator $\opB f(x) = - f''(x) - \a/x \, f'(x)$ on $L^2((0,\8), x^\a\, dx)$? It is known that if $\a+1 \in \NN$ then $\opB$ corresponds to $-\Delta$ on radial functions on $\Rd$ with $d=\a+1$, however $\opB$ exists and generates a semigroup for all $\a>-1$, which can be considered as the Laplace operator on spaces with non-integer dimensions.
In this paper we prove results similar to \cite{DZ_Studia2} and \cite{CzajaZienkiewicz_ProcAMS} for the Bessel operator with a potential. We were motivated to consider non-integer parameters $\a$ by the fact, that the Hardy space $H^1(\ll)$ admits different atomic decompositions in different dimensions, as it was mentioned above. Especially, we were interested in the dimensions $\a+1 \in (1,2)$, having in mind the difference of results for $\a=0$ and $\a=1$, see \cite{CzajaZienkiewicz_ProcAMS}, \cite{DZ_Studia2}, and \cite{DZ_Revista2}. Although we use the same scheme of proofs as in \cite{DZ_Studia2} and \cite{CzajaZienkiewicz_ProcAMS}, some technical difficulties appear. Indeed, in the space with the~weighted measure $x^\a \, dx$ the analysis is more delicate. One of the main problems is that the measure $\mu$~is not invariant under translation, and the induction argument used to prove Lemma \ref{lem:conditionD} becomes more complicated. Also, we added a precise explanation of the superharmonicity of $\phi_I$ (see \eqref{def_phi}) in the Appendix. Let us notice that the semigroup related to the Bessel operator $\opB$, which is used intensely in the proof of Theorem \ref{thm:maintheorem}, is given in terms of Bessel functions $I_\a$, but we shall use mainly the Gaussian estimates \eqref{eq:ptapprox1} for $\exp(-t\opB)$, which are obtained from asymptotics of $I_\a$. In particular, we need to prove an atomic characterization for local Hardy space related to $\opB$, see Subsection \ref{subsec22}.
Furthermore, for readers convenience, recall that with the operator $\opB$ on $L^2((0,\8), x^\a dx)$ we can relate another Bessel operator $\wt{\opB} f(x) = \Uu \opB \Uu^{-1} f(x) = -f''(x) + \frac{\a(\a-2)}{4x^2} f(x)$, where $\Uu : L^2((0,\8), x^\a dx) \to L^2((0,\8), dx)$ is an isometry given by $\Uu f(x) = x^{\a/2} f(x)$. Therefore, the $L^2$-theory of $\opB$ and $\wt{\opB}$ can be studied simultaneously. However, it seems that $\Uu$ (or even $\Uu^2$ which is an isometry of suitable $L^1$ spaces) cannot help in studies of Hardy spaces. To see this, one can look at the atomic decompositions of the Hardy spaces $H^1(\opB)$ and $H^1(\wt{\opB})$ given in \cite{BDT}. It appears that all the atoms for $H^1(\opB)$ have cancellation conditions with respect to $x^\a dx$ (see Section \ref{sec2} below), but some atoms for $H^1(\wt{\opB})$ are of the form $a(x) =\delta^{-1} \chi_{(0,\delta)}(x)$ with $\delta >0$, so they do not satisfy any cancellations. Moreover, $H^1(\wt{\opB})$ for each $\a>0$ is the same space, which can not be said about $H^1(\opB)$.
\subsection{Definitions}\label{ssec1.1}
For $\a>0$ let $(X, \rho, \mu)$ be a metric-measure space, where $X=(0,\8)$, $\rho(x,y) = |x-y|$ and $d\mu(x) = x^\a \, dx$. Denote $B(x,r) = \set{y\in X \ : \ \rho(x,y)<r}$ and observe that $X$ is a space of homogeneous type in the sense of~Coifman-Weiss \cite{CoifmanWeiss_BullAMS}, i.e. the doubling condition holds
\eqx{\lab{doubling}
\mu(B(x,2r)) \leq C \mu(B(x,r)),
}
where $C$ does not depend on $x\in X$ and $r>0$.
The classical Bessel operator is given by
\eqx{\opB f(x) = - f''(x) - \frac{\a}{x} f'(x).}
Slightly abusing notation, we shall also write $\opB$ for the densely defined, self-adjoint operator on $L^2(X, \mu)$ that corresponds to the differential operator above, see Subsection \ref{ssec2.1} for the semigroup generated by~$\opB$.
In this paper we consider the Bessel-Schr\"odinger operator $\opL$,
\eq{
\label{eq:operatorL} \opL f= \opB f + V\cdot f,
}
where $V \in L^1_{loc}(X,\mu), \, V\geq 0$. To be more precise, denote $\innprod{f,g} = \int f g \, d\mu$ and define a quadratic form
\eqx{
Q(f,g) = \innprod{f', g'} + \innprod{\sqrt{V}f, \sqrt{V} g},
}
with the domain
\eq{\label{DomQ}
\Dom(Q) = \rm{cl}\set{f\in C_c^1[0,\8) \ : \ f'(0^+)=0 } \cap \set{f\in L^2(X,\mu): \sqrt{V}f\in L^2(X,\mu)},
}
where $\rm{cl}(A)$ stands for the closure of the set $A$ in the norm $\norm{f}_{L^2(X,\mu)} + \norm{f'}_{L^2(X,\mu)}$.
The quadratic form $Q$ is positive and closed. Therefore, it defines a self-adjoint operator $\opL$ with the domain
\eq{\lab{DomL}
\Dom(\opL) = \set{f\in \Dom(Q): \exists h\in L^2(X,\mu) \ \forall g \in \Dom(Q) \quad Q(f,g)=\innprod{h,g}}.
}
For $f,h$ as above we put $\opL f:=h$. Let $\Kt{t} = \exp(-t\opL)$ be the semigroup generated by $\opL$. Denote by $B_s$ the Bessel process on $(X,\mu)$. By using the Feynman-Kac formula,
\eqx{
K_tf(x) = E^x\eee{\exp\eee{-\int_0^t V(B_s)\, ds}f(B_t)},}
one gets that $K_t$ has an integral kernel $\Kt{t}(x,y)$ and
\eq{\label{FKK}
0\leq \Kt{t}(x,y)\leq P_t(x,y),
}
where $P_t(x,y)$ is the kernel related to $P_t = \exp\eee{-t\opB}$, see Subsection \ref{ssec2.1}.
We define the Hardy space $\haL$ by means of the maximal operator associated with $\Kt{t}$, namely
\eq{
\label{eq:hardyLdefinition} \haL = \set{f\in \Ls{X} : \norm{f}_{\haL} = \norm{\sup_{t>0} \abs{\Kt{t}f}}_{\Ls{X}} < \8}.
}
The goal of this paper is to give an atomic characterizations of local type for $\haL$. Let $|\I{}|$ be the diameter of $\I{}$.
\defn{def_proper}{
Let $\ii$ be a collection of intervals that are closed with respect to the topology on $(0,\8)$. We~call a family $\ii$ {\it a proper section} of $X$ if:
\begin{enumerate}
\item[(a)] for $I,J \in \ii$, $I\neq J$, the intersection $I\cap J$ is either the empty set or a singleton,
\item[(b)] $X=\bigcup_{I\in\ii} I$,
\item[(c)] there exists a constant $C_0>0$ such that for $I,J \in \ii$, $I \cap J \neq \emptyset$ we have
\eqx{C_0^{-1}|I|\leq |J| \leq C_0 |I|.}
\end{enumerate}
}
Denote $\tau B(c,r) := B(c, \tau r)$. For an interval $I = B(x,r)$ (if $I=(0,2A)$ we take $I=B(A, A)$), let $c I:= B(x, c r)$. For a family $\ii$ as in Definition \ref{def_proper} we set $\iI{}:=\be \I{}$ for some fixed $\be>1$. By choosing any $\be<\min\left(2^{1/3},(1+C_0^{-1})^{1/3}\right)$ we have that $I^{***} \cap J^{***} \neq \emptyset$ if and only if $I\cap J \neq \emptyset$ and
\eq{\label{beta}
I^{***}\subset 2I.
}
We say that a function $a: X \to \CC$ is an $(\ii,\mu)${\it -atom} if:
\begin{enumerate}
\item[(i)] there exist $\I{} \in \ii$ and an interval $J\subset I^{**}$, such that: $\supp(a) \subset J$, $\norm{a}_\8 \leq \mu(J)^{-1}$, $\int a \, d\mu = 0$, \end{enumerate}
or
\begin{enumerate}
\item[(ii)] there exists $\I{} \in \ii$, such that $a(x) = \mu(\I{})^{-1} \mathbbm{1}_{\I{}}(x)$.
\end{enumerate}
The atoms as in $(ii)$ are called local atoms.
The atomic Hardy space $\hatom{}$ associated with the collection~$\ii$ is defined in the following way. We say that $f\in \hatom{}$ if
\eq{
\label{eq:hardyatomdefinition} f(x) = \sum_{n} \la_n a_n(x),
}
where $\la_n \in \CC$, $a_n$ are $(\ii,\mu)$-atoms, and $\sum_n \abs{\la_n} <\8$.
Set
\eq{
\label{eq:hardyatomnorm} \norm{f}_{\hatom{}} := \inf \sum_n \abs{\la_n},
}
where the infimum is taken over all possible representations of $f$ as in \eqref{eq:hardyatomdefinition}.
For a collection $\ii$ as above and $V\geq0$, $V\in L^1_{loc}(X,\mu)$ we consider the following two conditions:
\\- there exist constants $C, \e>0$ such that
\eqx{ \tag{D} \label{eq:D} \sup_{y \in \iiI{}} \int_X \Kt{2^k |I|^2}(x,y) \, d\mu(x) \leq Ck^{-1-\e} \qquad \text{for } \, I \in \ii,\, k\in\NN,}
- there exist constants $C,\de>0$ such that
\eqx{ \tag{K} \label{eq:K} \int_{0}^{2t} \int_X \Pt{s}(x,y) \mathbbm{1}_{\iiiI{}}(y) V(y) \, d\mu(y) \, ds \leq C\le\frac{t}{|I|^2}\)^{\de} \quad \text{for } \, x\in X,\, I \in \ii,\, t\leq |I|^2.}
\subsection{Statement of results} Our first main result is the following, cf. \cite[Thm. 2.2]{DZ_Studia2}
\thm{thm:maintheorem}{
Assume that a proper section $\ii$ and $0\leq V\in L^1_{loc}(X,\mu)$ are given, so that \eqref{eq:D} and \eqref{eq:K} hold. Then $\haL = \hatom{}$ and there exists a constant $C>0$, such that
\eqx{
\label{eq:hardynormseq} C^{-1} \norm{f}_{\hatom{}} \leq \norm{f}_{\haL} \leq C \norm{f}_{\hatom{}}.
}
}
In the second part we give an important application of Theorem \ref{thm:maintheorem}. Let us restrict ourselves to $\a\in (0,1)$. We prove that for {\bf any} $0\leq V\in L^1_{loc}(X,\mu)$ we can find a family $\ii(V)$ such that the assumptions of Theorem \ref{thm:maintheorem} hold, cf. \cite[Thm. 2.4]{CzajaZienkiewicz_ProcAMS} for the case $\a = 0$. To be more precise, let $\dd$ be a family of dyadic intervals on $(0,\8)$, that is $\mathcal{D}=\set{[k2^n,(k+1)2^n]:k\in\NN, n\in\ZZ}\cup\set{(0,2^n]: n\in \ZZ}$. Consider the family $\ii(V)$ that consists of maximal dyadic closed intervals $I$ that satisfy
\eqx{ \tag{S} \label{eq:S} \frac{\abs{2\I{}}^2}{\mu(2\I{})} \int_{2\I{}} V(y) \, d\mu(y) \leq 1.}
In Section \ref{sec4} we prove that $\ii(V)$ is a well defined proper section. The second main result is the following.
\thm{thm:maintheorem2}{}
Let $\a \in (0,1)$ and $0\leq V\in L^1_{loc}(X,\mu)$. Then the family $\ii(V)$ satisfies the assumptions of Theorem \ref{thm:maintheorem}.
\cor{coro_main}{
Let $\a \in (0,1)$ and $0\leq V\in L^1_{loc}(X,\mu)$. Then there is $C>0$, such that
\eqx{
C^{-1}\norm{f}_{\hh^1_{at}(\ii(V), \mu)} \leq \norm{f}_{\haL} \leq C \norm{f}_{\hh^1_{at}(\ii(V), \mu)}.}
}
The paper is organized as follows. In Section \ref{sec2} we study the atomic Hardy spaces related to $\opB$ and its local versions. This is used in a proof of Theorem \ref{thm:maintheorem}, which is provided in Section \ref{sec3}. Finally, in Section \ref{sec4} a proof of Theorem \ref{thm:maintheorem2} is given.
\section{Hardy spaces for the Bessel operator}\label{sec2}
\subsection{Global Hardy space for $\opB$}\lab{ssec2.1}
In this section we consider the case $\a>0$. Let $\Pt{t}=\exp(-t\opB)$ be the Bessel semigroup given by
\al{\nonumber
&\Pt{t}f(x) = \int_X \Pt{t}(x,y) f(y)\,d\mu(y),\\
&\label{eq:besselkernel} \Pt{t}(x,y) = (2t)^{-1}\exp\eee{-\frac{x^2+y^2}{4t}}I_{\ap}\le\frac{xy}{2t}\)(xy)^{-\ap},
}
where $I_\a(x)=\sum_{m=0}^{\8} \frac{1}{m!\Gamma(m+\a+1)}\le\frac{x}{2}\)^{2m+\a}$ is the modified Bessel function of the first kind. It is clear that $\Pt{t}(x,y) = \Pt{t}(y,x)$ and, since $\opB \mathbbm{1}_{(0,\8)}(x) = 0$, we have that
\eq{\label{contraction}
\int_X \Pt{t}(x,y) \, d\mu(x) = 1.
}
Let us recall that
\eq{\lab{ballmeasure}
\mu(B(x,t)) \simeq t(x+t)^\a.
}
It is known that the kernel $P_t(x,y)$ satisfies the two-side Gaussian estimates (see, e.g. \cite[Lem. 4.2]{DPW_JFAA}),
\eq{\label{eq:ptapprox1}
C^{-1} \mu(B(x,\sqrt{t}))^{-1} \exp\eee{-\frac{\abs{x-y}^2}{c_1t}} \leq \Pt{t}\xy \leq C \mu(B(x,\sqrt{t}))^{-1} \exp\eee{-\frac{\abs{x-y}^2}{c_2t}},
}
while the derivative satisfies
\eq{\label{gauss_prim}
\abs{\frac{\d}{\d x} \Pt{t}\xy} \leq C t^{-1/2} \mu(B(x,\sqrt{t}))^{-1} \exp\eee{-\frac{\abs{x-y}^2}{ct}}.
}
Let $H^1(\opB)$ be the Hardy space related to $P_t$, i.e. the space defined as in \eqref{eq:hardyLdefinition} with $\opL$ and $K_t$ replaced by $\opB$ and $P_t$, respectively. We call a function $a$ an $\mu$-{\it atom} if
\begin{enumerate}
\item[(iii)] there exists an interval $J\subseteq X$, such that: $\supp(a) \subset J$, $\norm{a}_\8 \leq \mu(J)^{-1}$, $\int a \, d\mu = 0$.
\end{enumerate}
The atomic Hardy space $H^1_{at}(\mu)$ is defined as in \eqref{eq:hardyatomdefinition} and \eqref{eq:hardyatomnorm} with $a_n$ being $\mu$-atoms.
\thm{thm:hardybessel}{
\cite[Thm. 1.7]{BDT} Let $\a >0$. There is $C>0$ such that
\eqx{ C^{-1}\norm{f}_{H^1_{at}(\mu)} \leq \norm{f}_{H^1(\opB)} \leq C \norm{f}_{H^1_{at}(\mu)}.}
}
\subsection{Local Hardy space for $\opB$}\label{subsec22}
For $\tau >0$ we define $h^1_\tau(\opB)$, {\it the local Hardy space related to $\opB$}, as the set of $L^1(X,\mu)$ functions for which the~norm
\eqx{
\label{eq:hardylocalnorm} \norm{f}_{\ha{\tau}} = \norm{\st{\tau} \abs{\Pt{t}f}}_{\Ls{X}}
}
is finite.
Let $\ii_{\tau}$ be a proper section of $X$ that consists of closed intervals of length $\tau$.
\thm{thm:atomiclocalnormseq}{
$(a)$ There exists $C>0$ such that for $\tau>0$ and an $(\ii_{\tau},\mu)$-atom $a$ we have
\eqx{\norm{a}_{\ha{\tau}} \leq C.}
$(b)$ Let $I$ be an interval such that $\supp(f) \subset I^{*}$ and $f\in\ha{|I|}$. Then
\eqx{ f = \sum_{n=0}^\8 \la_n a_n, \quad \sum_{n=0}^\8 \abs{\la_n} \leq C \norm{f}_{\ha{|I|}},}
where $a_0(x) = \mu(I)^{-1} \mathbbm{1}_I(x)$ and $a_n$ are $\mu$-atoms supported in $I^{**}$ for $n\geq 1$.
}
Let us remark that another characterization of $h^1_{\tau}(\opB)$, by mean of a local Riesz transform, was given in \cite[Thm. 2.11]{Preisner_JAT}. Theorem \ref{thm:atomiclocalnormseq} will be used to prove Theorem \ref{thm:maintheorem}.
\pr{
$(a)$ Obviously, if $a$ is $\mu$-atom, then the statement follows from Theorem \ref{thm:hardybessel}. Assume then that $a(x) = \mu(I)^{-1} \mathbbm{1}_I(x)$ and $|I|=\tau$. It is well known that \eqref{eq:ptapprox1} implies the boundedness of the maximal operator $\sup_{t>0} \abs{\Pt{t} f}$ on $L^2(X,\mu)$. Using this fact and the Schwarz inequality,
\eqx{
\norm{\sup_{t\leq \tau^2} \abs{\Pt{t} a}}_{L^1(I^{**},\mu)} \leq C \mu(I)^{1/2} \norm{\sup_{t>0} \abs{\Pt{t} a}}_{L^2(X,\mu)} \leq C \mu(I)^{1/2} \norm{a}_{L^2(X,\mu)} \leq C.
}
Denote by $c_I$ the center of $I$ and notice that $|x-y| \simeq |x-c_I|$ when $y\in I$ and $x\in (I^{**})^c$. By \eqref{eq:ptapprox1},
\alx{
\norm{\sup_{t\leq \tau^2} \abs{\Pt{t} a}}_{L^1((I^{**})^c,\mu)}
&\leq C \int_{(I^{**})^c} \sup_{t\leq \tau^2} \int_{I} \mu(B(x,\sqrt{t}))^{-1} \exp\eee{-\frac{|x-c_I|^2}{c_3 t}} \mu(I)^{-1} \, d\mu(y) \, d\mu(x)\\
&\leq C \int_{(I^{**})^c} \sup_{t\leq |I|^2} t^{-1/2} \frac{t}{|x-c_I|^2} \, dx\\
&\leq C \int_{(I^{**})^c} |I| |x-c_I|^{-2} \, dx \leq C.
}
$(b)$ Define $\la_0 := \int f d\mu$ and $g(x) = f(x) - \la_0 \mu(I)^{-1} \mathbbm{1}_I(x)$. Notice that
\eqx{
\abs{\la_0} \leq \norm{f}_{L^1(X,\mu)} \leq \norm{f}_{\ha{|I|}}.
}
Therefore, $\norm{g}_{L^1(X,\mu)} \leq 2 \norm{f}_{L^1(X,\mu)}$. Our goal is to prove that
\eq{\label{plk}
\norm{g}_{H^1(\opB)} \leq C \norm{f}_{\ha{|I|}}.
}
For $t\leq |I|^2$ using $(a)$ we obtain
\eqx{
\norm{\sup_{t\leq |I|^2} \abs{\Pt{t} g}}_{L^1(X,\mu)} \leq \norm{f}_{\ha{|I|}} +|\la_0|\norm{\mu(I)^{-1}\mathbbm{1}_I}_{\ha{|I|}} \leq C \norm{f}_{\ha{|I|}}.
}
For $t\geq |I|^2$ we shall use $\int g d\mu = 0$. By the symmetry of $P_t(x,y)$ and \eqref{gauss_prim},
\alx{
\norm{\sup_{t\geq |I|^2} \abs{\Pt{t} g}}_{L^1((I^{**})^c,\mu)} &\leq \int_{(I^{**})^c} \sup_{t\geq |I|^2} \abs{\int_{I^*}\eee{\Pt{t}(x,y) - \Pt{t}(x,c_I)}g(y) \, d\mu(y)} \, d\mu(x)\\
&\leq \int_{(I^{**})^c} \int_{I^*} \sup_{t\geq |I|^2} \frac{|y-c_I|}{\sqrt{t}}\mu(B(x,\sqrt{t}))^{-1} \exp\eee{-\frac{|x-c_I|^2}{c_2 t}}\abs{ g(y)} \, d\mu(y) \, d\mu(x)\\
&\leq C \norm{g}_{L^1(X,\mu)} |I| \int_{(I^{**})^c} \sup_{t\geq |I|^2} t^{-1} \exp\eee{-\frac{|x-c_I|^2}{c_2 t}} \, dx\\
&\leq C \norm{f}_{L^1(X,\mu)} |I| \int_{(I^{**})^c} |x-c_I|^{-2} \, dx \leq C \norm{f}_{\ha{|I|}}.
}
Likewise,
\alx{
\norm{\sup_{t\geq |I|^2} \abs{\Pt{t} g}}_{L^1(I^{**},\mu)} &\leq C \int_{I^{**}} \int_{I^*} \sup_{t\geq |I|^2} \mu(B(x,\sqrt{t}))^{-1} \abs{g(y)}\, d\mu(y) \, d\mu(x)\\
&\leq C \norm{g}_{L^1(X,\mu)} \int_{I^{**}} \sup_{t\geq |I|^2} t^{-1/2} \, dx \leq C \norm{f}_{\ha{|I|}}.
}
From \eqref{plk} we have that $g\in H^1(\opB)$, so using Theorem \ref{thm:hardybessel} we obtain $\la_k$ and $\mu$-atoms $a_k$ such that \break $g=\sum_{k=1}^\8 \la_k a_k$. Consequently,
\eqx{
f=\sum_{k=0}^{\8} \la_k a_k, \qquad \sum_{k=0}^\8 \abs{\la_k} \leq C\norm{f}_{h^1_{|I|}(\opB)},
}
where $a_0(x)=\mu(I)^{-1}\mathbbm{1}_I(x)$.
The only problem we have to deal with is that $a_k$ are not necessarily supported in $I^{**}$. Let $\psi$ be a function such that $\psi \equiv 1$ on $I^{*}$, $\psi \equiv 0$ on $(I^{**})^c$, and $\norm{\psi'}_\8 \leq C|I|^{-1}$. Then $f=\psi f$. To complete the proof we will show, that for every $\mu$-atom $a_k$ there exist sequences $\la_k^j\in\CC$ and $a_k^{j}$, such that each $a_k^j$ is either $\mu$-atom or $a_k^j =\mu(I)^{-1}\mathbbm{1}_I$, and $ a_k = \sum_j \la_k^{j} a_k^{j}$, $\sum_j |\la_k^{j}| < C$, with $C$ that not depend on $k$. Fix $a=a_k$ and an interval $J\subseteq X$ such that: $\supp(a) \subset J$, $\norm{a}_\8 \leq \mu(J)^{-1}$. Obviously $\supp(\psi a) \subseteq I^{**} \cap J$, and if $J\subset I^{*}$, then $\psi a = a$ is an $\mu$-atom. Furthermore, if $J\subset (I^{**})^c$, then $\psi a = 0$, so it suffices to consider the case that $J \cap (I^{*})^c\cap I^{**} \neq \emptyset$. Denote $K:=I^{**}\cap J$. Observe that $|K|\leq C_0|I|$ and let $N\in\NN$ be such that $1/2^{N+1}|I| \leq C_0^{-1}|K| \leq 1/2^{N}|I|$.
Define $\la := \int \psi a \, d\mu$ and notice that
\alx{ |\la| = \abs{\int_K \psi(x) a(x) \, d\mu(x)} & = \abs{\int_K a(x)(\psi(x) - \psi(c_{K})) \, d\mu(x)} \\
& \leq C |I|^{-1} \mu(K)^{-1} \int_K |x-c_{K}| \, d\mu(x) \\
& \leq C |I|^{-1} |K| \leq C 2^{-N}. }
Let us choose intervals $I_j$, such that $K =: I_0 \subset I_1 \subset \cdots \subset I_{N} \subset I^{**}$, where $|I_{j+1}|/|I_j|=2$ and $|I_N|\simeq |I|$. Then
\alx{ \psi a & = \left( \psi a - \la \mu(I_0)^{-1}\mathbbm{1}_{I_0} \right) + \sum_{j=1}^{N} \la \left( \mu(I_{j-1})^{-1}\mathbbm{1}_{I_{j-1}} - \mu(I_{j})^{-1}\mathbbm{1}_{I_{j}} \right) \\
& + \la (\mu(I_N)^{-1}\mathbbm{1}_{I_N} - \mu(I)^{-1}\mathbbm{1}_{I}) + \la \mu(I)^{-1}\mathbbm{1}_I = \sum_{j=0}^{N+2} b_j.\\ \quad &
}
Observe that:
\begin{enumerate}
\item[(1)] $\supp(b_0)\subseteq I_0$, $\int b_0 \, d\mu = 0$, and
\eqx{ \norm{b_0}_\8 \leq C(1+2^{-N})\mu(I_0)^{-1} \leq C \mu(I_0)^{-1},}
\item[(2)] for $j=1,...,N$ we have: $\supp(b_j) \subseteq I_j$, $\int b_j \, d\mu = 0$, and
\eqx{\norm{b_j}_\8 \leq C |\la| \mu(I_j)^{-1} \leq C 2^{-N} \mu(I_j)^{-1},}
\item[(3)] $\supp(b_{N+1}) \subseteq I^{**}$, $\norm{b_{N+1}}_\8 \leq C |\la| \mu(I)^{-1}$ and $\int b_{N+1} \, d\mu = 0$.
\end{enumerate}
We conclude that $b_j$ are multiples of $(\ii_{|I|},\mu)$-atoms and
\alx{
\norm{\sum_{j=0}^{N+2} b_j}_{H^1_{at}(\ii_{|I|},\mu)} \leq C \sum_{j=0}^{N+2} 2^{-N} \leq C.
}
}
\cor{cororor}{
There exists a constant $C>0$ such that for $\tau>0$ we have
\eqx{ C^{-1}\norm{f}_{\hatom{\tau}} \leq \norm{f}_{h^1_\tau(\opB)} \leq C \norm{f}_{\hatom{\tau}}.}
}
The right inequality in Corolarry \ref{cororor} follows easily from Theorem \ref{thm:atomiclocalnormseq}$(a)$. For the left inequality one uses Theorem \ref{thm:atomiclocalnormseq}$(b)$ with a suitable partition of unity and methods as in Lemmas \ref{lem:intfunctionwithPt} and \ref{lem:diffkt} below. We omit the details.
\section{Proof of Theorem \ref{thm:maintheorem}}\label{sec3}
\subsection{Auxiliary estimates}
For a proper section $\ii$ let $\set{\p{I}}_{I\in \ii}$ be a partition of unity associated with $\ii$, that is a family of $C^\8$ functions on~$X$, such that $\supp(\p{I})\subset I^{*},$
$ 0 \leq \p{I} \leq 1,\, \norm{\p{I}'}_\8 \leq |I|^{-1} $ and $\sum_{I\in\ii}\p{I}(x) = 1$ for all $x \in X$.
The perturbation formula states that
\eq{\lab{perturbation}
\Pt{t}\xy - \Kt{t}\xy = \int_0^t \int_X \Pt{t-s}(x,z) V(z) \Kt{s}(z,y) \, d\mu(z) ds.
}
To prove Theorem \ref{thm:maintheorem} we closely follow the proof of \cite[Thm. 2.2]{DZ_Studia2}. However, in our weighted space, some technical difficulties appear. Therefore we present all the details for convenience of the reader.
\lem{lem:intfunctionwithPt}{ For $I\in\ii$ and $f\in\Ls{X}$,
\eqx{ \norm{\st{4|I|}\abs{\Pt{t}(\p{I}f)}}_{\Ls{(\iiI{})^c}} \leq C \norm{\p{I}f}_{\Ls{X}}.}
}
\pr{
Denote by $c_{I}$ the center of the interval $\I{}\in\ii$. For $x\in(\iiI{})^c$ and $y\in\iI{}$ we have $\abs{x-y} \simeq \abs{x-c_I}$. Notice that \eqref{ballmeasure} implies
\eq{\label{uuu}
\mu(B(x,\sqrt{t}))^{-1} x^\a \leq t^{-1/2}.
}
Using \eqref{eq:ptapprox1} the left-hand side is bounded by
\spx{
\int_{(\iiI{})^c} \st{4|I|} & \int_{I^*} \mu(B(x,\sqrt{t}))^{-1} e^{-\frac{\abs{x-y}^2}{c_2t}} \abs{\p{I}(y)f(y)} \, d\mu(y) \, d\mu(x) \\
& \leq C \norm{\p{I} f}_{L^1(X,\mu)} \int_{(\iiI{})^c} \st{4|I|} x^{-\a} t^{-\frac{1}{2}} e^{-\frac{\abs{x-c_I}^2}{c_3 t}} \, d\mu(x) \\
& \leq C \norm{\p{I} f}_{L^1(X,\mu)} \int_{(\iiI{})^c} \st{4|I|}t^{-\frac{1}{2}} \frac{t}{|x-c_I|^2} \, dx \\
& \leq C \norm{\p{I} f}_{L^1(X,\mu)} \int_{(\iiI{})^c} |I| |x-c_I|^{-2} \, dx \\
& \leq C \norm{\p{I}f}_{\Ls{X}}.
}
}
\lem{lem:firstlemma}{
For $I\in\ii$ and $f\in\Ls{X}$,
\eqx{ \norm{\p{I} f}_{h^1_{|I|}(\opB)} \simeq \norm{ \st{|I|} |\Pt{t}(\p{I} f)|}_{\Ls{I^{**}}}. }
}
\newcommand{\ov}[1]{\bar{#1}}
Denote
\eqx{
\wt{f}_I:=\sum_{J:\I{}\cap J \neq \emptyset} \p{J} f, \qquad \qquad \ov{f}_I := f-\wt{f}_I = \sum_{J:\I{}\cap J = \emptyset} \p{J} f.
}
\lem{lem:diffkt}{
For $I\in\ii$ and $f\in\Ls{X}$,
\eqx{
\norm{\st{|I|}\abs{{\Kt{t}} \left(\p{I} \wt{f}_I\right) - \p{I} \Kt{t} \left(\wt{f}_I\right)}}_{\Ls{\iiI{}}} \leq C \sum_{J:\I{}\cap J \neq \emptyset} \norm{\p{J} f}_{\Ls{X}}.
}
}
\pr{
Denote $\wt{I} = \cup_{J:I\cap J \neq \emptyset} J^{*}$. Note that for $x\in\iiI{}$ and $y\in \wt{I}$, there is $\abs{x-y}\leq C |I|$ and
\eq{\label{eq:jjj}
\st{|I|}~t^{-\frac{1}{2}}~e^{-\frac{\abs{x-y}^2}{c_2t}}~\leq~C~\abs{x-y}^{-1}.
}
Using \eqref{FKK}, \eqref{eq:ptapprox1}, \eqref{uuu}, and \eqref{eq:jjj}
\spx{
\norm{\st{|I|}\abs{\Kt{t}\left(\p{I}\wt{f}_I\right)-\p{I}\Kt{t}\left(\wt{f}_I\right)}}_{\Ls{\iiI{}}} & \leq \int_{\iiI{}} \st{|I|} \int_{\wt{I}} \abs{\p{I}(y) -\p{I}(x)} \Kt{t}(x,y) \abs{\wt{f}_I(y)} \, d\mu(y) \, d\mu(x) \\
& \leq C \int_{\iiI{}} \int_{\wt{I}} \abs{x-y}\norm{\p{I}'}_\8 \st{|I|} t^{-\frac{1}{2}} e^{-\frac{\abs{x-y}^2}{c_2 t}} \abs{\wt{f}_I(y)} \, d\mu(y) \, dx \\
& \leq C \int_{I^{**}} |I|^{-1} dx \norm{\wt{f}_I}_{L^1(X,\mu)}.
}
}
\lem{lem:suppthreestars}{
Assume that $V$ and $\ii$ are given, so that \eqref{eq:D} holds. Then
\eqx{\sum_{I\in\ii} \norm{\sup_{t>0} \abs{\Kt{t} \left( \ov{f}_I\right) }}_{\Ls{\iiiI{}}} \leq C \norm{f}_{\Ls{X}}.}
}
\pr{
Denote $s_m=2^m |I|^2$ and let $m\geq 2$. By the semigroup property, \eqref{FKK}, \eqref{eq:ptapprox1},
\spx{
&\norm{\sup_{s_m \leq t \leq s_{m+1}} \Kt{t}(\abs{\p{J}f})}_{\Ls{X}} \\
& \leq \int_{X} \sup_{s_m\leq t \leq s_{m+1}} \int_X \int_X \Pt{t-s_{m-1}}(x,z) \Kt{s_{m-1}}(z,y) \abs{\p{J}(y)f(y)} \, d\mu(z) \, d\mu(y) \, d\mu(x) \\
& \leq C \int_X \abs{\p{J}(y)f(y)} \int_X \Kt{s_{m-1}}(z,y) \int_{X} s_m^{-1/2} \exp\expr{-\frac{|x-z|^2}{c_3 s_m}} dx \, d\mu(z) \, d\mu(y)\\
& \leq C (m-1)^{-1-\e} \cdot \norm{\p{J}f}_{\Ls{X}}.
}
In the last inequality we have applied \eqref{eq:D}. Using the above estimate and Lemma \ref{lem:intfunctionwithPt},
\spx{
&\sum_{I\in\ii} \norm{\sup_{t>0} \abs{\Kt{t} \left( \sum_{J:I\cap J=\emptyset} \p{J} f \) }}_{\Ls{\iiiI{}}} \\
& \leq \sum_{J\in\ii} \sum_{I:J\cap I = \emptyset} \norm{\sup_{t>0}\Kt{t}(\abs{\p{J}f})}_{\Ls{\iiiI{}}} \leq C \sum_{J\in\ii} \norm{\sup_{t>0}\Kt{t}(\abs{\p{J}f})}_{\Ls{(J^{**})^c}} \\
& \leq C \sum_{J\in\ii} \norm{\st{4|J|}\Kt{t}(\abs{\p{J}f})}_{\Ls{(J^{**})^c}} + C \sum_{J\in\ii} \sum_{m=2}^\8 \norm{\sup_{s_m\leq t \leq s_{m+1}}\Kt{t}(\abs{\p{J}f})}_{\Ls{(J^{**})^c}} \\
& \leq C \sum_{J\in\ii} \norm{\p{J}f}_{\Ls{X}} + C \sum_{J\in\ii} \sum_{m=2}^\8 m^{-1-\e} \cdot \norm{\p{J}f}_{\Ls{X}} \leq C \norm{f}_{\Ls{X}}.
}
}
\lem{lem:pertur}{
\eqx{\int_0^\8 \int_X V(z)\Kt{s}(\abs{f})(z) \, d\mu(z)\,ds \leq C \norm{f}_{\Ls{X}} }
}
\pr{
By integrating \eqref{perturbation} and using \eqref{contraction} we obtain
\eqx{ \int_0^t \int_X V(z) \Kt{s}(z,y) \, d\mu(z) \, ds \leq C.}
The proof is finished by setting $t\to \8$.
}
\lem{lem:diffptandkt}{
Assume that $V$ and $\ii$ are given, so that \eqref{eq:K} holds. Then for $I\in\ii$,
\eqx{
\norm{\st{|I|} \abs{(\Pt{t}-\Kt{t})(\p{I} f)}}_{\Ls{X}} \leq C \norm{\p{I} f}_{\Ls{X}}.
}
}
\pr{
By \eqref{FKK} and Lemma \ref{lem:intfunctionwithPt},
\eqx{ \norm{\st{|I|} \abs{(\Pt{t}-\Kt{t})(\p{I} f)}}_{\Ls{(\iiI{})^c}} \leq C \norm{\p{I} f}_{\Ls{X}}.}
Consider the integral on $I^{**}$. From \eqref{perturbation},
\eqx{(\Pt{t} - \Kt{t})(\p{I}f)(x) = \int_0^t\Pt{t-s}(V''\Kt{s}(\p{I}f))(x) \, ds + \int_0^t\Pt{t-s}(V'\Kt{s}(\p{I}f))(x) \, ds }
where $V=\mathbbm{1}_{\iiiI{}}V + \mathbbm{1}_{(\iiiI{})^c}V = V' + V''$. Repeating the argument from the proof of Lemma \ref{lem:intfunctionwithPt} and using Lemma \ref{lem:pertur} we obtain
\eqx{\norm{\st{|I|} \int_0^t\Pt{t-s}(V''\Kt{s}(\p{I}f)) \, ds }_{\Ls{\iiI}} \leq C \norm{\p{I}f}_{\Ls{X}}.}
To estimate the integral that contains $V'$ write
\eqx{ \abs{\int_0^t\Pt{t-s}(V'\Kt{s}(\p{j}f))(x) \, ds} \leq \int_0^{t/2}\Pt{t-s}(V'\Pt{s}(\abs{\p{j}f}))(x) \, ds + \int_{t/2}^t\Pt{t-s}(V'\Pt{s}(\abs{\p{j}f}))(x) \, ds}
\eqx{ = A_{1,t}(x) + A_{2,t}(x).}
Let $t_m = 2^{-m} |I|^2$. Similarly as in proof of the Lemma \ref{lem:suppthreestars} we obtain
\spx{
\norm{ \st{|I|} A_{1,t}}_{\Ls{X}} & \leq \sum_{m=0}^\8 \int_X \sup_{t_{m+1} \leq t \leq t_m} \int_0^{t/2} \int_X \Pt{t-s}(x,y)V'(y)\Pt{s}(\abs{\p{I}f})(y) \, d\mu(y) \, ds \, d\mu(x) \\
& \leq C \sum_{m=0}^\8 \int_X \int_0^{t_m} V'(y)\Pt{s}(\abs{\p{I}f})(y) \int_X t_m^{-\frac{1}{2}} e^{-\frac{\abs{x-y}^2}{{c_3 t_m}}} \, dx \, ds\, d\mu(y) \\
& \leq C \sum_{m=0}^\8 \int_0^{t_m} \int_X \mathbbm{1}_{\iiiI{}}(y)V(y)\Pt{s}(\abs{\p{I}f})(y) \, d\mu(y) \, ds \\
& \leq C \sum_{m=0}^\8 \left( \frac{2^{-m}|I|^2}{|I|^2} \right) ^{\de} \norm{\p{I}f}_{\Ls{X}}.
}
In the last inequality we have used \eqref{eq:K}. To estimate $A_{2,t}$ we proceed similarly noticing that for $t\in [t_{m+1}, t_m]$ and $s\in [t/2,t]$ we have $s \in [t_{m+2},t_m]$. The details are left to the reader.
}
\subsection{Proof of Theorem \ref{thm:maintheorem}}
\textbf{First inequality.} Assume that $f\in\haL$. Observe that $\p{I} f = \p{I}\wt{f}_I$ and
\eqx{
\Kt{t} \expr{\p{I} f} = \Kt{t} \expr{\p{I} \wt{f}_I} - \p{I} \Kt{t}\eee{\wt{f}_I} - \p{I} \Kt{t}\eee{\ov{f}_I}+ \p{I} \Kt{t}\eee{f}.
}
From Lemmas \ref{lem:diffkt} and \ref{lem:suppthreestars} we deduce that
\spx{
\sum_{I\in\ii} \norm{\st{|I|} \abs{\Kt{t}(\p{I} f)}}_{\Ls{\iiI{}}}
\leq & \sum_{I\in\ii} \norm{\st{|I|} \abs{\Kt{t}\left( \p{I} \wt{f}_I \right) - \p{I} \Kt{t} \left( \wt{f}_I \right) }}_{\Ls{\iiI{}}} \\
& + \sum_{I\in\ii}\norm{ \st{|I|} \abs{ \p{I} \Kt{t}\left( \ov{f}_I \)}}_{\Ls{\iiI{}}} \\
& + \sum_{I\in\ii} \norm{ \st{|I|} \abs{ \p{I} \Kt{t}f }}_{\Ls{\iiI{}}} \\
\leq& C \eee{\norm{f}_{\Ls{X}}+ \norm{f}_{\haL}} \leq C\norm{f}_{\haL}.
}
The above estimate, together with Corollary \ref{lem:firstlemma} and Lemma \ref{lem:diffptandkt}, lead to
\spx{
\sum_{I\in\ii} \norm{\p{I} f}_{\ha{|I|}}
& \leq C \sum_{I\in\ii} \norm{\st{|I|}\abs{(\Pt{t}-\Kt{t})(\p{I} f)}}_{\Ls{\iiI{}}} + C\sum_{I\in\ii} \norm{\st{|I|} \abs{\Kt{t}(\p{I} f)}}_{\Ls{\iiI{}}} \\
& \leq C \norm{f}_{\haL}. \\
}
Now we use Theorem \ref{thm:atomiclocalnormseq}$(b)$ for each $\p{I} f$ getting $(\la_n^I)_n$ and $(\ii, \mu)$-atoms $(a_n^I)_n$, so that
\eqx{\p{I}f(x)=\sum_{n} \la_n^I a_n^I(x), \quad \text{and} \quad \sum_{n} \abs{\la_n^I} \leq C \norm{\p{I} f}_{\ha{|I|}}.}
Notice that $\sum_{n,I}\norm{\la_n^Ia_n^I}_{\Ls{X}} < \8$, since $\norm{a_n^I}_{\Ls{X}}\leq 1$. Summing up for all $I\in \ii$ we finish the first part of the proof.
\textbf{Second inequality.} Let $a$ be an $(\ii,\mu)$-{\it atom}, such that $\supp(a) \subset \iiI{}$. There exists an integer $ m \geq 0$, independent of $I$, such that
\eqx{\inf \set{|I|^2 : J\in\ii, \, J\cap I \neq \emptyset} \geq 2^{-m}|I|^2.}
Denote $t_n=2^{-n}|I|^2$, $n\in\ZZ$.
Observe that
\eqx{\norm{\sup_{t\leq t_m} \abs{\Kt{t}(a)}}_{\Ls{X}} \leq \norm{\sup_{t\leq t_m} \abs{(\Kt{t}-\Pt{t})(a)}}_{\Ls{X}} + \norm{\sup_{t\leq t_m} \abs{\Pt{t}( a)}}_{\Ls{X}}\leq \norm{a}_{\Ls{X}}.}
In the last inequality we applied Lemma \ref{lem:diffptandkt} and Theorem \ref{thm:atomiclocalnormseq}$(b)$, since $a$ is also an $(\ii_{|I|},\mu)$-{\it atom}.
It suffices to estimate $\norm{\sup_{t>t_m}\abs{\Kt{t}a}}_{\Ls{X}}$. This is done by using \eqref{FKK} and \eqref{eq:K}. Indeed, using similar methods as in the proof of Lemma \ref{lem:intfunctionwithPt},
\spx{
\norm{\sup_{t>t_m} \abs{\Kt{t}a}}_{\Ls{X}} & \leq \sum_{n\leq m} \norm{\sup_{t_{n} \leq t \leq t_{n-1}}\abs{\Kt{t}a}}_{\Ls{X}} \\ & \leq \sum_{n\leq m} \norm{\sup_{t_{n+1}\leq t \leq 3t_{n+1}}\Kt{t}\abs{\Kt{t_{n+1}}a}}_{\Ls{X}} \\
& \leq C \sum_{n\leq m+1} \int_X \int_X \Kt{2^{-n}|I|^2}(x,y)\abs{a(y)} \, d\mu(y) \, d\mu(x) \\
& \leq C \norm{a}_{\Ls{X}}\leq C,
}
since the operator $\sup_{t_{n+1}\leq t \leq 3t_{n+1}} K_t$ is bounded on $L^1(X,\mu)$, see \eqref{eq:ptapprox1}.
\section{Proof of Theorem \ref{thm:maintheorem2}.}\label{sec4}
In the whole section we assume that $\a\in(0,1)$. Recall that for $0\leq a<b$,
\eq{\lab{asympt_2}
\mu((a,b)) = \frac{ b^{1+\a} - a^{1+\a}}{1+\a} \simeq
\begin{cases}
b^{\a+1} & 2a\leq b\\
(b-a) a^{\a} & 2a\geq b.
\end{cases}
}
If $I=(a,b)\subset(0,\8)$, by \eqref{asympt_2} and the Mean-Value Theorem we easily get
\eq{\label{costam}
\frac{|I|^2}{\mu(I)} \simeq b^{1-\a} - a^{1-\a}.
}
\lem{enlargement}{
For $\a\in(0,1)$ and $I\subset J \subset (0,\8)$,
\eqx{\frac{\abs{I}^2}{\mu(I)} \int_{I} V(y) \, d\mu(y) \leq \frac{\abs{J}^2}{\mu(J)} \int_{J} V(y) \, d\mu(y).}
}
\pr{
Obviously, since $V\geq 0$, it is enough to prove that
\eq{\lab{intinint}
\frac{\abs{I}^2}{\mu(I)} \leq \frac{\abs{J}^2}{\mu(J)},
}
provided that $I\subset J \subset (0,\8)$.
Let $a\leq b\leq c \leq d$ and $I=(b,c)\subset (a,d) =J$. Denote $\Gamma(x,y) = (y-x)^2/(y^{\a+1} - x^{\a+1})$. Now, \eqref{intinint} is equivalent to $\Gamma(b,c)\leq \Gamma(a,d)$. This is done in two steps.
{\bf Step 1:} $\Gamma(b,c) \leq \Gamma(a,c)$. Denote $a=sc$ and $b=tc$, where $0<s\leq t<1$. It is enough to prove
\eqx{
\frac{(1-t)^2}{1-t^{\a+1}} \leq \frac{(1-s)^2}{1-s^{\a+1}}.
}
By a simple calculus argument, the function $F_1(x) = (1-x)^2/(1-x^{\a+1})$ is monotonically decreasing for $x\in (0,1)$.
{\bf Step 2:} $\Gamma(a,c) \leq \Gamma(a,d)$. Similarly, let $c=t'a$, $d=s'a$, $1<t'\leq s'$. The function \break $F_2(x) = (x-1)^2/(x^{\a+1}-1)$ is monotonically increasing in $(1,\8)$, thus
\eqx{
\frac{(t'-1)^2}{(t')^{\a+1}-1} \leq \frac{(s'-1)^2}{(s')^{\a+1}-1}.
}
}
\prop{construction}{
Let $V\in L^1_{loc}(X,\mu)$, $V\geq 0$. Then the family $\ii(V)$ of maximal dyadic intervals satisfying \eqref{eq:S} is a proper section (see Definition \ref{def_proper}).
}
\pr{
For a closed dyadic interval $I$ consider $F(I)= |2I|^2 \mu(2I)^{-1} \int_{2I} V d\mu$ and denote by $\dI{}$ the smallest dyadic interval properly containing $\I{}$. Notice that $2I \subseteq 2\dI{}$ and, by Lemma \ref{enlargement}, we have $F(I)\leq F(\dI{})$. Also, for an increasing sequence of dyadic intervals $I_n \subset I_{n+1}$ we have $\lim_{n\to \8} F(I_n) = \8$ and $\lim_{n\to -\8} F(I_n) = 0$, see \eqref{costam}.
This justifies the choice of $\ii(V)$ as maximal dyadic intervals such that \eqref{eq:S} holds. What is left to prove is that $\ii(V)$ is a proper section, namely we need to show that for $I,J \in \ii(V)$, $I\cap J \neq \emptyset$ we have $|I| \simeq |J|$.
By contradiction, suppose that there exist $I_k$, $J_k$ such that $I_k \cap J_k \neq \emptyset$ and $|I_k|/|J_k| \to \8$. We can assume that $2J_k^d \subseteq 2 I_k$ for all $k$. Denote, $a_k = |I_k|^2 \mu(I_k)^{-1} |J_k|^{-2} \mu(J_k)$. By the choice of $\ii$,
\eqx{1\geq \frac{\abs{2I_k}^2}{\mu(2\I{k})} \int_{2\I{k}} V(y) \, d\mu(y) \geq \frac{\abs{2I_k}^2\mu(2J_k^d) }{\mu(2I_k)\abs{2J_k^d}^2} \frac{\abs{2J_k^d}^2}{\mu(2J_k^d)} \int_{2J_k^d} V(y) \, d\mu(y) \geq C^{-1} a_k.}
The proof will be finished when we show that $a_k \to \8$. This follows from \eqref{asympt_2} by considering several cases. Let $a,b,c$ be such that $0\leq a<b<c$.
{\bf Case 1:} $J_k = [a,b]$, $I_k = [b,c]$.\\
{\bf Subcase 1:} $4a\leq 2b\leq c$. Using \eqref{asympt_2} we have
\eqx{
a_k \simeq (c/b)^2 (b/c)^{1+\a} = (c/b)^{1-\a} \simeq (|I_k|/|J_k|)^{1-\a}.
}
{\bf Subcase 2:} $4a\leq 2b\geq c$. This subcase can hold only for finite $k$.\\
{\bf Subcase 3:} $4a\geq 2b\leq c$.
Using \eqref{asympt_2} we have
\eqx{
a_k \simeq \frac{c^2}{(b-a)^2} \frac{(b-a) {a^{\a}}}{c^{\a+1}} = \frac{{a^\a} c}{c^\a(b-a)} \geq \frac{c^{1-\a}}{(b-a)^{1-\a}} \simeq (|I_k|/|J_k|)^{1-\a}.
}
{\bf Subcase 4:} $4a\geq 2b\geq c$.
Using \eqref{asympt_2} we have
\eqx{
a_k \simeq \eee{\frac{c-b}{b-a}}^2 \frac{(b-a) {a^{\a}}}{(c-b){b^{\a}}} \simeq |I_k|/|J_k|.
}
{\bf Case 2:} $J_k = [b,c]$, $I_k = [a,b]$. Then
\eqx{
a_k \geq C \eee{\frac{|I_k|}{|J_k|}}^2 \frac{|J_k| b^\a}{|I_k| b^\a} \simeq |I_k|/|J_k|.
}
}
Recall that $\innprod{f,g} = \int_X f g \, d\mu$, so that $\sprod{-\Bb \phi, \psi}_\mu = \sprod{\phi, -\Bb \psi}_\mu$ for appropriate $\psi, \phi$. For $y>0$ the equation
\eqx{
\sprod{-\Bb \phi_y, \psi}_\mu = \psi(y)
}
has the solution given by $\phi_y(x) = \frac{1}{2(1-\a)}\abs{x^{1-\a} - y^{1-\a}}$. We shall use $\phi_y$ to construct superharmonic functions that will be crucial in the proof of \eqref{eq:D}.
\lem{lem:conditionD}{
Let $\a\in(0,1)$, $0\leq V\in L^1_{loc}(X,\mu)$, and $\ii(V)$ is as in Proposition \ref{construction}. Then
\eqx{\lab{DDD}
\int_X \Kt{2^n\abs{I}^2}(x,y) \, d\mu(x) \leq C 2^{-\frac{1-\a}{2} n}
}
for $y\in I^{**}$, $I\in \ii(V)$, and $n \geq 0$.
}
\pr{
Let $I$ be a dyadic interval such that
\eqx{
\frac{|2I|^2}{\mu(2I)}\int_{2I} V d\mu \leq 1, \qquad \frac{|2I^d|^2}{\mu(2I^d)}\int_{2I^d} V d\mu > 1.
}
From \eqref{beta} we have $I^{**}\subset 2I$. By a continuity argument there exists $J$ such that $2I\subset J \subset 2I^d$ and $\frac{\abs{J}^2}{\mu(J)} \int_{J}V \, d\mu = 1$. Let $J=(a,b)$ and observe that $|J|\simeq |I|$. Define
\eq{\label{def_phi}
\phi_I(x) = 1 + \frac{1}{2(1-\a)} \int_J V(y)\abs{x^{1-\a}-y^{1-\a}} \, d\mu(y).
}
Fix $z \in I^{**}$. By \eqref{costam} and the doubling condition,
\eq{\label{asdf}
\phi_I(z)\leq 1 + C \sup_{y,y'\in J} |y^{1-\a} - (y')^{1-\a}| \int_J V d\mu \simeq C.
}
Also, we claim that for $x\in X$,
\eq{\label{asdf2}
\phi_I(x) \simeq 1 +\frac{\mu(J)\abs{x^{1-\a}-z^{1-\a}}}{\abs{J}^{2}}.
}
Indeed, if $x$ is such that $|x^{1-\a} - z^{1-\a}|\leq C |J|^2 \mu(J)^{-1}$, with $C$ large enough, this follows exactly as in \eqref{asdf}. In~the opposite case $|x^{1-\a} - z^{1-\a}|\geq C |J|^2 \mu(J)^{-1}$, we have $|x^{1-\a} - z^{1-\a}| \simeq |x^{1-\a} - y^{1-\a}|$ for $y\in J$ and the claim follows.
Now we proceed to a crucial argument that uses superharmonicity. Observe that formal calculation gives
\eqx{
-\opL\phi_I(x) = -\opB\phi_I(x) - V(x)\phi_I(x) = V(x)(\mathbbm{1}_J(x) - \phi_I(x)) \leq 0
}
and, consequently,
\eqx{\dt \Kt{t}\phi_I(z) = \int_X \Kt{t}(z,x)(-\opL\phi_I(x)) \, d\mu(x) \leq 0.}
This leads to
\eq{\label{super}
\Kt{t}\phi_I(z) \leq \phi_I(z), \qquad t>0.
}
However, $\phi_I$ is not in $\Dom(\opL)$ (or even in $L^2(X,\mu)$), thus we provide a detailed proof of \eqref{super} in Appendix.
\newcommand{\rhoo}{\theta}
Denote
\eqx{\rhoo(t) = \int_X \Kt{t}(z,x) \, d\mu(x), \quad t>0.}
Our goal is prove that, there exists $c_0$ such that for every $n\in\NN$
\eq{\lab{induction}
\rhoo(2^n|I|^2) \leq c_0 2^{-\frac{1-\a}{2} n}.
}
This will follow by induction argument. Let $t>0$. By \eqref{FKK} and \eqref{eq:ptapprox1},
\eq{\label{twot}
\Kt{2t}(z,x) = \int_X \Kt{t}(z,y)\Kt{t}(y,x) \, d\mu(y) \leq C \rhoo(t) \mu(B(x,\sqrt{t}))^{-1}
}
and
\eqx{\label{est_m}
\rhoo(2t) = \int_{\abs{x^{1-\a}-z^{1-\a}} < R} \Kt{2t}(z,x) \, d\mu(x) + \int_{\abs{x^{1-\a}-z^{1-\a}} > R} \Kt{2t}(z,x) \, d\mu(x) = A_1 + A_2.
}
Here $R>0$ will be specified later. By \eqref{asdf2}, \eqref{super}, and \eqref{asdf} we have
\spx{
A_2 &\leq \frac{\abs{J}^2}{R\mu(J)} \int_{\abs{x^{1-\a}-z^{1-\a}} > R} \Kt{2t}(z,x) \frac{\mu(J)\abs{x^{1-\a}-z^{1-\a}}}{|J|^2} \, d\mu(x)\\
&\leq C R^{-1}|J|^2 \mu(J)^{-1} \int \Kt{2t}(z,x) \phi_I(x) \, d\mu(x)\\
&\leq C R^{-1}|J|^2 \mu(J)^{-1}.
}
To estimate $A_1$ we use \eqref{twot} and \eqref{ballmeasure},
\spx{
A_1 &\leq C \rhoo(t) \int_{\abs{x^{1-\a}-z^{1-\a}} < R} \mu(B(x,\sqrt{t}))^{-1} \, d\mu(x)\\
&\leq C \rhoo(t) t^{-1/2} \int_{\abs{x^{1-\a}-z^{1-\a}} < R} \, dx.
}
{\bf Case A:} $z^{1-\a} \geq 2R$, then
\eqx{\int_{\abs{x^{1-\a}-z^{1-\a}} < R} \, dx = (z^{1-\a} +R)^{\frac{1}{1-\a}} - (z^{1-\a} -R)^{\frac{1}{1-\a}} \leq C R z^{\a}.}
In this case
\eq{\lab{case1}
\rhoo(2t) \leq c_1 \left( \rhoo(t) t^{-1/2} R z^{\a} + R^{-1}|J|^2\mu(J)^{-1}\right).
}
{\bf Case B:} $z^{1-\a} < 2R$,
\eqx{\int_{\abs{x^{1-\a}-z^{1-\a}} < R} \, dx \leq C (z^{1-\a} +R)^{\frac{1}{1-\a}} \leq C R^{\frac{1}{1-\a}}.}
In this case
\eq{\lab{case2}
\rhoo(2t) \leq c_2 \left( \rhoo(t) t^{-1/2} R^{\frac{1}{1-\a}} + R^{-1}|J|^2\mu(J)^{-1}\right).
}
Now we proceed to the proof of \eqref{induction}. The first step, $\rhoo(|I|^2)\leq C$, follows simply by \eqref{FKK}. Assume that \eqref{induction} holds for some $n$. The proof is finished by considering four cases. Since the calculations are similar in all the cases we present a detailed argument only in Subcase 1.1.
\noindent
{\bf Case 1:} $\rho(0,I) \geq 2|I|$. In this case $\mu(I) \simeq |I| z^\a$.\\
{\bf Subcase 1.1:} $\rhoo(2^n|I|^2) \geq 2^{n/2}|I|^2 z^{-2}$. Observe that
\spx{
R_1 & := 2^{-1} \rhoo(2^n |I|^2)^{-1/2} (2^n |I|^2)^{1/4}|I|^{1/2} z^{-\a} \\
& \leq 2^{-1} 2^{-n/4}|I|^{-1} z \, 2^{n/4} |I|^{1/2}|I|^{1/2} z^{-\a} \\
& = z^{1-\a}/2.
}
Therefore we can use \eqref{case1} with $R=R_1$ and $t=2^n|I|^2$ together with the induction hypothesis, getting
\spx{
\rhoo(2^{n+1}|I|^2) & \leq c_1[\rhoo(2^n|I|^2)(2^n|I|^2)^{-1/2}2^{-1} \rhoo(2^n |I|^2)^{-1/2} (2^n |I|^2)^{1/4}|I|^{1/2} z^{-\a} z^\a \\
& + 2 \rhoo(2^n |I|^2)^{1/2} (2^n |I|^2)^{-1/4}|I|^{-1/2} z^{\a}|I|^2\mu(I)^{-1}]\\
& \leq c_3 (\rhoo(2^n |I|^2)^{1/2} 2^{-n/4} + \rhoo(2^n |I|^2)^{1/2} 2^{-n/4} |I|^{-1} z^{\a}|I|^2(z^\a|I|)^{-1}) \\
& = 2c_3 \rhoo(2^n|I|^2)^{1/2}2^{-n/4} \\
&\leq 2c_3c_0^{1/2} 2^{-n\frac{1-\a}{4}} 2^{-\frac{n}{4}} \leq c_0 2^{-(n+1)\frac{1-\a}{2}}.
}
The last inequality holds if we choose $c_0$ such that $c_0\geq (2c_3)^2 2^{1-\a}$.\\
{\bf Subcase 1.2:} $\rhoo(2^n|I|^2) \leq 2^{n/2}|I|^2 z^{-2}$. One easily checks that
\eqx{
R_2 := 2^{-1}\left( \rhoo(2^n |I|^2)^{-1} (2^n |I|^2)^{1/2}|I| z^{-\a}\right)^{\frac{1-\a}{2-\a}} > z^{1-\a}/2.
}
Putting $R=R_2$ and $t=2^n|I|^2$ into \eqref{case2} and using the induction hypothesis,
\spx{
\rhoo(2^{n+1}|I|^2) &\leq c_4 \left( \rhoo(2^n|I|^2) (2^n|I|^2)^{-1/2}|I|^{\frac{1}{1-\a}}z^{-\frac{\a}{1-\a}}\right)^{\frac{1-\a}{2-\a}} \\
&\leq c_4 \left( \rhoo(2^n|I|^2) (2^n|I|^2)^{-1/2}|I|\right)^{\frac{1-\a}{2-\a}} \\
&\leq c_4 \left(c_0(2^{-n})^{\frac{1-\a}{2} +\frac{1}{2}}\right)^{\frac{1-\a}{2-\a}}\leq c_0 2^{-(n+1)\frac{1-\a}{2}}.
}
In the last inequality we choose $c_0$ such that $c_0\geq c_4^{2-\a} 2^{(1-\a)(2-\a)/2}$. Notice that we have used $z\geq |I|$, which follows from $\rho(0,I) \geq 2|I|$.
\noindent
{\bf Case 2:} $\rho(0, I) \leq 2\abs{I}$. In this case $\mu(I) \simeq |I|^{\a+1}$. Notice that $z\leq 4 |I|$. \\
{\bf Subcase 2.1} $\rhoo(2^n|I|^2) \geq 2^{n/2}|I|^{2-\a} z^{\a-2}$. Observe that
\eqx{
R_3 := 2^{-1} \rhoo(2^n |I|^2)^{-1/2} (2^n |I|^{2})^{1/4} |I|^{\frac{1-\a}{2}} z^{-\a/2} < z^{1-\a}/2.
}
Putting $R=R_3$ and $t=2^n|I|^2$ into \eqref{case1} and using the induction hypothesis,
\spx{
\rhoo(2^{n+1}|I|^2) &\leq c_5 \rhoo(2^n|I|^2)^{1/2}2^{-n/4} z^{\a/2} |I|^{-\a/2} \\
&\leq c_5 c_0^{1/2} 2^{-n\frac{1-\a}{4}} 2^{-\frac{n}{4}} z^{\a/2} |I|^{-\a/2}\\
&\leq c_5 c_0^{1/2} 2^{\a} 2^{-n\frac{1-\a}{4}} 2^{-\frac{n}{4}}\\
&\leq c_0 2^{-(n+1)\frac{1-\a}{2}}.
}
The last inequality holds if we choose $c_0$ such that $c_0\geq c_5^2 2^{1+\a}.$\\
{\bf Subcase 2.2} $\rhoo(2^n|I|^2) < 2^{n/2}|I|^{2-\a} z^{\a-2}$. One easily checks that
\eqx{
R_4 := 2^{-1} \left( \rhoo(2^n |I|^2)^{-1} (2^n |I|^2)^{1/2}|I|^{1-\a}\right)^{\frac{1-\a}{2-\a}} > z^{1-\a}/2.
}
Putting $R_4$ and $t=2^n|I|^2$ into \eqref{case2} we obtain
\spx{
\rhoo(2^{n+1}|I|^2) &\leq c_6 \left( \rhoo(2^n|I|^2) (2^n|I|^2)^{-1/2}|I|\right)^{\frac{1-\a}{2-\a}} \\
&\leq c_6 \left(c_0(2^{-n})^{\frac{1-\a}{2} +\frac{1}{2}}\right)^{\frac{1-\a}{2-\a}}\leq c_0 2^{-(n+1)\frac{1-\a}{2}},
}
similarly as in subcase 1.2.
}
\lem{lem:conditionK}{
Let $\a\in(0,1)$, $0\leq V\in L^1_{loc}(X,\mu)$, and $\ii(V)$ is as in Proposition \ref{construction}. Then the pair $(V, \ii(V))$ satisfies \eqref{eq:K}.
}
\pr{
{\bf Case 1:} $\rho(0, I) \leq 2\abs{I}$. In this case $\mu(I) \simeq |I|^{1+\a}$. Using \eqref{eq:ptapprox1}, \eqref{ballmeasure} and \eqref{beta},
\alx{
\int_0^{2t}\int_{I^{***}} P_s(x,y) V(y)\, d\mu(y) \, ds \leq C \int_0^{2t} s^{-\frac{1+\a}{2}} ds \cdot \int_{I^{***}} V d\mu \leq C t^{\frac{1-\a}{2}} \frac{\mu(I)}{|I|^2}\leq C \eee{\frac{t}{|I|^2}}^{\frac{1-\a}{2}}.
}
{\bf Case 2:} $\rho(0, I) > 2\abs{I}$. In this case $\mu(I) \simeq |I|c_I^{\a}$, where $c_I$ denotes the center of $I$. Using \eqref{eq:ptapprox1}, \eqref{beta} and the doubling condition,
\alx{
\int_0^{2t} \int_{I^{***}} P_s(x,y) V(y)\, d\mu(y) \, ds & \leq C \int_0^{2t} \int_{I^{***}} \mu(B(y,\sqrt{s}))^{-1} V(y) d\mu(y) ds \\
& \leq C \int_0^{2t} s^{-\frac{1}{2}} c_I^{-\a} ds \cdot \int_{I^{***}} V(y) d\mu(y) \\
& \leq C t^{\frac{1}{2}} \frac{\mu(I)}{|I|^2} c_I^{-\a} \leq C \eee{\frac{t}{|I|^2}}^{\frac{1}{2}}.
}
}
Combining Lemmas \ref{lem:conditionD} and \ref{lem:conditionK} we obtain Theorem \ref{thm:maintheorem2}.
\section*{Appendix}\lab{App}
The goal of this Appendix is to give a precise proof of the formula \eqref{super}. In order to do this we need to pay careful attention to boundary conditions near zero, c.f. \eqref{DomQ} and \eqref{DomL}. Recall that $J=(a,b)$, $0\leq a <b$, and $\int_J V d\mu = \mu(J)|J|^{-2} =: c_J>0$. Moreover, by \eqref{def_phi}, $\phi_I \in L^\8_{loc}[0,\8)$, $\abs{\phi_I(x) -\phi_I(0)} \leq C x^{1-\a}$ at $x=0$, and
\eq{\label{eq:eqq}
\phi_I'(x) =\frac{1}{2} x^{-\a} \begin{cases} - c_J, & x\leq a\\
\int_a^x V(y) \, d\mu(y) - \int_x^b V(y) \, d\mu(y), & x\in(a,b)\\
c_J , & x\geq b,
\end{cases}
}
\eq{\label{eq:eq}
-\phi_I''(x) - \frac{\a}{x} \phi_I'(x) = - \mathbbm{1}_J(x) V(x).
}
Since $\phi_I$ is not bounded in the infinity, we shall need some cut-off functions:
\eqx{
\eta_{n}\in C_c^\8(X), \quad \eta_{n} \rightarrow \mathbbm{1}_X, \quad \chi_{[0,n]}(x) \leq \eta_{n}(x) \leq \chi_{[0,n+1]} (x), \quad \norm{\eta_{n}'}_\8+\norm{\eta_{n}''}_\8 \leq C.
}
\lem{lem:lemat}{
Let $n\in \NN$, $\psi\in \Dom(Q)$, $\psi\geq 0$, $\supp(\psi) \subseteq [0,n]$. Then $ \phi_I \eta_{n+1} \in \Dom(Q)$ and
\eq{\label{ineq}
Q(\psi,\phi_I\eta_{n+1}) \geq 0.
}
}
\pr{
To simplify the notation we denote $\eta = \eta_{n+1}$. Let us first prove that $\phi_I\eta \in \rm{Dom}(Q)$. It is clear that $\sqrt{V} \phi_I\eta \in L^2(X, \mu)$, since $\rm{supp} \, \eta $
is compact, $\phi_I, \eta \in{L^\8_{loc}(X)} $ and $V\in L^1_{loc}(X)$. Observe that $\phi_I \eta \in C^1(X)$, but near zero we only have $\abs{(\phi_I \eta)'(x)} \leq C x^{-\a}$ . Let $\tau_k\in C_c^\8 (X)$ be such that
\eqx{
\chi_{[2/k,\8)}(x) \leq \tau_k(x) \leq \chi_{[1/k,\8)}(x) \ \text { and} \ \norm{\tau_k'}_\8 \leq Ck.
}
Define a sequence $\kappa_k =
\phi_I\eta \tau_k + \phi_I(0) \eta (1-\tau_k)$. It is clear that $\kappa_k \in C_c^1[0,\8)$, $\kappa_k'(0^+) =0$, and $\kappa_k \to \phi_I \eta$ in $L^2(X,\mu)$. Moreover,
\eqx{
\kappa_k' = (\phi_I\eta)' \tau_k + \phi_I(0) \eta' (1- \tau_k) + ( \phi_I- \phi_I(0))\eta \tau_k' = (\phi_I\eta)' \tau_k + r_k.
}
Observe that $\eta' (1- \tau_k)=0$ and
\spx{
\norm{ ( \phi_I- \phi_I(0))\eta \tau_k'}_{L^2(X,\mu)}^2
\leq C \int_{1/k}^{2/k} x^{2(1-\a)} \norm{\tau_k'}^2_{\8} \, d\mu(x)\leq C k^{\a-1},
}
thus $\norm{r_k}_{L^2(X,\mu)} \to 0$. Now, $(\phi_I \eta - \kappa_n)' = (\phi_I\eta)' (1-\tau_k) -r_k \to 0$ in $L^2(X,\mu)$ since $(\phi_I\eta)' \in L^2(X,\mu)$. This finishes the proof that $\phi_I\eta \in \rm{Dom}(Q)$.
What is left is to check \eqref{ineq}. Since $\phi_I \geq 1$, $\psi\geq 0$, and $-\phi_I'(x)x^\a|_{x=0^+} = c_J/2 >0$ the proof follows from a formal calculation based on \eqref{eq:eq}, namely
\eq{\label{eq:equation}
\int_X \psi'(x) (\phi_I(x) \eta(x))'\, d\mu(x)
= -\int_X \psi(x) \mathbbm{1}_J(x) V(x)\, d\mu(x) -\psi(x)\phi_I'(x)x^{\a}|_{x=0^+} .
}
In the rest of the proof, for reader's convenience, we provide a detailed argument for \eqref{eq:equation}. The main problem is to deal with the boundary $x=0$. Since $\psi \in \rm{Dom}(Q)$, we can find $\psi_k \in C_c^1[0,\8)$, $\psi'_k(0^+)=0$, such that $\psi_k \to \psi, \ \psi_k' \to \psi'$ in $L^2(X,\mu)$. We can additionally assume that $\supp(\psi_k)\subseteq [0,n+1]$.
By integrating by parts,
\alx{
\int_X \psi' (\phi_I \eta)' d\mu &= \lim_{k\to \8} \int_X \psi_k' \kappa_k' \, d\mu = \lim_{k\to \8} \int_X \psi_k' (\phi_I \eta)' \tau_k \, d\mu \\
&= - \lim_{k\to \8}\int_X \psi_k \eee{\eee{\phi_I \eta }'' + \frac{\a}{x}\eee{\phi_I \eta }'} \tau_k \, d\mu - \lim_{k\to \8} \int_X \psi_k \eee{\phi_I \eta}' \tau_k' \, d\mu\\
&= A_1 + A_2
}
Since, $\eta = 1$ on $\supp(\psi_k)$, by \eqref{eq:eq} we get $A_1 \to - \int_X \psi \mathbbm{1}_J V \, d\mu$ as $k\to \8$.
Notice that $ \eee{\phi_I \eta}'(x) = \phi_I'(x) = - x^{-\a} (c_J/2 - p(x))$ on $(1/k,2/k) $, where $p(x)=0$ when $a>0$ or $p(x)=\int_0^x V \,d\mu$ when $a=0$. We shall consider only the latter case, getting
\alx{A_2 &= \limk \int_{1/k}^{2/k} \psi_k \eee{\frac{c_J}{2} - p(x)} \tau_k' \, dx\\
&= \limk \eee{ \frac{ c_J}{2} \psi_k(2/k) - \psi_k(2/k) p(2/k) - \int_{1/k}^{2/k} \psi_k' \eee{\frac{c_J}{2} - p(x)} \tau_k \, dx + \int_{1/k}^{2/k} \psi_k V \tau_k \, d\mu(x)}\\
& = A_3+A_4+A_5+A_6.
}
From the fact that $\psi,\psi_n, \psi', \psi_n' \in L^2(X)$ one can deduce that $\psi, \psi_n \in C[0,\8)$. Moreover, by the Cauchy-Schwarz inequality,
\eqx{
\abs{\psi_k(x)-\psi_k(0)} = \abs{\int_{0}^{x} \psi_k'(y) dy} \leq C \norm{\psi_k'}_{L^2(X,\mu)} x^{(1-\a)/2} \leq C x^{(1-\a)/2}.
}
As a consequence we get that $\psi_k(0) \to \psi(0)$ and $A_3 \to \phi(0) c_J/2 > 0$. The proof is finished by noticing that $A_4,A_6 \to 0$ and
\spx{
|A_5| &\leq C \int_{1/k}^{2/k} \psi_k'(x) dx \leq C \sup_j \norm{\psi_j'}_{L^2(X,\mu)} k^{(\a-1)/2} \to 0.
}
}
Recall that $z\in J$ is fixed and denote
$
\vt(u) = \Kt{u} \phi_I(z).
$
Our goal is to prove that $\vt(t+s)\leq \vt(t)$ for $t,s>0$.
Denote $k(x):=\Kt{t}(x,z) = \Kt{t/2}(\Kt{t/2}(\cdot,z))(x)$. Since the semigroup $\Kt{t}$ is analytic and $\Kt{t/2}(\cdot,z)\in L^2(X,\mu)$ (see \eqref{FKK} and \eqref{eq:ptapprox1}), we have $k\in \Dom(\opL)\subset \Dom(Q)$.
First, observe that
\alx{
(\Kt{u}(k)\eta_{n-1})'(\phi_I\eta_{n+1})' & = (\Kt{u}(k)\eta_{n-1})'\phi_I' = \Kt{u}(k)'\phi_I'\eta_{n-1}+ \Kt{u}(k)\phi_I'\eta_{n-1}'.
}
Using this,
\sp{\label{spsp}
Q(\Kt{u}(k),\phi_I\eta_{n-1}) & = \int_X \Kt{u}(k)'(\phi_I\eta_{n-1})' \, d\mu + \int_X V \Kt{u}(k) \phi_I \eta_{n-1} \, d\mu \\
& = \int_X {\Kt{u}(k)'\phi'_I\eta_{n-1}} \, d\mu + \int_X \Kt{u}(k)'\phi_I \eta'_{n-1} \, d\mu + \int_X V \Kt{u}(k) \phi_I \eta_{n-1} \, d\mu \\
& = \int_X (\Kt{u}(k)\eta_{n-1})'(\phi_I\eta_{n+1})' \, d\mu - \int_X \Kt{u}(k) \phi'_I \eta'_{n-1}\, d\mu + \int_X \Kt{u}(k)'\phi_I \eta'_{n-1} \, d\mu \\
& \quad + \int_X V (\Kt{u}(k) \eta_{n-1}) (\phi_I \eta_{n+1}) \, d\mu \\
& = Q(\Kt{u}(k)\eta_{n-1}, \phi_I \eta_{n+1})\\
& - 2\int_X \Kt{u}(k) \phi'_I \eta'_{n-1}\, d\mu - \int_X \Kt{u}(k) \phi_I \eta''_{n-1} \, d\mu - \int \Kt{u}(k) \phi_I \eta'_{n-1} \frac{\a}{x} \, d\mu\\
& = Q(\Kt{u}(k)\eta_{n-1}, \phi_I \eta_{n+1}) - 2B_1 - B_2-B_3.
}
\prop{decreasing}{ For fixed $I$ and $z\in J\supseteq 2I$ the function ${\vt(u)= \Kt{u} \phi_I(z)}$ is non-increasing.}
\pr{ Observe that $\left(\Kt{s}(k) - k\right) \phi_I$ is in $L^1(X, \mu)$. Using \eqref{spsp},
\spx{
\vt(t+s) - \vt(t)
& = \int_X \left(\Kt{s}(k) - k\right)(x) \phi_I(x) \, d\mu(x) \\
& = \limn \int_X (\Kt{s}(k) - k)(x) \phi_I(x) \eta_{n-1}(x) \, d\mu(x) \\
& = \limn \int_X \left( \int_0^s (-\opL) \Kt{u}(k)(x) \, du \right) \phi_I(x) \eta_{n-1}(x) \, d\mu(x) \\
& = - \limn \int_0^s \int_X \opL \Kt{u}(k)(x) \phi_I(x) \eta_{n-1}(x) \, d\mu(x) \, du \\
& = - \limn \int_0^s Q(\Kt{u}(k),\phi_I\eta_{n-1}) \, du \\
& = - \limn \int_0^s \left( Q(\Kt{u}(k)\eta_{n-1}, \phi_I \eta_{n+1}) - 2 B_1 - B_2 -B_3\right) \, du.
}
Having in mind Lemma \ref{lem:lemat} it is enough to show that $\int_0^s B_i \, du \to 0$ as $n\to \8$ for $i=1,2,3$. This follows from Lebesgue's Dominated Convergence Theorem and the estimates we have already established. For example, for $B_2$ observe that $\abs{\phi_I(x) \eta_{n-1}''(x)} \leq C |x|^{1-\a}\mathbbm{1}_{{[n-1,n]}}(x)$ for $n\geq N$ with $N$ large enough. Then a majorant is
\alx{
\int_0^s \int_X &\sup_{n\geq N} \abs{\Kt{u}(k)(x) \phi_I(x) \eta_{n-1}''(x)}\, d\mu(x) \, du \\
&\leq C \int_0^s \int_X \frac{|x|^{1-\a}}{\mu(B(x,\sqrt{t+u}))} \exp{\eee{-\frac{|x-z|^2}{c_2(t+u)}}} x^\a\, dx \, du\\
&\leq C \int_0^s \int_X \frac{|x|^{1-\a}}{(t+u)^{1/2}} \left(\frac{x}{x+\sqrt{t+u}}\right)^{\a} \exp{\eee{-\frac{|x-z|^2}{c_2(t+u)}}}\, dx \, du\\
&\leq C t^{-1/2} s \int_X |x|^{1-\a} \exp{\eee{-\frac{|x-z|^2}{c_2(t+s)}}} \, dx \\
&\leq C(I,t,s).
}
The integrals with $B_3$ and $B_1$ goes similarly. For the latter one we use {$\abs{\phi_I'(x)} \leq C x^{-\a}$.}
}
\noindent
{\bf Acknowledgments:} The authors would like to thank Jacek Dziuba\'nski and the referees for their helpful comments.
\bibliographystyle{mn}
\providecommand{\WileyBibTextsc}{}
\let\textsc\WileyBibTextsc
\providecommand{\othercit}{}
\providecommand{\jr}[1]{#1}
\providecommand{\etal}{~et~al.}
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 9,195
|
\section{Introduction}
Twitter is one of the most popular worldwide social media platforms with over 330 million monthly users and approximately 500 million tweets every day\cite{twitter}.
Their recommendation system has to meet many challenges such as fast recommendation time and adaptation to rapidly changing hashtags and real-world news. A central concept of Twitter recommendations is fairness: the quality of the recommendations should be equally high regardless of the popularity of authors, tweets, and individual languages. All these challenges are reflected in this year's competition hosted by Twitter - the ACM RecSys 2021 Challenge. The aim of the challenge is to predict whether a given user will react to a given tweet. The competing systems are required to provide fair recommendations and work in a strictly resource-constrained setting, requiring very small model size and fast inference.
Our approach takes the second place on the final leaderboard. The model relies on smart input feature preparation with our Efficient Manifold Density Estimator (EMDE) \cite{emde} and Fourier Feature Encoding (see \S\ref{fourier}) methods (among others), used to represent the tweet text, historical interactions of users and their Twitter account status. The underlying model is a simple feed-forward neural network. Our contributions are the following:
\begin{enumerate}
\item We adapt the EMDE to a new domain and obtain a successful result. Previously, EMDE has achieved top results in session-based and top-k recommendations \cite{emde}, and various challenges: KDD Cup 2021 (predictions in an academic graph) \cite{kddcup}, WSDM Challenge 2021 (recommendation of travel destinations) \cite{daniluk2021modeling}, and SIGIR eCom Challenge 2020 (multimodal retrieval) \cite{sigir_ecom}.
\item We introduce the Fourier Feature Encoding which allows to express continuous features in a numerically stable way without normalization.
\item We analyze the effectiveness of various input features, such as user cluster assignments, user account features, and tweet text features.
\item We analyze the fairness concept in detail, taking into account tweet languages and user popularity. We find that our model mostly gives fair recommendations.
\item We show how our design choices lead to the creation of a very efficient system, with single prediction taking about 4ms on a single CPU without a GPU card.
\end{enumerate}
\section{RecSys 2021 Challenge}
The ACM RecSys 2021 Challenge \cite{DBLP:conf/recsys/AnelliKFBTPL0XH21} focuses on the real-world task of predicting whether a user will engage with a given tweet. The competing models must predict the probability of engagement for each of the four reaction types: \textit{Like}, \textit{Reply}, \textit{Retweet} and \textit{Quote}.
The released data set \cite{belli20212021} consists of more than 1 billion data points from 28 consecutive days between 4 February 2021 and 4 March 2021.
The data set has a tabular structure with 20 raw features.
The first three weeks were used for training, while the last week was randomly divided into validation and test sets.
At the beginning of the challenge, the models were evaluated on the public leaderboard that used the validation data set. Two weeks before the end of the competition, the validation targets were released as an extra data source for training and a new test leaderboard was introduced. This changed the competition setting as the validation set is sampled from the same time distribution as the final test set, while the training set is sampled from the past.
It imitates the real production environment, where models can be fine-tuned in real-time to adapt to current trends.
The data set contained both positive and negative examples of different types of engagement between a given tweet and the users. The available features describe two types of users, which we further call \textit{engaging user} and \textit{engaged user} for consistency with Twitter nomenclature. \textit{Engaged} user denotes the creator of the given tweet, while \textit{engaging user} denotes the user who reacts (or chooses not to react) to the tweet. Each tweet is represented as set of tokenized wordpiece ids from multilingual BERT model from 66 different languages such as English, Japanese or Thai. Moreover, the data contains information about hashtags, present media, links domains, number of followers of engaged and engaging users, verification of both user's accounts and information if engaged user follows engaging user.
This is the second edition of the challenge hosted by Twitter, with the previous RecSys Challenge 2020 being very similar in terms of data structure and targets. Four main changes were introduced in the 2021 edition. First, the challenge brings the problem even closer to Twitter's real recommendation systems by introducing strict latency constraints. Each model needed to be uploaded and evaluated in a separate test environment provided by the organizers, with only 1 CPU core, no GPU, $64$GB RAM size, $20$GB disk space, and a time limit of 24 hours for making all predictions, which gives about 6ms per single tweet prediction. It encouraged teams to develop novel solutions that can be easily applied in the production environment.
Second, the size of training data was increased from one week to three weeks.
Third, the data density was increased in terms of the graph where users are considered to be nodes and interactions as edges.
Additionally, recommendation fairness was cast as a vital part of the challenge.
The metrics used for performance evaluation were relative cross entropy (RCE) and the average-precision (AP) for each type of engagement. The fairness concept was included in the metrics by dividing authors into 5 groups according to their popularity on the platfom and evaluating separately for each group. The final score was computed as the average of the scores across each group.
\section{Related Work}
The last year's challenge - the Twitter Recsys Challenge 2020 - was very similar to the 2021 edition, with one of the core differences being that it allowed to build more complex models, due to no constraints on latency and smaller data volume. The winners \cite{10.1145/3415959.3415996} introduced an XGBoost model with extensive feature engineering, encoding features with techniques such as Target Encoding, or difference lag (time difference for datetime features). They found that experimental detection of adequate feature transformations and feature combinations was of chief importance. The runner ups \cite{volkovs2020predicting} included extensive feature engineering as well (467 features total). They introduce a Transformer model for comparing the embedding of current tweet and a collection of historical user tweets, overall creating a mixed architecture of deep learning and gradient boosting trees. The 3rd place team \cite{10.1145/3415959.3415994} aimed to exploit correlations between various target variables with a two-stage approach using LightGBM models.
Overall, top past challenge solutions often included a simple (usually non-neural) classifier model with the bulk of work put into feature engineering. This is the approach whose merit is further confirmed by our submission in a much more resource-constrained setting.
\section{Approach}
The first step of our approach is to obtain a tweet text representation by fine-tuning a DistilBERT model \cite{sanh2019distilbert} on tweets, and using EMDE \cite{emde} to represent text as a \textit{sketch} (a compressed, fixed-size representation of the tweet meaning). We then extract features that describe tweet content, the engagement history of engaging and engaged users, and their account status features. Text sketch and features are fed into a simple shallow feed-forward neural network which is trained to predict each type of engagement.
Since the time distribution of validation and training sets is different, we train the model in two stages: 1) using training set, and then 2) fine-tuning the model on validation set, which was released two weeks before the end of the competition. Note that the training set contains 3 weeks of historical data, but validation set is taken from the same time distribution as final test set.
\begin{figure}[ht]%
\subfigure[A non-overlapping 24 hours sliding window was applied to the training period. Engagements from a single day were used as training targets, while all the rest are used for feature extraction.]{
\includegraphics[width=82mm]{data_partition.png}
}
\subfigure[A validation set was randomly divided into 10 parts. Each of them was used as training targets, while the rest 9 parts were used for feature extraction.]{
\includegraphics[width=82mm]{data_partition_valid.png}
}
\caption{Data Partitioning. First we train our model using training set, then we fine-tune it using validation set which was sampled from the same time distribution as hidden test set.}
\label{fig:data_partition}
\end{figure}
\subsection{Data Partitioning}
The data partitioning process is depicted in Figure \ref{fig:data_partition}.
Inspired by \cite{volkovs2020predicting}, we apply a non-overlapping 24 hours sliding window to the training period. As a result, all training examples are divided into day-sized chunks based on engagement time for positive examples or tweet creation date for negative samples. Engagements from a single day were used as training targets, while all the rest are used for feature extraction. Note that temporal structure was broken here, because we use features from the future interactions to predict current engagement. This partitioning results in creating 21 training parts that were shuffled before training.
The local evaluation of our model was performed on a 10\% random sample from the provided validation set.
Additionally, we apply similar partition procedure to the validation set whose targets were released by the organizers, but instead of partitioning the data per day, we randomly divided it into 10 parts. Each of them was used as training targets, while the rest 9 parts were used for feature extraction.
\subsection{Feature extraction}
In order to describe users and target tweets, we apply feature engineering with a particular focus on historical engagements of tweet creator and engaging users. For each data point, we compute the following features:
\begin{itemize}
\item \textbf{Interactions between engaged and engaging user}. These features summarize historical engagement between the author of a tweet and engaging user. We simply count the number of engagements for each reaction between the engaged and engaging users on historical part of data.
\item \textbf{Engaged user interactions.} We extract features that summarize the history of interactions of engaged user by counting the number of interactions of each type of engagement which this author received, which describes the total level of engagement they received.
\item \textbf{Engaging user interactions.} Similarly to the engaged user features, we count the number of each type of reaction which the engaging user gives to any other user.
In addition, we calculate the number of interactions of engaging user with the tweets that are in the same language as the target one. We also incorporate historical interactions between the engaging user and hashtags from the current tweet.
\item \textbf{Interactions between engaging user and users similar to the engaged user.} We count the number of interactions between engaging user and the users which are similar to engaged user. Similar users are detected in the following way: first, for each user $U$, a set of users $F_U$ who follow the user $U$ is created. Then, for each user pair $(A,B)$ we compute their follower similarity score as Jaccard similarity between the follower sets $F_A$ and $F_B$ using an efficient Python set similarity implementation\footnote{\url{https://github.com/ekzhu/SetSimilaritySearch}}. If the similarity score falls above a threshold (selected experimentally), we mark the two users as similar. Clusters of similar users are precomputed at the training stage and reused during inference.
\item \textbf{Interaction with tweets.} Since some tweets from the validation set also appear in the test set, we count the number of interactions with a specific tweet id from the validation set.
\item \textbf{Account status.} We extract features that describe the accounts of a both engaged and engaging users such as: number of followers, number of following users, binary flag indicating if the account of a user is verified, time since account creation.
\item \textbf{Interaction clusters.} We perform community detection on directed graphs of engaged-engaging user interactions utilizing Leiden algorithm\citep{Traag2019} implemented in \textit{leidenalg}\footnote{\url{https://github.com/vtraag/leidenalg}} package. We calculate Modularity Vertex Partitions with default settings on 4 graphs spanned by all 4 engagement types. For each graph we have 2 features: a 0/1 variable indicating if both users belong to the same partition, and the inverse of partition cardinality (zero when users belong to different partitions). This allows us to capture potentially complex communities of mutual interaction between users, with large communities having lower interaction strength.
\item \textbf{Tweet Content.} Features summarizing the content of a target tweet, e.g: number of hashtags, language of a tweet, the presence of additional content (image, video, gif, links, media, domains), the type of tweet (quote, retweet or top level), time of a tweet.
\end{itemize}
\begin{figure*}[ht]%
\centering
\includegraphics[width=150mm]{emde-recsys.png}
\caption{EMDE used to represent tweets as token sketches. The matrices $M$ hold ids of regions to which each data point had been assigned after a density-aware LSH space partitioning. The matrices $B$ hold sparsified versions of matrices $M$. Matrices $B$ (representing each word) are summed and normalized to form an aggregate sketch (jointly representing the meaning of all words in a tweet). Aggregate sketches are used directly as input to a downstream model.}
\label{fig:emde-tokens}
\end{figure*}
\subsection{Fourier Feature Encoding}
\label{fourier}
All numerical features such as number of interactions, number of followers, time since account creations are encoded with our Fourier Feature Encoding, which is a way to sidestep the necessity of explicit normalization of model inputs. Instead of feeding a single input feature, we transform it into a 16-dimensional vector. First, an input numeric value is divided by a few numbers representing increasing scale levels (in our case, 8 scale levels represented by powers of 2). Then, each of the 8 results is fed to $sin$ and $cos$ functions. The numeric significance of the feature is thus represented on multiple levels, and in a numerically stable way as any number is brought to a small numeric interval defined by trigonometric functions. Fourier Feature Encoding does not usually offer significant performance gains but it facilitates the usage of continuous features. An example of values obtained from the algorithm is shown in Figure \ref{fig:fourier}. Below we present a code snippet of the Fourier Feature Encoding.
\begin{lstlisting}[language=Python]
import numpy as np
def multiscale(x, scales):
return np.hstack([x.reshape(-1,1)/pow(2., i) for i in scales])
def encode_scalar_column(x, scales=[-1, 0, 1, 2, 3, 4, 5, 6]):
return np.hstack([np.sin(multiscale(x, scales)),
np.cos(multiscale(x, scales))])
\end{lstlisting}
\begin{figure}%
\subfigure[The sine component]{
\includegraphics[width=40mm]{fourier_sine.png}
%
}
\subfigure[The cosine component]{
\includegraphics[width=40mm]{fourier_cosine.png}%
}
\caption{Values obtained from the Fourier Encoding for 8 scale levels (powers of 2: $2^1$,...,$2^8$), and input value of 100.}
\label{fig:fourier}
\end{figure}
\subsection{Tweet representation}
Due to the inference time limit, embedding tweets on the fly with Transformer-based models such as BERT was not feasible. Thus, we utilized a more time-efficient approach representation of tweet contents based on pretrained DistilBERT \cite{DBLP:journals/corr/abs-1910-01108} vectors fed to EMDE \cite{emde}.
EMDE uses a density-aware manifold partitioning method in order to create meaningful manifold regions. Each region holds samples which are similar according to distance metric expressed within the embedding space which spans the manifold. EMDE can aggregate item sets of arbitrary size into a fixed-side structure which works well with simple feed-forward networks, and much of the computational effort can be done once, before training and inference.
First, in order to acquire high quality token embeddings, we fine-tune a generic pretrained DistilBERT multilingual model \textit{distilbert-base-multilingual-cased} from the Huggingface library \cite{wolf-etal-2020-transformers}. We fine-tuned the DistilBERT model using the training set, then applied the portion of the validation set which is available for training. In order to construct EMDE sketches, we needed to obtain per-token embeddings. We achieved this by embedding all tweets with our DistilBERT model and computing the average per-token embedding. We fit EMDE on a single manifold spanned by all token vectors, irrespective of the language. As a result, we obtain multiple data-aware space partitionings, with similar tokens located in the same regions frequently (per analogy to Locality-Sensitive Hashing \cite{emde}). EMDE sketches have the property of additive compositionality, which allows us to create an aggregate tweet content representation with simple summation of all tweet token sketches. The aggregate per-tweet sketches are finally L2-normalized. The procedure is displayed in Figure \ref{fig:emde-tokens}.
\subsection{Model}
We base our model architecture on our previous design choices which proved appropriate for recommendation with EMDE \cite{emde}, various previous challenges \cite{daniluk2021modeling,sigir_ecom,kddcup}, and which form the backbone of our commercial system. We train a three-layer residual feed-forward neural network with 1500 neurons in each hidden layer, with leaky ReLU activations and batch normalization. The input of the network consists of the following feature vectors, which are simply concatenated:
\begin{enumerate}
\item Width-wise L2-normalized sketch that represents the text of a tweet.
\item Numeric features such as number of historical interactions, number of followers etc. which are encoded with our Fourier Feature Encoding.
\item Categorical features such as current day, time of tweet and language of tweet that are represented by embedding layers.
\end{enumerate}
The output of the network consists of 4 neurons that represents predictions of like, reply, retweet and quote engagements for a target tweet. The model is trained by optimizing a binary cross entropy loss function for each engagement type.
\section{Experiments}
In this section, we first describe our training setup (§\ref{training}), then we present our final result on the leaderboard (§\ref{results}), and analyze the effectiveness and challenges of our method (§\ref{ablation}).
\subsection{Training}
\label{training}
We train our model on single Tesla V100 16 GB GPU card. Training takes circa 24 hours on one billion training data points. Then, we fine-tune the model on the released validation set, which takes about 60 minutes.
We use AdamW optimizer \citep{loshchilov2017decoupled} with first momentum coefficient of 0.9 and second momentum coefficient of 0.999 \footnote{Standard configuration recommended by \citep{kingma2014adam}} with an initial learning rate of $0.0001$, weight decay of $0.01$ and a mini-batch size of 256 for optimization. The learning rate was linearly decayed.
The final model was trained for 2 epochs on the training set and then fine-tuned for 3 epochs on the validation set.
\subsection{Results}
\label{results}
Table \ref{tab:results} presents results on the final leaderboard. Our team achieves 2nd place in this competition with comparable performance on both AP and RCE metrics.
\begin{table*}[ht]
\centering
\resizebox{\textwidth}{!}{
\begin{tabular}{l|lllllllll}
\textbf{Method} & \textbf{AP Retweet} & \textbf{RCE Retweet} & \textbf{AP Reply} & \textbf{RCE Reply} & \textbf{AP Like} & \textbf{RCE Like} & \textbf{AP Quote} & \textbf{RCE Quote} & \textbf{Time Taken} \\ \hline
NVIDIA & 0.4614 & 29.5127 & 0.2649 & 26.6123 & 0.7216 & 23.6124 & 0.0692 & 17.6868 & 23 hours \\
\textbf{Synerise AI} & 0.4514 & 28.5222 & 0.2559 & 25.7468 & 0.7046 & 22.0994 & 0.0662 & 16.9245 & 18 hours \\
LAYER6 AI & 0.4317 & 27.4239 & 0.2490 & 25.3526 & 0.6836 & 19.8578 & 0.0660 & 16.8696 & 13 hours \\
test\_lightgbm & 0.4060 & 25.0928 & 0.2118 & 22.6491 & 0.6636 & 17.9193 & 0.0520 & 14.0357 & 5 hours \\
final1 & 0.3940 & 24.0142 & 0.2077 & 22.1539 & 0.6559 & 16.9609 & 0.0459 & 12.6722 & 19 hours
\end{tabular}}
\caption{Performance of top five teams on the final challenge leaderborad. Our approach \textbf{Synerise AI} takes 2nd place.}
\label{tab:results}
\end{table*}
\begin{table*}[ht]
\centering
\resizebox{\textwidth}{!}{
\begin{tabular}{l|llllllll}
\textbf{Model inputs} & \textbf{AP Retweet} & \textbf{RCE Retweet} & \textbf{AP Reply} & \textbf{RCE Reply} & \textbf{AP Like} & \textbf{RCE Like} & \textbf{AP Quote} & \textbf{RCE Quote} \\ \hline
Interactions & 0.405 & 25.488 & 0.170 & 19.219 & 0.697 & 18.682 & 0.0593 & 14.594 \\
\textbf{+} tweet content features & 0.427 & 27.382 & 0.221 & 24.614 & 0.7296 & 23.047 & 0.0636 & 16.177 \\
\textbf{+} users account features & 0.429 & 27.679 & 0.222 & 24.869 & 0.733 & 23.223 & 0.0641 & 16.373 \\
\textbf{+} interaction clusters & \textbf{0.431} & \textbf{27.970} & \textbf{0.227} & \textbf{25.183} & \textbf{0.734} & \textbf{23.676} & \textbf{0.0663} & \textbf{16.770} \\ \hline
\textbf{-} emde & 0.424 & 27.285 & 0.213 & 24.156 & 0.730 & 23.069 & 0.0643 & 16.068
\end{tabular}}
\caption{Ablation study results. Features are incrementally added (+) or removed (-) from the model input.}
\label{tab:ablation}
\end{table*}
\subsection{Ablation studies}
\label{ablation}
In order to understand the effects of crucial parts of the training process, we conduct additional experiments. Due to the long training time, we train our models only on $90\%$ of validation set, but use all interactions from the training set as input features. The remaining part of the validation set was used for our final evaluation.
To ensure a comparable number of parameters of all models, we adjusted the hidden size to have roughly the same total number of model parameters.
The ablation results are summarized in Table \ref{tab:ablation}. In the \textit{Interaction} setting our model takes as input only historical engagement information of target users. By including tweet content features (sketch representation of tweet text; presence of video, image, gif; type of tweet), we observed the average $11.8\%$ increase of the average precision score and $17.4\%$ increase of relative cross entropy score. The biggest improvement is for \textit{reply} reaction, which may lead to the conclusion that information about the content of the tweet, e.g. information that the author is asking a question, is particularly important for reply engagement.
Adding user account features (number of followers, verification, binary flag indicating if engaging user follows the author of a tweet) improves the AP and RCE scores by $0.5\%$ and $1\%$, respectively. In addition, we verify the impact of features based on user communities. They increase the average AP score from $0.362$ to $0.365$, and the average RCE score from $23.0$ to $23.4$.
In the next set of experiments, we compare another representation of tweet text. We replace EMDE with an average of all tokens' embeddings learned by our fine-tuned DistilBERT model. Note that embedding the tweets on the fly during inference breaks the latency constraints, so we use precomputed and per-token averaged embeddings. The results are presented in Table \ref{tab:ablation}.
Elimination of EMDE decreases metrics significantly by $2.8\%$ for AP score, and $2.3\%$ for RCE on average.
\begin{figure}%
\subfigure[Reply]{
\includegraphics[width=40mm]{reply.png}
%
}
\subfigure[Like]{
\includegraphics[width=40mm]{like.png}%
}
\subfigure[Retweet]{
\includegraphics[width=40mm]{retweet.png}
%
}
\subfigure[Quote]{
\includegraphics[width=40mm]{quote.png}%
}
\caption{Average precision scores for each type of reaction in terms of the popularity of the author of a tweet. Each data point was assigned to one of the 200 quantile groups based on the author's number of followers in test set. The popularity increases with the X-axis.}
\label{fig:fairness}
\end{figure}
We also verify the impact of the fairness concept by dividing the test examples into 200 quantile groups based on the popularity of the author of a tweet (computed as the number of the author's number of followers). The results are visualised in Figure \ref{fig:fairness}. We can observe that there is no strong trend in the average precision score for \textit{reply}, \textit{retweet} and \textit{quote} reactions. However, the performance of predicting \textit{likes} increases with the popularity of the author.
Additionally, we evaluate our model for different tweet languages. Figure \ref{fig:languages} shows the visualization of average precision score in terms of language popularity. Generally, less popular languages achieve scores similar to most popular languages (with some exceptions). A strong outlier is Thai language, which achieved significantly lower score in spite of being fairly popular. We hypothesize that this is due to the fact that our DistilBERT had been initially trained on 104 languages, however Thai was excluded according to Huggingface documentation\footnote{\url{https://huggingface.co/distilbert-base-multilingual-cased}}.
\begin{figure}[ht]%
\includegraphics[width=70mm]{lang.png}
\caption{Evaluation for different languages of a tweet. The visualization shows the average precision score (y-axis) in terms of the popularity of the language (x-axis), which was computed as number of tweets in particular language.}
\label{fig:languages}
\end{figure}
\section{Summary}
In this paper we present our model which achieves 2nd place in the ACM RecSys Twitter 2021 Challenge. Our model is a 3-layer feed-forward neural network, which ingests tweet text representation encoded with EMDE along with numerical and categorical features that describe users and target tweet. The model is very efficient and adheres to the strict latency constraints within the competition - a single prediction takes about 4ms on a single CPU without a GPU card.
\bibliographystyle{ACM-Reference-Format}
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 286
|
ITMB Maps Tanzania - East Africa is one of the gems of eco-tourism, with its well-protected national parks and wildlife preserves. Tanzania is the best of the region, hosting the massive Serengeti Preserve, Mount Kilimanjaro, the amazingly fertile Ngorogoro Crater. Lake Nanyiki, and the vast grasslands of Selous. These are all shown on our map, as well as Lakes Victoria and Tanganyika, with roads, petrol stations, accommodation, and medical clinics included, whenever known. The map has an excellent inset map of Dar es Salaam, and insets of Arusha and Stone Town in Zanzibar. Printed on paper, double-sided. Map size: 27" X 39"
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 8,264
|
Although many travelers believe they know it all when it arrives to heading abroad, but there is a lot much more to think about, particularly when travelling to new places you know little to absolutely nothing about. Here You will discover essential touring suggestions from flights, when to get your tickets, packing tips, and other common guidelines of thumb. You will also be uncovered to much more particular Australian travel suggestions, like local regulations, guidelines, regulations, or unusual customs you may be unfamiliar with. Most travelers appear like travelers-fanny packs, maps, cameras tied about their necks, and the notorious Hawaiian shirts, but you don't have to go without a general concept as to what to anticipate.
The best factor about Northern Virginia is that it is very near to the money metropolis, Washington D.C. At the exact same time, it is outfitted with all the fundamental amenities that arrive in handy with remaining in the suburbs. That means you can discover the ideal combination of activity and solitude in any of the Fairfax homes for sale. You can discover reduction from the continuous noise and humdrum of city life without becoming completely dissociated with it.
Jerome, AZ is the Place to Go for Halloween – If you are looking for a location to spend Halloween, allow me recommend to you the hillside mining ghost town of Jerome, Arizona. Jerome is situated in the beautiful Verde Valley not far from Sedona and within driving distance from Phoenix. Right here are some factors why Halloween would be a great time to visit Jerome. Full Article.
The women were scheduled to Traveling to the Philippines to Wilmington last Saturday, but the journey was postponed to Wednesday due to snow that day. Richter said the additional time off was good for her group because it gave extra time to permit several gamers who had been sick or injured to recover.
No 1 knows what happened during the last reversal. But specialists think that a polar reversal will make the Earth spin in the opposite direction. It will produce horrendous situations like earthquakes, tsunamis, and huge electrical storms in each part of the globe.
Reynolds stated 2,500 copies of his low cost flyer have been printed and many of them are being dispersed via the Friday Folder district elementary college students consider house. Fans can show the flyers for $1 off admission to both video games Saturday for all family members who want to attend.
If you are considering obtaining whether or not to get pain medication, know that you can't make ideas prior to a beginning. Things occur very fast throughout labor. Trust your doctor's advice, and don't decide yourself for your choices. There's absolutely nothing wrong with obtaining the epidural.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 252
|
Дом де Леон () — знатный бретонский род, представители которого правили в виконтстве Леон.
История
Первым достоверно известным основателем дома де Леон стал виконт Эрве I (II). Во время его правления виконтство Леон достигло высшего могущества. В 1139 году Эрве прибыл в Англию с целью поддержать короля Стефана. Стефан женил Эрве на своей внебрачной дочери Сибилле и наградил его титулом графа Уилтшира. Эрве участвовал в гражданской войне против императрицы Матильды на протяжении двух лет, до 1141 года, когда был лишен титула и владений в Англии Матильдой, ставшей английской королевой, после чего ему пришлось вернуться в Бретань.
После смерти Эрве I в 1168 году его преемником стал его старший сын Гиомар III. В 1171 году младший сын Эрве I, Аймон, ставший виконтом Леона, поссорился со своим братом Гиомаром III, а 21 января 1171 года, согласно средневековому хронисту Роберту де Ториньи, был им убит. В 1179 году замок виконтов Леона был захвачен англо-нормандскими противниками его отца. В этом же году Гиомар III скончался.
Захватчики виконтства, герцог Бретани Жоффруа II Плантагенет и его отец король Англии Генрих II Плантагенет позволили сыновьям Гиомара, Гиомару IV и Эрве II (III) получить некоторые владения их отца. Гиомару IV досталось одиннадцать имений. Больше упоминаний о братьях у хронистов не упоминается. Их сестра Жанна вышла замуж за Эда II, графа де Пороэт, и ставшего впоследствии герцогом Бретани.
В мае 1231 года упоминается некий Гиомар, принесший оммаж королю Франции Людовику IX. 7 июня 1254 года упоминается виконт Леона Эрве III (IV), который скончался в 1264 или 1265 году. Его сыном был Эрве IV (V), упоминаемый в сентябре 1273 года.
Сестрой последнего была Эми де Леон, которая вышла замуж за сеньора де Динан Ролана III. В октябре 1276 года Ролан продал приданое своей жены герцогу Бретани Жану I. Эрве IV имел единственную дочь, также по имени Эми, мужем который был Прижен де Коэтмен, виконт де Тонкедек. 7 июня 1298 года Прижен продал и свою часть виконтства, сыну Жана I, Жану II, наследовавшему своему отцу в Бретани. В то время Эрве IV был еще жив, но после его смерти виконтство Леон окончательно перешло к герцогам Бретани.
Генеалогия
I. Эрве I (II) (ум. 1168) — виконт Леона и граф Уилтшир
II. Гиомар III (ум. 1179) — виконт Леона
III. Гиомар IV (ум. после 1179) — виконт Леона
III. Эрве II (III) (ум. после 1179) — виконт Леона
I. Гиомар V (ум. после мая 1231) — виконт Леона
I. Эрве III (IV) (ум. 1264/1265) — виконт Леона
II. Эрве IV (V) (ум. после 7 июня 1298)) — виконт Леона
III. Эми (ум. после октября 1276); муж — Прижен де Коэтмен (ум. после 1300), виконт де Тонкедек
II. Эми (ум. после 7 июня 1298); муж — Ролан III де Динан (ум. после 1 мая 1303), сеньор де Динан
Ссылки
Виконты Леона
Династии средневековой Европы
Бретонские династии
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 2,335
|
Tomography is an imaging technique for reconstructing an object from projection images. Projection images are obtained by a scanner consisting of a radiation source and a detector. Due to imperfections in the scanner setup, measurement errors are introduced in the projections, which lead to errors in the reconstructed image. In this thesis we developed several techniques to make reconstruction algorithms more robust to these imperfections, such that errors in the projection data have a smaller effect on the reconstruction. We consider errors caused by: misalignment, low radiation dose, unknown background intensities and nonlinearities in the projection acquisition. The key concept that is used for correcting errors is consistency optimization of the projection data and the reconstruction. By using a forward projection model, projections from a reconstruction can be simulated. If the model (and the reconstruction) is accurate the simulated projections match the observed projections. Consistency is therefore a measure which we can use to estimate parameters of the model. For example, misalignment can be corrected for by introducing geometric parameters for the position of the detector and perturbations in the projection angles. Once the parameters correspond to the experiment the consistency is maximized and reconstruction errors are reduced.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 6,749
|
Q: "Request method 'POST' not supported" after put security annotations I have 2 controllers:
a)
@Override
@RequestMapping(value = "/portal/form.html", method = RequestMethod.GET)
@Transactional
public String form(final Message message, final Model model) {
return "portal/form";
}
b)
@Override
@RequestMapping(value = "/portal/form.html", method = RequestMethod.POST)
@Transactional
@PreAuthorize("#message.id!=null ? hasPermission(#message, 'WRITE') : hasRole('ROLE_ADMIN')")
public String form(@Valid final Message message, final Model model) {
if (message.getId() == null) {
someService.save(message);
AclManager.create(message);
} else {
someService.update(message);
AclManager.update(message);
}
return "redirect:result.html";
}
Before I put annotations of security in Controller "b" all was well.
Now when I go to the page of controller "a" and I fill out a form, click a button that causes the controller "b" and I get "HTTP Status 405 - Request method 'POST' not supported".
Why is this happening and how to fix it?
UPD:I helped add to the login controller RequestMethod.POST
A: It's hard to debug without your spring security configuration, but I'm guessing when you go to /portal/form.html spring redirects you to a login page using HTTP POST while your login page controller handler is only mapped to HTTP GET. Try mapping the login page handler to POST method.
A: I also happened to run into your described symptoms when migrating my application's usage of Spring Security 3.2.0.RELEASE from an XML-based configuration to the JavaConfig-based configuration.
In my particular case, the message cause of HTTP Status 405 - Request method 'POST' not supported was Spring Security's default activation of CSRF protection when using JavaConfig-based configuration.
To remove this cause, one can:
*
*either update the application to follow Spring Security's CSRF protection mechanism
*or deactivate Spring Security's CSRF protection mechanism.
To deactivate the CSRF protection mechanism, in the WebSecurityConfigurer do something like the following:
@Configuration
@EnableWebSecurity
public class SecurityConfig extends WebSecurityConfigurerAdapter {
@Override
protected void configure(HttpSecurity http) throws Exception {
http
.csrf()
.disable() //FIXME
.authorizeRequests() ....
}
// ...
}
|
{
"redpajama_set_name": "RedPajamaStackExchange"
}
| 2,458
|
Ellis Radcliffe - 2010 - 2015
Ellis gained 5 A-C at GCSE, he played a huge part in the building and maintenance of the Upper Batley High School Barnyard. He then went onto the Grimsby Institute College based in Grimsby where he is studying Public Services-Uniform for Marines. Ellis is hoping to join the Marines at the end of his 2 year course
Arshad Ranguni 2007-2012
Gained 6 A-C and a Distinction in Computing Arshad then went onto study Applied Science and Double Sport at Mirfield Grammer Sixth form. He is currently at Leeds Beckett University where he is now studying Sport and Exercise Science. Arshad is hoping to become a Physiotherapist or Nurse.
Andrew Aston 2009-2014
7 A to C's Studying at St John Fisher's Academy 6th Form PE A Level A Level Phycology A level Product Design ICT Hoping to go on and study at Leeds University
Faizaan Naseer - September 2009 – August 2013
6 A to C's B in English Literature and B in Art Studying at Huddersfield New College Law, Art and English Literature. Hoping to go on to study at Leeds Beckett College
Nathan Blakeley 2009-2014
7A to C's Since leaving BBEC Nathan is attending Batley School of Art at Kirklees College. He is studying Art and Design level 3 He hopes to go onto Manchester or Leeds Uninversity
Huzaifah Maniyar 2005-2009
11 GCSE's A-C Huzaifah continued his education at Batley Grammar from 2009 to 2012 and gained 3 A levels, he then went onto Bradford University to study a degree in Phsycology. Huzaifah is currently doing work experience here at Batley Upper High School, and hopes to go into teaching Physcology when he's competed his degree.
Mohammed Emtiyaz Yoosoof 2007-2012
14.5 GCSE all A*-C grades From Batley Business & Enterprise College he went to Greenhead College where he is currently taking A levels in Economics, English Language and Politics. He hopes to go to London School of Economics to study Political Science in Sept 2014.
Irfan Tai 2006-2011
17.5 GCSE all A*-A grades Best ever student with the highest GCSE score. From Batley Business & Enterprise College he went to Greenhead College where he studied Maths, Physics and Chemistry. Two A* and 1 A grade at A level. Now in his first year at the University of Manchester studying Petroleum Engineering.
Kabir Hussain - September 2008 – August 2013
12.5 GCSE's – 13 A - G's Huddersfield New College - Extended diploma in ScienceFood Science To make some TV presentation, Working in Television, hopefully with Nigella Lawson Published his 2nd book "Loss of a Loved One", hopefully a series 1, 2 and 3. Also, completed a Gold Award in Performing Arts.
Yasin Hussain - September 2008 – August 2013
15 GCSE's – 13 A* - C's Huddersfield New College - Sociology Sport Physiology [2 year course] Hoping to study at Wolverhampton University Intend to join the Fire Services / Voluntary work "My experience was nearly as amazing as me!"
Paul Senior 1976-81
Paul completes an apprenticeship in Brickwork reaching Level 3 advanced craft, he then worked for Kirklees Council before setting up his own building business. After gaining a teaching qualification he went on to teach construction, brickwork and painting and decorating at Wakefield College.
Adam Patel - September 2008 – August 2013
Adam Patel - September 2008 – August 2013 12 GCSE's – 12 A* - C's Wakefield College - Chemistry / Biology / Psychology Hoping to study Medicine at Manchester University
Fayaz Kasia - September 2006 – August 2011 – Head Boy
14 GCSE's – 12 A* - C's Greenhead College - 4 A Levels Currently having a Gap year – he took part in a BBC TV programme which was shown in March 2014. Fayaz has advised viewers to watch it on BBC iPlayer instead. The programme will also be part of the resources available to the GCSE exam boards and teachers for their delivery of Religious Education. This resource will be available for the next 5 years via BBC Learning to help students prepare for their GCSE Religious Educations exam. While preparing for this show, Fayaz was, along with other students involved in an inspirational visit to a Hospice in Kinross, Scotland, he attended a Healing Service in a Pentecostal church in Hull, East Yorkshire. Fayaz feels that being involved in this type of project; he is giving back to education. Fayaz is a seasoned television personality; he was also involved in helping to profile the Downs Syndrome charity in the television programme - Emmerdale. Fayaz aims to have inspires individua
Jaabir Gharda 1998-2003
Jaabir gained 10 Gcse's @ A-G, he took a year out then returned to us in the 6th Form, studying towards A levels. The next stage of his career was also at BBEC, he became one of the schools Apprentice ICT Technicians, he worked within the school for 2½ years in conjunction with Kirklees college . Following on from this, he did a Foundation Degree at Leeds Metropolitan University, followed by an Honours Degree in Computing and ICT and then a Masters in information Technology also at Leeds Metropolitan University. He is currently working with a private company in Huddersfield as a software engineer. Quote; Thanks to Mrs Brackley in Business Studies, he and the cohort were one of the first students to receive the AIM Higher experience.
Noel Johnson 1979-83
After leaving school Noel completed an apprenticeship in Engineering at Dewsbury College and went to work in Engineering for several years, he then went onto complete a teacher training course and then a Masters in further education and taught Engineering at Dewsbury College for 7 years. After joining Hull city Council on their Marine Engineering programme he became head of training for the council. To further his career he set up his own private training company Prospect Training based in Hull, who have recently opened a training provision in Batley. He also sits on the board of the Association of Employment and Learning Providers who represent learning providers and advise the government.
David Miller - 1992 to 1994
David attended which was then Batley High School for Boys in 1987 to 1992 , he left with eight GCSEs, ranging from Bs and Cs. He went onto Wakefield College in 1992 to 1994 to study BTEC National Diploma in Media Production Triple Distinction. (David was the first student from college media and Performing Arts department to achieve this). In 1994 to 1998 David attended Edinburgh Napier University (BA Hons Journalism 2:1 classification). RMIT University Melbourne 1996 (BA Hons Journalism international scholarship with distinction). He did his work experience at the Yorkshire Post, Los Angeles Times, Melbourne Herald Sun and The Heidelberger, a suburban weekly in Melbourne. His career started at East Lothian Courier in Haddington, near Edinburgh. David wrote for national media at a news agency before joining the Toronto Sun in Canada and Press Association in Glasgow. David is now currently freelance, and The Press in Batley is one of his clients.
Darren Pearson - 1983- 1988
After leaving in 1988 Darren enrolled in A Youth Training Scheme and worked for a Kitchen Manufacturer here in Batley, when the company closed he went on to work for fox's biscuits and Rentokil Initial, Whilst working for these company's Darren worked continually on evenings and weekend trying to build up his own business as a Kitchen and Bathroom installer, he did this for five years, and in 2004 he took the plunge and officially stared his own business. 12 years later his company as gone from strength to strength, and even branches out to lots of different counties Darren came for a tour of the school and said "It was great to have a walk down memory lane, the school looked so much bigger and brighter than I remembered it" he went on to say, "I'm glad they have got rid of the big clock, if you misbehaved you had to stand underneath it for what seemed forever, not that I would know, I never stood under it!"
Mohammed Amin Daji 1989 – 1994
'Life is a Full Circle' – Mohammed left the school after finishing his GCSEs and went onto study an Islamic Theology course. He returned to education in 2006 where he completed the Access Course at Dewsbury College. Mohammed then enrolled at the University of Bradford to complete my BA (Hons) English and graduated in 2011. He then immediately went onto a teacher training course and became a qualified teacher. Now, 22 years after leaving the school, I have returned as a teacher, having a strong passion and desire to educate 'tomorrow's successful men!'
Haroon Lunat
Haroon began his journey at Upper Batley High School in September 1988 and left in July 1993. He left school with 4 GCSE's, Haroon said" I didn't settle in very well at high school, and when I look back now I can see how the school tried their best to support me, I ended up working various jobs that had very little future prospects, these ranged from warehouse, labouring and driving jobs, which I had little pleasure in doing." He went on to say "I realised I had to go back to education to improve my chances of a better occupation." Haroon wanted to work with young people and help them with issues that hold them back from progression in life, so he started Kirklees College in 2008 and finished the access course to higher education. This got him in to Huddersfield University in 2009 where he completed the Youth & community degree and graduated in 2012 when he then started working for Kirklees council as a youth worker .Haroon visits UBHS as part of our YISP Programme and say
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 4,067
|
Joseph Viateur Dandurand, dit Léo Dandurand, (né le à Bourbonnais en Illinois et mort le à Montréal au Canada) est propriétaire dans la Ligue nationale de hockey (LNH) et dans la Ligue canadienne de football.
Biographie
Dandurand, né aux États-Unis, a déménagé au Canada avec sa famille à l'âge de seize ans et a fréquenté le Collège Sainte-Marie de Montréal. Il a été un arbitre dans la LNH et a été très engagé dans le hockey mineur dans la région de Montréal.
Copropriétaire des Canadiens de Montréal
Le , Dandurand et ses partenaires Joseph Cattarinich et Louis Létourneau ont acheté le club de hockey les Canadiens de Montréal de la veuve de George Kennedy pour la somme de 11 000 $. Durant l'ère où Dandurand a été le propriétaire de l'équipe, le CH a remporté la Coupe Stanley dans la LNH à trois reprises (en 1924, en 1930 et en 1931). Létourneau a vendu sa part du club en 1930 et Dandurand and Cattarinich ont continué d'en être les propriétaires jusqu'en 1935, où ils l'ont vendu pour la somme de .
Homme d'affaires dans les courses de chevaux
En plus du hockey, Dandurand et ses partenaires ont été hautement engagés dans les courses de chevaux. En 1932, ils ont acheté l'Hippodrome de Montréal. Ils ont possédé dix-sept hippodromes à Québec, en Ontario, à New York, en Ohio, au Delaware, en Illinois, en Utah et en Louisiane en 1938. Léo Dandurand était aussi un promoteur de boxe et de lutte à Montréal en plus d'être le directeur de l'équipe de baseball les Royals de Montréal.
Fondateur des Alouettes de Montréal
En 1946, Dandurand a fondé les Alouettes de Montréal, une équipe de football, avec Eric Cradock et Lew Hayman. L'équipe a joué dans la Interprovincial Rugby Football Union, qui plus tard est devenue la division Est de la Ligue canadienne de football.
Durant ses vieux jours, Léo Dandurand a été le propriétaire d'un fameux restaurant au centre-ville de Montréal.
Décès
Il est mort à Montréal le d'une crise cardiaque à l'âge de 74 ans, 12 jours avant son . Sa sépulture est située dans le Cimetière Notre-Dame-des-Neiges, à Montréal.
Hommages et distinctions
Il a été introduit au Temple de la renommée du hockey en 1963.
Le trophée Léo-Dandurand est une récompense de la Ligue canadienne de football remise chaque année au meilleur joueur de la première ligne de la division Est.
Liens externes
Site officiel Temple de la renommée du hockey
Références
Naissance en juillet 1889
Naissance dans le comté de Kankakee
Canadiens de Montréal
Entraîneur des Canadiens de Montréal
Membre du Panthéon des sports du Québec
Membre du Temple de la renommée du hockey
Champion de la Coupe Stanley
Décès en juin 1964
Décès à Montréal
Décès à 74 ans
Élève du collège Sainte-Marie de Montréal
Dirigeant de football canadien
Personnalité inhumée au cimetière Notre-Dame-des-Neiges
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 4,127
|
During our visit at the Bonsai Village Omiya we visited all the nurseries that are all in walking distance - unfortunately you are not allowed to take pictures in some of them but where it was allowed I did. The quality of the trees are STUNNING - so are the prices!
Some famous materpieces are there to be found as well... can you spot them on the pictures? Please do not hesitate to contact me if you need any guidance on how to get to the village from Tokyo.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 274
|
There were 7 posts published in April.
The 3D-printed mushrooms based on the images are an attempt to move the plantation to another location, despite its frailty. The mushrooms are printed during the exhibition.
Images made of an abandoned mushroom plantation found inside the mountain. The mouldy mushrooms were extremely fragile and could not be touched or taken out of the cave to photograph in daylight.
A collection consisting of 4 documents found in a small archive in the Dutch village Ulenstraten kept by a former employee. The documents have been treated with the chemical 1,2-indanedione to reveal the hidden fingerprints. Many different types of paper – although looking clean – can carry a thin film of dirt and sweat particles that show the personal contact that people had with the object.
Reconstruction of the golf course inside the complex based on information from archive material, as well as a wide variety of recollections from by former employees.
Many geologists assumed that most rocks had been around since the origins of Earth, well before life formed on this planet. However, recent discoveries show that the planet as we know it could not have existed, if it wasn't for the strange, twisty relationship of living organisms and cold, lifeless rock. This project is a visual research based on this co-evolution.
Currently, scientists are searching for a better understanding of that close interplay of life and rocks, in order to predict our impact on the earth's system. Geologists use thin sections of stone to investigate their mineral composition. I started looking at them as little drawings left behind for geoscientists by previous forms of life.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 39
|
Figure 3 Comparing the structural changes in site 2 of the hGH and hGHv-hGH-R ternary complexes. Hormone residues are labeled in black and receptor residues in red. Arrows point to the carbonyl oxygen atom of the peptide bond of Trp169 that flips its conformation. The other arrows point to the reorientation of the side chains of Trp169 and Trp104.
to the receptor. It is probable that the small movements of Asp116 that are a consequence of the repacking of the four-helix bundle in hGHv effectively trigger the new H-bonding scheme where by the carboxylate side chain makes new H bonds to the receptor through the indole nitrogens of the side chains of Trp104 and Trp169, as well as to the side chain of Arg43. As seen in Fig. 3, the Trp side chains undergo a significant reorganization to facilitate the formation of these H bonds. These observations clearly have fundamental importance to our understanding of the inherent efficiencies of these cytokine hormones and receptors as binding entities even outside of evolutionary control.
The concept that the two spatially distinct binding sites on cytokine hormones are structurally and functionally coupled as displayed in the hGHv complex is novel, and the process whereby new binding surfaces are synthesized by indirect through molecule effects has been termed functional cooperativity . In this mechanism, it is not only the mutations in one site that affect the other site. A set of concerted changes also occurs among the hormone and the receptor ECDs. The finding of strong cross-molecular interaction induced during receptor dimerization establishes a new molecular recognition paradigm and opens up fundamental new areas of investigation relating to the mechanisms of biological regulation by protein-protein associations. However, it remains an open question as to how general this is and whether evolution actually uses this strategy to influence the receptor signaling of GH/PRL systems in biologically important ways.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 1,509
|
Carnon is een badplaats aan de Middellandse Zee in de Franse gemeente Mauguio (Hérault).
Carnon heeft 7 km aan zandstranden en een jachthaven. De plaats is populair bij kitesurfers.
Geschiedenis
Carnon is gelegen op de schoorwal die de Middellandse Zee scheidt van de lagune Étang de l'Or. Ter hoogte van Carnon was er een verbinding (grau) tussen de zee en de lagune. Carnon bestond in de middeleeuwen uit enkele vissershutten maar won aan belang omdat er over de schoorwal een handelsweg ontstond. In de 19e eeuw werd het Canal du grau de Pérols gegraven. De plaats ontwikkelde zich toeristisch vanaf 1830 toen er een veerdienst kwam tussen het vasteland en Carnon. In 1921 kwam er een brug naar Pérols en werd de badplaats bekend onder de naam Carnonville-la-mer. De badplaats was vooral in trek bij inwoners van Montpellier, zeker nadat er in 1933 een dagelijkse busverbinding kwam. Vanaf de jaren 1970 kwam er hoogbouw en een jachthaven aan zee.
Bronnen
Carnon, ouverture à la mer (mauguio-carnon.com) Geraadpleegd op 24 februari 2023.
Brochure: Mauguio Carnon, 7000 ans d'histoire, Ville de Mauguio Carnon
Mauguio
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 7,523
|
#
# Le salut de
l'arrière-pays
## Figures légendaires,
récits imaginaires
et humour crucifère
du Nord de l'Ontario
Du même auteur
_De face et de billet_ , Sudbury, Éditions Prise de parole, 2002, 249 p.
_Bozo l'orignal vampire_ , roman jeunesse, Sudbury, à compte d'auteur, 2008, 63 p.
_Flagabou l'enfantôme et la bombe à bonbons_ , roman jeunesse, Sudbury, à compte d'auteur, 2009, 58 p.
_Cinquante exemplaires de cet ouvrage_
_ont été numérotés et signés par l'auteur._
# Le salut de
l'arrière-pays
## Figures légendaires,
récits imaginaires
et humour crucifère
du Nord de l'Ontario
## Collectif
sous la direction
de Normand Renaud
Éditions Prise de parole
Sudbury 2010
**Catalogage avant publication de Bibliothèque et Archives Canada**
Le salut de l'arrière-pays : figures légendaires, récits imaginaires et humour crucifère du Nord de l'Ontario / sous la direction de Normand Renaud.
Basé sur une série d'émissions radiophoniques diffusées en 2004-2005 à l'occasion du 25ème anniversaire de la station de radio CBON.
ISBN 978-2-89423-246-0
1. Canadiens français – Ontario (Nord) – Biographies. 2. Radio communautaire – Ontario (Nord). 3. Ontario (Nord) – Anthologies. I. Renaud, Normand, 1956- . II. CBON (Station de radio : Sudbury, Ont.)
FC3094.8.F85S35 2010 971.3'1004114 C2010-906799-1
Distribution au Québec: Prologue • 1650, boul. Lionel-Bertrand
• Boisbriand (QC) J7H 1N7 • 450-434-0306
Ancrées dans le Nouvel-Ontario, les Éditions Prise de parole appuient les auteurs et les créateurs d'expression et de culture françaises au Canada, en privilégiant des œuvres de facture contemporaine.
La maison d'édition remercie le Conseil des Arts de l'Ontario, le Conseil des Arts du Canada, le Patrimoine canadien (programme Développement des communautés de langue officielle et Fonds du livre du Canada) et la Ville du Grand Sudbury de leur appui financier.
Œuvre en page de couverture : Christian Pelletier, _L'hiver approche_ , photographie et retouche par l'artiste
Conception de la page de couverture et mise en pages : Olivier Lasser
Tous droits de traduction, de reproduction et d'adaptation réservés pour tous pays.
Imprimé au Canada.
Copyright © Radio-Canada, sauf pour les textes suivants :
« Watershe », de Céleste Dubé, « La parade », de Danièle Vallée, « Madame Fillion », de Guylaine Tousignant, « L'hurluberlu d'Earlton », de Gaston Tremblay, « Sabine au fonds des bois », de Marguerite Andersen, « Tout feu tout flamme », de Richard Léger, « Avance », de Robert Dickson et « Tamara de Fauquier », de Melchior Mbonimpa.
Éditions Prise de parole
C.P. 550, Sudbury (Ontario) Canada P3E 4R2
<http://www.prisedeparole.ca>
ISBN 978-2-89423-246-0 (Papier)
ISBN 978-2-89423-337-5 (PDF)
ISBN 978-2-89423-563-8 (ePub)
### NOTE DE L'ÉDITEUR
Nous sommes au tout début de l'année 2003. J'avais été invitée par Alain Dorion, alors directeur de la station CBON, la Première Chaîne de Radio-Canada dans le Nord de l'Ontario, à une rencontre autour d'un projet dont il m'avait donné peu de détails. Je me souviens simplement qu'il souhaitait mettre en place quelque chose de spécial pour souligner le 25e anniversaire de CBON, et qu'il m'avait communiqué son enthousiasme.
Je me suis donc retrouvée autour de la table, dans son bureau, avec Alain Dorion et Normand Renaud, l'initiateur-idéateur du projet. Et c'est _Le salut de l'arrière-pays_ qu'ils ont exposé devant moi.
Projet auquel j'ai immédiatement adhéré. Pour l'intelligence et la finesse de sa formule, pour ses objectifs, de révélateur de la richesse qu'ont à offrir les communautés du Nord de l'Ontario. Tout ça dans une formule ambulante : des émissions enregistrées en direct devant public dans chacune des communautés retenues.
Outre les artisans de Radio-Canada, le projet mettrait à contribution des auteurs de l'Ontario français pour la création des contes littéraires.
Les Éditions Prise de parole ont donc plongé avec grand bonheur dans cette aventure. Nous avons obtenu du Conseil des Arts du Canada le financement requis pour recruter les auteurs des contes littéraires; nous avons également réussi à intéresser la Fondation Langellier de la Société Saint-Jean-Baptiste à ce projet. Que ces deux organismes soient remerciés.
_Le salut de l'arrière-pays_ , « petit café ambulant, littéraire et chantant », a pris la route le 20 février 2004, direction Gogama. Suivront Verner, Chapleau, Earlton, Iroquois Falls, Spanish, Sturgeon Falls. Dernier arrêt, Fauquier, 4 mars 2005.
L'aventure se poursuit grâce au livre que vous tenez entre vos mains. Grâce aussi, à un disque que Stéphane Paquette termine en ce moment, que je vous invite également à découvrir.
À tous les créateurs et artisans qui ont contribué au succès du _Salut de l'arrière-pays_ , je dis un grand, grand merci!
_denise truax_
_Éditions Prise de parole_
### PRÉFACE
_Le salut de l'arrière-pays :_
contes et légendes du Nord
de l'Ontario
Gogama, Verner, Chapleau, Earlton, Iroquois Falls, Spanish, Sturgeon Falls et Fauquier... autant de petites communautés du Nord de l'Ontario qui ont inspiré une légende — c'est à dire l'histoire d'un héros populaire — et un conte, dont la toute première présentation a été sous la forme d'une soirée communautaire avec des chansons originales et de l'humour satirique. Mais avant d'être le recueil de textes que vous tenez entre les mains, _Le salut de l'arrière-pays_ a aussi été une série de huit émissions radiophoniques produites par CBON, la Première Chaîne de Radio-Canada dans le Nord de l'Ontario. Normand Renaud, le concepteur du projet, dit s'être inspiré de _Vinyl Café_ de Stuart McLean (CBC Radio One) et _A Prairie Home Companion_ de Garrison Keiller (Minnesota Public Radio), deux émissions radiophoniques très populaires et respectées qui misent sur la narration et la fiction ainsi que sur l'engagement envers les communautés non métropolitaines. Enregistrées devant public entre le 20 février 2004 et le 4 mars 2005, et diffusées dans la semaine suivant leur captation, ces soirées ont permis de réunir, par la magie des ondes, non seulement les gens des huit communautés visitées, mais tout le Nord de l'Ontario. La qualité de cette création radiophonique a été remarquée. En juin 2004, avec l'émission enregistrée à Verner, _Le salut de l'arrière-pays_ a remporté devant 33 concurrents le prix de la meilleure émission régionale lors de la remise des Prix de la radio 2003-2004 de Radio-Canada à Montréal. Le jury — composé de Jean-Pierre Desaulniers (président), professeur au département des communications à l'Université du Québec à Montréal, Richard Joubert, animateur de Radio-Canada à la retraite et Martine Thériault, coordonnatrice de RADAR, le Réseau atlantique de diffusion des arts de la scène —, a justifié son choix comme suit : « Le Salut de l'arrière-pays sait bâtir un lien exceptionnel entre la tradition et la modernité. Il n'enfouit pas la région dans une complaisance nostalgique, mais dégage de vrais liens entre la liberté des premiers défricheurs et la liberté des vrais communicateurs que sont les membres de cette équipe de radio. un concept solide, bien pensé, des objectifs réels de développement social, un vent d'impertinence et une joie rafraîchissante. Bref, un monde en santé. »
Ancré dans son milieu et engagé envers sa région, le projet souhaitait exprimer une prise de conscience régionale face au déclin démographique. Il se voulait une réponse à la difficile situation socioéconomique du Nord de l'Ontario où plusieurs communautés dépendent d'une industrie forestière en déclin et où l'on assiste, impuissant, à l'exode des jeunes vers les grands centres avec l'espoir de trouver un emploi et un avenir. Devant le déclin économique et démographique, il n'était pas question, pour Normand Renaud, de baisser les bras. Chaque soirée — « un petit café ambulant, littéraire et chantant », comme l'annonçait l'animateur en guise d'introduction —, puis chaque émission ont affirmé le dynamisme et la richesse de l'imaginaire de chacune des huit communautés. Dans un tel contexte, généreux, dynamique et innovateur, l'expression « vitalité régionale » prenait tout son sens. Ainsi, pour parvenir à ces fins, _Le salut de l'arrière-pays_ a eu recours à l'histoire, à l'humour, à la chanson et à la littérature, en particulier à la légende et au conte. De toute évidence, dans l'esprit de son concepteur et de ses artisans, les formes de la littérature orale étaient celles qu'il fallait mettre à contribution pour intervenir efficacement dans le contexte socioéconomique qui les préoccupait.
##### **Huit légendes de Normand Renaud**
Dans chaque communauté visitée, Normand Renaud a choisi — ou plutôt déniché! — un homme plus grand que nature, dont il a fait un héros, bien que, de son aveu même, les personnages aient été retenus davantage pour leurs qualités romanesques que leurs qualités morales. Car, « [q]uand _Le salut de l'arrière-pays_ visite un village, c'est pour trouver dans son histoire un symbole de ses richesses humaines, qui valent plus encore que ses richesses naturelles. C'est pour y chercher des images et des personnages qui puissent rappeler un passé dont l'avenir pourrait se nourrir », clamait l'animateur. Il n'y a pas d'avenir sans passé et sans mémoire. Ces huit portraits s'appuient donc sur des témoignages et des documents, des sources orales et écrites. Au dire de l'auteur, ces portraits ne comportent aucune invention de sa part; toute affirmation factuelle au sujet des personnages vient de ses sources et aucune n'a été inventée pour bonifier le récit. Mais s'ils sont tous bien ancrés dans le réel, ils sont aussi magnifiés par le temps — sept des huit héros appartiennent au passé — ainsi que par la verve de Normand Renaud qui amplifie leur présence, si bien qu'ils deviennent rien de moins que des personnages légendaires. N'est-ce pas là la définition même de la légende, qui suppose un fait historique ou réel qui est par la suite amplifié par l'imagination populaire? Les personnages qui ont retenu l'attention de Normand Renaud sont donc légendaires par leur source dans le passé, mais aussi par leur résonnance dans la conscience.
Ainsi, Gogama se voit représenté par Jos Laflamme — « _the Wolfman of the North_ »— un « indompté », un « esprit libre » qui fascine par la force de son caractère délinquant, un dompteur de loups dont la grande célébrité de son vivant est aujourd'hui presque oubliée. À Verner, le père Charles Alfred Paradis, « prêtre rebelle, justicier fougueux, colonisateur visionnaire, tempérament volontaire, brasseur de m... mémorables controverses », est inspirant parce qu'il a osé avoir pour le Nord de l'Ontario des rêves qui ne cadraient pas avec les volontés des autorités et parce qu'il a résolument tenu tête à ses détracteurs. À Chapleau, c'est un personnage de roman, François Duvalet — l'émigré français qu'imagine Maurice de Goumois — , qui fait ressentir l'emprise qu'a exercé sur les esprits le pays neuf nord-ontarien, « un monde régénéré, débarrassé de toute contrainte, où un homme n'est que ce qu'il est, et non pas qui il est » (et dont le récit fait porter à Chapleau une auréole littéraire plus pertinente que celle que lui donne Louis Hémon, dont le seul rapport avec cette ville est sa mort accidentelle en ce lieu). À Earlton, c'est l'homme fort Wilfrid Paiement qui séduit par la convergence de sa force physique, de sa personnalité attachante et de son dynamisme de bon vivant. À Iroquois Falls, c'est un homme du présent, engagé auprès de sa communauté, qui est devenu un modèle : Denis « Bine » Lachapelle incarne si franchement et généreusement le père Noël que la figure mythique et l'homme bien réel se confondent. À Spanish, les recherches de Renaud au musée local lui ont permis de découvrir Jules Couvrette, un conteur d'histoires invraisemblables uniques en leur genre qui impressionnent parce qu'elles survivent jusqu'à nos jours dans la tradition orale locale alors que le souvenir de l'homme lui-même est oublié. À Sturgeon Falls, le père Joseph Eugène Marchand — « le gros père Marchand » — , vicaire de la paroisse Sacré-Cœur, s'est imposé par l'improbable mélange de son imposante corpulence, de sa personnalité sans prétention et de son humour populacier, qualités qui ont su humaniser tout à fait les valeurs de l'Église. Enfin, à Fauquier, Bucko (Lucien) Beauchamp, un bûcheron dont l'habileté extraordinaire a été remarquée dans tous les chantiers où il est passé, de Smooth Rock Falls à Thunder Bay, fascine du fait que ce petit homme est tout le contraire du stéréotype du bûcheron à la Paul Bunyan : « Il n'avait pas besoin d'être grand, notre géant. Il était Bucko Beauchamp. »
En faisant sortir de l'ombre ces « remarquables oubliés » — bien avant ceux de Serge Bouchard —, Normand Renaud a rassemblé une galerie de personnages dont l'effet inspirateur est d'autant plus frappant que leurs portraits font souvent fi des figures attendues et des valeurs traditionnelles, garantes de la cohérence sociale. Ce faisant, l'auteur a répondu originalement à l'impératif que Charles Nodier lançait au XIXe siècle : « Hâtons-nous de raconter les délicieuses histoires du peuple avant qu'il ne les ait oubliées. »
##### **Huit contes d'auteurs franco-ontariens**
Si les légendes de Normand Renaud permettent de faire sortir des personnages de l'oubli et de les ériger en modèles de personnalités indépendantes et originales, les huit contes inspirés de chacune des communautés visitées font émerger des thèmes humains qui découlent des milieux périphériques aux prises avec l'exode démographique. Plusieurs des contes mettent en scène des jeunes qui auraient voulu ou qui voudraient vivre dans le Nord de l'Ontario, mais qui ont dû quitter leur région faute d'emploi; d'autres, au contraire, montrent le trajet d'un jeune qui renoue avec le Nord ontarien ou qui adopte cette communauté en partant d'ailleurs. Ce dont tous ces contes nous convainquent, c'est qu'il fait bon vivre dans le Nord de l'Ontario, au contraire de ce que l'exode de la population peut faire croire. Le conte, cette « parole rassembleuse » parce que publique, est on ne peut mieux choisi comme moyen pour nous toucher. Huit auteurs franco-ontariens se sont prêtés au jeu de la parole conteuse mise au service de la cause régionaliste. La seule contrainte était qu'ils devaient passer quelques jours dans la communauté dont ils allaient s'inspirer pour écrire leur conte. On retrouve ainsi dans chaque texte un ancrage spatial qui met littéralement en scène le lieu d'où l'on parle, qu'il s'agisse d'un nom de rue, du magasin du coin, du numéro d'une route, etc., bref, qui fait que l'on s'y reconnaît. Chacun était libre d'orienter son récit comme bon lui semblait, tout en connaissant le contexte de l'émission dans laquelle il devait s'insérer, soit l'évocation du déclin démographique et économique et le lien à renouer entre les réalités prometteuses d'hier et les réalités inquiétantes d'aujourd'hui.
Ainsi, « Wathershe » de Céleste Dubé, qui s'est inspirée de Gogama, met en scène une jeune fille qui souhaite quitter sa petite communauté où elle étouffe, mais qui finit par prendre conscience que son destin était de rester à Gogama et d'y connaître un vrai bonheur. « La parade » de Danièle Vallée, qui s'inspire de Verner, nous ramène dans le passé, en posant un regard tendre et nostalgique sur l'enfance vécue en un milieu villageois d'antan qui reproduisait parfaitement le modèle de la société canadienne-française dans le nord de l'Ontario. « Madame Fillion » de Guylaine Tousignant, qui s'inspire de Chapleau, met en scène un personnage dont le retour à la ville de son enfance, qu'elle achevait presque d'oublier, est d'abord un acte de rupture avec la personne qu'elle est devenue dans une ville métropolitaine. Or, la trajectoire du personnage aboutit dans une communauté voisine (Biscotasing), bien plus petite encore que sa ville d'origine, comme s'il s'agissait de passer d'un extrême démographique à l'autre pour mieux pousser à la limite le refus critique de l'effet aliénant que la grande ville peut exercer sur la personnalité. « L'hurluberlu d'Earlton » de Gaston Tremblay nous ramène aux origines, à la vie sur la terre, tant au passé qu'au présent, nous rappelle que la modernisation a eu ses défauts puisqu'elle a profondément modifié la façon de faire les choses et en a obligé plus d'un à s'exiler; il met en scène le désir de revenir chez soi, mais en même temps l'impossibilité de le faire. Dans « Sabine au fond des bois » (Iroquois Falls), Marguerite Andersen nous convie à une quête de soi qui passe par la quête de l'autre. Les chutes qui ont donné son nom au village n'existent plus, tout comme la petite communauté juive qui y a disparu en ne laissant que de faibles traces, mais Sabine redécouvre ces présences du passé et renoue avec elles pour ancrer son bonheur dans le présent. « Tout feu tout flamme » de Richard Léger s'inspire de Spanish; sur fond d'histoire d'amour et dans un registre fantastique, il réactualise et amplifie la légende d'une énigmatique présence espagnole parmi les Amérindiens de ce coin de pays (qui a donné son nom à la rivière et à la communauté), rendant cette légende de nouveau présente et vivante. Dans « Avance », Robert Dickson, qui s'inspire de Sturgeon Falls, utilise le pouvoir critique du conte en rappelant que les Franco-Ontariens peuvent s'enorgueillir d'avoir su faire valoir leurs droits par la revendication solidaire, en même temps qu'il célèbre l'amitié, la fidélité à soi et la fierté dans le quotidien. Enfin, « Tamara de Fauquier » de Melchior Mbonimpa, qui s'inspire de Fauquier, nous plonge dans une réflexion sur les origines, la mémoire — celle qui nous reste, celle qui nous fait défaut et qu'il faut activement chercher à redécouvrir — et le métissage, thème très contemporain, mais déjà bien présent dans le passé du Nord de l'Ontario qui a accueilli des pionniers d'un peu partout qui rêvaient d'un avenir meilleur.
Toutes ces légendes et tous ces contes, ancrés dans l'ici-maintenant des gens du Nord, tournés vers l'avenir tout en étant conscients du passé, sont vivants; leur oralité, même dans la version écrite, nous interpelle. Si l'on affirme qu'il s'agit de contes régionaux — étiquette qui peut comporter une connotation négative ou réductrice —, il ne faut pas par là sous-estimer leur portée universelle. Tous ces auteurs ont démontré à l'envi que, pour peu qu'on s'y penche, ces petites communautés nord-ontariennes recèlent dans leur imaginaire et leur histoire d'amples et inspirantes richesses : la chaleur humaine, l'amour, l'amitié, les rêves d'une vie meilleure sont les aspirations de l'humanité entière; or elles retrouvent dans ce décor nord-ontarien des couleurs fraîches et des forces vives qui raniment l'espoir et nourrissent les rêves. On en arrive presque à oublier que le projet littéraire du _Salut de l'arrière-pays_ est issu, tout au contraire, d'une prise de conscience inquiétante devant les tendances démographiques et économiques. C'est dire que le pari littéraire du concepteur et de ses collaborateurs est gagnant.
##### **Chansons originales et humour**
Aux côtés des légendes et des contes, chaque _Salut de l'arrière-pays_ comporte une séquence humoristique sous forme de pseudo-entrevue où la satire se fait loufoque et la parodie grinçante pour dénoncer l'insuffisance et la superficialité d'un certain discours sur le développement économique en région. Ces sketchs mettent en scène un agent de développement économique régional qui présente à une animatrice de la métropole des rapports aux contenus et aux titres invraisemblables. L'humour d'Alain Harvey et de Pierre-Mathieu Tremblay affirme clairement que le Nord de l'Ontario n'est pas dupe du peu de promesses, de la visée réductrice et de la vision stéréotypée de la panacée que serait l'écotourisme, solution que l'on propose trop souvent comme futur ressort économique à des communautés qui tiraient jusqu'à tout récemment leur vitalité et leur fierté de leur vocation industrielle. Satiristes efficaces, leurs délires débridés sont troublants par leur familiarité.
Quant aux chansons, elles servent efficacement de liens ou de transition entre les différents segments de l'émission. Si les légendes et les contes relèvent du récit et du plaisir de se raconter des histoires, et que les sketchs humoristiques nourrissent l'aspect critique du _Salut de l'arrière-pays_ , les chansons y apportent une touche d'émotion, tout en restant ancrées dans le quotidien. Elles contribuent à souligner des états d'âme : tristesse devant le constat que le Nord n'est plus la terre promise (« Les petites fraises d'été ») ou en pensant à ceux qui sont partis trouver ailleurs leur bonheur («Ta silhouette qui s'éloignait »), inquiétude devant le déclin et même peur de devoir s'exiler (« La onze, la onze »), désillusion de ce fils qui est parti travailler dans la grande ville dans l'espoir de réaliser ses rêves (« Votre fils au loin »), voire une certaine détresse dans l'expression de ce besoin de rêver, de croire que la vie sera belle là où on est né. D'autres chansons mettent en scène le retour au pays de l'enfance et les sentiments qui l'accompagnent (« Entre ailleurs et autrefois »), ceux qui partent, mais surtout ceux qui sont « [f ]aits pour rester », ou font le portrait d'un pays de promesses (« À la gare de Chapleau ») et de liberté (« Par un chemin de bois »). Dix de ces seize chansons gagneront à être entendues plutôt que lues, grâce aux nouvelles interprétations endisquées que prépare l'auteur-compositeur-interprète Stéphane Paquette. Dans sa version radiophonique, chaque épisode du _Salut de l'arrière-pays_ comportait aussi une chanson humoristique, mais celles-ci n'ont pas été retenues pour la publication.
_Le salut de l'arrière-pays_ est certainement un projet régional, tant par son sujet que par son premier mode de diffusion. Si on veut le qualifier de régionaliste, il faut prendre bien soin de définir le terme, souvent associé à un repli sur soi et à une idéologie terroiriste (les romans de la terre, par exemple). Car, si _Le salut_ — comme ses artisans ont pris l'habitude de le désigner — met en scène des personnages du passé, ce n'est pas pour prôner un retour à cette vie passée, mais plutôt pour mettre en valeur les richesses humaines de ces communautés et susciter la fierté. Que ce soit en puisant dans les forces de la mémoire comme le font les légendes, en fouillant les possibilités de l'imaginaire comme le font les contes, ou en nous plongeant au cœur du quotidien et des émotions avec les chansons, il s'agit de montrer comment il est possible de vivre en région au présent. Régionaliste, la série s'inscrit bel et bien dans une logique du centre s'opposant à la périphérie, comme en témoignent avec éloquence les sketches humoristiques : ce sont les petites communautés qui cherchent à s'affirmer vis-à-vis les grandes villes et les métropoles. Hors du grand centre, point de salut? Rien n'est moins sûr, affirment chacun des textes de la série, leurs _salutations_ revendiquant un statut d'égal à égal, leur _salut_ passant par une critique de l'uniformisation de la société d'aujourd'hui qui prône le « _Big is beautiful_ ». Le régionalisme du _Salut de l'arrière-pays_ est critique — il cherche à renverser les préjugés envers les petites communautés —, engagé — il nous force à réfléchir, sinon à agir — et, surtout, contemporain : il constitue une réponse au marasme qui envahit les petites communautés en affirmant haut et fort qu'il est possible de vivre, au présent, dans ces communautés.
Comme les paroles s'envolent et que les écrits restent, il était tout à fait approprié d'offrir la possibilité aux lecteurs et lectrices de plonger ou de replonger dans l'univers du _Salut de l'arrière-pays_ par la voie de l'écrit, bien après leur représentation et leur diffusion. De cette façon, Gogama, Verner, Chapleau, Earlton, Iroquois Falls, Spanish, Sturgeon Falls et Fauquier saluent non seulement le Nord de l'Ontario, mais toutes les régions dites périphériques du pays auxquelles elles peuvent servir d'inspiration, ou à tout le moins proposer des visages familiers dans lesquels elles pourront se reconnaître. Car il faut bien convenir, comme _Le salut de l'arrière-pays_ nous y invite, que ces milieux sont riches en valeur et en chaleur humaine.
_Johanne Melançon_
_Professeure agrégée_
_Département d'études françaises_
_Université Laurentienne_
### COLLABORATIONS
_Normand Renaud –_ Conception du projet – Direction de la publication
###### **Création**
_Alain Harvey –_ Rédaction et interprétation des sketchs comiques
_Stef Paquette –_ Composition et interprétation des chansons
_Normand Renaud –_ Rédaction et lecture des portraits de personnages – Paroles des chansons régionalistes (sauf * ci-dessous) et chansons comiques (non retenues dans cette publication) – Animation, harmonies, percussions
_Pierre-Mathieu Tremblay –_ Rédaction et interprétation des sketchs comiques
###### **Auteurs et lecteurs**
_Marguerite Andersen –_ « Sabine au fond des bois »— Lu à Iroquois Falls par _Miriam_
_Cusson_
_Céleste Dubé –_ _«_ Watershe »— Lu à Gogama par l'auteure
_Robert Dickson –_ « Avance »— Lu à Sturgeon Falls par _Stéphane Gauthier_
_Richard Léger –_ « Tout feu tout flamme »— Lu à Spanish par l'auteur
_Melchior Mbonimpa –_ « Tamara de Fauquier »— Lu à Fauquier par Mélanie Tremblay
_Guylaine Tousignant –_ « Madame Fillion »— Lu à Chapleau par l'auteure – *Paroles de chansons : « un étranger a pris ta place » et « Petite Revanche »; coécriture de « L'été qui s'étend » avec Normand Renaud.
_Gaston Tremblay –_ « L'hurluberlu d'Earlton »— Lu à Earlton par _Stéphane Gauthier_
_Danièle Vallée –_ « La parade »— Lu à Verner par l'auteure
###### **Aide à la recherche**
_Pierre Albert –_ Pour le portrait de Bucko Beauchamp de Fauquier
_Wayne LeBelle_ – Pour le portrait du père Marchand de Sturgeon Falls
###### **Aide à l'édition**
_Michel Ouellette –_ Révision de l'ébauche du livre
_Denise Truax –_ Relations avec les auteurs de contes littéraires
###### **Production des émissions**
_Alain Dorion –_ Directeur de station initiateur du projet
_Roger Corriveau –_ Sonorisation et enregistrement
_Alain Harvey –_ Logistique des soirées et des visites d'auteurs dans les communautés
_Guylaine Tousignant –_ Communications
_Éric Robitaille –_ Réalisation
###### **Épisodes**
_Gogama_ – Enregistrement le vendredi 20 février 2004 à 20 h au restaurant Ruby's; diffusion le samedi 28 février 2004 à 10 h
_Verner_ – Enregistrement le vendredi 19 mars 2004 à 20 h au restaurant Trans-Canada; diffusion le samedi 27 mars 2004 à 10 h
_Chapleau_ – Enregistrement le vendredi 23 avril 2004 à 20 h au restaurant Aux trois moulins; diffusion le samedi 15 mai 2004 10 h
_Earlton_ – Enregistrement le vendredi 11 juin 2004 à 20 h au restaurant le Grand Boulevard; diffusion le samedi 18 juin 2004 à 10 h
_Iroquois Falls_ – Enregistrement le vendredi 22 octobre 2004 à 20 h au Main Street Café; diffusion le samedi 6 novembre 2004 à 10 h
_Spanish_ – Enregistrement le vendredi 26 novembre 2004 à 21 h à Vance's Motor Inn; diffusion le samedi 4 décembre 2004 à 10 h
_Sturgeon Falls –_ Enregistrement le vendredi 11 février 2005 à 19 h au Centre communautaire; diffusion le samedi 19 février 2005 à 10 h
_Fauquier_ **–** Enregistrement le vendredi 4 mars 2005 à 20 h au Centre communautaire; diffusion le samedi 12 mars 2005 à 10 h
###
### Bienvenue au Salut!
Le « Salut de l'arrière-pays », c'est le signe de la main que les villages envoient aux villes, leurs voisines, les deux se ressemblant bien plus qu'on le pense. Car ces petites communautés qui cherchent leur avenir sont à l'image de vastes régions du pays qui cherchent leur avenir, malgré le dépeuplement qu'indiquent les recensements, malgré l'exode de la jeunesse particulièrement.
Le « Salut de l'arrière-pays », c'est l'espoir en l'avenir affirmé en récits et en chansons. C'est une expression de la confiance en la force du cœur et de l'esprit des gens d'ici. C'est un acte de présence à l'ensemble du Nord de l'Ontario, notre coin du monde qui nous tient à cœur.
Quand le « Salut de l'arrière-pays » visite un village, c'est pour trouver dans son histoire des symboles de ses richesses humaines, qui valent plus encore que ses richesses naturelles. C'est pour y chercher des images et des personnages qui puissent rappeler un passé dont l'avenir pourrait se nourrir. Gens du Nord, voici des hommages à nos villages.
_Des photos dans un album_
_Des souvenirs de bons coups_
_Des « Salut! Je t'en conte une bonne »_
_Et puis comment ça va chez vous_
_Un pays qu'on a vu grandir_
_Des histoires à redire_
_À tous ceux qui veulent que ça continue_
_On dit salut!_
###
### Salut Gogama!
Gogama est un village qui impressionne avant même que vous ne l'ayez vu ni connu. C'est déjà assez impressionnant juste de voir sur une carte routière le petit point noir qui l'identifie. C'est le seul point sur la carte à des centaines — des centaines!— de kilomètres à la ronde.
Gogama, c'est le nom qu'on lance quand on a besoin du symbole même d'une petite place isolée. « L'Université de Gogama », ont dit les sarcastiques en réplique à ceux qui ont rêvé de fonder une Université française dans le Nord de l'Ontario. « Gogama, capitale du Nord », a-t-on lancé à ceux qui ont rêvé de faire du Nord de l'Ontario une province distincte. Mais le nom Gogama ne veut pas dire « petite place isolée ». On n'est jamais sûr des noms d'origine amérindienne, mais on pense que Gogama signifie « poisson qui saute ». Cette image du poisson qui saute, qui apparaît soudain et qui disparaît aussitôt, c'est une image qui symbolise bien la courte histoire des hauts et des bas de la ville de Gogama.
Quand on voit le village d'aujourd'hui, avec ses quatre cents habitants, on a peine à croire que Gogama a déjà été une ville de quatre mille habitants. Quatre mille habitants — il y avait autant de monde ici qu'à Hearst! Dans ce temps-là, dans les années trente et quarante, Gogama était, elle aussi, une ville de scieries qui en produisaient du bois, _en veux-tu, en v'là_! Mais dans ce temps-là, on n'y venait pas en voiture. Le seul lien avec le monde extérieur, c'était le train. une ville sans même une route pour s'y rendre, c'était ça, le Nord de l'Ontario, à une époque encore récente. Il a fallu attendre les années 1970 pour que la route 144 perce la forêt et relie Timmins et Sudbury, en passant par Gogama.
Peu à peu, ce qui devait arriver est arrivé : le bon bois a commencé à se faire rare. Puis, le coup de grâce a été donné par le grand feu de forêt des années cinquante. Après ce désastre, le moulin de Poupore s'est réinstallé à Skead et Gogama est devenu une ville fantôme. Ranimons un de ces fantômes pour évoquer l'époque où Gogama était prospère. Réveillons le souvenir de Jos Laflamme!
###### Jos Laflamme
Jos Laflamme, _Wolf Man of the North_ , vedette internationale, a été un personnage haut en couleur pour qui Gogama a été une ville bien assez grande pour le laisser être lui-même. Il était un bonhomme pas commode, pas très recommandable, mais admirable à sa façon, justement pour ces raisons. Pour comprendre l'homme, il faut comprendre l'époque. Il faut, pour commencer, imaginer ce petit monde à l'écart du grand monde. Mais jamais assez à part pour que les lois de l'État et de l'Église ne s'y fassent sentir.
La paroisse Notre-Dame-du-Rosaire de Gogama a été fondée en 1922. Le premier curé, le père Achille Cournoyer, y est demeuré jusqu'en 1939. Dans les premiers temps, il célébrait la messe dans la salle de billard de monsieur Arthur l'Abbé. Je les imagine le dimanche, les gars dont l'imagination divaguait pendant le sermon en se rappelant leurs parties de billard chaudement disputées dans ces mêmes lieux. Transformer le salon de billard en église, admettons que c'est une manière de se débarrasser de lieux mal famés. Ce n'est pas le genre de coup qu'on peut refaire souvent, sinon il y aurait bien trop d'églises en ville. On n'a pas essayé ce coup-là chez Jos Laflamme. Chez lui, on s'y est pris par la loi de l'État. Mais la justice n'en est pas venue à bout. C'est que, voyez-vous, Jos Laflamme était dompteur de loups.
Avant d'en arriver aux loups, il faut d'abord que je vous raconte que Jos Laflamme faisait un petit commerce marginal, mais lucratif, de vendeur non autorisé de boissons alcoolisées — il était _bootlegger_. La grande vedette internationale était le petit délinquant local. Dans les archives du _Sudbury Star_ des années trente, des articles rapportent des procès où Jos Laflamme était accusé de contrebande d'alcool. Il a réussi à éviter la prison à cause de vices de procédure. Ce n'était pas ce genre de bon coup qui allait le rendre docile devant la cour.
Une fois, selon le _Sudbury Star_ , Jos Laflamme a été convoqué comme témoin dans une affaire d'incendie criminel pesant contre une certaine madame Fortin de Gogama. Le procureur de la Couronne le questionnait manifestement pour montrer sa faible crédibilité. Il a demandé, à Jos Laflamme, s'il avait fait commerce illicite d'alcool dans le passé.
— Ça, ça a rien à voir avec l'histoire de madame Fortin, qu'il a répliqué.
Le procureur a insisté.
— Avez-vous fait de la contrebande, oui ou non?
L'insinuation était embarrassante. Mais la réponse de Jos Laflamme est partie comme une flèche:
— Tu devrais bien le savoir, monsieur le procureur. T'es venu chez moi bien assez souvent.
C'est la réplique d'un homme qui n'a pas froid aux yeux. S'il n'avait pas peur des avocats, c'est qu'il avait mis à sa main bien pire qu'eux. Jos Laflamme a tenu les rênes d'attelages d'orignaux domptés et de loups domptés. J'ai bien dit domptés, pas apprivoisés. Parce qu'au dire de Jos Laflamme, on n'apprivoise jamais un loup. On ne peut que le dominer.
_Wolfman of the North_ , tel était son surnom, à Jos Laflamme, quand il se donnait en spectacle. Son attelage de loups a parcouru les rues d'Ottawa, de Toronto, de Buffalo, de Boston et de New York en plein sur la Broadway. Le monde a changé. Aujourd'hui, personne n'irait donner pareil spectacle dans ces villes du Sud peuplées d'écologistes de banlieue conducteurs de « véhicules sportifs utilitaires ». Mais, dans le temps, pour voir ce gros homme fort à la forte barbe, habillé en trappeur du Grand Nord, les jambes rayées par les griffes et les crocs des loups, on se massait dans les rues l'hiver comme pour la parade du père Noël.
Une fois, cependant, une mémorable fois, Jos Laflamme a eu de la misère avec ses loups. Mais il faut dire que lui aussi, il leur avait donné de la misère. Fin janvier 1939, Jos Laflamme devait descendre à Sudbury pour faire voir ses loups. Mais ce matin-là, il faisait moins trente, trop froid pour les faire monter dans un wagon de train non chauffé. Et le train, je l'ai dit plus tôt, était la seule manière d'entrer à Gogama ou d'en sortir. La seule? Non, pas la seule. Il y avait aussi l'avion de brousse. Jos Laflamme s'y connaissait, lui qui, une fois, avait survécu plusieurs jours en forêt après l'atterrissage forcé d'un avion de brousse qui avait perdu la moitié de son hélice. En tout cas, ce matin-là, Jos Laflamme s'est dit: on va les amener en ville en avion, ces loups-là.
Une fois dans le ciel, le rapport de force entre les loups et leur maître a été mis à l'épreuve. Les loups se sont affolés. Ils voulaient sortir de là. Imaginez la scène. Jos Laflamme devait repousser les loups vers l'arrière de l'avion pour les empêcher de se jeter sur le pilote. Ceux qui ne se ruaient pas sur lui grattaient frénétiquement le fuselage de l'avion de brousse qui, comme vous le savez peut-être, est fait de toile. La folle envolée jusqu'à Sudbury a pris quarante minutes. Au journaliste du _Sudbury Star_ qui l'a interviewé après l'atterrissage, Jos Laflamme a déclaré ceci: « En toutes mes longues années de dompteur de loups, ce matin, c'est la seule fois où j'ai vraiment eu peur. »
Mais la journée n'était pas finie. Ses loups devaient encore tirer leur traîneau à travers le centre-ville de Sudbury. Les rues Durham et Elgin étaient bondées. Les spectateurs débordaient des trottoirs et s'entassaient si serré qu'il ne restait presque plus de place pour laisser passer l'équipage sauvage. Le journal rapporte que pour partager cette mer humaine, Jos Laflamme faisait claquer son fouet en criant à la foule : « Ce fouet-là, c'est pour mes loups, mais si vous ne vous tassez pas, vous allez y goûter! » Jos Laflamme a fini par perdre le contrôle de ses loups, qui étaient sans muselière. Heureusement, ils n'ont fait rien de plus grave que de se réfugier derrière un hangar près de la voie ferrée.
Mais ce n'est pas tout. Le lendemain, Jos Laflamme était invité à la radio CKSO de Sudbury, juste pour y faire prendre sa photo avec un loup. Jos Laflamme installe son loup muselé sur la table devant les micros. Pour que la photo soit plus jolie, il lui enlève sa muselière. Imprévisible comme toujours, le loup choisit ce moment-là pour changer d'humeur, babines dressées, crocs dénudés, en tirant sur son collet de chaîne. Puis soudain, il s'échappe de son collet. L'animateur a juste le temps de sortir du studio. Ce qui s'ensuit, c'est une lutte entre le loup et le dompteur, qui tient une chaise comme on le fait devant les lions au cirque. Jos Laflamme veut l'approcher de dos, mais toujours le loup lui montre ses crocs. Sur ces entrefaites, quelqu'un à CKSO a l'idée d'ouvrir les micros. Alors la musique du Metropolitan Opera est soudainement interrompue par des grognements de loup et des commandes que Jos Laflamme crie en anglais, en français et en ojibwé. L'animateur excité se met à décrire en direct ce combat, qui prend fin quand Jos Laflamme réussit enfin à saisir le loup dans une prise de lutteur pour lui enfiler sa muselière. Cinq minutes plus tard, le loup est redevenu tout calme. Donc qu'est-ce qu'il fait, Jos Laflamme? Il lui enlève sa muselière! Et on la prend, la fameuse photo, comme si rien de tout ça n'était arrivé.
Voilà le genre d'homme qu'on pouvait être dans les années cinquante, quand on était Jos Laflamme, _Wolf Man of the North_ , et qu'on habitait Gogama, Nord de l'Ontario. C'est une époque pas si lointaine. Des habitants de Gogama vous diront qu'ils l'ont vu, Jos Laflamme, dans les rues du village dans son traîneau tiré par des orignaux. Ils se souviennent de ses loups qu'il gardait dans sa cour derrière la maison, où il gardait aussi dans un hangar ses réserves d'alcool. L'époque a beau n'être pas lointaine, ces souvenirs sont d'un autre monde. Jos Laflamme, vedette internationale, a disparu dans l'anonymat. Comme lui, des milliers d'habitants de Gogama sont partis, laissant derrière eux un village en déclin à la place d'une petite ville si sauvagement animée qu'un contrebandier pouvait se moquer de la justice et que le hurlement des loups se mêlait à celui des scieries.
Disparaître sans laisser de traces après avoir été si extraordinaire, ce n'est pas une fin juste pour une vie d'homme. Non plus pour une région qui épuise ses énergies et ses richesses naturelles à alimenter en matières premières des industries installées ailleurs. un jour, le Nord de l'Ontario va devoir apprendre à déjouer l'économie qui lui réserve ce rôle ingrat et éphémère. Nous allons devoir apprendre à mieux nous y prendre pour nous enraciner et durer. Or, je devine qu'une partie du secret du dur désir de durer, c'est peut-être de savoir rappeler les sentiers par où le passé nous a menés et les personnages plus grands que nature qui les ont ouverts. Jos Laflamme n'était peut-être pas un homme qu'on aurait aimé fréquenter ou recevoir dans son salon. Mais c'était un homme plus grand que nature, qui nous apprend qui nous avons été et qui nous sommes encore un peu. Avec une semblable force de volonté aujourd'hui, de quoi serions-nous capables? Les excentriques, les indomptés, les esprits libres, les Jos Laflamme ne doivent pas être d'une autre époque. Il nous en faudrait, aujourd'hui, des personnages ayant cette force de caractère et cette originalité. On devait au moins se souvenir de ceux qu'on a vus passer.
###### Les petites fraises d'été
_À l'âge où les rêves, c'est toute la vie_
_Il a pris la route des espoirs permis_
_Croyant qu'un avenir lui était promis_
_Dans un pays jeune comme lui_
_Il a eu une bonne job en sortant du secondaire_
_A marié sa blonde, est devenu un père_
_A bâti une maison dans une petite ville prospère_
_Quand le Nord était jeune comme lui_
Les jours sont courts et l'hiver est long
On a pris ce qu'on a pu quand les temps étaient bons
On a tant donné, où est-ce donc tout allé
Peux-tu encore espérer
Le retour des petites fraises d'été?
_Les chemins de bois s'allongent de plus en plus_
_Il faut rouler deux heures juste pour se rendre au début_
_Où les machines avancent, elles laissent la terre toute nue_
_Elles rasent toute une forêt par mois_
_Aujourd'hui, ils te demandent d'investir des milliers_
_De t'endetter à la banque juste pour travailler_
_C'est la grosse abatteuse qu'il faut pour bûcher_
_Pas les scies à chaînes comme autrefois_
Les jours sont courts et l'hiver est long
On a pris ce qu'on a pu quand les temps étaient bons
On a tant donné, où est-ce donc tout allé
Peux-tu encore espérer
Le retour des petites fraises d'été?
_Y en a plus de travail où deux bons bras suffisent_
_Une bonne tête, un bon dos et de la débrouillardise_
_On ne vient plus ici comme en terre promise_
_À cause des bonnes payes_
_Il a dit à son gars d'aller faire ses études_
_Dans un gros collège d'une ville du Sud_
_C'est pas que le Nord est rendu trop rude_
_C'est que c'est plus pareil_
Les jours sont courts et l'hiver est long
On a pris ce qu'on a pu quand les temps étaient bons
On a tant donné, où est-ce donc tout allé
Peux-tu encore espérer
Le retour des petites fraises d'été
###### Watershe
Quand on est jeune, les vacances d'été représentent souvent la plus grande des libertés. Mais pour une jeune fille qui habite Gogama, c'est pas nécessairement le cas. Dans le fond, les vacances m'attristaient. J'aimais ça, moi, quand j'étais jeune, aller passer la semaine à Timmins, suivre des cours à Thériault. C'était la seule chance que j'avais de sortir du village, sauf quand j'allais faire les épiceries avec ma mère, une fois par mois. L'été, j'étais prisonnière des lacs, de la forêt pis des maringouins, c'est pas mêlant! Si t'aimais pas la chasse ou la pêche, il y avait pas grand-chose à faire. Ça fait que je faisais pas grand-chose.
À l'âge de 17 ans, j'ai réussi à me poigner une job d'été au restaurant _Watershed_. Ça s'appelle le _Watershed_ parce qu'on est sur la hauteur des terres entre deux bassins versants. D'un côté, l'eau coule vers le nord, jusqu'à la baie d'Hudson; de l'autre, vers le sud, jusqu'aux Grands Lacs. En tout cas, ce nom me donnait des idées de départ. En plus, il y avait de l'argent à faire, pis ça changeait le mal de place. Dès le début, je me suis dit que si jamais je ramassais assez d'argent, je quitterais Gogama pis jamais j'y remettrais les pieds. Travailler au _Watershed_ , ça me permettait de rêver à une vie meilleure, une vie plus excitante que celle que je vivais.
Un soir de juillet, je venais tout juste de nettoyer la dernière table du restaurant pis je préparais les poubelles quand tout d'un coup quelqu'un a frappé à la porte. J'ai répondu. Il voulait savoir si le restaurant était toujours ouvert. un grand gars aux yeux bleus perçants. J'ai dit non. Ça faisait une demi-heure que j'avais barré la porte. « C'est de valeur, j'ai tellement faim », qu'il m'a dit avec un petit sourire en coin. C'était clair qu'il n'avait pas du tout envie de partir. Ça se voit, vous savez! Pis c'est drôle, mais c'est comme si je voulais pas qu'il parte. Je sais pas ce qui m'a pris, mais sans même hésiter j'ai ouvert la porte. Il est entré et il s'est assis. Timidement, je lui ai donné le menu et j'ai dit que je pouvais lui préparer n'importe quoi. Il a feuilleté rapidement et sans réfléchir a commandé le _cheeseburger platter_. Pour le _burger_ , il y avait pas de problème, mais ça faisait déjà deux heures que j'avais éteint la friteuse. Il aurait eu à attendre que l'huile soit assez chaude pour que je puisse lui faire des frites. Il m'a dit que c'était correct, qu'il pouvait patienter. Pis moi, ben, disons que, soudainement, j'étais pas pressée de partir.
En attendant que l'huile chauffe, il a offert de m'aider à sortir les poubelles. J'ai accepté. Des sacs plein les bras, il m'a suivi derrière le _Watershed_. J'ai déposé les miens contre le mur et il a fait pareil. Pis là, avant même que je puisse le remercier, il m'a embrassée. C'est vrai! En plein sur la bouche. Il m'a chuchoté à l'oreille qu'il me trouvait belle. Pis là il m'a tourné le dos, il a ramassé une vieille échelle qu'il y avait au sol, il l'a placée contre le bord du toit et s'est mis à grimper. « un vrai fou! » que je me suis dit. Arrivé au sommet, il s'est tourné vers moi et m'a demandé de le suivre. Sur le coup, je pouvais pas répondre, j'étais tellement nerveuse, j'avais le cœur pris dans la gorge. En tout cas, poussée par la curiosité, je l'ai suivi.
Arrivée en haut, son regard était braqué sur moi comme celui d'une bête sauvage qui examine sa proie. La seule différence, c'est qu'il y avait aucune violence dans son regard. C'est drôle, mais je me sentais bien avec lui. Je l'ai embrassé, ou il m'a embrassée (enfin! j'oublie), pis là j'me suis dirigée vers le devant du toit, là où se trouvaient les grandes lettres qui formaient le mot WATERSHED. Je me suis assise pis j'ai placé mes talons contre le D pour éviter de glisser. Assis à côté de moi, il s'est mis à me dire comment il trouvait ça beau la forêt la nuit. Vous me direz que c'est absurde, mais pour la première fois je trouvais ça beau, moi aussi.
Il m'a parlé de tout et de n'importe quoi. Je sais même pas si j'écoutais les mots qui lui sortaient de la bouche. Pis il m'a embrassée encore et, dans le temps de le dire, il était sur moi. Moi, ben, je me suis laissé faire. Quand il est venu, au moment même, tout ce que j'entendais c'était sa respiration. Pis tout à coup, un bruit fracassant. Il s'est enlevé de sur moi et s'est dirigé vers le bord du toit. « _Fuck_ , calice! » Il arrêtait pas de hurler. « _Fuck_ , calice! » Pis il est descendu à toute vitesse. En remettant mes p'tites culottes et ma robe d'été, je me suis dirigée vers le bord du toit et c'est là que j'ai aperçu ce qu'il avait vu. Le D de WATERSHED était tombé sur le _hood_ de son camion. Je sais pas, mais j'imagine que j'ai donné un coup de talons sur le D quand on était en train de... En tout cas, il était en crisse! Je peux comprendre pourquoi, ça avait complètement poqué sa _paint job_.
Mais c'est ce qui s'est passé après que je pourrai jamais comprendre. En beau maudit, il a enlevé le D de sur son camion et m'a regardée comme pour me dire que c'était ma faute que sa _paint job_ était _fuckée_. Sans dire un mot, il a pris le volant pis il est parti dans un nuage de poussière, comme ça, direction Timmins. Je suis restée là sur le toit, à le regarder s'éloigner. Lorsque son camion est enfin disparu, j'ai pris mon courage à deux mains pis je suis descendue de ma perche. J'aurais pu monter dans mon char pour le suivre, mais je l'ai pas fait. Mieux valait l'oublier tout de suite. En tout cas, essayer! J'ai ramassé le D pis, en pleurnichant, je l'ai jeté dans le champ d'à côté.
Ce soir-là, j'ai très mal dormi, couchée à plat ventre parce que j'avais les fesses tellement irritées par la toiture. En plus, j'ai pleuré presque toute la nuit. Et le lendemain au travail c'était pas mieux. Loin de là! Ma patronne m'a prise de côté pour me dire qu'elle était pas impressionnée par mon travail. Ça d'l'air que quand je suis partie la veille, j'ai oublié d'éteindre la friteuse. Il y a pas eu d'incendie, mais quand même, elle était pas contente de moi, mais pas du tout! Elle m'a dit qu'elle trouvait que j'avais la tête ailleurs, pis que si j'étais pas capable de me déniaiser, elle serait obligée de me _clairer_. C'est vrai que j'avais la tête ailleurs. Si seulement le corps avait suivi! En tout cas...
Avec le temps, j'ai fini par comprendre qu'il est loin d'être facile de se faire oublier, surtout quand on veut. Vous voyez, quand on habite un village comme Gogama, tout le monde connaît les potins des autres. On peut rien dire, rien faire sans que tout le monde soit au courant. C'est pourquoi les gens préfèrent souvent vivre leur petite vie en silence, en retrait, parce que c'est plus facile que de subir le jugement de son voisin. Je suis convaincue de ça parce que c'est justement ce que j'ai choisi de faire. M'enfermer dans le silence. Ma petite prison bien à moi.
J'avais 17 ans et... j'étais enceinte. Pis le pire c'est que je connaissais même pas le nom du père. Je savais que je devais pas parler de cette relation, question de pas faire aller les mauvaises langues. Vous vous rendez compte? « Elle est enceinte d'un homme sans nom qui sera jamais le père de son enfant et patati, patata! » Non! Personne du village, ni même de ma famille, pouvait savoir ce que j'avais fait. On peut même pas péter ici sans que tout le monde vienne sentir. Pis moi, ben, j'avais pas le goût que tout le monde se mette le nez dans mes affaires, que tout le monde se mette à parler de moi. À dire que j'étais une créature sans moralité, une débauchée, une moins que rien qui voit sa vie changer parce qu'un beau soir, la quatrième lettre de l'alphabet décide de s'écraser sur un camion! Non, ça m'intéressait pas.
Deux ou trois semaines après « l'incident », ben, je me suis accotée avec un gars du village. Celui-là avait un nom et je le connaissais. Robert, c'était mon voisin quand j'habitais sur la Cedar. Je lui ai dit que l'enfant était le sien. Il a dû me croire parce que pas longtemps après, on s'est mariés. Le village au complet y était. Pis quand la p'tite est née, toutes les femmes ont organisé le plus gros des _baby shower_ imaginables. Aujourd'hui, je travaille encore au _Watershed_. Et après toutes ces années, pouvez-vous imaginer qu'ils ont jamais remplacé le D?
L'autre jour, Robert a emmené la p'tite à souper au restaurant. Il est tellement bon pour elle et pour moi. Pendant qu'ils attendaient leurs repas, la p'tite faisait ses devoirs à la table, un projet au sujet des Amérindiens. Tout à coup elle s'est mise à rigoler. Elle a trouvé le mot _weebashi_ dans son petit dictionnaire ojibwé. Ça veut dire : « celui qui est poussé par le vent ». Elle a dit : « M'man, _weebashi_ ça sonne un peu comme _watershe_ , comme ce qui est écrit sur le toit du restaurant ».
Soudain, la porte s'est ouverte. Il est entré et s'est assis à la même table où il s'était assis, 13 ans avant. Je l'ai reconnu tout de suite. Son visage avait vieilli, mais c'était lui. On oublie pas facilement des yeux comme les siens. Je me suis approchée de sa table, menu en main et le cœur serré. Arrivée devant lui, je m'attendais à ce qu'il me dise quelque chose, n'importe quoi. Mais non, rien, il m'a même pas reconnue. C'est drôle, mais j'étais soulagée. Du coin de l'œil j'ai vu Robert et la p'tite. C'est là que j'ai compris pour la première fois que les événements de ma vie avaient respecté la direction d'un courant naturel, que j'aurais pas pu être ailleurs qu'ici.
J'ai bien peur que bientôt ma p'tite va commencer à trouver le temps long, qu'elle va commencer à rêver à des ailleurs. Dans le fin fond, j'aimerais ça qu'elle reste. Mais si elle se sent poussée par le vent et qu'elle décide de partir, je sais que je pourrai pas l'arrêter, qu'elle va devoir suivre le courant qui l'entraîne.
_Céleste Dubé_
###### _Incitatif pédant_ : Go! Go! Gogama!
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau; vous êtes à l'écoute d' _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
_Il va répondre, mais elle enchaîne._
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez notamment écrit _Marathon : un avenir unijambiste_. Vous avez également publié un rapport sur la communauté de Sandwich, dans le Sud-Ouest ontarien, rapport intitulé _Sandwich : l'affaire est ketchup_. Vous venez maintenant nous présenter votre plus récent rapport, savamment intitulé _Go! Go! Gogama!_
_Il va répondre, mais elle enchaîne._
Vous êtes titulaire d'une maîtrise en histoire. Votre mémoire de maîtrise s'appelait _Le saucisson hongrois_ _— Témoin de l'évolution d'un continent pluriethnique_ , un mémoire qui fait autorité dans le monde de la charcuterie. Vous êtes trois fois champion du monde de corde à danser.
_Il va répondre, mais elle enchaîne._
Vous êtes ambidextre. Des deux mains à part de ça! Et on pourrait continuer comme ça longtemps, mais ça serait exagéré.
_Il va répondre, mais elle enchaîne._
Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous avez étudié la communauté de Gogama dans la perspective de diversifier son économie et, à la lecture de votre rapport...
_Elle montre le rapport, un volume gigantesque_
On se rend bien compte que les pistes ne manquent pas.
###### Gregory Rotheingham
Tout à fait, Annie-France. Nos nombreuses études sont catégoriques : Gogama a tout ce qu'il faut pour provoquer un boum économique.
###### Annie-France Brazeau
Un boum économique?
###### Gregory Rotheingham
Un boum économique.
###### Annie-France Brazeau
Boum?
###### Gregory Rotheingham
Boum! Ou si vous voulez, une véritable explosion des fondements commercialotouristiques de Gogama.
###### Annie-France Brazeau
Ah, là c'est clair! Et quels sont les grands axes que vous avez identifiés pour faire de Gogama la nouvelle New York?
_Gregory lui fait signe de faire une comparaison moins ambitieuse._
La nouvelle Ottawa?
_Gregory lui fait signe de faire une comparaison moins ambitieuse._
La nouvelle Sudbury?
_Gregory lui fait signe de faire une comparaison plus ambitieuse._
Bon, enfin, quelles sont vos pistes, Gregory Rotheingham?
###### Gregory Rotheingham
Premièrement, les études de marché montrent un énorme potentiel touristique.
###### Annie-France Brazeau
Ben là, Gregory Rotheingham, un peu de sérieux.
###### Gregory Rotheingham
Non, non, non, écoutez bien, Annie-France. En ressuscitant certains sites patrimoniaux du riche passé de notre merveilleux hameau, on pourrait facilement avoir un circuit historique d'intérêt nord-américain à Gogama.
###### Annie-France Brazeau
Un circuit? Et par où commencerait cette visite?
###### Gregory Rotheingham
Par la salle de billard!
###### Annie-France Brazeau
_en grimaçant_
Étonnant!
###### Gregory Rotheingham
Vous savez qu'à une certaine époque, à Gogama, la messe était célébrée dans la salle de billard?
###### Annie-France Brazeau
Et qu'est-ce que vous en feriez, de la salle de billard?
###### Gregory Rotheingham
Pour attirer les touristes, c'est incontournable : on ferme l'église, on rebâtit la vieille salle de billard et on y célèbre la messe.
###### Annie-France Brazeau
Vous allez avoir un problème : au billard, une des baguettes s'appelle « le diable ».
###### Gregory Rotheingham
Ah oui, il faudrait la sortir.
###### Annie-France Brazeau
Je vois très bien la scène d'ici: « Sortez le diable! Le curé s'en vient! »
###### Gregory Rotheingham
Mieux encore, Annie-France, j'affirme que Gogama peut réussir le même pari que Las Vegas.
###### Annie-France Brazeau
Quoi? Des bingos à longueur de journée? Des nuits de poker endiablées?
###### Gregory Rotheingham
Non. Les gens de partout vont se marier à Las Vegas. Ils ont maintenant une alternative : on invite tout le monde à se marier à Gogama dans le magnifique décor d'une salle de billard d'époque.
###### Annie-France Brazeau
Bien sûr, avec le slogan « Mariez-vous dans une salle de billard : c'est plus facile de casser. »
###### Gregory Rotheingham
C'est pas tout. On pourra aussi visiter la bibliothèque dans un wagon.
###### Annie-France Brazeau
Un wagon de train?
###### Gregory Rotheingham
Oui. Jusqu'au début des années quatre-vingt, la bibliothèque de Gogama était dans un wagon.
###### Annie-France Brazeau
Elle ne devait pas ouvrir souvent à l'heure.
###### Gregory Rotheingham
Ça fait partie du charme. Imaginez un peu, on ramène le wagon rempli de tous mes rapports et de mes études. On attirerait les fonctionnaires de partout sur la planète pour venir admirer mon œuvre.
###### Annie-France Brazeau
Un wagon rempli de rapports, ça va être palpitant. Il faudrait peut-être penser ajouter un wagon-lit.
###### Gregory Rotheingham
Je comprends pas.
###### Annie-France Brazeau
Ah, pas de doute, avec votre wagon-bibliothèque, la relance serait sur les rails.
###### Gregory Rotheingham
Bon, en tout cas, notre tournée historique finirait par la visite de la maison de Jos Laflamme. Vous savez, Jos Laflamme est un peu notre Mickey Mouse à Gogama! Alors pourquoi ne pas créer un gigantesque parc d'attractions inspiré par Jos Laflamme?
###### Annie-France Brazeau
Un parc d'attractions?
###### Gregory Rotheingham
Oui, on place trois ou quatre loups en plastique avec des grognements enregistrés à l'entrée et là les gens entrent et sont accueillis par une mascotte déguisée en Jos Laflamme. On aura aussi des manèges, comme des montagnes russes avec des traîneaux à loup, un carrousel d'orignaux pour les grands et un carrousel de castors pour les petits. On aura des kiosques où les gens peuvent faire du tir au poignet contre un ours, ou faire une course dans un labyrinthe avec une moufette. Évidemment, les touristes devront signer un formulaire qui nous dégage de toute responsabilité. On va aussi vendre des t-shirts, des casquettes à panache et des queues de castor.
###### Annie-France Brazeau
Gregory Rotheingham, vous parlez visiblement d'investissements importants. Comment s'assurer qu'on ne gaspille pas l'argent public?
###### Gregory Rotheingham
Vous savez, Annie-France, on ne me paye pas un salaire faramineux pour jeter l'argent des contribuables par les fenêtres. J'ai déjà dépensé, de façon responsable, des dizaines de millions quand j'étais responsable du programme fédéral de commandites.
###### Annie-France Brazeau
Vous courrez à la ruine.
###### Gregory Rotheingham
Pas du tout, notre campagne promotionnelle ne coûtera pratiquement rien. Saviez-vous que près d'une dizaine de policiers de la Police provinciale de l'Ontario sont postés à Gogama?
###### Annie-France Brazeau
Quoi, une dizaine de policiers pour une population de 400 habitants? Je pense que le message est assez clair. Si vous circulez sur la 144, à l'approche de Gogama, ralentissez!
###### Gregory Rotheingham
Justement, on va profiter de leur expertise. Mais plutôt que de donner des _tickets_ , les policiers vont donner des rabais de 50 % sur le prix d'entrée à toutes nos activités.
###### Annie-France Brazeau
Au nombre de contraventions qui sont émises à Gogama, votre succès est assuré. Alors visiblement, la tournée historique de Gogama s'adresse à un public familial. Mais il y a aussi d'autres publics que vous désirez cibler.
###### Gregory Rotheingham
Bien sûr! Les jeunes. Vous savez, Annie-France, en juin dernier, un train a déraillé à Gogama. Dix-sept wagons chargés de houblon se sont déversés dans le lac. Y'en a qui ont appelé ça une catastrophe écologique. Moi, j'appelle ça une opportunité!
###### Annie-France Brazeau
Fascinant, une opportunité. Pour quoi au juste?
###### Gregory Rotheingham
Pour offrir le premier lac de bière au monde! Avec la quantité de houblon dans le lac, l'eau est devenue brune. On a donc fait des tests pour se rendre compte qu'on avait un lac de bière à 5,6 % d'alcool. Imaginez l'attrait pour des jeunes de 19 à 25 ans! Pendant la semaine de relâche de février, oubliez la Floride, on s'en va à Gogama!
###### Annie-France Brazeau
Mais comment allez-vous vous adapter aux tendances des grandes destinations touristiques des étudiants?
###### Gregory Rotheingham
Ça se passe en février, alors c'est garanti que la bière va être froide. On va aménager des cabanes à pêche munies de pompes avec un compteur pour siphonner le lac. On aura aussi la populaire baignade polaire, le _Polar Beer Dip_. Également, on va recycler les serres de la pépinière avec d'immenses lampes de bronzage et organiser un tournoi de volley-ball de plage en bikini.
###### Annie-France Brazeau
Oui et je crois que j'ai trouvé la chanson thème de votre festival : « Elle a un tout petit, petit bikini, au lac Mi-ni-si-na-kwa oui oui! »
###### Gregory Rotheingham
À la fin de la journée, les jeunes vont tout voir en double. Alors ça devient facile de leur faire croire qu'ils ont pêché trop de poissons et qu'ils ont dépassé la limite. Et bang! la police leur donne des _tickets_!
###### Annie-France Brazeau
Je suis sûre que les étudiants vont beaucoup apprécier de pêcher des poissons qui ont des bedaines de bière.
###### Gregory Rotheingham
Et n'oubliez pas que le lac de bière est là toute l'année! Au mois de juin, il y aura un grand Festival de la bière. Tous les 16 juin, on recréera le déraillement en déversant 17 wagons de bière dans le lac. Tous les 16 juin, on recréera le déraillement en déversant 17 wagons de bières dans le lac.
###### Annie-France Brazeau
Le lac de bière ne vous fait pas. Vous parlez en double, Gregory Rotheingham.
###### Gregory Rotheingham
Et à la fin de la journée, les policiers donnent des _tickets_!
###### Annie-France Brazeau
Bien sûr. Mais vous avez aussi un projet de relance économique avec une usine de homard.
###### Gregory Rotheingham
Ah oui! Le _Gogama Lobster_...
###### Annie-France Brazeau
Il y a du homard à Gogama?
###### Gregory Rotheingham
Ben non, le _Gogama Lobster_ est fait avec du brochet.
###### Annie-France Brazeau
Le _Gogama Lobster_ , c'est du brochet. Ça, c'est une idée qui doit venir du lac de bière. Remarquez qu'après le café décaféiné et la bière sans alcool, pourquoi pas du homard sans homard? Expliquez-moi ce que c'est exactement, le _Gogama Lobster_.
###### Gregory Rotheingham
C'est une recette locale faite avec de minces filets de brochet tranchés en morceaux et bouillis pendant sept minutes dans l'eau salée. Les morceaux forment comme des frisettes qu'on fait revenir dans du beurre à l'ail. On va faire du _Gogama Lobster_ en canne et le vendre partout dans le monde. Attention, Capitaine High Liner!
###### Annie-France Brazeau
Et vous n'avez pas peur d'avoir des problèmes avec l'appellation « homard » de votre produit?
###### Gregory Rotheingham
Non, on a mis une petite étoile à côté du nom et au bas de l'emballage, en petits caractères, on a écrit : 100 % brochet.
###### Annie-France Brazeau
Exactement, comme on a fait sur votre contrat, sauf que vous, au bout de la petite étoile, c'est écrit : bénévolat.
###### Gregory Rotheingham
Ce qui sera beau, Annie-France, c'est qu'une telle usine créerait des centaines d'emplois, donnerait de la vigueur à l'économie et, tous les jours, à la sortie de l'usine, les policiers donneraient des _tickets_.
###### Annie-France Brazeau
Ah, que de beaux projets pour les Gogamois!
###### Gregory Rotheingham
Gogamiens!
###### Annie-France Brazeau
Gogaméens... Gogamaltèque...
###### Gregory Rotheingham
Peut-être Gogamais?
###### Annie-France Brazeau
Ça vaudrait la peine de faire un rapport sur le sujet, Gregory Rotheingham.
###### Gregory Rotheingham
Trop tard, j'ai déjà un sujet pour mon prochain rapport : les grandes fatigantes en visite hors de leur habitat naturel.
###### Annie-France Brazeau
Ben alors, voilà, je vous remercie, Gregory Rotheingham. Je vous rappelle que votre rapport _Go! Go! Gogama!_ est un incontournable du monde du rapport et on vous retrouvera très bientôt.
###### Gregory Rotheingham
Merci Annie-France.
###### Votre fils au loin
_Collé un timbre sur ma lettre_
En pensant payer le prix
_De la belle vie dans la grande ville_
_Que je m'étais choisie_
_Tu prendras soin de toi, mon gars_
_Que mon père m'avait dit_
_Tu sais que quoi qu'il arrivera_
_T'as une place ici_
J'ai signé: Votre fils au loin
Prenez bien soin de vous
J'ai écrit: Votre fils va bien
J'me sens ici chez nous
_De temps en temps, je téléphone_
_Oui, je sais bien, c'est pas souvent_
_Mais une lettre, c'est la bonne manière_
_De dire ce qui est important_
_Je travaille fort, mais je tiens le coup_
_Pas du tout envie de lâcher_
_Mais je pense quand même à vous_
_Eh oui, parfois vous me manquez_
J'ai signé: Votre fils au loin
Prenez bien soin de vous
J'ai écrit: Votre fils va bien
J'me sens ici chez nous
_Je voyais loin, je pensais grand_
_Mais dans la capitale_
_Des rêves, des rêves, il y en a tellement_
_Qu'enfin, on les voit mal_
_Je prends l'autobus chaque matin_
_La foule est tellement dense_
_Le défi c'est de gagner mon pain_
_Sans perdre mon espérance_
J'ai signé: Votre fils au loin
Prenez bien soin de vous
J'ai écrit: Votre fils va bien
J'me sens ici chez nous
###
### Salut Verner!
Les voyageurs sur la route 17 en provenance de Sudbury voient soudain la forêt reculer, l'horizon s'ouvrir et les champs s'étendre dans une plaine destinée à être une prairie laitière. une rivière étroite comme une ruelle serpente dans la plaine; elle frôle et câline la Transcanadienne. La Veuve, disent en français les panneaux de la province. Ce nom laisse imaginer la femme, en taisant son nom. Certains disent que c'était à l'origine la rivière Beuve. On aurait donc corrigé la majuscule. Tant mieux pour la poésie du pays.
Entre la route et la rivière, c'est Verner, village « fier de ses droits, fidèle à ses traditions », comme c'est écrit, en français, sous les armoiries du canton. La population de Verner, quinze cents habitants, est à forte majorité française, comme en témoignent les beaux noms de plusieurs commerces et institutions : le restaurant L'Écho, la Cuisine du soleil, la Villa du bonheur, la rue Principale, l'école Marguerite-d'Youville et l'église Saint-Jean-Baptiste. À côté de l'église à Verner, il y a une plaque de la Société historique de l'Ontario. Mine de rien, c'est le détail le plus incongru du paysage. Car cette plaque en plein cœur de Verner rend hommage à un homme qui ne croyait pas en Verner.
Charles Alfred Paradis : prêtre rebelle, justicier fougueux, colonisateur visionnaire, tempérament volontaire, brasseur de m... mémorables controverses. C'était un homme qui savait croire et surtout faire croire en des pays qui n'existaient pas encore. Pourtant, son village de Domrémy, le grand rival de Verner, n'existe plus aujourd'hui.
Notre homme avait déjà fait bien du chemin avant d'arriver dans le Nipissing. Il faut savoir par où il est passé pour comprendre qui il était à son arrivée.
###### Charles Paradis
Charles Paradis, né en 1848 dans le Kamouraska, est devenu oblat à la surprise de ses maîtres. Jeune homme trapu, mais expansif, ses maîtres le voyaient, et je cite, comme « un fou, un imaginaire, un buté, un écervelé en théorie et en pratique ».
Au début, il a enseigné l'art à Ottawa. Il en avait le talent, comme en témoignent ses magnifiques paysages nordiques conservés aux Archives nationales. Puis en 1880, à 32 ans, il a fait son premier voyage en Nouvel-Ontario et accompli son premier miracle : il a attrapé, à l'aide d'une perche de cèdre ancrée dans une roche, la plus grosse truite jamais sortie du lac Temagami. Soixante livres, monsieur! presque aussi longue que son canot, croyez-en la parole du gérant du poste de traite. C'est comme ça qu'un pays vous saisit l'esprit.
Le revoici donc en 1882, nommé missionnaire au Témiscamingue. Cette fois, il y vit une histoire à chanson de mensonges : au lieu d'être transporté par le bateau, il a transporté le bateau. Charles Paradis et ses compagnons ont toué et halé sur l'Outaouais la coque du Mattawan, le premier bateau à vapeur du lac Témiscamingue. Ça se dit bien, jusqu'à ce que les câbles cassent dans des rapides de six milles.
Ensuite, Charles Paradis a traversé en canot ce pays neuf jusqu'à Moose Factory. Il a étudié son climat et ses terres pour défaire les préjugés. Selon lui, tout ce pays était propice à l'agriculture, même les plaines de la baie de James. Avec son compagnon, le père Nédelec, il s'est tant disputé qu'il est revenu seul. Sa frasque a coûté 100 $ en dépenses aux Oblats. Le petit père se faisait payer une réputation.
À son retour, Charles Paradis était devenu celui qu'Arthur Buies a décrit comme « ce modeste oblat dévoré d'une ambition patriotique que ne pouvait contenir la sphère où l'emprisonnaient ses humbles fonctions de missionnaire, et dont l'intelligence, écartant les voiles de l'avenir, lui avait révélé ce que pouvait devenir [...] la vaste contrée où il avait jusque-là exercé obscurément son ministère ». Le rapport du père Paradis au ministre des Travaux publics, 70 pages écrites à la main, vous pouvez le lire dans le site Internet des Archives nationales du Québec.
Le problème étant le transport, la solution serait simple: de la dynamite à l'embouchure du lac Témiscamingue, pour l'abaisser; un barrage sur la rivière des Outaouais, pour la relever. Tout ça noierait les rapides et permettrait la navigation. On ne l'a pas laissé faire. Mais retenez la dynamite, elle reviendra. Quant aux _mosus_ de moustiques, Paradis avait la solution là aussi : son _Maringouinifuge_ , formule brevetée. En tout cas, son rapport a si bien vanté le pays que les Oblats ont créé la Société de colonisation du Témiscamingue. Mais ils ont mis à sa tête un autre que lui. Le turbulent père Paradis inquiétait ses supérieurs. À son grand désarroi, ils l'ont écarté en l'envoyant dans le Maniwaki. Leur prudence leur a rapporté quoi? un immense scandale : l'affaire Gilmour, où le père Paradis a acquis une réputation nationale.
Simplifions une histoire complexe. Le gouvernement du Québec avait concédé des terres au père Paradis. Lui, il les vendait aux colons, qui à ce titre avaient le droit d'y bûcher. Mais ces mêmes terres avaient aussi été concédées à la compagnie forestière Gilmour. Préférant le bras de fer aux recours judiciaires, Gilmour marque de ses sceaux le bois bûché par les colons en leur donnant une compensation dérisoire. Indigné par cette injustice, le père Paradis remplace les sceaux de Gilmore par les siens. Gilmour crie au vol. Scandale : un prêtre est sous mandat d'arrêt! La presse du Québec s'empare de cette affaire qui oppose les lois contradictoires de deux gouvernements successifs. Les journaux rouges sont pour Paradis, les journaux bleus sont contre. Si vous comprenez le dicton — le ciel est bleu, l'enfer est rouge —, vous voyez bien que ce Paradis est de la mauvaise couleur. Il dérange ses supérieurs et tant pis pour le vœu d'obéissance. L'affaire atteint les hauts tribunaux et Charles Paradis en sort vainqueur, grand héros du petit peuple.
Mais le fougueux père a dit une phrase de trop. Il a dit que Ross, l'ancien premier ministre bleu, lui avait offert un pot-de-vin de 15 000 $ contre sa trahison des colons. C'est le genre de chose qu'il ne faut jamais dire parce qu'on ne pourra jamais le prouver. Ross se plaint vigoureusement auprès du supérieur des Oblats, le père Augier. Celui-ci demande à Paradis de se rétracter. Pensez-vous qu'il allait se rétracter? Conséquence : le voici maintenant banni plus loin encore, à Buffalo, dans l'état de New York, où il s'installe dans un presbytère... sans permission. Les policiers viennent l'en expulser. C'est au tour du père Paradis de se plaindre vigoureusement du traitement de ses supérieurs. Il ira se défendre jusqu'à Paris et à Rome. Mais à la fin, on l'expulse de l'ordre des Oblats. Jusqu'à la fin de ses jours, le père Charles Paradis ne fera plus confiance aux autorités religieuses. Mais on dit aussi que jusqu'à la fin de ses jours, le père Augier n'a jamais pu entendre prononcer le mot _paradis_ sans tiquer.
À 42 ans maintenant, Charles Paradis, expulsé du Témiscamingue, du Maniwaki, du Canada puis des Oblats, garde quand même des amis en haut lieu. Ceux-ci convainquent l'évêque de Peterborough de lui faire une petite place aux confins de son diocèse. Le père Paradis sera agent colonisateur des environs du lac Temagami. En 1893, il arrive dans le Nouvel-Ontario avec des compagnons. Il s'installe d'abord dans une baie du lac Temagami qu'il appelle Sacré-Cœur et qui s'appelle aujourd'hui Sandy Inlet. Il imagine ces rives parsemées de fermes prospères. Mais il sait qu'il faudrait commencer par coloniser plus au sud, aux abords de la voie ferrée. Donc, à deux milles à l'ouest de Verner, il construit une chapelle, une résidence et une scierie. Il nomme l'endroit Domrémy, en évoquant Jeanne d'Arc, la sainte qui a bravé les autorités, comme lui. L'été, il se rend de Domrémy au lac Temagami par les rivières. Mais pour l'accès en hiver, il ouvre en fendant la forêt un chemin de 26 milles en ligne droite qui suit la limite du comté. Cet ancien chemin est aujourd'hui presque entièrement envahi par la végétation, mais les gens de Verner le connaissent encore. Ils l'appellent « la marche du père Paradis ».
Jusqu'à ce moment-là, la colonisation le long du Canadien Pacifique avait été surtout l'affaire des Jésuites. Les curés locaux n'ont pas apprécié la présence de ce prêtre autonome. Son coup d'éclat du printemps de 1895 n'a pas aidé l'affaire. Ce printemps-là, le père Paradis est allé dans les chantiers du Michigan y retrouver des Canadiens français expatriés au péril de leur langue et de leur foi. Notre homme savait faire croire en des pays à venir; en voici la preuve. Le père Paradis a soulevé une vague de rapatriement impliquant environ 2 000 personnes. Trois ou quatre cents sont arrivés en premier. Instantanément, Domrémy éclipsait Verner. Mais peu avait été mis en place pour les recevoir. Certains sont rentrés au Michigan, désillusionnés. Beaucoup ont travaillé avec vaillance pour s'installer, leur prêtre bien-aimé trimant dur à leur côté. Mais une minorité s'est agitée, sous l'incitation du curé Desaulniers. Il a dénoncé Paradis au gouvernement. Il l'a accusé de vouloir détruire Verner et Warren. Il l'a même accusé de fraude, ce qui était un bien gros mot pour dire que Paradis, à son avis, investissait mal les fonds de la colonisation. Des enquêteurs venus voir sur place ont reçu une pétition d'appui au père Paradis signée par la majorité des rapatriés. Et on a pu prouver que des signatures sur la contre-pétition de Desaulniers étaient fausses. Cependant, un envoyé de l'évêque, le père Laurent, a vu les choses autrement : « Tout son projet pour le lac Témagami est une chimère et si ce coin de pays n'était pas si reculé, aucun homme sain d'esprit ne l'écouterait. » Il a décrit le père Paradis comme un rusé malhonnête au discours violent contre ses opposants.
Puis, encore une fois, le père Paradis a posé un geste de trop. En mai 1895, l'église de Verner étant abandonnée, il l'occupe. Le père Ferron de Sturgeon Falls délègue deux paroissiens pour aller la reprendre. Résultat : une échauffourée à coups de bâton. Le père Paradis fait arrêter les deux paroissiens. L'évêque s'en mêle et finit par lui retirer tous ses droits dans la région, sauf sur le lac Temagami, comme missionnaire des Amérindiens. L'évêque discipline aussi les curés Ferron et Desaulniers pour... abus d'alcool.
Mais le père Paradis avait encore des amis au gouvernement. En 1896, il devient agent de la Société de colonisation ontarienne. L'année suivante, un grand malheur : un feu à Domrémy détruit sa chapelle, sa scierie et son moulin à grain. Le père Paradis est criblé de dettes. Le nouveau curé Langlois de Verner le fait brièvement emprisonner pour ça. un autre enquêteur gouvernemental vient voir. Paradis s'en tire encore sans blâme. Puis, dans sa période la plus noire, le père Paradis a reçu un bel appui, mais de l'extérieur. Le journal _La Patrie_ de Montréal envoie un journaliste. Un long article signé « Henri » vante les projets du père Paradis. Il dit que Domrémy sera un haut lieu du Nipissing quand le père aura fait harnacher les chutes de la Temagami pour bâtir un chemin de fer électrique jusqu'au lac. Il dit même que l'emplacement est mieux choisi que Verner, parce que les chemins provinciaux sont normalement à la limite des comtés. Malgré tout, le couperet est tombé sur ces rêves. En 1899, la Société de colonisation provinciale a cessé de financer son agent controversé. Le dynamisme fougueux du prêtre autonome a perdu. La règle et l'ordre des curés conventionnels ont gagné.
Le père Paradis a quand même continué de recruter des colons. À son dire, il a rapatrié huit mille Canadiens français, éparpillés de Mattawa à Sault-Sainte-Marie. Certains doutent qu'il y en ait eu autant, mais ce qui est sûr, c'est qu'il y a en eu beaucoup. On a continué de le persécuter, sous prétexte maintenant que le Canadien Pacifique lui faisait un traitement de faveur. Après 1905, Charles Paradis s'est replié au loin sur sa ferme du lac Temagami où il tenait un orphelinat. En 1909, son esprit d'entreprise a resurgi : pour financer l'éducation des enfants, disait-il, il prospectait dans le nord, près de Timmins. Il y a même trouvé un peu d'or. À soixante ans, notre homme a fait à raquettes soixante milles de jour et de nuit pour enregistrer un _claim_ à Sudbury avant un rival. S'il n'avait pas changé son projet de prospection, disait-il, c'est lui qui serait tombé le premier sur le grand gisement de Hollinger.
Maintenant, vous souvenez-vous de la dynamite et de lacs à vider? Eh bien, en 1909, le prospecteur Paradis et ses hommes ont dynamité l'embouchure du lac Frederick House. Désastre: le lac s'est presque vidé! Le petit père se vantait d'avoir donné de bonnes terres alluviales à la colonisation. Le ministre responsable n'était pas impressionné. Il a refusé l'offre quand le père Paradis s'est porté volontaire pour aller remettre le lac à sa place. Plus tard, le chemin de fer menant à Timmins a été bâti sur les terres qu'il a dégagées.
À la mission du Sacré-Cœur à Sandy Inlet, les années ont passé. La colonisation sur le lac n'était pas permise. Le père Paradis y faisait figure folklorique en vendant ses légumes aux touristes. En 1924, un journaliste du _Star Weekly_ y a joint le vieillard de 76 ans. Il a trouvé là un homme vif, animé de l'amour du pays, dents blanches intactes, nez crochu, visière en carton au front, barbe blanche en deux longues pointes sur sa poitrine. Cet homme lui a dit à peu près ceci : « Le monde n'est plus ce qu'il était. Trop de gens s'entassent dans les villes, rêvent de richesse et ne sont jamais satisfaits. Ils ne réfléchissent à rien, mais pensent tout savoir. Je veux pour eux le bonheur de la vie simple que j'ai connue dans la nature. »
Franchement, ce que le père Paradis a vécu dans la nature nord-ontarienne, c'est tout sauf le bonheur de la vie simple. Plus tard cette année-là, un incendie a détruit ses dernières possessions. Il a perdu ses nombreuses aquarelles, des tableaux précieux qu'il avait achetés à Rome et un dictionnaire français-ojibwé qu'il rédigeait depuis des années. une de ses dernières consolations a été de se rendre au Collège Sacré-Cœur de Sudbury pour voir des enfants de ses colons obtenir leur diplôme. Peu après, Charles Paradis s'est rendu à Montréal, où il est mort en 1926.
Aujourd'hui la végétation a envahi sa ferme du lac où des ruines pourries demeurent. un camp d'été pour jeunes a occupé le site. Et Domrémy? Eh bien, allez-y pour voir, à deux milles à l'ouest de Verner, juste à la limite du comté de Caldwell. Il n'y a pas de village là, même pas une affiche. De l'œuvre et des rêves d'un homme plus grand que nature, il reste la descendance de ses rapatriés. Quant à sa marque sur le paysage, il a laissé un sentier en forêt et une plaque, au milieu d'un village qui l'a craint, à côté d'une église dont les curés l'ont malmené.
Verner a prospéré, grâce à d'autres personnalités fortes qui ont suivi des voies plus conventionnelles. Mais les voies conventionnelles mènent ailleurs aujourd'hui, comme le prouvent l'exode de nos jeunes et le plafonnement de nos industries. Les leaders légitimes de nos jours ne semblent pas savoir en changer le cours. Si nos rêves d'hier avaient été plus audacieux, le Nouvel-Ontario s'en porterait-il pire ou mieux? Si demain nous amenait un autre fonceur de la trempe de Charles Paradis, serait-il à sa place ici? Verner, village fier de ses droits, fidèle à ses traditions, lesquelles au juste s'ouvrent sur l'avenir?
###### Par un chemin de bois
_Quand c'est le temps de s'en aller, il faut pas discuter_
_Notre cachette, tu la connais_
_Un coin à notre goût, notre petit coin à nous_
_Au bord du lac dans la forêt_
_La solitude à deux, où veux-tu trouver mieux_
_Pour voir ta vie t'appartenir_
_Toi et moi, seuls, ma belle, dans un monde naturel_
_Ça fait du bien d'y revenir_
C'est souvent comme ça que ça va, dans la vie
À force de faire ce qu'on te dit, tu sais plus t'es qui
La place où être bien, c'est pas toujours très loin
Mais l'affaire, c'est que souvent
On ne s'y rend pas facilement
_C'est un mauvais petit chemin, mais on y passe bien_
_Avec la machine qu'on a_
_Et une fois arrivés et tout bien installés_
_On fait ce qu'on veut, y a personne là_
_Personne pour surveiller, pas de voisins pour chialer_
_Si notre feu fait de la boucane_
_On peut en faire du bruit, si jamais ça nous l'dit_
_Sans s'inquiéter de qui ça tanne_
C'est souvent comme ça que ça va, dans la vie
Où c'est le monde qui mène, il faut que tu fasses ce qu'on dit
Pour tout faire à sa tête, c'est pas si loin peut-être
Mais l'affaire c'est que bien souvent
On ne s'y rend pas facilement
_Chez nous, c'est un village, mais des fois c'est une cage_
_On ne fait pas tout ce qu'on veut_
_Y a bien assez que la semaine, c'est le travail qui mène_
_Un jour on vivra mieux_
_Quand la retraite viendra, on va pas rester là_
_On passera tout notre temps_
_Où tout est vert et bleu, où on respire mieux_
_Sous le soleil, à notre camp_
C'est souvent comme ça, qu'ils disent, dans la vie
À force de faire ce qu'on te dit, tu sais plus t'es qui
C'est au bout d'un mauvais chemin que tu tiens ton destin
Même si pour ça, on s'en sort pas...
Il faut passer par un chemin de bois
###### La parade
J'ai toujours aimé regarder passer les gros chars. C'était ici, à Verner. J'étais toute petite et regarder passer le train des passagers, c'était ma parade à moi. J'imaginais que dans les wagons, c'était plein de beaux personnages déguisés et costumés, comme sur les chars allégoriques de la parade de la Saint-Jean-Baptiste.
Un jour, ma tante Madeleine qui habitait la grand-ville est venue nous visiter. Elle est arrivée par le train. On l'attendait en face de la gare quand on a entendu le sifflet du train qui annonçait son arrivée. Tout de suite à travers le boisé, j'ai vu le gros œil jaune de la locomotive qui fonçait sur nous. Le convoi s'est alors mis à modérer et par la fenêtre de son wagon, ma tante Madeleine nous faisait des signes de la main et de grands sourires, comme une princesse de contes sur un char allégorique. Deux minutes plus tard, elle sortait par une portière dans un nuage de vapeur, comme une apparition de la Sainte Vierge. Elle avait une belle robe blanche et bleue, des gants et un chapeau de paille blancs. Elle sautillait en marchant vers nous, au milieu d'autres voyageurs tristes, en habits noirs et en habits gris. Elle brillait comme le petit mouton blanc de la parade de la Saint-Jean-Baptiste. Elle était si jolie, ma tante Madeleine. Je l'aimais beaucoup.
Elle est restée chez nous juste quelques jours, puis elle a repris le train et n'est plus jamais revenue. Je n'ai jamais su pourquoi, mais j'ai surpris ma mère raconter à la voisine que sa sœur Madeleine était une femme d'affaires influente qui brassait de grosses affaires avec des hommes très très importants à Sudbury, à Toronto, à Montréal, même à New York. C'est pour ça qu'elle n'avait plus le temps de venir nous visiter. J'ai aussi entendu mon père chuchoter à ma mère en ricanant que les grosses affaires que brassait ma tante Madeleine s'appelaient Léon, Georges, Armand, Henri, Pierre, tous des hommes très importants. Oui, j'ai tout entendu, mais je n'ai rien compris du tout!
Le train des passagers a continué de passer chez nous encore quelques années, en sifflant très fort, mais sans jamais ramener ma tante Madeleine. Souvent, moi, au son du sifflet, je courais dehors pour envoyer la main aux gros chars, au cas où ma belle ma tante serait parmi les passagers qui vont à Sudbury et qu'elle verrait mon salut. Plus tard, le maire et des hommes d'affaires sérieux de Verner ont décidé de fermer la gare. Non seulement, ils l'ont fermée, la gare, mais ils l'ont démolie, comme si le maire ne voulait plus voir personne arriver au village ou voir personne quitter le village. Finis les gros chars. Finis les passagers. Adieu ma tante Madeleine! Commençaient mes grandes déceptions.
Je n'avais plus rien à regarder passer, sauf, une fois par année, la parade de la Saint-Jean-Baptiste. Mais c'était tellement beau chez nous, à Verner, la parade du 24 juin, c'était l'événement de l'année. Il y avait au moins quinze tracteurs avec des beaux rubans qui tiraient les chars allégoriques sur lesquels trônaient des gens du village habillés en personnages colorés, beaux comme ma tante Madeleine. Jamais gris, jamais noirs. À chaque année, ce jour-là, même si c'était un jour de semaine, toute la famille s'habillait en dimanche pour aller s'installer en face de l'hôtel Commercial pour voir passer la parade.
Moi, c'était toujours le dernier char allégorique que j'attendais avec impatience. Celui qui amènerait le petit saint Jean-Baptiste aux cheveux blonds tout bouclés naturellement, avec de grands yeux bleus et vêtu d'une peau de mouton. Et, à ses pieds sur de la paille, un vrai mouton en vie, bouclé comme l'enfant et tellement blanc qu'on aurait dit que les organisateurs l'avaient trempé dans l'eau de Javel. Depuis au moins cinq ans que je venais à la parade avec mes parents, j'étais complètement fascinée par ces petits étrangers blonds, frisés naturel et aux grands yeux bleus. Je me demandais bien d'où ils venaient ces beaux enfants-là, parce que chez nous dans le village, c'était plutôt rare les blonds aux yeux bleus. Je me disais que, comme pour les petits moutons blancs, on en faisait sans doute l'élevage dans une ferme pas loin de chez nous.
Cette année-là, je devais bien avoir sept ans. C'était deux jours avant la parade. À l'école, pendant la récréation, mon amie Diane Beaudry m'a demandé :
— Sais-tu qui va faire le petit saint Jean-Baptiste, à la parade samedi?
— Non, je ne le sais pas et toi non plus tu ne le sais pas, que je lui ai répondu. C'est comme un mystère, personne ne le sait sur la terre!
— C'est mon frère! qu'elle m'a annoncé en se vantant comme un paon, la Diane Beaudry.
— Ton frère? Ton frère? Ton frère Jean-Pierre? Jean-Pierre Beaudry va personnifier le petit saint Jean-Baptiste à la parade?
Elle a fait signe que oui, avec un petit air précieux. Je n'en revenais pas. Jean-Pierre Beaudry! Non, non, non. Oui, il avait bien les cheveux blonds et les yeux bleus, mais il avait les cheveux raides comme de la paille. Et puis, je me demandais bien comment ce petit garçon si ordinaire avec des écorchures sur les genoux, qui me criait toujours des noms et me lançait des roches pourrait personnifier un saint du ciel, le patron même de notre paroisse Saint-Jean-Baptiste. Et dire que par-dessus le marché, Jean-Pierre Beaudry avait déjà eu des poux, sa sœur me l'avait dit!
— Je ne te crois pas. Saint Jean-Baptiste, ça ne peut pas être ton frère, que je lui ai dit en tournant les talons.
— Tu verras bien! Tu verras bien! qu'elle m'a lancé en faisant sa fraîche et en rajoutant, et puis, dis-le pas, c'est un secret.
En rentrant chez nous, je l'ai dit tout de suite à ma mère, son secret. Ma mère, qui est comme une sainte et qui ne dit jamais rien de mal contre son prochain, n'a rien dit avec sa voix, mais elle a grimacé si fort que c'était pire que si elle avait hurlé. Après, elle m'a serrée dans ses bras en me disant « T'en fais pas, c'est une blague qu'elle t'a racontée, ton amie Diane ».
Le matin du 24 juin, on s'est tous habillés en dimanche et j'ai eu le cœur serré jusqu'à temps qu'on parte toute la famille. On a longé la rue Principale vers l'hôtel Commercial. On marchait vite pour avoir notre place. En passant devant l'église, j'ai vu que les portes étaient grandes ouvertes. Évidemment, que je me suis dit, le bedeau veut que saint Jean-Baptiste voie passer sa parade. C'est là qu'il m'est passé une idée folle par la tête. J'ai grimpé les marches de l'église deux par deux, je suis rentrée dans l'église et j'ai marché sur la pointe des pieds jusque dans le chœur. Il y avait une grande fresque peinturée au plafond qui montrait saint Jean-Baptiste occupé à verser de l'eau sur la tête de Jésus pour le baptiser. Sur un petit coussin en satin cousu avec du fil doré, il y avait un beau petit mouton blanc qui les regardait. Vite, vite, vite, j'ai prié saint Jean-Baptiste de faire un miracle pour empêcher ce mal élevé de Jean-Pierre Beaudry de le représenter sur le dernier char allégorique. Je ne voulais pas trop le déranger, mais j'espérais qu'il m'ait bien entendue et que ça lui tente de faire un miracle, vu que c'était sa fête!
Là, j'ai couru rejoindre les autres devant l'hôtel Commercial. Les parents étaient debout sur le trottoir, nous les petits, debout sur la galerie de l'hôtel, on voyait par-dessus la tête des grands. C'était plein de monde sur la rue Principale. Ça parlait fort, ça riait dans tout le village. Toutes les _madames_ avaient mis du rouge à lèvres et du fard à joues agencés à leurs belles toilettes. Les grands comme les petits buvaient de la liqueur et mangeaient des bonbons. Même la commère du village ne rouspétait pas ce jour-là.
Tout à coup, à l'autre bout de la rue Principale, en face de l'école, la fanfare a accordé ses instruments, puis a ouvert la marche. Le défilé s'est mis en branle. Les tracteurs ont réchauffé leurs moteurs pour tirer les chars allégoriques, la fanfare jouait, les majorettes marchaient comme des soldats en bottillons blancs en faisant tourner leurs bâtons qui brillaient au soleil. Tout le monde était là. Le curé, le bedeau, le maire, la mairesse, les marguilliers, les chevaliers de Colomb, les dames de Sainte-Anne, le Cercle des fermières, tous maquillés et costumés comme des figurines de gâteau de noces et bien installés sur de splendides chars allégoriques qui se sont mis à défiler devant nous. Cette année, c'était encore plus beau que d'habitude. Tout le monde applaudissait fort en lançant d'éclatants grands bravos aux participants. Ça riait et ça criait si fort qu'on n'a même pas entendu l'angélus et ses douze coups de midi.
Et finalement, au loin, en face du couvent des sœurs, sur le dernier char, j'ai vu briller la laine blanche du petit mouton et les boucles d'or du petit saint Jean-Baptiste. Ils approchaient. Mon cœur battait plus fort que la grosse caisse de la fanfare. Tout à coup, je l'ai reconnu... Non, ce n'est pas vrai... Oui, c'est vrai... C'était lui, pas d'erreur, c'était bien lui, Jean-Pierre Beaudry! C'était Jean-Pierre Beaudry, tout blond, tout bouclé, comme une vraie carte postale. J'étais tellement déçue. Déçue comme quand ma tante Madeleine n'est plus revenue chez nous, comme quand le train des passagers s'est arrêté de passer à Verner, comme quand le maire a démoli la gare. Ma mère s'est retournée vers moi en me disant : « Tu vois, ma petite fille, ce doit être sa mère qui l'a frisé hier soir. » Je pense qu'elle aussi a eu envie de pleurer en essuyant une larme qui avait coulé sur ma joue. Et, comble de malheur, le moteur du tracteur a calé et le convoi de la parade s'est arrêté juste devant nous avec Jean-Pierre Beaudry qui souriait et m'envoyait la main en faisant bondir ses fausses bouclettes blondes. J'ai fermé les yeux plutôt que de lui tirer la langue. J'étais bien élevée, moi.
Mais, comme il y a une justice pour les saints Jean-Baptiste, il est arrivé un beau petit nuage gris foncé, presque noir qui s'est arrêté juste au-dessus du char allégorique et tout à coup _pshhhhh_ , une grosse pluie froide s'est abattue sur l'imposteur. Tant pis pour lui, que j'ai pensé. On ne joue pas avec l'image des saints. Le petit mouton n'a pas défrisé lui, mais Jean-Pierre Beaudry... À ce moment précis, je l'ai vraiment reconnu avec ses cheveux raides qui lui collaient de chaque côté du visage. Ô miracle! j'ai pensé. Saint Jean-Baptiste vient de verser de l'eau sur la tête de Jean-Pierre Beaudry.
Tout de suite, le soleil est revenu et le tracteur a reconduit les deux frisés tout mouillés jusqu'au bout du village. Jean-Pierre braillait, le mouton bêlait, la foule bougonnait. Tout le monde était désappointé. « Même pas un vrai saint Jean-Baptiste, même pas un vrai p'tit blond frisé naturel », que chacun répétait. Pour la première fois dans le village, la parade était un peu gâchée.
On est rentrés chez nous, en silence, en suivant papa et maman qui ne riaient plus. Je pense bien que j'étais la seule à rire aux éclats par en dedans. En passant devant l'église, j'ai monté les marches quatre par quatre, cette fois, pour aller remercier et féliciter saint Jean-Baptiste. « Le plus beau miracle de la Bible, de l'Évangile de l'Ancien et du Nouveau Testament », que je lui ai dit. J'étais tellement fière de notre saint patron. J'étais fière de moi aussi. Je me sentais comme ma chère ma tante Madeleine, parce que comme elle, comme une femme d'affaires influente, je venais de brasser une grosse affaire avec l'homme le plus important de Verner : saint Jean-Baptiste!
_Danièle Vallée_
###### Incitatif pédant: Vers un Verner vert
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau, vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui, malgré la demande générale, Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez notamment écrit _Bing Bang à Pénétang_. Vous venez maintenant nous présenter votre plus récent rapport, savamment intitulé _Vers un Verner vert_.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Vous avez écrit un livre sur votre traversée des Grands Lacs en pédalo, livre intitulé _Donnez-moi un rasoir, S.V.P., ça presse_. Vous revenez tout juste d'un congrès international de médecine à Kingston, où vous êtes allé présenter votre méthode « guérir les tendinites en jouant du violoncelle ». L'accueil qu'on vous a réservé a été plutôt froid et vous avez dû quitter le congrès sous forte escorte policière. Votre couleur préférée est le fuchsia.
_Il va répondre. Elle enchaîne._
Votre chiffre chanceux est 457 982.
_Il va répondre. Elle enchaîne._
Et on pourrait continuer longtemps comme ça, mais ça serait exagéré. Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous avez étudié la communauté de Verner dans la perspective de diversifier son économie et, à la lecture de votre rapport,
_Présentant un rapport gigantesque_
...on se rend bien compte que vous ne manquez pas d'imagination.
###### Gregory Rotheingham
Tout à fait, Annie-France. Comme à notre habitude, nous avons dépensé des milliers de dollars avec les plus grands spécialistes de... de... de... du « dépensage » de milliers de dollars.
###### Annie-France Brazeau
Et si j'en juge par le titre de votre rapport, _Vers un Verner vert_ , on devine que vous suggérez à cette région agricole de prendre le virage environnemental.
###### Gregory Rotheingham
Absolument, Annie-France. Il existe déjà une économie agricole florissante à Verner, et si les résidents suivent les conclusions de l'étude de notre ministère, ils deviendront un exemple mondial d'une économie génétiquement modifiée 100 % biologique.
###### Annie-France Brazeau
Une économie génétiquement modifiée 100 % biologique. Très tendance mais...
_jeu de mains_
le génétiquement modifié et le biologique ne sont-ils pas à l'opposé l'un de l'autre?
###### Gregory Rotheingham
Très tendancieuse, votre remarque, et non ils ne sont pas opposés l'un à l'autre. Verner rassemblera des écologistes et des futuristes parce que la terre, madame Brazeau,
_s'emportant_
la terre, elle est organique et le fruit de cette terre, même s'il est un peu aidé par des éprouvettes, sera toujours organique!
###### Annie-France Brazeau
Raisonnement éprouvant. Mais concrètement, quels sont les projets que vous proposez?
###### Gregory Rotheingham
Eh bien, vous savez que l'Ontario est la province la plus polluée au Canada.
###### Annie-France Brazeau
C'est toujours bon d'être numéro un quelque part. En Amérique du Nord, on est numéro deux derrière le Texas.
###### Annie-France Brazeau
_amèrement envieuse_
Toujours le Texas!
###### Gregory Rotheingham
Eh bien, cette pollution s'arrête à Verner, Annie-France. Verner sera la reine de l'économie verte.
###### Annie-France Brazeau
Vous êtes un monarchiste, un roi de la patate avec ça. Gregory Rotheingham, les habitants de Verner ontils l'expertise nécessaire pour accomplir une pareille révolution?
###### Gregory Rotheingham
Absolument! Dès le XIXe siècle, le père Paradis a breveté une recette appelée le _Maringouinifuge_ , qui éloignait les moustiques. une recette 100 % biologique, à base de vessie de mouffette, de racine d'herbe à poux et de ratte de belette. Nous allons ouvrir une usine de _Maringouinifuge_ et, avec l'appellation biologique, c'est un succès garanti!
###### Annie-France Brazeau
Gregory, le mot biologique — quoique très « tendance » — a ses limites d'attraction. une mouffette même biologique, c'est une mouffette. Mais, dites-moi Gregory, cette recette de _Maringouinifuge_ est-elle efficace?
###### Gregory Rotheingham
Oui, oui oui. J'en ai mis avant l'émission pis j'suis allé dehors : j'ai pas vu un moustique.
###### Annie-France Brazeau
On est en mars, Gregory. Ce n'est pas très concluant pour les maringouins. Par contre, pour les animatrices, l'effet répulsif est sans contredit efficace.
###### Gregory Rotheingham
J'en suis ravi.
###### Annie-France Brazeau
Poursuivez.
###### Gregory Rotheingham
Et des inventeurs, à Verner, on en trouve aussi à notre époque. une industrie de Verner a mis sur le marché un autre gadget écologique.
###### Annie-France Brazeau
Peut-on parler de génie « vénérien »?
###### Gregory Rotheingham
Je dirais plutôt le génie « ver-ne-rois », Annie-France. Le génie « vénérien, » vous ne voudriez pas le côtoyer de trop proche.
###### Annie-France Brazeau
_d'un ton suggestif_
C'est vous qui le dites.
###### Gregory Rotheingham
Le « Block-Elec », Annie-France!
_le montrant_
C'est un bidule qu'on branche à sa rallonge de voiture l'hiver. La voiture est branchée, son moteur réchauffe, mais si la température grimpe au-dessus de moins 12, l'alimentation est coupée! Ça permet de faire des économies d'énergie.
###### Annie-France Brazeau
Oui. L'autre façon de faire des économies, c'est de brancher sa voiture seulement s'il fait moins 15.
_clin d'œil_
Mais, vous savez, avec le réchauffement de la planète, cette invention risque de devenir inutile bientôt.
###### Gregory Rotheingham
Certainement, Annie-France. C'est pourquoi les inventeurs locaux sont déjà à l'œuvre pour modifier le « Bloc-Elec ». Quand la température atteindra les 30 degrés Celsius, c'est l'air climatisé qui va démarrer. Et en option, il y aura un distributeur de _popsicle_.
###### Annie-France Brazeau
Bon, mais ce sont des inventions déjà existantes dont vous me parlez. Avez-vous de nouveaux projets?
###### Gregory Rotheingham
Certainement. Il suffit de poursuivre dans la voie tracée par nos pionniers. Dans la région, les champs produisent des céréales, dont le soya. Il faut simplement pousser cette industrie-là plus loin.
###### Annie-France Brazeau
Je meurs d'envie de savoir comment vous allez faire... Achevez-moi!
###### Gregory Rotheingham
J'ai un plan. Vous savez que le soya est employé dans la fabrication du tofu. Mais, actuellement, le tofu est seulement utilisé en alimentation.
###### Annie-France Brazeau
C'est vrai qu'il y a pas beaucoup de tofu dans les balles de ping-pong.
###### Gregory Rotheingham
Le secret, c'est d'augmenter les besoins de l'industrie en tofu! Comme ça, le prix des récoltes va augmenter.
###### Annie-France Brazeau
Vous êtes « la Main invisible du marché » à vous seul. Comment allez-vous faire?
###### Gregory Rotheingham
En multipliant les utilisations du tofu.
_Il montre le plan._
###### Annie-France Brazeau
Vous vous êtes trompé de page, Gregory Rotheingham, vous nous montrez actuellement un futon.
###### Gregory Rotheingham
_triomphant_
C'est ça! Le futon en tofu!
###### Annie-France Brazeau
_abasourdie_
Le futon en tofu, ben oui! Pis, tant qu'à y être, moi, je pourrais me mettre une jupe hawaïenne et m'inventer un vocabulaire à la base des lettres K, W, u et G.
###### Gregory Rotheingham
Attendez, imaginez les applications d'un tel matériau.
###### Annie-France Brazeau
Kwug. Guwk. Kuwg.
###### Gregory Rotheingham
Non, écoutez, on créerait des meubles mous et moelleux. Imaginez simplement l'avantage des meubles en tofu dans les garderies ou dans les institutions psychiatriques.
###### Annie-France Brazeau
Ah, c'est ça, vous préparez votre retraite! Mais vous savez, le tofu, ça goûte pas grand-chose, alors j'ai un slogan pour vous : « Les meubles en tofu, pour tous ceux qui n'ont pas de goût! »
###### Gregory Rotheingham
_Il grogne._
###### Annie-France Brazeau
Mon sofa au tofu est passé date! Bon, maintenant Gregory, vous allez me faire plaisir, nous allons revenir sur la planète Terre, n'est-ce pas?
###### Gregory Rotheingham
Le retour à la terre. C'est une bonne idée ça!
###### Annie-France Brazeau
Avez-vous d'autres projets pour Verner?
###### Gregory Rotheingham
_de nouveau enthousiaste_
Imaginez, Annie-France, si on...
###### Annie-France Brazeau
_l'interrompant_
J'vous rappelle qu'on est sur la planète Terre.
###### Gregory Rotheingham
Bien, j'avais pensé installer un laboratoire d'ingénierie génétique.
###### Annie-France Brazeau
Génial, enfin une bonne idée!
###### Gregory Rotheingham
Oui, et dans le laboratoire, les scientifiques croiseraient certaines espèces animales pour en créer de nouvelles.
###### Annie-France Brazeau
Intéressant. À quel genre de croisement pourrait-on s'attendre?
###### Gregory Rotheingham
Ben, je me disais que si on croisait une vache avec une poule, on pourrait obtenir un nouvel animal.
###### Annie-France Brazeau
Une « voule »?
###### Gregory Rotheingham
Oui, ou une « pache » ou une « pouche ». Mais le plus important, c'est que quand on trairait la « voule », on obtiendrait un mélange à omelette. Ou pour le temps des Fêtes, du lait de poule.
###### Annie-France Brazeau
Le lait de poule, c'est effectivement très tendance. On pourrait aussi mélanger une mouffette et une morue pour en faire une bête particulièrement puante, voire une arme de destruction massive.
###### Gregory Rotheingham
On pense aussi croiser des porcs-épics avec du bœuf, des piments et des oignons.
###### Annie-France Brazeau
Qu'est-ce que ça donnerait?
###### Gregory Rotheingham
Des brochettes!
###### Annie-France Brazeau
Audacieux!
###### Gregory Rotheingham
On pense aussi utiliser des sangsues et des squelettes pour faire des tatouages écologiques.
###### Annie-France Brazeau
J'ai presque envie de croire à vos élucubrations. Gregory Rotheingham, avec vous le développement économique n'est pas une science : c'est une science-fiction, quoique vos effets spéciaux soient plus près de _King Kong_ que du _Seigneur des Anneaux_. Savez-vous, plutôt que d'intituler votre rapport _Vers un Verner vert,_ vous auriez dû l'appeler _2004, l'Odyssée de Verner_.
###### Gregory Rotheingham
_vexé_
J'aurais aussi pu l'appeler _Invasion de vaches folles montréalaises à Verner_.
###### Annie-France Brazeau
Bon, eh bien, Gregory Rotheingham, c'est tout le temps que nous avions à vous accorder. Ah...
_Elle éternue._
De toute façon, votre présence réveille en moi des symptômes de grippe aviaire.
###### Gregory Rotheingham
Oui, mais n'oubliez pas qu'à Verner, c'est une grippe aviaire biologique.
###### Annie-France Brazeau
Je vais donc pouvoir mourir en santé. Gregory Rotheingham, merci.
###### Ta silhouette qui s'éloignait
_J'ai vu ta silhouette qui s'éloignait_
_Ton dernier salut de la main_
_Maintenant, j'avoue que j'espérais_
_Que tu allais rebrousser chemin_
_Je sais, je ne te reverrai plus ici_
_Je sais, tu as le droit de faire ta vie_
_Je sais, tu crois au monde qui t'entoure_
_Je connais aussi les risques que tu cours_
Il te faut tenter ta chance
Il te faut croire en l'avenir
Tu as droit à l'espérance
Je ne veux pas te retenir
Tu as le droit à tes erreurs
Il te faut ta liberté
Je te souhaite de tout cœur
Du succès bien mérité
_Je crois voir ton sourire dans une publicité_
_Ce n'est pas toi, je sais, mais j'aime me tromper_
_Je pense t'apercevoir à la télévision_
_Il vaut mieux que ça soit moi, pas toi, qui aie des illusions_
_Quand tu as dû partir, je t'ai fait mes bons vœux_
_C'est pas moi qui t'empêcherais de faire ce que tu veux_
_Je garde nos souvenirs, je vais toujours te dire_
_Que tu as tout ce qu'il faut, que tu vas réussir_
Il te faut suivre ta route
Et y perdre ton innocence
Te mesurer à tes doutes
Et voir grandir ta confiance
Tu es libre, c'est ta vie
Ton étoile t'a réclamée
Sans demander d'où tu es partie
Ni qui t'y a bien aimée
Ni qui t'y a bien aimée
_J'ai vu ta silhouette qui s'éloignait_
_Ton dernier salut de la main_
###
### Salut Chapleau!
Pour entrer à Chapleau, il faut passer sur un haut viaduc qui enjambe les rails en faisant d'abord une boucle complète à gauche, suivie d'un virage en épingle à droite, suivi d'un carrefour en Y. Pour en sortir, évidemment, c'est le contraire. Autrement dit, les touristes sont étourdis! Par contre, à Chapleau, la toponymie des rues est simplifiée : les rues nord-sud ont des noms de gouverneurs généraux et les rues est-ouest ont des noms d'arbres. Ce qui fait qu'à Chapleau, comme à Sudbury, on peut se retrouver à l'angle des rues Elm et Elgin.
Dans le Nord de l'Ontario, il y a peu de communautés qui ont un nom bien français comme Chapleau. C'est un peu plus étonnant quand on apprend que cette petite ville de l'Ontario porte le nom d'un premier ministre du Québec, tout comme Cartier, à 130 milles d'ici sur la voie ferrée. Des voisins, quoi! Joseph-Adolphe Chapleau et Georges-Étienne Cartier, deux Canadiens français du Québec, honorés sur la voie ferrée en Ontario. On y croyait parfois, au principe des deux peuples fondateurs, à l'époque où le chemin de fer concrétisait le Canada.
Comme tant d'autres villages, Chapleau est né du chemin de fer. Mais comme peu d'autres, Chapleau s'est développé en centre ferroviaire grâce aux ateliers du Canadien Pacifique. Avec le chemin de fer, les chemins de bois ont fait de Chapleau une ville forestière, ce qui fait qu'à Chapleau on peut se retrouver dans un restaurant qui s'appelle « Aux trois moulins ». Comme ailleurs, Chapleau a vu ralentir ses deux moteurs économiques. La ville a moins de trois mille habitants, alors qu'elle en a déjà eu plus de quatre mille.
Chapleau a des titres de renommée internationale. Dans les environs, il y a la réserve faunique la plus étendue au monde. Il y a aussi, depuis deux ans, un festival de la nature qui fait si bien les choses en grand qu'il gagnera sa renommée mondiale, donnez-lui le temps. Et dans son cimetière, se dresse la pierre tombale d'un Français qui a signé un chef-d'œuvre littéraire canadien, vendu à des millions d'exemplaires partout au monde en des douzaines de traductions. L'auteur de _Maria Chapdelaine_ , Louis Hémon, repose à Chapleau. Il est mort près d'ici, en 1913, frappé par un train, ses derniers huit dollars cachés dans sa botte, son petit poêle dans son _pack-sack_ , alors qu'il partait pour l'Ouest. Il n'était pas encore le romancier mondialement célèbre qu'il deviendrait dix ans plus tard. Et le pays de Maria Chapdelaine, ce n'est pas Chapleau; c'est Péribonka, au lac Saint-Jean.
Mais ce que moins de gens savent — et c'est dommage — c'est que Chapleau est aussi le pays du héros d'un autre roman trop bien écrit pour être resté si peu connu. Le roman _François Duvalet_ a été publié à Québec en 1954, et réédité en Ontario par les Éditions le Nordir en 1989. Maurice de Goumois, son auteur, est un Canadien né en France, dans la région du Valais, justement. L'auteur est passé par Chapleau dans les années vingt. Ça, on le sait. À quel point son roman reflète son expérience, ça, on ne le sait pas.
Le roman brosse le portrait des années vingt, une époque où le Nouvel-Ontario avait l'air d'une terre d'espoir pour un jeune Parisien. Il raconte la vie de François Duvalet, un gars vaillant qui a trimé dur dans un pays dur, qu'il a aimé plus que celui-ci ne l'a aimé. C'est l'histoire d'un homme en quête de succès et d'un pays qui le lui refuse. Ses paysages et ses personnages, ses ambiances et ses espérances sont toujours celles du Nord ontarien d'aujourd'hui. À mes yeux, ce roman est une allégorie de notre destinée collective. Malgré notre vaillance, notre coin de pays se dépeuple. Chaque nouveau recensement le dit. Est-ce donc vrai que c'est juste ailleurs qu'il y a de l'espoir pour les gens de valeur?
###### _FrançoIs Duvalet_
À Paris, dans les années trente, François Duvalet est petit salarié dans une banque. Le Canada, c'est sa façon de fuir la médiocrité, mais aussi sa liaison avec une femme. Car malgré lui, il plaît aux femmes, le beau François. Il aura à le regretter plus d'une fois.
À Montréal, on lui offre du travail dans les chantiers du Nord de l'Ontario. Le voici donc au bout du monde, au bout d'un lac, au fond d'une forêt dense aux odeurs d'humus et de résine. Pour rompre avec son passé, il fait un geste théâtral : il lance au loin sur le lac son ridicule chapeau melon. Ce sera son dernier geste habile avant longtemps. François Duvalet est si mauvais bûcheron qu'il faillit s'y tuer. Il veut s'en aller, mais il n'a pas gagné le coût de son transport, de son linge et de ses bottes. Pour ça, la compagnie exige les gages d'un mois d'ouvrage. Donc il devient — quel déshonneur! — aide-cuisinier dans la _cookerie_. un soir, le chef Jo-Jo se permet de caresser le dos de François qui se lavait en lui disant : « T'as une vraie peau de fille, Frenchy ». Ça lui vaut, à Jo-Jo, un coup de poêle à frire sur la tête. François ne perd pourtant pas son emploi ce jour-là. Il le perd le surlendemain, quand dans la noirceur et la fatigue, il se trompe en versant le vinaigre des betteraves dans le baril des pruneaux. On le congédie, mais non sans lui exiger de remettre son dû à la North Star Lumber Company, si jamais il gagne un jour de quoi le payer.
Après trente milles de marche solitaire dans un sentier forestier, François Duvalet atteint Chapleau la nuit tombée. Ici, gens de Chapleau, je cite la description de votre ville des années trente que ce roman conserve.
Chapleau [...] n'est donc [...] ni un bourg ni un village, mais tout simplement un préau ferroviaire, où tous les hommes bien nés sont au service du chemin de fer. [...] Là tous les trains s'arrêtent pour changer de locomotive et de personnel roulant. Et comme ces relais se font à n'importe quelle heure du jour ou de la nuit, Chapleau arbore un faux air de métropole qui ne dort jamais. [...] Cette sorte de cour en plein bois, oasis imprévue, est semée d'ateliers et de remises, de montagnes de charbon et de tours d'eau. Obscurcie par les vomissements de fumées et de vapeurs, constamment animée par le grincement des roues sur les rails, par les rames de wagons qui s'entrechoquent et par des locomotives qui s'époumonent au son de leur cloche de bronze, elle ressemble à la forge de Vulcain.
Voilà à quoi Chapleau ressemblait, il n'y a pas si longtemps.
À ses débuts à Chapleau, la chance sourit à François Duvalet. un commerçant aisé et rusé originaire du Luxembourg prend ce jeune Français comme protégé. Fin calculateur, Jacob Druten voit même en ce garçon vaillant son futur gendre et gérant. Car en pleine crise des années trente, jamais il ne pourrait vendre son magasin. En jouant de ruse et de billard, Druten gagne un emploi pour François. Il ira décrasser le mâchefer, au marteau et au ciseau, au fond des fourneaux des locomotives, sous les ordres de Boucher, qui insiste pour qu'on prononce Butcher. Sale et suant, François travaille vaillamment. C'est ici, mes amis, que son mauvais sort de beau garçon le rejoint, sous la forme d'une Anglaise osseuse et froide. Ici je cite le roman à nouveau.
Français, il ignorait sans doute qu'il appartenait à la race de ceux qui passent pour des virtuoses de l'amour parmi les femmes d'un climat plus rude. Et ce n'était pas de sa faute si son physique et une certaine grâce de manières correspondaient à l'image que l'on se fait d'un grand amoureux. [...] Ce fut ainsi que l'épouse de Jacob Druten se transforma subitement en femme de Putifar. [...] Dorothy Druten fit irruption dans la chambre du Français. [...] C'est alors qu'il vit qu'elle n'avait presque rien sur elle et comprit qu'elle s'offrait sans vergogne. La situation était impossible; il recula d'horreur. La femme de son bienfaiteur, et sous son propre toit par-dessus le marché! Mais comment dire tout cela en anglais?
En tout cas, notre pauvre François repousse ses avances et, ce faisant, se fait la pire des ennemies : une femme dédaignée et compromise. Elle ne pourra plus endurer la présence de ce Français chez elle, ni le projet de mariage de son mari pour sa fille. Le vieux Druten, ne se doutant de rien, comprend seulement qu'il faudra du temps. Donc, pour écarter François temporairement, il l'équipe pour la trappe. Au bon endroit, ça peut être très payant.
François est ainsi plongé, inexpérimenté, en une épreuve d'endurance hivernale au fond d'une cabane qui a l'allure d'une sépulture. La fatigue et le froid le font halluciner. En trois semaines, il a attrapé un lièvre. François ravale son orgueil et va demander la charité à une famille de cantonniers isolés, au Mille-Onze, dans une belle forêt de bouleaux. C'est Courtemanche qui lui donne l'idée de s'essayer à vendre du bois de chauffage. Sa fille Isoline, elle, a pour idée que François est bien de son goût. Vous voyez venir un problème, non? Pourtant François, lui, ne le voit pas.
En tout cas, il est meilleur _jobeur_ que trappeur. À Chapleau, il réussit à se faire prêter un cheval contre son engagement à livrer son bois au revendeur Buxton. Mais le pauvre George, on n'aurait jamais cru ça d'un cheval de Chapleau : il a peur des trains! Pourtant il faut charroyer le long de la voie. Pour le faire avancer, il faut lui chauffer l'arrière-train avec un bout d'écorce en flammes. Pour le faire passer sur le viaduc à Chapleau, il faut lui couvrir la tête d'un sac et le faire tourner sur lui-même plusieurs fois pour le désorienter. C'était de toute beauté de le voir faire ce petit ballet chevalin. Décidément, François Duvalet fait preuve de débrouillardise. Notre Français se canadianise. Mais les hommes sont plus malcommodes que son cheval. Il s'en passe des choses, que je dois vous compter trop vite.
Deux engagés de François amènent des femmes de mœurs légères au Mille-Onze et ne veulent même plus sortir de leur cabane pour travailler. François doit amener la police pour les en expulser. Buxton décide de ne pas prendre le bois de François, la crise économique lui semblant motif valable pour ne pas honorer sa parole donnée. François se décide de vendre lui-même son bois aux habitants de Chapleau. Il fait des livraisons partout. Mais personne ne finit par payer. Puis l'engagé des Courtemanche devient jaloux fou de François parce qu'Isoline lui fait de beaux yeux. François vient à un cheveu de commettre un meurtre par autodéfense. Son réconfort doux-amer dans tout ce malheur, c'est la rencontre d'une Française comme lui, belle et cultivée, souffrant de la médiocrité de son existence. Mais Clarisse est l'épouse malheureuse d'un mari anglais.
On comprend comment, dans ce tourment, Chapleau semble pour François un endroit misérable. Et pourtant, il sent qu'il y appartient. Voici comment il se l'explique :
Parfois le pays lui apparaissait comme une fondrière. une fois engagé, comment en sortir? Cette immensité perdue, cette nature à peine touchée, la vie même qui en prenait le langage, avaient quelque chose qui pénétrait tout l'être. On s'y attachait d'autant plus solidement qu'elle ne cessait de promettre un monde régénéré, débarrassé de toute contrainte, où un homme n'est que ce qu'il est, et non pas QUI il est.
C'est un portrait flatteur de notre coin de pays. Mais voyons si ça sera vrai pour François Duvalet.
Toujours sans-le-sou après tant d'efforts, il est engagé pour une livraison par canot à un poste de la Compagnie de la Baie d'Hudson. Il y rencontre un commis qui, sans le moindre humour, lui dit ne pas vivre isolé : son plus proche voisin est à juste cinquante milles. Le Canada façonne des hommes immenses. François veut leur ressembler. Puis il s'aventure jusqu'à Fort William pour l'entretien des voies ferrées. Cette fois-ci, c'est la finlandaise Katrinka qui lui fait de l'œil. Et c'est le communiste Igor qui en devient fou de jalousie, au point de s'en tirer une balle dans la tête. François se questionne sur son talent pour inspirer le crime passionnel, d'autant plus que ce pauvre Igor, il a fallu l'enterrer trois fois. Je n'ai pas le temps de vous conter pourquoi.
François retourne à Chapleau, où il lance une petite entreprise de livraison depuis la gare. Enfin un travail un peu payant. Il se fait des économies. Mais c'est le moment choisi pour que Ned Druten entre dans le portrait. Lui, c'est le frère fou de Dorothy Druten, un détraqué messianique revenu de New Liskeard. Il est fou, mais rusé. Il s'arrange pour voler le traîneau de François tout en faisant arrêter François pour ce même vol. Puis il en vient aux agressions en l'accusant d'avoir essayé de déshonorer sa sœur. Ned a beau être fou, cette accusation trouble Jacob Druten. François comprend que le temps est venu de quitter Chapleau. Par ses manigances, la femme qu'il a dédaignée aura eu sa vengeance.
Le voici donc à Toronto, d'où il cherche à rentrer en France. Mais il lui reste une chose à faire. Vous vous souvenez de sa dette envers la North Star Lumber Company? Eh bien, il tient à l'acquitter. C'est par ce geste-là que sa vie changera. Dans le bureau de la North Star Lumber à Toronto, la secrétaire ne comprend pas ce qu'il fait là. L'affaire simple se complique. Le grand patron s'en mêle. Celui-ci est impressionné par l'honnêteté de son ancien bûcheron, mais plus encore par le récit de sa vie. Après tout, racontée d'un souffle comme je viens de le faire, l'aventure devient risible et néanmoins admirable. C'est ainsi que François Duvalet se voit offrir un voyage en Angleterre pour surveiller les cargaisons de bois. De là, il pourra rentrer en France s'il le désire. Mais s'il le veut, il aura à son retour au Canada le poste d'inspecteur des comptes de tous les chantiers de la North Star Lumber. Question de s'assurer, dit le patron en riant, qu'aucun bûcheron ne parte sans payer ses dettes.
Et voilà où s'achève l'histoire de François Duvalet. Elle a deux morales, il me semble. L'une est classique. C'est celle de l'honnêteté récompensée et de la valeur personnelle reconnue. Mais l'autre, mine de rien, donne un tour neuf à un vieux cliché. Depuis Balzac et Stendhal au moins, les romans disent que c'est à Paris, ou dans les métropoles, que le jeune provincial de valeur doit faire ses preuves. Mais voici un roman où un Parisien a tout fait pour prouver sa valeur dans un lointain arrière-pays où l'humanité qui l'inspire est ce à quoi il aspire, et qui se voit refuser le succès tant qu'il n'a pas pris, malgré lui, le chemin de la métropole. C'est tellement ironique. C'est pour ça que c'est la plus vraie des deux morales. En écrivant dans les années cinquante son roman des années trente, Maurice de Goumois parlait déjà de notre défi des années 2000. Nous voyons tant de nos jeunes vaillants partir à la quête du succès dans les grandes villes du Sud. François Duvalet, lui, a tant voulu vivre ici, dans ce pays. Pourquoi n'a-t-il pas réussi?
Gens de Chapleau, sachez que votre ville a inspiré un roman authentique et sympathique. _Maria Chapdelaine_ vous salue de loin, mais _François Duvalet_ vous salue de près.
###### À la gare de Chapleau
_C'était un pays de pins_
_C'est un pays de trains_
_Un pays de canots de toile et de castors_
_De rivières aux noms indiens_
_De familles venues de loin_
_De chercheurs de job et de chercheurs d'or_
_Et puis un soir, une fille du lac St-Jean_
_Avec son jeune frère, puis ses deux parents_
_Était descendue, nouveau début_
_À la gare de Chapleau_
_En ne sachant pas sa place dans cette vie-là_
_Ça se passait souvent comme ça, dans ce pays-là_
_C'était un pays où un vagabond_
_Pouvait mourir gelé, pris dans un wagon_
_Sur le sentier clandestin de son destin_
_C'était un pays où un feu de forêt_
_Pouvait embraser, quand il s'enrageait_
_L'horizon de bord en bord, et encore plus loin_
_Et puis un soir, un gars de l'Abitibi_
_Dans un train rempli de plein d'autres gars comme lui_
_Était descendu, le corps fourbu_
_À la gare de Chapleau_
_Rentrant de son combat contre ce feu-là_
_Il s'en allait chez lui, passant par ce lieu-là_
_Et bien sûr, la fille, le gars_
_Se sont croisés ce soir-là_
_Elle attendait sa malle, et lui, son train_
_Et au cœur d'un pays rude, fulgurante certitude_
_Ils se découvraient un amour canadien_
_Quand est venue l'annonce « tout le monde à bord »_
_Quand ils allaient se quitter, chacun de son bord_
_Il lui a dévoilé une intention osée_
_À la gare de Chapleau_
_J'aimerais bien te revoir, si tu veux, c'est promis_
_Dans un an, jour pour jour, je reviendrai ici_
_Ils se sont écrit des douzaines de lettres_
_Ils se sont promis tout ce qu'il fallait se promettre_
_Et la fille y était, comme il l'espérait_
_À la gare de Chapleau_
_Voilà une belle histoire qui finit bien_
_Dommage qu'on ne voyage plus tellement en train_
###### Madame Fillion
L'an: 1886. Jour précis: oublié. Le Canadian Pacific Railway, le CPR, déménage. Sans cérémonie, la compagnie de chemin de fer laisse sombrer dans l'oubli Biscotasing, village situé à une centaine de kilomètres de Chapleau, qui jouait depuis environ six ans le rôle qu'on conférait désormais à son voisin. Il faut dire que pour les grands bâtisseurs du chemin de fer — « D'un bout à l'autre du pays! D'est en ouest! Peuplons nos régions! Colonisons le pays! unissons le Canada! » — , il faut dire que Bisco, avec ses bordels et _saloons_ qu'on disait assez nombreux pour assouvir la soif et les désirs charnels de... « ah! au moins 100 000 hommes par jour! », n'était plus de grand intérêt pour la compagnie, ses gants blancs et sa royauté bleue. Salut Bisco la terrible! Bonjour Chapleau la paisible.
J'aurais aimé être de ce temps, dans le Nord de l'Ontario. Je me serais établie à Bisco. Tout naissait à cette époque. Tout était possible. Les limites, on les oublie. On se permet de faire des folies. Je me serais nommée _Madam_ du _Finest Biscotasing Brothel, pleasing Northern Ontario men since 1880_. Mais, je suis pas née à la fin du 19esiècle. Je suis née le 25 février 1972. Baptisée le 1er mars de la même année : Bernadette Éléonore Émilie Marie Tremblay. J'ai reçu ma première communion le 14 avril 1979, ma première confession le 3 juin 1980, ma confirmation le 8 octobre 1985. Le reste, je l'ai oublié.
Il m'a fallu 11 ans pour me donner le goût de rentrer chez moi, à Chapleau. Le hasard a voulu que je croise Louis Hémon dans un mots croisés. Depuis que je suis toute petite, je m'amuse à faire des mots croisés. Des mots croisés que je termine jamais. Mon séjour à Toronto a été une période propice à l'accumulation de cahiers de mots croisés pas terminés. J'en traînais toujours un dans mon sac pour occuper mes heures passées sous terre, à transiter dans le métro de Toronto : 11 ans de transit, à 6 jours par semaine, ça en fait des mots croisés. Surtout quand on les termine pas. À y repenser, ça me donne la nausée. J'peux même plus dire « mots croisés » sans que ça me lève le cœur. N'empêche que c'est à cause d'un mots croisés que je me souviendrai toujours du jour où je me suis enfin décidée.
C'était en l'an 2002. Jour précis: le 16 février. Je sortais de la clinique de soins psychologiques. Je me dirigeais sous terre pour prendre le métro, encore une fois et toujours vers l'ouest. J'étais claquée. Je ne souhaitais que de revoir mon condo et profiter du confort de mon lit pour dormir, dormir, dormir, ah oui! dormir.
Vertical, numéro 17, dix lettres. Indice : grand écrivain d'origine française, mort à Chapleau en 1913? À Chapleau? Tiens, tiens... Et c'était un grand écrivain? En êtes-vous certain? Grand comparé aux plus grands? Ou grand comparé aux plus petits? S'il était si grand que ça, je vois vraiment pas ce qu'il faisait là-bas! À ma connaissance, c'était la première fois qu'on parlait de Chapleau dans un mots croisés. Mon village natal. Ce jour-là, et pour la première fois de ma vie, je me suis donnée comme objectif d'avoir de l'ambition. Je la sentais monter en moi. Je comprenais pas ce qui se passait. Je n'ai jamais eu d'ambition.
Arrivée au condo, au lieu d'aller me coucher, j'ai fait une recherche dans Internet. Pourquoi? Je l'sais pas. Mais si je terminais pas ce mots croisés, je serais obligée de constater que ma vie était un échec total. C'est le constat habituel de tout psychologue sur sa propre vie. Et je suis psychologue. Ma profession me faisait terriblement peur cette journée-là.
Google. Chapleau. _Search_. Résultats de la recherche : 61 400 liens trouvés. 61 400 liens sur Chapleau? Louis Hémon, auteur de _Maria Chapdelaine_ , mort à Chapleau le 8 juillet 1913, heurté par une locomotive du Canadien National. Louis Hémon... L-O-U-I-SH-E-M-O-N. Louis Hémon est mort à Chapleau? C'est vrai. Comment ai-je pu l'oublier?
Ce jour précis, jour du premier mots croisés terminé, j'ai décidé que j'en avais assez de Toronto. Je bâtirais ma nouvelle vie à Chapleau, dans le Nord de l'Ontario. J'effectuerais un retour aux sources! J'ai vendu mon condo au centre-ville pour une somme faramineuse. Je me suis acheté une maison, sans même aller la voir sur place, une maison sur la rue Dufferin à Chapleau, pour une fraction du prix que j'avais obtenu pour mon condo à Toronto. Inutile de vous dire que mes poches étaient remplies de dollars canadiens. « On peut faire le troc à Chapleau avec des dollars canadiens? » Si ma mémoire est bonne, oui. Peu importe...
Ce qui comptait pour moi ce 16e jour du mois de février de l'an 2002, c'est que j'allais enfin vivre une vie paisible, dans le silence, près des forêts, des lacs et des rivières, une vie au cœur de la vie sauvage. Ouain...
Je suis à Chapleau depuis deux ans maintenant. Aujourd'hui, j'ai l'impression que le temps n'existe plus. Les échecs non plus. Je dis ça comme ça, mais j'y ai bien réfléchi. Si je n'ai retenu qu'une seule leçon depuis mon arrivée à Chapleau, c'est que les hommes ne naissent jamais trop tard. Ils naissent toujours trop tôt. Je suis née trop tôt, à Chapleau.
Un jour, un homme de train m'a dit que si le chemin de fer avait été un échec au Canada c'est parce que : « Y voulaient pas que ça fonctionne. _Goddamn right_ qu'y voulaient pas! » Dans le fond — et ils s'en sont rendu compte trop tard —, on avait choisi le mauvais axe. Il aurait fallu se concentrer davantage sur la construction d'un réseau nord-sud et « non, non, non, non, surtout pas sur le peuplement des régions "reculées" du Canada. » un projet voué à l'échec, même avant son commencement.
Quand je suis partie de Toronto pour venir m'installer à Chapleau, c'était avec la ferme intention d'y établir ma pratique de soins psychologiques et de bien-être mental. En préparant mes boîtes de déménagement, je suis tombée sur la lettre d'une adolescente de Chapleau. C'est ainsi qu'elle se nommait : adolescente de Chapleau. Sa lettre, datée de 1987, était adressée au Groupe d'études sur le développement des régions rurales et reculées, une initiative du gouvernement canadien :
Comme jeune citoyenne de Chapleau, je réclame le droit à l'activité, ayant déjà vécu une bonne partie de mon adolescence dans un état profond d'inactivité. C'est, comment pourrais-je dire, ennuyant. Je voudrais bien occuper mon temps à faire autre chose que des graffitis sur les maisons de la rue Dufferin, en face de la voie ferrée, mais il n'y a rien d'autre à faire à Chapleau. J'ai l'impression de sombrer petit à petit dans un état de grave dépression. Lorsque cela m'arrive, j'écris et je dors. Pourriez-vous m'aider? Précision : je n'aime ni les _skidoos_ , ni les 4 par 4, ni la pêche, ni la chasse, ni les randonnées en nature, ni les parties de _shuffleboard_ au sous-sol de la Légion, ni l'école, ni mes parents, ni mes amis.
Faites ce que vous pouvez. Je retourne me coucher. J'attendrai votre réponse.
Signé,
Adolescente de Chapleau
Bernadette Éléonore Émilie Marie Tremblay, t'étais pas du monde à 15 ans! Puis, à vrai dire, t'as pas changé tant que ça, même avec un doctorat en psychologie de la Queen's university.
L'adolescente de Chapleau, je l'ai retrouvée assez rapidement merci! J'arrivais au village et je me dirigeais tranquillement en voiture vers la voie ferrée. Soudain, du coin de l'œil, j'ai aperçu la petite maison rose, au bout de la rue. Je me suis approchée... Non! C'était la petite maison rose de madame Fillion. Cette femme que je détestais tellement quand j'étais jeune : « Pilez pas sur mon parterre, mes mal élevés! Voulez-vous que je sorte mon balai? un p'tit coup s'a tête, ça vous dompterait peut-être! » Madame Fillion, propriétaire du 17, rue Dufferin, la maison que j'ai achetée. Je me suis rapprochée... non, non, non! Madame Fillion avait retouché ici et là les murs avec un rose un peu plus foncé que l'original. En m'approchant encore un peu, question de voir le laid de plus près, j'ai entrevu des traces de graffitis à travers la peinture. À 15 ans, je trouvais ça quétaine le rose, puis j'avais rien d'autre à faire à Chapleau. Sa petite maison rose, je l'avais couverte de graffitis pour m'occuper et pour l'embellir. J'avais donc acheté pour presque rien la maison de madame Fillion! Elle avait décidé de déménager à Toronto pour être plus près de ses enfants. Ils ne lui avaient pas rendu visite depuis au moins huit ans. Elle a fini par comprendre que c'était à elle d'aller les rejoindre.
Mon adolescente de Chapleau, je l'ai guérie avec le temps et... mes dollars canadiens. une semaine après mon arrivée, j'ai fait parvenir un chèque à madame Fillion, d'un montant raisonnable. un chèque signé Bernadette Éléonore Émilie Marie Tremblay, née le 13 février 1972, à Chapleau. un chèque, avec une note sur un _post-it_ rose : « Les graffitis sur votre maison, Madame Fillion, c'est moi. Je viens de m'en souvenir, et je suis contente de les revoir. »
Je me suis guérie et j'ai ouvert ma pratique de soins psychologiques et de bien-être mental. Les lundi, mercredi et vendredi matins, on la trouve au _Stedman's_. Les mardi, jeudi et samedi matins, _Aux trois moulins_. En semaine, je fais la compilation de tous les potins du village, et les dimanches, à l'heure du brunch, je laisse traîner la liste à potins sur toutes les tables des restaurants. Quand quelqu'un tombe sur un potin qui le concerne, ce qui arrive tout le temps, au lieu de se mettre à sacrer, il se met à rire comme un fou et les autres suivent.
Malgré ses ambitions, l'homme ne réussira jamais à être autre chose que de son temps. C'était vrai en 1880. C'est toujours vrai en 2004. Est-ce qu'on pourrait faire autrement avec une espérance de vie de 75 ans? J'ai rapidement compris que de regretter, ça ne sert à rien. Que d'oublier, ce n'est pas si grave que ça. Il y a toujours des p'tites maisons roses, des madame Fillion ou des mots croisés pour nous le rappeler, quand on décide de s'en souvenir.
Des gens qui regrettent Chapleau, j'en ai pas rencontré beaucoup. En fait, je me demande comment j'ai fait pour oublier pendant aussi longtemps ma ville natale. Aujourd'hui, je peux vous dire que la seule chose qui me dérange à Chapleau, c'est les têtes d'orignal accrochées partout. Dans les restaurants, les magasins, les entrées de maison, même sur les murs des patios de roulottes au lac Racine. Puis même à ça, quand je vois les enfants sortir du lac, accrocher leurs costumes de bain trempes aux panaches pour qu'ils sèchent au soleil, ça me fait sourire et je me dis que toute chose dans la vie trouve un jour son utilité propre.
Cet été, je quitte Chapleau. Je déménage à Bisco. Mes parents, à la retraite, ont quitté Chapleau pour le soleil de la Floride. Les deux ont la bouche grande ouverte depuis que je leur ai annoncé ma migration de l'arrière-pays vers l'arrière-arrière-pays. Ils comprennent pas que les presque riens m'attirent. Que, pour moi, le presque rien qu'est Bisco, c'est le grand espace dont j'avais besoin pour faire des folies. Non, je ne rêve pas de faire revivre les bordels de Bisco (ben, peut-être...), ni d'être pop-psychologue du seul resto en ville. Puis dans le fond, qui sait... Je pourrais me remettre à faire des mots croisés.
_Guylaine Tousignant_
###### _Incitatif pédant_ : Tour du Chapleau
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau, vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez notamment écrit _Vers un Verner vert_. Vous avez également publié un rapport sur la communauté de Blind River dans lequel vous préconisez l'ouverture d'un terrain de _paintball_ pour non-voyants, rapport judicieusement intitulé _Bang Bang Braille Braille à Blind River_.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Vous avez prononcé une conférence sur la rénovation immobilière à la prestigieuse Université Queen's, conférence intitulée _Wet Paint_. Malheureusement personne n'a osé ouvrir la porte pour entrer dans la salle de peur de se beurrer les mains. Vous collectionnez les disques de danse exercice de Jane Fonda — pour les articles.
_Il va répondre. Elle enchaîne._
Vous avez, dans votre jeunesse, enregistré un 45-tours country intitulé « Chéri, je veux t'embrasser, mais ma main droite est à Cincinnati ».
_Il va répondre. Elle enchaîne._
Et on pourrait continuer longtemps comme ça, mais je ne suis pas sûre qu'on a envie d'en apprendre plus. Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous êtes prolifique. Vous venez nous présenter une autre de vos études, sur Chapleau cette fois-ci. Et avec votre grande perspicacité, vous nommez votre nouveau rapport, _Chapleau, une nature naturelle à exploiter naturellement_. J'imagine qu'on a dû vous payer assez cher pour trouver ça.
###### Gregory Rotheingham
On m'a payé en nature. Vous savez, Annie-France, pour réaliser cette étude, j'ai parcouru le territoire de la municipalité en long et en large.
###### Annie-France Brazeau
Vous avez fait le « tour du Chapleau ».
###### Gregory Rotheingham
Avec la plus grande réserve faunique au monde, Chapleau va devenir le plus important pôle d'attraction naturelle de la planète. L'écotourisme, madame, voilà ce qui va nous mettre sur la carte!
###### Annie-France Brazeau
Regardez, au nord-ouest de Sultan, Chapleau est déjà sur la carte. On vous a payé pour rien.
###### Gregory Rotheingham
Des milliers de touristes vont débarquer à chaque année, madame Brazeau, tout ça grâce à l'écotourisme!
###### Annie-France Brazeau
À l'écotourisme?
###### Gregory Rotheingham
À l'écotourisme!
###### Annie-France Brazeau
Bon, on a déjà l'écho, il manque juste les touristes. C'est vrai que l'écotourisme est un concept très tendance, mais le marché est tellement saturé. Comment allez-vous donc vous distinguer?
###### Gregory Rotheingham
Pour les touristes, voir un orignal, un ours noir, un castor ou une mouffette, c'est aussi exotique que de voir un dinosaure. On va donc faire ici le Parc jurassique du Nord de l'Ontario.
###### Annie-France Brazeau
Le Parc « pathétique »!
###### Gregory Rotheingham
Non, le Parc « funique ».
###### Annie-France Brazeau
Vous voulez dire le Parc « faunique »?
###### Gregory Rotheingham
Non, non, le Parc « funique ». Parce que la faune, c'est le fun!
###### Annie-France Brazeau
Vous avez même le slogan. Étonnant!
###### Gregory Rotheingham
Tout d'abord, la réserve faunique actuelle est beaucoup trop grande et les touristes s'y perdent tout le temps. Alors, on va fournir des quatre-roues aux visiteurs...
###### Annie-France Brazeau
Vous allez fournir quatre-roues? Y a-t-il un véhicule qui vient avec ça?
###### Gregory Rotheingham
Non. Des quatre-roues, des véhicules tout-terrain... téléguidés. À partir d'une centrale, on va téléguider les touristes vers les plus belles destinations qu'on aura aménagées.
###### Annie-France Brazeau
J'imagine que ces destinations seront féeriques.
###### Gregory Rotheingham
Absolument! Le premier arrêt, c'est le dépotoir municipal!
###### Annie-France Brazeau
_dégoûtée_
Pourquoi?
###### Gregory Rotheingham
Pour voir les ours noirs dans leur environnement naturel! On va aussi financer la gestion des déchets à Chapleau : au lieu de vendre des sacs de cacahuètes pour nourrir les animaux, les gens vont acheter un sac de vidanges pour attirer les ours. Et, comme on attend des millions de touristes, on va pouvoir régler le problème des déchets de Toronto. La métropole va nous les donner et nous on va les vendre!
###### Annie-France Brazeau
Gregory Rotheingham, vous rendez-vous compte que la première mesure que vous proposez dans la réserve faunique est une importation massive de déchets? Les seuls touristes que vous allez attirer, c'est les adeptes de _La petite vie_. Tous les « pôpas » de l'Amérique du Nord vont se garrocher à Chapleau.
###### Gregory Rotheingham
Un autre arrêt qui sera un succès incontesté sera la chasse à l'orignal.
###### Annie-France Brazeau
Oh!
###### Gregory Rotheingham
On arrête sur le bord d'un magnifique marais, avec les roseaux qui ballottent au vent, les grenouilles qui coassent et, tout à coup, au loin, un gros _buck_ apparaît. Pour une somme dérisoire, les touristes pourront tirer l'orignal.
###### Annie-France Brazeau
Gregory Rotheingham, où avez-vous la tête? Vous suggérez d'amener les gens dans une réserve faunique pour y tuer la faune!
###### Gregory Rotheingham
Il mourra pas pour vrai. On va dompter un orignal et en faire un grand acteur. Les touristes n'auront que des fusils à blanc et, quand ils vont tirer l'orignal, il va mourir d'une façon interminable et atroce.
###### Annie-France Brazeau
Pourquoi l'orignal agonisant ne pourrait-il pas donner des conseils aux plus jeunes?
_prenant la voix de l'orignal_
« Regarde toujours des deux côtés de la rue avant de traverser. »
_reprenant sa voix_
Prenez des notes, Gregory.
###### Gregory Rotheingham
C'est une très bonne idée. Mais vous oubliez une chose, Annie-France.
###### Annie-France Brazeau
Quoi donc?
###### Gregory Rotheingham
Un orignal, ça ne parle pas.
###### Annie-France Brazeau
Vous... vous voulez dire que cet orignal qui venait tous les soirs à la fenêtre de ma chambre lorsque j'avais 13 ans et qui m'a patiemment appris les rudiments du parchési n'était pas un vrai orignal? Oh!
###### Gregory Rotheingham
On amènera aussi les touristes à la pêche. Mais comme c'est parfois long et plate, on va avoir un gros maskinongé mécanique qui va attaquer les touristes en chaloupe, pour leur donner des sensations fortes.
###### Annie-France Brazeau
Bon, après le _Parc jurassique_ , v'là _Les dents de la mer_!
###### Gregory Rotheingham
Bonne idée! On pourrait racheter le requin de _Jaws_ et le maquiller comme un maskinongé et hop! le tour est joué. Ce qui est magnifique dans ce projet, c'est qu'avec un poisson mécanique, nous aurons un environnement contrôlé. Il n'y a aucun risque d'accident.
###### Annie-France Brazeau
Bravo, les touristes vont se faire dévorer en toute sécurité.
###### Gregory Rotheingham
Nos études démontrent aussi qu'il nous faut viser de nouveaux marchés. On va donc attirer les Indiens.
###### Annie-France Brazeau
Les Autochtones?
###### Gregory Rotheingham
Non, non! Les Indiens des Indes. On va organiser à Chapleau un tournoi international de fakirs. Il va y avoir différentes épreuves.
###### Annie-France Brazeau
Par exemple?
###### Gregory Rotheingham
« La roulade sur porcs-épics. » On va remplir une salle de porcs-épics et les fakirs devront se coucher sur les porcs-épics et rouler jusqu'à l'autre bout de la salle. Le gagnant est celui qui se rend au bout avec le maximum d'épines dans le corps.
###### Annie-France Brazeau
Vous savez ce qu'il y a de bon avec un tournoi international de fakirs? Ils vont tous se battre pour être le clou du spectacle!
###### Gregory Rotheingham
On va aussi satisfaire les amateurs de culture. Vous savez que l'auteur Louis Hémon est mort à Chapleau?
###### Annie-France Brazeau
Oui, frappé par un train.
###### Gregory Rotheingham
_emballé_
Le clou de la tournée de Chapleau sera « le spectacle multimédia Louis Hémon ».
###### Annie-France Brazeau
Jusqu'à maintenant, on pouvait penser que vous étiez fou et /ou tordu, Grégory, mais cette idée de multimédia vous rachète. C'est tellement tendaaance!
###### Gregory Rotheingham
Ce sera formidable. On va installer des estrades le long de la voie ferrée. Au loin, on voit venir sur la voie ferrée cet écrivain vagabond et juste quand il arrive devant les estrades, bang! il se fait frapper par un train et va revoler 200 mètres dans les airs!
###### Annie-France Brazeau
Mais c'est épouvantable!
###### Gregory Rotheingham
Ben non, mais non, on embauchera un cascadeur. Mais le plus beau, c'est que quand on meurt, on voit le film de notre vie. Alors, dans le train, on va projeter la vie de Louis Hémon... EN SURROUND! Quatre représentations par jour!
###### Annie-France Brazeau
C'est la première fois de ma vie que je regrette d'avoir des oreilles.
###### Gregory Rotheingham
Mais voyons, Annie-France, les gens vont venir de partout pour voir ça.
###### Annie-France Brazeau
Ben, justement, comment allez-vous transporter tous ces touristes? Ça va vous prendre des autoroutes à quatre voies et un aéroport international tant qu'à y être.
###### Gregory Rotheingham
Non, on va plutôt miser sur les forces de Chapleau, qui a toujours été une ville ferroviaire. On va construire un train à grande vitesse, un TGV, Toronto-Chapleau.
###### Annie-France Brazeau
Vous savez que ça circule vite, ces trains-là. Est-ce qu'on va dire aux gens de se méfier des trains?
###### Gregory Rotheingham
Oui, on va installer une grosse pancarte sur laquelle ça va être écrit : Gare!
###### Annie-France Brazeau
Bien beau, tout ça, mais un auteur qui se fait frapper par un train, c'est un peu mince côté culture.
###### Gregory Rotheingham
Ah ben, ça, de la culture, on en a. un gros festival de la nature triculturel où l'on rassemble les cultures francophone, anglophone et autochtone.
###### Annie-France Brazeau
Là, Gregory Rotheingham, vous exagérez. Ça ne marchera jamais, c'est impossible de faire travailler des Français, des Anglais et des Autochtones sur un même projet. Regardez le Canada : 126 années d'échec.
###### Gregory Rotheingham
Ben, à Chapleau, c'est trois ans de succès. Ça existe déjà depuis trois ans. Des gens qui fêtent et s'amusent ensemble et ça attire déjà près de 1 500 personnes chaque année.
###### Annie-France Brazeau
Gregory, franchement, ai-je l'air d'une cruche? Vos projets de Parc « funique », de Festival du fakir, ou de collision multimédia passent toujours, ça pourrait être branché. Mais l'harmonie linguistique et culturelle, on va rire de vous.
###### Gregory Rotheingham
Mais écoutez, est-ce si difficile à croire qu'on peut avoir un lunch avec un _fish & chip_, une poutine et un castor rôti?
###### Annie-France Brazeau
Les trois, ensemble, je les trouve effectivement assez dur à avaler. Comme toutes vos idées de développement économique d'ailleurs. Heureusement, c'est tout le temps que nous avons pour l'entrevue. Gregory Rotheingham, merci.
###### Gregory Rotheingham
De rien.
###### Annie-France Brazeau
De rien, oui, en effet.
###### Faits pour rester
_Quand tes jours sont pleins de détours_
_Et que tu sais plus t'y prendre_
_Quand la vie te joue des tours_
_Et que tu t'es laissé prendre_
_Ce sera là que tu sauras_
_La valeur de tes espoirs_
_Perds-les pas de vue, mon gars_
_Tu n'as pas le choix que d'y croire_
Car y en a, mon gars, qui sont faits pour partir
Et y en a, mon gars, qui sont faits pour bâtir
Si le Sud t'appelle, je vais pas protester
Mais y en a, ma belle, qui sont faits pour rester
_Quand ta nuit est pleine d'étoiles_
_L'espérance vient sans effort_
_Pour guider ton idéal_
_Y a pas juste l'étoile du Nord_
_Quand ce ciel qui te sourit_
_C'est celui qui t'a vue naître_
_Il n'exige pas de merci_
_Mais tu peux le reconnaître_
_Quand elles brillent, ma fille, les promesses d'avenir_
_Elle t'appelle, ma belle, celle que tu vas choisir_
_Si tu suis ton cœur, tu auras bien choisi_
_Puis y en a du cœur, dans ce coin de pays_
_Et à chaque nouveau matin_
_Un café chaud pour la route_
_Un petit bec sur les deux joues_
_Faut que j'y aille, même si ça me coûte_
_Croise tes doigts puis cogne du bois_
_Elle va bien notre petite vie_
_J'avoue que ça m'inquiète parfois_
_Faudra-t-il partir aussi?_
Car y en a, mon gars, qui sont pris pour partir
Quand il faut, ma fille, pour le bien d'la famille
Si le Nord nous lâche, faudra bien s'en aller
Mais moi, je lâche pas, et je veux y rester
###
### Salut Earlton!
Même si un gigantesque bison de métal surplombe la route 11 près d'Earlton, en vérité, c'est la vache qui règne sur la région. Par ici, les fermes laitières sont si perfectionnées qu'il y en a même une où l'étable se nettoie tout seul! Plusieurs fois par jour, le plancher incliné est lavé par une vague d'eau qui emporte le fumier. C'est à Earlton que vous trouverez cette merveille : une étable qui se _flushe_! Et à côté de ça, un carrousel ultramoderne en acier étincelant où les vaches font leur petit tour de manège deux fois par jour pour la traite. Je vous le dis, elles croient qu'elles vivent à Disneyland. Dans le Nord de l'Ontario, c'est seulement ici, dans le petit Claybelt, que les champs s'étendent vraiment à perte de vue. Au printemps, on se croirait sur une mer d'argile. Dans cette grande plaine, ça se comprend qu'on y élève des vaches, des bœufs, des bisons, des chevaux percherons, des animaux forts et massifs. Donc ce n'est peut-être pas étonnant non plus que ce coin de pays ait été la patrie d'un homme aussi fort qu'eux : Wilfrid Paiement.
Au dire de ceux qui l'ont connu, il était fort, oui, mais en plus, il avait du bagout. Plus d'un journaliste sportif en a été fasciné. Assis dans les gradins, Wilfrid Paiement volait la vedette à ses fils hockeyeurs de la Ligue nationale par sa volubilité. Quand il se mettait d'humeur à vous conter une de ses histoires, vous aviez droit à une de ces histoires. J'espère qu'en vous contant des bouts de sa vie, j'aurai des bouts de sa verve.
###### Wilfrid Paiement
Wilfrid Paiement est né au Québec en 1910. Il avait sept ans à son arrivée à Earlton. Il était écolier de dix ans quand il a gagné son premier match de lutte... contre sa maîtresse d'école. Selon Wilfrid, elle avait perdu les pédales. Elle venait de renverser des écolières en les disciplinant à coups de planche. Le jeune Wilfrid l'a saisie par les poignets et n'a pas lâché prise avant qu'elle ne s'épuise. À la fin, elle est tombée sans connaissance. Wilfrid a promis au directeur de ne plus jamais toucher la maîtresse.
C'est à l'école qu'il a connu Rosilia. Avec elle, il aura onze filles et cinq gars. Madame Paiement a dit un jour à un journaliste sportif que son mari était fort, mais sans malice. S'il avait été mauvais, a-t-elle dit, elle ne lui aurait pas donné seize enfants.
En tout cas, l'écolier était devenu une pièce d'homme de six pieds, deux cent soixante-dix livres. Tour de poitrine, 62 pouces. Tour de biceps, 22 pouces et demi. Tour du cou, 20 pouces. Wilfrid Paiement pouvait, comme si de rien n'était, ramasser un tonneau d'essence, un « quart » de 45 gallons bien plein, et le charger dans un camion. Bien des gens l'ont vu faire ça. Il faisait des tours de force dans les foires régionales. Il pouvait soulever un cheval en grimpant dans un poteau ou à partir d'une plate-forme.
Il n'a jamais bu d'alcool ni de café, juste un petit _crème soda_ de temps en temps. Mais du lait, ça par exemple! Au repas, il pouvait en boire deux pintes et manger les trois quarts d'une livre de beurre. Et il aimait les sauces crémeuses. une chance qu'il habitait un pays laitier! Quand il vous serrait la main, la vôtre s'y perdait comme dans une grappe de bananes. Ses photos montrent un visage empreint de confiance tranquille et de joie de vivre. Il aurait fait un bon bonhomme carnaval. Mais on n'aurait pas demandé au bonhomme carnaval de se battre avec un ours.
En 1939, Wilfrid Paiement était à l'exposition de Québec. Ses tours de force lui avaient valu l'amitié de célèbres hommes forts comme Victor Delamarre et Paul Baillargeon. C'est un d'eux qui l'a incité : « Envoye, Frid, envoye! » un défi physique, ça l'attirait, c'était plus fort que lui. C'était une des rares choses plus fortes que lui. Ce jour-là, un promoteur de manèges mettait les visiteurs au défi de se battre contre un ours. Les badauds pouvaient gager sur l'issue du combat. Les organisateurs connaissaient leur ours. Ils ont parié en sa faveur à vingt contre un. Mais les amis de Wilfrid connaissaient l'homme qui a vu l'ours. Ensemble, ils ont misé plus de mille dollars sur lui, une somme immense à l'époque. L'ours portait une muselière et d'épaisses mitaines, mais même sans crocs ni griffes, vous aviez affaire à 690 livres de muscles poilus qui ne veulent pas se coller le dos au plancher.
La cloche a sonné et le face-à-face s'est engagé. Vous connaissez la prise de l'ours? Mettons que les circonstances s'y prêtent. En vrai lutteur, Wilfrid Paiement a saisi la bête en une embrassade par-derrière et l'a soulevée de terre. L'ours battait l'air de toutes ses pattes. Wilfrid Paiement enfonçait sa tête dans le dos de l'ours. Il poussait, il poussait. Lentement, l'ours a manqué de souffle. Il s'est étouffé. Notre homme n'a pas lâché prise tant qu'on ne lui a pas montré la couleur de l'argent qu'il avait gagné. À ce point-là, il pendait de la gueule de l'ours un bon huit pouces de langue toute bleue. Il était blessé à mort. Il avait l'épine dorsale écrasée. Il a fallu achever la pauvre bête. Wilfrid Paiement avait tué un ours de ses mains nues. Et ses amis s'en sont mis plein les mains, en vendant la peau de l'ours avant de l'avoir tué.
Ce pauvre ours n'avait pas couru après son malheur. Mais une fois, Wilfrid Paiement a dompté des adversaires qui l'avaient cherché, et qui l'ont trouvé. En 1937, à 27 ans, Wilfrid Paiement était _jobeur_. Il fournissait en bois la compagnie A. J. Murphy. Ses six cents hommes, pour la plupart venus du Québec, étaient logés, nourris et blanchis par sa femme et ses filles employées au camp. Dans ce temps-là, ses bûcherons étaient payés 11 cents la bille de seize pieds. un jour, à ce qu'on dit, des têtes chaudes ont réclamé 3 $ la bille — méchante augmentation! Ils insistaient à coups de gestes d'intimidation et de vandalisme. Les Murphy ont menacé de faire venir la police. Cette intervention aurait peut-être été plus tragique que l'incident qui a suivi. Mais jamais on ne l'a su, car la police n'est pas venue.
Au petit matin vers cinq heures, la forêt résonnait de la clameur d'une centaine hommes de vilaine humeur à la porte du bureau de Wilfrid Paiement. Sa femme et ses enfants y logeaient, effrayées avec raison. Soudain, ces hommes font irruption dans la pièce d'en avant en sacrant et en exigeant que le peureux de Paiement se montre. Wilfrid laçait ses bottes dans la pièce d'en arrière. C'était ses « bottes à coups de pied au derrière ». Il est sorti et a demandé qui l'avait traité de peureux. un grand costaud de six pieds quatre a dit : c'est moi. Wilfrid l'a saisi à la gorge, de sa grosse main épaisse comme un rôti de bœuf. Il l'a soulevé de terre et l'a sorti en défonçant la porte. Notez bien que dans ce temps-là, on bâtissait les portes de camp en planches épaisses, des deux par huit. Rendu dehors, Wilfrid a fait bondir son homme sur le fessier au moins vingt fois sur une souche plate. Ça l'a tranquillisé. Mais pendant ce temps, un autre s'approchait pour lui sacrer un coup de deux par quatre. Quelqu'un a crié. Wilfrid a pu saisir la planche. La plupart se sont enfuis. Mais quatorze se sont retrouvés blessés à l'hôpital de Haileybury.
Wilfrid Paiement a raconté plus tard, avec tristesse, comment il a appris que l'homme qu'il avait rassis en était resté infirme. Il l'a su d'une religieuse croisée à la messe par hasard bien plus tard dans le Sud de l'Ontario. Elle lui a conté l'histoire de son frère qui venait de mourir, sans laisser voir si elle savait qu'elle s'adressait à l'homme qui lui avait infligé son malheur. Wilfrid Paiement avait défendu sa famille et sa peau. De ça, il n'avait pas de regrets. Aussi, disait-il, il s'était fait un nom. Et dans le Nord, quand tu te fais un nom, ce n'est pas à recommencer. Voilà une histoire qui rappelle qu'un syndicat civilise les conflits de travail et qu'on n'a pas à regretter l'époque où il n'y en avait pas.
Dans les années quarante, Wilfrid Paiement était devenu éleveur de chevaux trotteurs et propriétaire de l'hippodrome Châteauguay, à Earlton. Les courses étaient parfois suivies de matchs de lutte et quand il manquait un athlète au programme, Wilfrid se portait volontaire. Parmi ses centaines de trotteurs, il y en a eu deux qui n'ont jamais été défaits à la course d'endurance de cinq milles. Doc Vino et Bannock Bay, ça, ça courait! Et le pire, c'est qu'ils traînaient le gros Wilf Paiement dans le sulky!
Une fois, en novembre 1950, Wilfrid se trouvait à l'hippodrome de Québec quand un cheval nommé Alvin Hanover est tombé mort en pleine course. Le tracteur qui aurait dû venir dégager la piste ne démarrait pas. Wilfrid Paiement a réglé le problème. Il a traîné à l'écart un cheval de 1 350 livres. une autre fois, en revenant de Blue Bonnets, à Montréal, une roulotte qui transportait des chevaux a fait une crevaison. Pressé, il faut croire, Wilfrid Paiement a soulevé la roulotte pendant qu'on changeait le pneu.
Jeune homme, Wilfrid Paiement remportait déjà des championnats de tir au poignet. C'est un plaisir qu'il s'est offert jusque dans l'âge d'or. En 1979, à 69 ans, il a gagné huit matchs d'affilée à Toronto contre des adversaires de partout au monde. Le huitième était un Allemand super poids lourd. Ils sont restés barrés presque quatre minutes, sans que rien ne bouge, puis tranquillement, Wilfrid... a senti... son adversaire... ramollir. En 1984, à 74 ans, il est revenu de Londres avec le championnat de non pas une, mais deux catégories : poids lourd et super poids lourd. La même année, à Edmonton, il a désarticulé le bras d'un de ses adversaires. Pourtant, il ne consacrait généralement pas plus de deux jours à se mettre en forme avant les matchs. L'entraînement, il n'aimait pas ça.
Wilfrid Paiement a quitté ce monde en 1992 à 82 ans. La mort emportait un homme doté de force musculaire, mais aussi de force de caractère. Jeune homme, il a été le policier du village; il avait le physique de l'emploi, vous me direz. Pour bâtir l'église, il a aidé à la collecte. Comme entrepreneur, il a su réussir en affaires, mais a su aussi aider ses proches. En somme, il a été un de ces hommes du genre qu'il faut pour bâtir une communauté. Ce genre de force n'a rien à voir avec la force physique. Cette autre force, il l'avait aussi.
En somme, Wilfrid Paiement a été un homme remarquable. Et pourtant, n'eût été la célébrité de ses fils hockeyeurs et de la distraction de quelques journalistes sportifs, nous n'aurions aujourd'hui à peu près aucun témoignage écrit de sa vie. Ironiquement, plus d'un de ces journalistes a terminé son article en disant qu'il faudrait bien écrire un livre sur la vie de cet homme-là. Mais ce livre-là, on ne l'a pas écrit. Le Nord de l'Ontario ne semble pas avoir besoin du souvenir de ses hommes plus grands que nature. C'est ainsi que nous perdons notre mémoire, celle de l'époque où nos forêts et nos terres attiraient des dizaines de milliers de familles.
Aujourd'hui, les rêves de belle vie mènent la jeunesse ailleurs qu'ici. Le canton d'Armstrong où se trouve Earlton a perdu 12 % de sa population depuis cinq ans. En cela, il ressemble à toutes les communautés, grandes et petites, du Nord de l'Ontario. Notre coin de pays se dépeuple. L'exode va grandissant sans qu'on sache le contrer. Pour le contrer, d'abord il faut en avoir la volonté. Or, le cœur se nourrit d'amour et d'art, et l'art se nourrit de métaphores. Nous en avons besoin, du souvenir de notre homme fort.
###### Un étranger a pris ta place
un étranger a pris ta place
Il est assis sur ton banc de bois
Tout près, ici, à côté de moi
Le temps d'un verre, le temps d'une bière
Qu'est-ce que je vais faire, lui dire de se taire?
C'est lui qui est là, ce soir, petit frère
_T'es pas d'ici, t'arrives d'ailleurs?_
_T'es de passage, t'es un truckeur?_
_L'autre avant toi parlait de tracteurs_
_Et avant lui, un autre vendeur_
_Des fois, je te parle encore, des fois_
_Tout le temps, tout seul, tout bas, petit frère_
_Dans le fond, Gaston, reviens pas en arrière_
_T'as bien fait, t'as bien fait de lâcher la terre_
_C'est pas la terre, c'est les affaires_
_Cent cinquante têtes à deux cents piastres_
_Peux-tu croire ça? Moi, je le prends pas_
_Je me parle tout seul, je te parle tout bas_
_Traite-moi de fou, mais moi je vends pas_
_T'es pas d'ici, tu comprends pas_
_C'est mieux comme ça. Bois vite, puis va!_
Un étranger a pris ta place
Il est assis sur ton banc de bois
Tout près, comme toi, à côté de moi
Avant qu'il vienne, c'était ta place
Mais toi aussi, t'étais de passage
Le temps d'un verre, le temps d'une bière
_Dans le fond, Gaston, reviens pas en arrière_
_Il faut être fou pour vivre de la terre_
_Dans le fond, Gaston, reviens pas en arrière_
_Il faut être fou pour vivre de la terre_
_Comme le grand fou qu'est ton grand frère_
###### L'hurluberlu d'Earlton
Maudites chaises de bingo! À la taverne, au moins, il y a des vraies chaises de bois. C'est quoi l'idée d'installer des chaises de sous-sol d'église dans un restaurant?
Es-tu gêné? Non! Mais, alors... pourquoi tu me réponds pas comme du monde? On est à Earlton ici, pas à Sudbury. Inquiète-toi pas, il y a pas de maniaques dans le Grand Boulevard. Juste du monde ben ordinaire, comme moi, comme nous autres. Crois-moi, je les connais tous et puis là, j'ai des choses à leur dire!
Eux autres là-bas, dans le petit coin noir, c'est mon neveu et la fille de Rosaire. Dans l'autre coin, c'est la gang de réguliers de l'hôtel. Ils viennent au village pour passer la soirée, les gars pour boire de la bière, les filles pour danser ou pour se parler dans la face. D'habitude, tout le monde s'amuse comme des petits fous. Mais ce soir, je me suis retrouvé tout seul entouré de chaises vides, à téter tout seul ma bière... à parler aux chaises vides. Bout de christ, j'ai tellement gueulé que le _bouncer_ m'a mis à la porte. C'est là que j'ai décidé de venir prendre une petite bière avec vous autres. C'est vendredi soir! _The more the merrier_! Ici, on fête en gang.
Je parle pas trop fort, c'est toi qui a les oreilles délicates. Tu devrais venir avec nous autres pour une _run_ sur une plate-forme de _drillage_. Bout de christ, je te jure que ça leur ferait la job à tes petites oreilles de femmelette.
Tu vois? Je te l'avais dit! Ils sont pas regardants. Par ici, tout le monde est bienvenu. T'as qu'à payer une ronde pis y vont tout oublier, même ton accent de moumoune.
Ah, eh, eh... fâche-toi pas! C'est pas de ma faute si tu parles comme une bonne sœur!
Moi, tu veux savoir ce que je fais ici? Ben... c'est une longue histoire.
Mon père, c'était Pierre Lachance. Le fils de pépère Pierre et pis, tenez-vous bien, le père de mon grand frère Pierre. Dans notre famille, la terre passe de Pierre en Pierre. Les filles, les Ti-Paul et les Gros-Jean comme moi doivent se trouver un Pierre ou encore mieux une Pierrette pour se faire une place dans la vallée.
Mon grand-père Lachance est arrivé à Earlton en 1923, quelques mois après le grand feu. Il avait son héritage en poche, ce n'était pas grand-chose, mais il a débarqué du train au moment où il y avait plus de vendeurs que d'acheteurs. Il y a rien comme du _cash_ pour faire descendre le prix. Il s'est acheté une terre le lendemain et a fait venir sa femme le surlendemain. Puis là, ben, pour vous dire franchement, il s'est employé à les faire profiter tous les deux. Ma grand-mère est arrivée à Earlton sans bagages sauf pour son linge de corps et son petit tabouret de laitière, une belle tranche de pin blanc dans laquelle son père avait planté trois petites pattes de tremble. En 39, au début de la guerre, mes grands-parents avaient déjà cinq grands gars pour les aider à faire le train. Ils ont profité de la Deuxième Guerre pour les établir, les plus jeunes dans le troisième rang et le plus vieux, mon père Pierre, sur la terre paternelle.
Quand j'entendais le grand-père raconter l'histoire de la famille, ben, je prenais goût à travailler plus fort en croyant qu'un jour, ce serait mon tour. Vous savez, la senteur du fumier, les bêtes qui bardassent dans les brancards, les longues journées à soigner les animaux, ben... j'ai toujours aimé ça. Et puis sur une nouvelle terre, sur _ta_ terre, t'as pas de boss, même pas de bonhomme pour te dire quoi faire.
Ce que mon grand frère et mon père faisaient, je m'arrangeais pour le faire trois fois mieux. Je commençais le train du matin avant eux autres et je rentrais pas à la maison avant qu'ils soient attablés tous les deux. Et puis à un moment donné, Ti-Paul, le plus jeune des gars, s'est mis à faire exactement comme moi. Après quelques années, on est devenus inséparables. Lui et moi, quand on s'attelait à une job, on travaillait ensemble comme les rouages d'une seule et unique machine. Faut croire que ça dérangeait le père de nous voir nous éreinter à construire le bien de la famille et, pour mettre les choses au clair, il disait souvent à haute voix devant tout le monde : « Pierre, tu es Pierre et sur cette Pierre, je cultiverai ma terre! »
C'est pas qu'il voulait pas nous aider, mais c'est comme s'il voyait venir le futur. C'est dans le milieu des années soixante-dix que les choses se sont compliquées. Comme tous les autres habitants, on a décidé de moderniser la ferme, d'acheter des gros équipements et même d'acheter la terre de notre voisin. On produisait plus, c'est sûr, et à quatre on avait l'impression qu'on allait prendre assez d'avance pour faire de la place pour tout notre monde. Mais là, en plus de nourrir la famille, on devait rembourser le capital emprunté et payer des intérêts de plus en plus gros. Chaque soir, le père sortait ses grands livres de chiffres et il passait la soirée à gratter pour réduire les dépenses, à tout réinventer pour diversifier les revenus, mais malgré tout il arrivait pas à équilibrer ses comptes. À un moment donné, sa première priorité est devenue de réchapper sa ferme. Il avait plus assez d'argent ou de temps pour songer à établir ses fils sur leur propre terre.
Au fur et à mesure que le temps passait, il est devenu évident que les choses s'amélioreraient pas. Tout empirait dans les affaires comme dans la famille. À la maison, la vie est devenue... ben... un nid de guêpes. Les problèmes d'argent, c'est une chose, mais les chicanes de famille, c'est autre chose. La mère a toujours trouvé un moyen de partager le pâté chinois, mais quand la chicane a poigné dans la famille, c'était pas évident de remettre la paix.
Vous savez, il faut beaucoup plus que du foin ou des recettes de bonne fermière pour nourrir un banquier. Il faut lui donner la part du lion, rien de moins. C'est ce qui arrive quand tu invites un cochon à dîner : il faut pas s'attendre à ce qu'il apprenne des bonnes manières du jour au lendemain. C'est pas tellement ce qu'il fait qui dérange, mais plutôt l'effet qu'il a sur les autres. À la fin des années soixante-dix, le taux d'intérêt a monté en haut de quinze pour cent. À ce niveau-là, il y en a plus de crème sur la table, il y a plus de beurre sur le pain et les femmes n'ont que de l'amertume pour allonger la soupe.
Par un soir que ça bardait plus que d'habitude, on s'est levés de table, Paul et moi, et puis là... au lieu de repousser nos chaises à leur place, comme il se doit, on les a, une fois pour toutes, rangées le long du mur avec celles des invités. La mère a tout compris, le père aussi, mais que vouliez-vous qu'ils fassent? Ils étaient trop occupés à réchapper la terre pour voir aux besoins de Ti-Paul et moi, et puis, de toute façon, avec tous les nouveaux équipements qu'on avait achetés, il y avait plus de travail pour quatre hommes sur la terre.
Ti-Paul et moi, on est descendus à Sudbury pour travailler dans les mines, comme on dit, dans le trou. Ç'a pas l'air évident, mais, vous savez, pelleter du fumier ou pelleter du sable de remplissage, c'est échanger quatre trente sous pour une piastre. En dessous de la terre, c'est pas comme sur une terre, tu peux pas changer de tas de branches à volonté. Quand t'as fini une job, y faut que t'attendes que le _shift boss_ passe. On s'amusait à voir dans combien de temps on pouvait finir la job, mais il faut dire que les autres mineurs trouvaient pas ça drôle.
C'est à ce moment-là que le boss nous a offert d'aller travailler du côté de la production, là où les casseurs de records se méritent des bonis plutôt que des insultes. Nous autres, on demandait pas mieux, on était pas là pour se chicaner avec les gars, mais plutôt pour mettre notre plan en branle. C'était pas ben ben compliqué, notre affaire : on voulait sauver notre argent pour faire un dépôt sur une première terre, la cultiver pour la payer et ensuite l'hypothéquer pour en acheter une deuxième. On était convaincus qu'on arriverait à faire ça avant d'avoir cinquante ans : trois ans dans les mines pour sauver le dépôt, quinze ans de travail sur la terre pour payer la première et dix ans pour payer la deuxième. On pensait qu'à deux, on pourrait faire ce que notre grand-père avait fait seul. Sans problème!
Ça fait qu'on a pris un appartement en ville et l'année suivante nos trois sœurs sont venues vivre avec nous le temps de leurs études. À un moment donné, la plus grande partie de la famille vivait à Sudbury. Pendant que les filles faisaient leur école normale, nous autres on faisait notre magot. On prenait nos vacances pour rentrer les foins du père, on passait les fêtes en famille. En fait, on se préparait à retourner au village, les gars sur une terre bien à eux et les filles comme maîtresse dans une école de par chez nous.
Après leur « graduation » les filles ont tout fait pour se chercher de l'ouvrage, mais elles ont passé l'été à attendre. À Earlton, les inscriptions dans les écoles n'augmentaient plus, le baby-boom était fini, les familles étaient moins grosses, et puis, la plupart des jeunes rangeaient leur chaise contre le mur avant de partir pour la ville. À la fin du mois d'août, les filles ont accepté un poste dans une école d'immersion d'Ottawa. Ça fait que là, ç'a été notre tour à Paul et moi, de retenir nos émotions pendant que nos sœurs rangeaient leurs chaises le long du mur des invités.
Pendant ces trois années-là, ça nous a pas coûté ben cher vivre, donc Paul et moi, on avait accumulé un peu plus de 20 000 $ en banque. C'était tout un motton, mais le prix des terres avait plus que doublé et notre magot était devenu trop petit. En 1982, le taux d'intérêt a dépassé les 22 pour cent. On a eu beau dire au banquier qu'on était des bons travailleurs, il nous a répondu en grognant que ça avait rien à voir. La banque venait tout juste de fermer la manufacture de roulottes Boler, donc pour nous autres ça voulait dire qu'il y avait plus de job de manufacture au village pour faire arriver nos fins de mois.
En quelques années, l'agriculture était devenue un christ de gros business. Pour embarquer, il fallait du gros _cash_ , du _cash_ pour acheter une grosse terre, du _cash_ pour des équipements, du gros _cash_ pour les quotas et beaucoup de change pour les _et cætera_. Le gérant de la caisse a pris le temps de nous expliquer tout ça en détail. Il nous a dit qu'on aurait beau travailler fort, l'inflation galopante grugerait notre capital plus vite qu'on le construirait. Il nous restait plus rien à faire que sauter une maudite brosse. En fait, à vrai dire... j'ai sauté une brosse au village, dans la taverne de l'hôtel Lasalle. J'ai pris une chambre en haut puis j'ai passé une semaine à boire. C'est pas comme si j'étais le premier à faire ça, je vous en passe un papier, il y a des centaines de bûcherons qui ont bu leur paye dans cette taverne, mais au moins ils buvaient en gang. Moi, j'étais entouré de chaises vides.
Paul est retourné à Sudbury pour voir sa blonde pendant que l'appartement était vide. Par le temps que je suis retourné en ville, j'avais perdu ma job et la blonde de Paul avait pris ma place dans l'appartement. Ils parlaient de mariage, d'enfant, de futur, ça fait que je leur ai donné le restant du magot. J'ai pris le temps de dire bonjour à Ti-Paul, d'embrasser sa fiancée et de ranger ma chaise contre le mur de la visite. Ç'a pas été long que j'ai poigné une job de _diamond driller_ avec la Boart Longyear de North Bay. J'ai fait ma première _run_ au Chili, trois mois dans un camp avec 25 gars, faut le faire. Douze semaines à douze heures par jour, huit heures pour dormir et puis quatre heures pour manger, pour se laver ou pour jouer aux cartes. Quatre vingt-dix jours à chercher de l'or, suivis de trente jours de vacances à Manhattan, à New York, New York. Eh christ! C'est là que j'ai pris le goût des voyages et des grandes villes : trois mois au Pérou, au Gabon, en Afrique du Sud, en Guyane française, en Australie, en Russie, en Mongolie, en Alaska, au Nunavut, suivis d'un mois à Paris, à Londres, à Rome, à Barcelone _,_ à Tokyo, à Moscou, à Berlin, à Johannesburg et à Casablanca. En fait, ce sont mes dernières vacances à Montréal qui ont été les meilleures. J'ai passé un mois dans une suite exécutive du Ritz-Carleton de la rue Sherbrooke. Eh, bout de christ, il faut le voir pour le croire : une salle à manger privée, un salon avec un foyer, une grande chambre à coucher, deux salles de bains et une facture de 550 piastres par soir! Mais... _Easy come, easy go_!
J'ai fait le tour de la Terre de même une bonne dizaine de fois, mais c'est comme si j'avais entrepris de labourer ma terre à partir des bords en m'en allant vers le milieu. J'ai commencé ma spirale au Chili et en Afrique du Sud, pour me rapprocher de plus en plus de l'Europe et de l'Amérique du Nord, et puis du Canada et du Nouvel-Ontario. À cette heure, je travaille à la baie James et je passe mes vacances à Earlton, dans une des chambres de luxe de l'hôtel La Salle. Quatre murs, un plafond, un plancher, une douche et une toilette dans le corridor pour trente piastres par soir. Et puis là, t'as pas besoin de la voir pour le croire. Entre un petit bol de caviar au Café de Paris de la rue Sherbrooke et une grande assiette de poutine au Grand Boulevard de la 10e Rue, il y a tout un monde, croyez-moi!
Il y a ben des choses qui sont pas correctes en ville, mais c'est pas pour ça que je reviens toujours passer mes vacances au village. C'est les petites choses que j'aime. Je reviens parce qu'il y a que des grandes tables communes dans le restaurant. Le matin, on est pas obligé de manger ses œufs tout seul, on peut s'asseoir avec les gens qu'on connaît. C'est la même chose dans l'hôtel Lasalle : s'il y a un petit nouveau qui se présente, on l'invite à prendre un verre. Et puis dans le village, quand il y a un besoin communautaire, les gens de la place organisent une corvée ou bien ils font tirer au sort un _truck_ et une grosse roulotte pour faire de l'argent.
Je reviens ici parce que je suis un gars de terre glaise et que je suis pas tellement heureux sur les trottoirs de ciment et pas tellement à l'aise sur les planchers de marbre. Dans tous mes voyages, ce que j'ai aimé le plus, c'est les _runs_ dans les chantiers, avec les gars, à travailler fort, à jouer aux cartes et puis à gueuler contre les boss. Les maudits ci, les maudits ça, vous savez ce que je veux dire! À l'ouvrage, j'étais tellement heureux. Des fois, quand je me fermais les yeux le soir, j'avais l'impression d'être avec mon grand-père dans un chantier de bûcherons sur les rives du lac Témiscamingue, puis là, immanquablement, je m'ennuyais comme lui et j'avais le goût de rentrer à la maison pour faire le train du printemps.
La vie est drôle. Moi, je veux revenir vivre au village et puis, mon grand frère, lui, y veut prendre sa retraite. Il veut vendre la terre de mon père pour s'en aller dans un condominium à Ottawa, près de ses enfants. Son petit dernier, mon jeune neveu qui est en train de minoucher sa blonde dans le coin là-bas, veut rien savoir des vaches, il préfère les ordinateurs. Il dit qu'il est un cybercow-boy qui enfourche sa souris tous les matins pour aller traire son troupeau de puces électroniques. C'est certainement pas lui qui va prendre la relève sur la ferme. J'ai pensé racheter la terre de mon frère. Avec mes _runs_ j'ai tout de même un christ de gros magot qui dort à la caisse, mais la terre est rendue tellement grosse, tellement différente de ce que j'ai connu, de ce que je sais faire, que je saurais pas par où commencer. À cette heure, y a trop de chiffres, de recherche dans Internet, de satellites et de banquiers. C'est comme si le monde a changé, pis moi je suis resté poigné dans le passé. À vrai dire, il y a trop de machines et pas assez de sueur. Je suis trop vieux pour tout recommencer.
De toute façon, c'est pas comme s'il va avoir de la misère à vendre son bien, parce qu'aujourd'hui comme hier, il y a plus de preneurs que de vendeurs. Les gens de par ici ont peur des mennonites. Il paraît que lorsque ces gens-là entrent dans une vallée, ils achètent toutes les terres, une après l'autre. Ils ont leur propre financement, leur propre fournisseur, ils se tiennent en gang puis ils s'entraident. Je pense que c'est pas de ce monde-là qu'on devrait avoir peur, mais plutôt de ce qu'on est en train de devenir, parce qu'au fond les mennonites sont exactement ce que nos grands-pères étaient.
Comme eux autres, on a relevé le défi, fallait bien; on a défriché nos terres, on avait pas le choix; mais nous autres on s'est ouvert sur le monde, parce que c'était notre devoir; on pouvait pas rester enfermé pendant que l'univers se déployait devant nous à la télévision. Bout de christ, y a du monde qui meurt de faim aux nouvelles de six heures. C'est plus possible de se renfermer sur soi-même, on a pas été élevé de même. On s'est embarqué de plain-pied dans le progrès, dans le vingtième siècle, pour jouer pleinement notre rôle.
On a les deux pieds ancrés dans la terre glaise, comme nos vieux. Ils ont eu de la misère noire à tous les jours de leur règne, parce que toute bonne chose a un prix, c'est la vie, il y a du bon pis il y a du mauvais. On vit mieux que nos parents, mais il y a eu un prix à payer. À chaque génération sa misère. Aujourd'hui, on paie le progrès à coup de chaises vides. À mesure qu'on devient plus efficace, on élimine une job. Chaque chaise vide dans l'hôtel Lasalle est un ami qui est parti, chaque banc d'école abandonné est un enfant qui apprend ailleurs, chaque banc d'église vide est une famille de Canadiens errants.
Il y en a qui disent que je suis trop direct, que je dis des choses qui se disent pas, mais les gens d'ici m'endurent parce que je suis l'hurluberlu du village. Je suis pas dangereux, peut-être un peu tannant, mais pas méchant. Quand je prends un verre de trop, je leur parle dans la face et parfois, s'il y a personne pour m'écouter, je parle aux chaises vides. Parler aux chaises, ça poigne pas. Pas par ici en tout cas. un gars qui parlerait aux vaches, ça pourrait peut-être aller si ça donnait du meilleur lait. Il y en a qui comprennent pas ça du tout, ils voudraient m'envoyer en pension permanente sur la côte à North Bay. Mais si la plupart m'endurent, c'est qu'ils savent que je parle aux gens qui étaient ici, à leurs frères qui travaillent dans les mines, à leurs sœurs qui enseignent ailleurs, à tous les enfants du village qui travaillent pour le gouvernement à Ottawa. Il y a même quelques personnes qui savent que je parle à ceux qui vont venir et à ceux qui vont revenir parce que ce village est indestructible. Il y aura toujours une place de choix sur la terre pour ceux qui nourrissent les multitudes.
Ça fait que, quand tu retourneras à Sudbury, dis salut à mon frère de Sudbury, salut à la plus grande partie de Earlton qui vit maintenant ailleurs. Salut vous autres! Salut, mon Ti-Paul, passe donc à l'hôtel Lasalle, il y a en masse de chaises! Des chaises de fer, des bancs d'églises, des chaises d'écoliers, des chaises pour la visite et peut-être même, dans le hangar de notre mémoire, le petit tabouret de pin blanc de mémère Lachance.
_Gaston Tremblay_
###### Incitatif pédant **:** Le gros tourisme à Earlton
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau, vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez notamment écrit _Chapleau, une nature naturelle à exploiter naturellement_ , vous avez également publié un rapport pour relancer la communauté d'Ajax, à l'est de Toronto, dans lequel vous préconisez l'ouverture d'une usine d'équipement sportif, rapport intitulé _Ajax : jackstrap au max_.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Vous avez atteint le sommet de l'Everest en transportant un immense tuyau sur vos épaules. Quand les sherpas vous ont ramené au camp, à moitié frigorifié, vous avez expliqué que vous vouliez installer une gouttière sur le toit du monde.
_Il va répondre. Elle enchaîne._
Vous avez reçu un prix international pour l'ensemble de votre œuvre, prix qui vous a été retiré après qu'on a découvert que le jury était composé de votre grand-père et de vos cousins germains, Gontrand et Bilbo.
_Il va répondre. Elle enchaîne._
Dans son édition de janvier 2004, le magazine _Cosmopolitan_ dressait la liste des dix hommes les plus sexy — liste dont vous ne faites pas partie.
_Il s'apprête à répondre. Elle enchaîne._
Et on pourrait continuer longtemps comme ça, mais je suis pas sûre qu'on ait envie d'en apprendre plus. Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous m'enchantez en me faisant découvrir Earlton, dans cette magnifique région du Québec, le Témiscamingue.
###### Gregory Rotheingham
Non, non, Annie-France, vous vous pensez de l'autre côté du lac.
###### Annie-France Brazeau
Ah oui, en Ontario, je vois. Juste en banlieue de Toronto. Qu'est-ce que vos études approfondies nous apprennent sur cette région, prolifique Gregory?
###### Gregory Rotheingham
Eh bien, voyez-vous, Annie-France, nos recherches ont identifié un marché qui est complètement inexploité et qui pourrait, ici même à Earlton, être le cœur d'une nouvelle tendance au niveau touristique : le gros tourisme.
###### Annie-France Brazeau
Ah oui, l'agrotourisme, c'est tout à fait charmant et bucolique. Je me souviens d'une belle expérience en fin d'après-midi en banlieue de Montréal. Vivre comme nos grands-parents; récolter les légumes biologiques, cueillir des vers de terres, ah!
###### Gregory Rotheingham
Ben non, madame Brazeau, pas l'agrotourisme, ça tout le monde le fait déjà. Non, nous, on va faire le « gros tourisme » : du tourisme destiné aux gros mangeurs.
###### Annie-France Brazeau
Et quel est le titre du rapport que vous venez nous présenter?
###### Gregory Rotheingham
_en montrant le rapport_
_À Earlton, gâtons les gloutons aux gueuletons et gagnons le motton._
###### Annie-France Brazeau
Étonnons... Euh? Étonnant! Du tourisme engraissant pour engraisser vos profits. Ça donne une nouvelle dimension au slogan « _Think big_ ». Elvis Gratton serait fier de vous, Gregory. C'est d'ailleurs le seul qui va être fier de vous.
###### Gregory Rotheingham
Le Témiscamingue québécois aussi. Eux, ils ont déjà les « Foires gourmandes », nous, on aura les « Gourmands en foire ». Vous savez, depuis trente ans, les études parlent du fléau de l'obésité. Nous, on voit ça comme un choix de vie et une clientèle sans cesse grossissante. Bref, du gros _cash_.
###### Annie-France Brazeau
Johnny Cash n'était pas gros. Elvis, je l'admets! Mais pas Johnny Cash! Ah, je viens de comprendre.
###### Gregory Rotheingham
Enfin!
###### Annie-France Brazeau
Mais ça n'empêche pas que votre projet va à l'encontre de toutes les tendances actuelles. À vous entendre, le bon docteur Atkins doit se retourner dans sa tombe. Remarquez que s'il se retourne une couple de fois, il pourrait maigrir davantage.
###### Gregory Rotheingham
Parlons-en du docteur Atkins. Il a ostracisé tous ces gens « bien enveloppés » avec ses messages dénigrants. À Earlton, on va les accueillir, ces touristes au portefeuille aussi volumineux que leur tour de taille. Ici, fini le _Guide alimentaire canadien_ , place au _Guide alimentaire nord-ontarien_!
###### Annie-France Brazeau
Le _Guide alimentaire nord-ontarien_ , qu'est-ce que c'est?
###### Gregory Rotheingham
C'est un de mes projets. J'ai identifié quatre grands groupes alimentaires : les produits pannés, les colas et leurs substituts, les matières grasses, et les légumes et les frites.
###### Annie-France Brazeau
Ah, il y a quand même des légumes.
###### Gregory Rotheingham
Oui. En fait, c'est du ketchup. Mes recherches m'ont amené à découvrir que c'est fait à base de tomates.
###### Annie-France Brazeau
Vos recherches?
###### Gregory Rotheingham
Oui. Bon... J'ai lu les ingrédients.
###### Annie-France Brazeau
Avec de tels projets, tous les grands guides de voyages vont vous tomber dessus à bras raccourcis.
###### Gregory Rotheingham
Vous savez, on ne recherche pas la clientèle qui court les restaurants quatre fourchettes. De toute façon, je me demande quel genre de plats ils servent pour avoir besoin de les manger avec quatre fourchettes!
###### Annie-France Brazeau
Euh? Beuh...
###### Gregory Rotheingham
_indigné_
Oui, mais y avez-vous pensé? Quatre fourchettes! Pour manger avec tes quatre fourchettes, y faut que tu te fasses greffer au moins deux autres bras!
###### Annie-France Brazeau
Calmez-vous, Gregory. On va vous inscrire à un cours de restauration. Revenons sur la planète Terre, monsieur quatre-bras, monsieur la demi-pieuvre. Expliquez-moi plutôt, Gregory, en quoi vos forfaits seront différents des autres forfaits touristiques.
###### Gregory Rotheingham
Eh bien, tout d'abord, on va construire un nouvel hôtel qui sera énorme!
###### Annie-France Brazeau
Est-ce que c'est cet hôtel que je vois sur vos feuilles?
###### Gregory Rotheingham
Tout à fait.
###### Annie-France Brazeau
Je pense que vous devriez changer de nom, ça ne m'apparaît pas trop prometteur d'appeler un hôtel _Le No Vacancy_.
###### Gregory Rotheingham
C'est un détail. On parle d'un hôtel avec des lits renforcés, des chaises de trente-six pouces de large pour que l'ensemble du fessier soit bien appuyé. Et au menu du restaurant de l'hôtel, le spécial Wilf Paiement : entrée de poutine, ragoût de pattes et _cheese-cake_ sans un regard de travers.
###### Annie-France Brazeau
Une douzaine d'œufs et une livre de bacon pour déjeuner, j'imagine?
###### Gregory Rotheingham
Non, non. Pas plus que deux œufs. un œuf d'émeu et un œuf d'autruche.
###### Annie-France Brazeau
Oui, c'est bien beau de loger et de nourrir vos volumineux touristes, mais à part manger, qu'est-ce que vous leur faites faire?
###### Gregory Rotheingham
Ah, plein de choses! Bon, il faudra adapter un peu. Par exemple, on a ici une ferme laitière ultramoderne avec un carrousel pour traire jusqu'à douze vaches. On va installer des sièges et on aura un carrousel adapté aux dimensions de nos touristes!
###### Annie-France Brazeau
C'est grossier!
###### Gregory Rotheingham
Et la tuyauterie qui sert habituellement à la traite sera modifiée en distributeurs : un pour la crème fouettée, un autre pour le foie gras, un pour le creton en crème, _et cetera_.
###### Annie-France Brazeau
C'est dégradant! Bon, encore de la nourriture. J'imagine que c'est présent à toutes vos activités?
###### Gregory Rotheingham
Oh oui! C'est au cœur de nos forfaits touristiques. Vous savez que les épluchettes de blé d'Inde sont très populaires. Ben nous, on va faire des épluchettes de livres de beurre.
###### Annie-France Brazeau
Oui, saupoudré de quelques grains de maïs! Voyons, toutes vos activités ne peuvent pas être destinées qu'aux gros mangeurs. Il doit bien y avoir des activités grand public.
###### Gregory Rotheingham
Absolument! Au mois de juillet, on s'intègre au Earlton Steam Show, cette exposition d'équipement agricole à vapeur. Alors, au lieu de gaspiller la précieuse vapeur des engins, on va l'utiliser pour _steamer_ des hot-dogs.
###### Annie-France Brazeau
Voyez, encore de la nourriture!
###### Gregory Rotheingham
À chaque année, on accueillera une grande compétition de lutte sumo. Enfin un endroit où ces formidables athlètes se sentiront chez eux. Ils seront un modèle pour les jeunes touristes qui hésitent à prendre du poids.
###### Annie-France Brazeau
Mais c'est tout à fait criminel d'encourager les enfants dans la voie de l'obésité! Et vous n'avez pas peur qu'au milieu des volumineux touristes, vos lutteurs de sumo n'aient l'air un peu chétifs?
###### Gregory Rotheingham
Non, ils apprécieront la chance de passer incognito, à part le fait qu'ils sont Japonais.
###### Annie-France Brazeau
Mais, dites-moi, vos projets comprennent-ils de grands happenings?
###### Gregory Rotheingham
Absolument! Il existe déjà le concours de beauté Miss Monde. Nous, on accueillera annuellement le concours international Miss Gloutonne d'Earlton.
###### Annie-France Brazeau
Le concours de beauté des femmes de forte taille.
###### Gregory Rotheingham
Tout à fait! Mais jamais plus d'une concurrente à la fois sur scène!
###### Annie-France Brazeau
Mais sérieusement, Gregory, votre concours, Miss Gloutonne d'Earlton, va-t-il bêtement récompenser la plus grosse mangeuse?
###### Gregory Rotheingham
Non, non. On va aussi s'intéresser à la personnalité des concurrentes. On va leur poser des questions et on va leur donner une note sur 100.
###### Annie-France Brazeau
Des questions. Comme quoi?
###### Gregory Rotheingham
Euh... Êtes-vous pour ou contre la crème 35 %?
###### Annie-France Brazeau
POUR! Euh? CONTRE! J'ai échoué... Combien me donnez-vous?
###### Gregory Rotheingham
Hum... 35%.
###### Annie-France Brazeau
Finalement, c'est pas tellement différent des autres concours de beauté; je ne gagne jamais. Mais là, dites-moi, Gregory, on est à la campagne, allez-vous profiter des attraits de la nature?
###### Gregory Rotheingham
Bien sûr, on va en offrir de la nature et en gros format à part ça. Comme tout sera plus gros à Earlton, on va créer un jardin botanique où on va exposer des bonsaïs géants.
###### Annie-France Brazeau
Des bonsaïs géants? Ça existe déjà, Gregory. On appelle ça des arbres. Vous voulez créer une forêt?
###### Gregory Rotheingham
Et c'est pas tout : on va transformer le zoo d'Earlton. On va se débarrasser des animaux blessés qu'on soigne au centre pour accueillir des éléphants, des hippopotames, des grizzlys.
###### Annie-France Brazeau
Gigantesque projet. Des baleines avec ça?
###### Gregory Rotheingham
Non, les baleines, on va les mettre dans le grand trou de la mine Adams. Surtout que les gens ont refusé mon dernier plan, de remplir le trou avec des déchets, qui était pourtant une très bonne idée1.
###### Annie-France Brazeau
En effet, espérons que votre projet tient mieux compte des sensibilités locales que ce projet de dépotoir, qui a tant soulevé de controverse dans la région.
###### Gregory Rotheingham
N'ayez crainte, Annie-France. Le gros tourisme, ça va passer comme un couteau dans le beurre!
###### Annie-France Brazeau
C'est tout le temps que nous avons. Gregory Rotheingham, merci.
###### Gregory Rotheingham
Je vous en prie. Allez, Annie-France, je vous offre un beigne à l'érable. Prenez, prenez. J'insiste.
###### Annie-France Brazeau
Ah, et pourquoi pas!
1. À la fin des années 1990 et au début des années 2000, les gens du Témiscamingue ontarien ont lutté vigoureusement contre le projet d'un consortium qui voulait acheminer les déchets de Toronto par train pour les enfouir dans cette mine à ciel ouvert située à Kirkland Lake.
###### Printemps du nord
_Pas de photos, pas de champagne_
_Ni tour de piste glorieux_
_Pour couronner cette campagne_
_Un autre a couru mieux_
_Plus une voix pour l'acclamer_
_Dans l'hippodrome silencieux_
_Ils ont emporté le trophée_
_Un autre a couru mieux_
_Printemps du Nord était son nom_
_Issu d'une lignée de champions_
_Porteur d'espoirs et d'ambitions_
_Distance, vitesse et position_
_Printemps du Nord_
_Dans l'hippodrome déserté_
_Le lourd silence n'est pas entier_
_Résonne son trot saccadé_
_Et le roulement de son sulky_
_Le mille en une minute cinquante_
_Il va le faire, il l'a rêvé_
_Il aura son heure triomphante_
_Il va y arriver_
_Et seul en piste le cœur battant_
_Il tient la gauche au dernier droit_
_Rien à son flanc et rien devant_
_Ce sera ça, son jour de joie_
_Distance, vitesse et position_
_Porteur d'espoir et d'ambition_
_Issu d'une lignée de champions_
_Printemps du Nord était son nom_
_Printemps du Nord_
###
### Salut Iroquois Falls!
Dès ses débuts, le destin de cette communauté chaleureuse où il fait bon vivre était couché sur papier. Car Iroquois Falls est une de ces petites villes nord-ontariennes qu'une grande papetière a mise au monde. C'est une ville de compagnie, comme on dit, prévue et planifiée, comme en témoignent les rues courbées bordées de maisonnettes coquettes du secteur résidentiel, qui ont dans leur diversité un air de parenté. C'est ici que l'Abitibi Pulp and Paper Company est née en 1912. La naissance de la compagnie a donc accompagné la naissance de la ville.
Iroquois Falls a bien failli devenir une ville fantôme. L'Abitibi projetait de fermer sa grande machine numéro 1, la dernière de ce moulin, malgré le fait que le syndicat des employés, la CEP (le Syndicat canadien des communications, de l'énergie et du papier), était prêt à l'acheter. On a eu la frousse, mais on a évité le pire. Depuis ce temps-là, la ville brasse des idées de diversification économique. On veut produire de l'alcool de glace. On veut chauffer des planches d'épinette pour leur donner l'allure du cèdre. On en explore, des idées originales. Car, avant cette frousse-là, d'autres mises à pied étaient déjà venues friper la petite ville du papier. Des six mille personnes qui vivaient ici en 1991, il en reste environ 5000 aujourd'hui. De ceux-là, presque la moitié sont des Franco-Ontariens et à ce qu'on me dit, presque tous sont connus par leur sobriquet. un de ceux-ci est « Bine ». C'est le sobriquet d'un personnage dont l'engagement civique défie la logique économique aveugle qui perd le Nord. Bine, c'est le plus célèbre des personnages du Grand Nord. Car Bine, c'est le père Noël!
###### « Bine » Lachappelle
Le père Noël n'habite pas le pôle Nord. Il habite Iroquois Falls, Ontario. Puis c'est pas juste moi qui le dis, c'est un garçon de six ans qui habite ici lui aussi et qui est un spécialiste en cette matière, comme le sont tous les enfants de six ans. une fois, ses parents l'ont amené voir le père Noël à Timmins. Mais il l'a dit à sa mère : « C'est pas lui le vrai père Noël. Le vrai, c'est celui qui est chez nous à Iroquois Falls. » Il en est bien sûr parce qu'il l'a bien vu, et pas juste une fois non plus.
Le père Noël d'Iroquois Falls vient même voir les enfants l'été sur la plage du parc, où il accoste en chaloupe pour le Noël en juillet, enveloppé dans son costume rouge et sa barbe blanche par une chaleur de 27 degrés. Il leur explique, aux enfants, qu'il a un télescope grâce auquel il les observe pour s'assurer qu'ils sont gentils. C'est comme ça qu'un bon jour, un autre garçon de quatre ans d'Iroquois Falls a reçu un coup de téléphone du père Noël, qui lui a dit :
— Ho! Ho! Daniel, je vois avec mon télescope que tu veux jamais rien manger. Il faut que tu manges pour grandir. Tu sais, ta maman a bien peur que tu finisses par tomber malade si tu manges pas.
— OK, père Noël... Oui, je vais essayer... Oui, je le promets.
Vous voyez, le père Noël d'Iroquois Falls connaît l'histoire de chaque enfant en ville. Et tous les enfants de la ville sont là sans faute, à bord du train spécial qui, chaque année, part d'Iroquois Falls et roule jusqu'à Porquis. Ils s'en vont chercher le père Noël, qui les attend debout dans la forêt, près de la voie ferrée. C'est la fête à bord, vous pouvez en être sûr, sur le chemin du retour.
Mais en plus de ça, certains enfants ont même la chance de l'apercevoir chez eux, dans leur cour, à leur porte. une année, il avait bien vu avec son télescope une famille qui n'avait pas les moyens d'acheter des cadeaux de Noël. Donc, dans la nuit de Noël, il est allé leur en porter. Puis il n'a même pas chicané la fillette de quatre ans si excitée de voir le père Noël arriver chez elle à bord d'une motoneige qu'avant qu'on puisse l'en empêcher, elle est sortie en courant, petits pieds nus dans la neige.
Parfois le père Noël d'Iroquois Falls amenait ses propres enfants à lui, costumés en lutins, dans ses tournées de Noël. Ça leur faisait un Noël assez différent de celui d'autres enfants. Leur Noël, ils le donnaient aux autres enfants. une année, la mère du père Noël est décédée juste avant le temps des Fêtes. Il avait le moral bas, ça se comprend. Il n'avait plus tellement envie d'être père Noël. Mais il avait promis d'aller voir un enfant malade du cancer. Et il a tenu sa promesse.
Ce ne sont pas juste les enfants qui aperçoivent le père Noël d'Iroquois Falls. Les adultes le voient aussi. Il est le père Noël de l'hôpital. Il est le père Noël de la police. Il est le père Noël du moulin pour les fêtes d'employés. Il est le père Noël du bazar annuel de sa paroisse Saints-Martyrs-Canadiens et d'autres aussi. Il est le père Noël de toutes les écoles. Il est le père Noël de presque tous les rassemblements de familles, même celles qu'il ne connaît pas tant que ça. Parce que tout le monde sait que ça ne serait pas la même fête sans lui. Comprenez-vous, il est le père Noël d'Iroquois Falls.
Un soir, la chorale qui répétait à l'église l'a vu entrer tout essoufflé dans ses bottes noires et son costume rouge, en retard pour la répétition. Entre deux de ses apparitions de père Noël, il tenait absolument à se préparer pour chanter l' _Adeste fideles_ à la messe de minuit. Il a repris son souffle. Il a répété son chant sacré. Puis il est reparti en coup de vent comme il était arrivé. Il n'avait même pas pris le temps d'enlever son costume.
Les adultes vous diront aussi que le père Noël d'Iroquois Falls a tendance à faire des migraines. Donc quand le gros _rush_ du temps des fêtes approche, il doit faire un petit sacrifice et cesser de boire autant de Pepsi. Ils vous diront aussi que le père Noël d'Iroquois Falls ne chante pas juste des chants sacrés. Il connaît aussi des chansons salées. Il faut le voir chanter « Envoye! envoye! la p'tite jument! » en se roulant par terre pendant les soirées de fête. Ça vous met les larmes aux yeux. Moi, je ne l'ai jamais vu faire ça. Mais j'ai ri aux larmes la fois que je l'ai vu gagner haut la main le concours du plus grand conteur de blagues du Nord de l'Ontario, à Génier, en 1998.
Donc voyez-vous, moi aussi j'ai aperçu le père Noël d'Iroquois Falls, même si je ne vis pas ici. La seule et unique fois que je l'ai eu en entrevue pour la radio, c'était pour aider la vente des ampoules du grand arbre de Noël devant l'hôpital général Anson, ces petites lumières à la mémoire de nos proches disparus. C'est là que, par accident, il m'a révélé son secret, sans faire exprès. Il m'a confié qu'il était le père Noël d'Iroquois Falls. Et laissez-moi vous apprendre que cet homme-là, le clown des parties, le boute-en-train de toutes vos fêtes, le gros comique qui a de la joie de vivre plein son casque, eh bien, mettez-le devant un micro de la radio et... il est gêné comme un pet! En revanche, placez-le devant un appareil photo avec son casque dur pour sa photo de carte d'identité à l'usine de pâte et papier, puis là, qu'est-ce qu'il fait? Eh bien, il enlève son dentier, pour faire une belle photo professionnelle.
Étant donné qu'il est le père Noël, il a le pouvoir d'être partout et de connaître tout le monde. On dit qu'il a un genre de don : celui de voir les gens par en dedans, par les sentiments. C'est comme ça qu'il sait toujours qui a besoin de son aide. Mais il ne sait jamais accepter la moindre cenne noir en paiement pour ça. Même sans son costume, il est toujours le même. On l'a aperçu dans l'équipe des repas communautaires qu'on prépare pour les familles qui ont des funérailles et où il sait remonter le moral de tout le monde. On l'a aperçu au restaurant chinois, derrière le comptoir, en train d'aider à verser le café quand la serveuse est débordée tant la place est occupée. On l'a aperçu en train de peindre les murs d'un autre restaurant, trois jours de suite, à l'étonnement de la commerçante, nouvelle en ville, qui avait bien besoin d'aide pour ouvrir ses portes le jour prévu.
On l'a aussi aperçu assis en auto aux côtés d'une jeune femme qui voulait apprendre à conduire, mais qui n'y arrivait pas avec d'autres, tant elle était nerveuse. Eh bien, avec lui elle a réussi, et si bien que le mot s'est passé. Au fil des années, il a ainsi dû apprendre à conduire à au moins la moitié de la jeunesse d'Iroquois Falls. Il s'était informé un peu pour voir s'il pouvait en faire un commerce, mais c'était compliqué. Alors, il a tout simplement continué bénévolement.
Et quand, un bon jour, le père Noël a eu besoin d'un gros cadeau pour un membre de sa propre famille, alors là, son aide donnée aux autres lui a été rendue. Son petit neveu gravement atteint de la maladie de Crohn avait besoin de traitements expérimentaux à Toronto, des traitements très coûteux et non couverts par le régime de santé publique. Eh bien, le père Noël d'Iroquois Falls, qui connaît tout le monde, a littéralement fait jouer ses contacts. Il a réuni sept groupes musicaux locaux qui ont fait ensemble un grand spectacle-bénéfice. La soirée a rapporté huit mille dollars, parce que bien sûr, une bonne partie de la ville s'est déplacée pour l'appuyer. D'habitude, notre homme aurait certainement su remercier tout le monde abondamment. Mais à la fin de cette soirée-là, sur scène avec son neveu malade, l'émotion était si vive que dans le micro, il a à peine pu dire deux mots.
Mais la fois où Iroquois Falls a le mieux remercié son père Noël n'a pas été celle-là. C'est arrivé lors d'une mémorable soirée organisée exprès pour lui. Ce soir-là, 600 personnes se sont entassées serré dans la grande salle syndicale qui normalement n'en contient qu'un maximum de 400. Il n'y restait plus au centre qu'un étroit passage d'un mètre de large pour que l'invité d'honneur se rende sur scène, sous un tonnerre d'acclamations. une petite ville du Nord rendait hommage à un homme en qui elle voyait le symbole de ce qu'il y a de plus grand chez elle et en elle. Les gens d'Iroquois Falls l'ont fait pleurer, le père Noël, ce soir-là. Il répétait qu'il ne comprenait pas qu'on doive le fêter, puisqu'il ne se croit pas si différent des autres. Et au fond, il avait un peu raison.
Iroquois Falls a sa bonne part de gens de valeur, des gens dont l'engagement professionnel et le bénévolat sont des richesses humaines aussi précieuses que des richesses naturelles. C'est une source d'énergie aussi puissante que des chutes sur la rivière Abitibi. On met des villes au monde quand on sait la harnacher. Cette énergie est si puissante qu'annuellement, depuis dix ans, des douzaines d'hommes d'Iroquois Falls, bien virils et bien en chair, s'habillent en robe, en collants, en soutien-gorge et en perruque. Ils se maquillent, se font belles et font la parade sur scène, juste pour financer l'autobus adapté au transport de personnes en fauteuil roulant. On me dit que la soirée est toujours d'une beauté troublante.
C'est par ce genre d'esprit communautaire que le Nord ontarien compense aujourd'hui et qu'il vaincra demain ses problèmes de région dite périphérique, que les grandes métropoles siphonnent et que les grandes tendances économiques défavorisent. Cette petite ville du papier a un cœur d'or et mérite un avenir plein d'argent. Elle saura d'autant mieux le faire advenir, ce grand avenir, qu'elle a su reconnaître à présent un homme qui symbolise son grand cœur. Je vous parle, depuis le début, de celui que tout le monde ici appelle Bine — qui, en passant, a été cette année couronné reine de la soirée des femmes. Selon moi, ce modeste héros local mérite aussi le titre de héros régional. Son nom est Denis Lachapelle. On l'appelle Bine. Il est le père Noël d'Iroquois Falls.
###### Petite revanche
_Que les desseins de nos vies ne nous ressemblent plus_
_Que les forêts ne cachent que les morts du passé_
_Que les chutes soient fantômes dans un corps passager_
_Que les hommes n'arrivent plus à la porte fatigués_
_Que la deux, que la trois, que la quatre ne tournent plus_
_Qu'aucun mot, pas un sacre n'accompagne le café_
Qu'il ne reste que des miettes d'amertume à cracher
_Que nous aurions encore juste à tout endurer_
Ah oui, je resterai
Je ne partirai pas
Je n'oublierai pas
Ah non, ah non, je ne partirai pas
_Que le temps à tuer ne passe plus comme avant_
_Que les heures et les jours coulent et meurent sans promesses_
_Que la fin n'arrive plus comme elle était prévue_
_Qu'il n'y ait plus vraiment récompense à la fin_
Ah oui, je resterai
Je ne partirai pas
Je n'oublierai pas
Ah non, ah non, je ne partirai pas
_Pour Lauzon, Camiré, Lagacé et Boucher_
_Pour ti-Guy, ti-Pit, ti-Gus et ti-Diable_
_Pour les boss, les foremen, l'overtime, les jobs sales_
_Pour les hommes qui ont rêvé de partir sans le pouvoir_
Ah, je resterai
Je ne partirai pas
Je n'oublierai pas
Ah non, ah non, je ne partirai pas
###### Sabine au fond des bois
Iroquois Falls! Je croyais y passer quelques jours, puis repartir vers la grande ville, Toronto. Est-ce le hasard qui a tout changé? Fait-il bien les choses? À vous de juger. Moi, je vais simplement vous conter ce qui m'est arrivé.
Quand je suis venue ici, il y a deux mois, je m'attendais à pouvoir admirer des chutes. Le premier soir, j'ai demandé au propriétaire du motel où elles étaient. D'abord, il n'a rien compris, puis, il m'a appris que, hautes de quinze pieds, elles avaient été en 1912 domptées par un barrage et avaient donc disparu. « Pas grave, ajouta-t-il en voyant mon visage déçu, une usine hydroélectrique et une papetière, à l'époque, ça valait bien les chutes. » Moi, je trouvais que le nom de la ville ne tenait point sa promesse.
Il y a eu une bonne surprise aussi : à Toronto, les gens m'avaient dit que les moulins à papier empestent l'air. Or, je ne sais pas comment Abitibi Consolidated se débrouille, mais ça ne sent pas mauvais à Iroquois Falls. Bravo!
Bien sûr, je n'étais pas venue seulement pour voir des chutes inexistantes et examiner la qualité de l'air. Non. J'avais terminé mon B.A. en histoire et j'avais trouvé un sujet de thèse de maîtrise qui semblait me convenir à merveille. une de mes profs, une femme un peu folle, mais intéressante, avait accepté que je fasse des recherches en vue d'une thèse sur la signification historique, sociale et politique des magasins de vêtements usagés. J'avais alors décidé de visiter les villes du Nord de l'Ontario afin de déterminer si elles avaient de tels magasins et s'il y avait des différences marquantes entre ceux-ci et les magasins de Toronto.
Je connais bien les magasins de vêtements et d'objets usagés. Tout ce que je possède, meubles, équipement de cuisine et de sport, tout mon habillement vient de là. Robes, jupes, t-shirts, blouses, vestes et manteaux, je fouille dans les rayons et trouve toujours de quoi ajouter à ma collection. Je fouille et j'achète, je ne peux pas résister, mes placards débordent! J'ai l'œil pour les bonnes aubaines. Regardez-moi donc cette jupe trois dollars, le châle deux dollars... Et le pull, pure laine, qu'en dites-vous, deux dollars quatre-vingt-quinze? C'est au Value Village de Timmins que je l'ai trouvé. Je vous le recommande, ce magasin-là! Surtout qu'il n'y en a pas à Iroquois Falls, de Value Village. Alors, un petit voyage, ça vaudrait peut-être le coup?
J'ai passé les premières semaines de mon été à arpenter — dans une vieille voiture, cadeau de ma mère — cet arrière-pays si beau et si tranquille. Je me vois encore. Moi, la citadine, en train de traverser l'océan vert des forêts sans fin, les yeux grands ouverts devant l'immense ciel bleu et ses merveilleux nuages. Le bruit des pneus sur l'asphalte des routes me rappelait le bruit des vagues que l'on entend lorsqu'on est couché sur une plage. Je me sentais une âme de surfeur debout sur sa planche, alors que j'étais assise dans cette vieille voiture. J'essayais de ne pas trop rêver afin de maintenir l'équilibre entre la vitesse et l'espace qui semblaient m'emporter.
J'aurais voulu voir un orignal, un de ceux que les panneaux annoncent à intervalles réguliers. Je n'en ai vu aucun. Disparus, eux aussi? L'orignal anime-t-il le mythe d'un Nord soi-disant inaccessible? Mais, jeune de mes vingt-trois ans, j'étais heureuse. J'étais sûre que l'été serait merveilleux. Et c'est vrai... Il y a eu Antoine.
Mais je m'égare... J'ai tendance à faire ça. À parler d'une chose, puis à prendre un chemin de traverse, raconter autre chose. Je suis navrée... Excusez-moi. Voilà, retournons à Iroquois Falls.
Au fond, pourquoi iroquois et pas ojibwé? Dans le prospectus que j'ai trouvé dans ma chambre, ça disait que les Iroquois avaient la méchante habitude de s'attaquer à d'autres tribus. Après une invasion d'un paisible village ojibwé près de Sturgeon Falls, ils auraient forcé quelques survivants à les guider vers un village plus au Nord. Pas si bêtes, ces guides! une fois sur la rivière, ils ont, à l'approche des chutes, sauté à l'eau en emportant les pagaies, laissant les Iroquois mourir la mort violente qu'ils méritaient. une autre version — féministe — conte qu'une vieille femme ojibwée aurait conduit les guerriers iroquois dans leur canoë vers les chutes, aurait au moment décisif lancé les pagaies dans l'eau et aurait ri très fort en mourant avec l'ennemi. Je la vois d'ici, cette femme, trouvant sa mort dans la joie de la victoire! Youpi!
Des Ojibwés paisibles, des Iroquois agressifs... Et, tenez-vous bien, en 1906, trois immigrants juifs y ont péri. Il faut ajouter Pierre de Troyes à notre liste, même si lui ne s'est pas tué ici en 1686, s'il est seulement passé par là en allant avec cent soldats et un aumônier vaincre les _British_ à la baie James. Elle est chargée, l'histoire d'Iroquois Falls!
Bon, revenons au présent. Quand on regarde pour la première fois le plan de la ville, on est stupéfait. Vraiment, je n'exagère pas! Vous connaissez des villes canadiennes qui possèdent une rue Synagogue? Que je sache, il n'y en a pas à Toronto ni à Montréal. La rue Synagogue, sur le plan de la ville d'Iroquois Falls, ça saute aux yeux!
J'aurais aimé la trouver, cette synagogue! Où? Rue Synagogue, pardi! Mais, incapable de trouver l'édifice, je me suis adressée à un monsieur, un grand aux cheveux gris, assez vieux, qui venait de descendre de sa bicyclette. Il a rajusté ses lunettes en me regardant. La synagogue? Ben, elle n'existe plus, remplacée par une station-service dans les années soixante. Disparue, tout comme les chutes!
Il était gentil, ce monsieur, un ancien curé. Il m'a offert la chaise pliante qu'il transportait sur son porte-bagages, je me suis assise, il s'appuyait contre sa bicyclette, puis il m'a fait tout un discours sur les Juifs à Iroquois Falls. C'est en 1904 que les premiers, Louis Abramson et sa femme Mooshe, les Kideckel, les Crotin et les Perkus quittent la Russie antisémite pour s'installer à Krugerdorf, où le gouvernement fédéral leur donne des terres. D'autres les ont suivis, sont allés vivre à Englehart pour participer à la construction du chemin de fer — ça leur rapportait un dollar par jour — ou bien pour travailler la terre, eux aussi. Terre fertile, familles fertiles! Mooshe et Louis ont eu treize enfants et Mooshe, sage-femme, a assisté en l'absence de médecin à plus de cinquante autres naissances.
La ville a prospéré grâce à Frank H. Anson, fondateur de la papetière. En 1925 déjà, nos immigrés ont des maisons avec électricité et eau courante, des salles de bains. Ils ouvrent des magasins, rue Railway, à ce qui s'appelle alors Ansonville. Il y a une synagogue, une petite école où les jeunes apprennent l'hébreu, un théâtre. Et ils sont amis avec tout le monde, francophones ou anglophones! La vie est belle, Iroquois Falls devient « la ville modèle du Nord ».
Ah, j'oublie de vous raconter l'histoire des trois Juifs morts dans les chutes! Le 26 juin 1906, Ben Perkus et son fils Morris quittent Krugerdorf, vont en canoë jusqu'aux chutes, les contournent en faisant du portage, continuent en canoë, puis font le reste du chemin à pied, jusqu'à la gare d'Englehart. Ils vont accueillir des amis qui arrivent de Russie, via Montréal. On s'embrasse, on dîne chez des amis, on boit un verre. C'est la fête.
Pour le retour, un jeune Juif anglais, qui avait envie d'aller voir Krugerdorf, se joint à eux. Les cinq embarquent. Hélas, quand ils arrivent au centre de la rivière, un très fort courant saisit le canoë. Ils pagaient comme des fous pour regagner la rive, mais, impétueux, le courant les emporte vers les chutes. Les nouveaux arrivés ont de la chance : projetés à l'eau, ils réussissent à nager vers la terre ferme. Les trois autres périssent.
Douze jours plus tard on retrouve leurs pauvres corps au bord de l'eau, à cinq kilomètres au nord. un fermier, dont les terres touchent ici la rive, offre un terrain pour les inhumer. Ainsi est créé le cimetière hébraïque de Krugerdorf, qui existe encore aujourd'hui, à une centaine de kilomètres au sud-est d'Iroquois Falls. Entouré d'arbres, loin de tout, loin aussi des antisémites qui vandalisent les cimetières juifs, il est difficile à trouver — j'en ai fait l'expérience! — surtout qu'il n'y a aucun panneau indicateur sur la route.
D'après le curé, il reste aujourd'hui une seule famille juive à Iroquois Falls. C'est eux qui tiennent le petit magasin de bonbons et de livres d'occasion, rue Ambridge. Sarah et Max, des juifs conservateurs, non, progressifs, je dirais. Max rit des fondamentalistes américains tout en pesant mon sachet de _gummibears_. Sarah hausse les sourcils quand son mari avance des propos athées. Parti vers des climats plus cléments, leur fils, devenu orthodoxe, enseigne les études juives dans une Université américaine. Sera-t-il enterré à Krugerdorf?
Ils me montrent une photo de ce fameux cimetière et, tout à coup, je décide d'y aller.
Je le leur dis, ils me regardent avec intérêt.
— Elle est Juive, s'écrie Max.
Je proteste. Je ne veux pas m'approprier ce qui est le leur.
— Vous ne le trouverez pas, ce cimetière, me dit-il et sa femme renchérit
— Deux heures de route, précise-t-elle. Il se fait déjà tard. Attendez au moins jusqu'à demain.
À eux deux, ils m'expliquent le chemin. Direction sud-est. Le cimetière est sur la route 11, entre Iroquois Falls et Kirkland Lake, là où la 11, la 101 et la 560 forment une sorte de triangle. J'ai déjà la tête qui tourne. Devrais-je leur demander de me faire un dessin? Mais j'ai l'obscur pressentiment qu'une telle requête ne serait pas la bienvenue. Il faut que je cherche l'endroit sans trop d'assistance et même sans assistance aucune. Trouver le cimetière des Juifs du Nord, c'est mon but, ma tâche, ma quête.
Inutile de vous dire que le lendemain, ça a été la catastrophe. Deux heures de route, avait-elle dit? Trois heures et je tournais toujours en rond, je ne savais plus si j'étais sur la 11 ou sur la 101, la 560 peut-être, je ne savais plus où était le nord, où était l'est. Quel triangle? Cramponnée à mon volant, je me croyais dans un labyrinthe.
Midi. La faim me tiraille et je n'ai rien à manger. Que faut-il faire quand on est perdu?
J'arrête la voiture. Je descends. Je regarde autour de moi. Je cherche un chemin qui m'amènerait vers le lieu sacré, à pied s'il le faut. Rien. Zéro. Mais voilà, je vois une bicyclette, de l'autre côté de la route. Serait-ce celle du curé? Elle me semble plus neuve. Mais ça doit être lui. un curé, ça peut posséder deux bicyclettes, non? Et qui d'autre qu'un croyant fidèle abandonnerait un vélo en parfait état dans un fossé? Pourtant, holà, cent kilomètres à bicyclette, n'est-ce pas trop pour un monsieur d'un certain âge?
Je m'approche et — y a-t-il encore des miracles? — voici l'entrée du cimetière! Vaste, solennel, comme intouchable. une sorte de clairière où s'alignent des pierres tombales grises et roses, marquées de lettres que je ne saurais déchiffrer. Vais-je oser y pénétrer? Ai-je le droit de poursuivre ceux et celles qui ont trouvé refuge ici? Au loin, un oiseau se manifeste.
J'ai tellement envie de partager mon émotion. Je regarde autour de moi, je cherche le propriétaire de la bicyclette. À gauche de l'entrée, je vois une petite tente. Entre deux arbres, une corde à linge. un t-shirt et un jeans y prennent le soleil.
Et voilà que je vois un homme couché dans l'herbe, immobile, mais pas mort comme tous ceux que l'on a amenés ici, non, il est tout simplement en train de faire un petit somme. Je vois qu'il est nu, ou presque. Une petite serviette protège son sexe des regards indiscrets. L'inconnu est jeune, il a les cheveux bruns, bouclés. Décidément, ce n'est pas le curé. J'aurais envie de me déshabiller moi aussi, de me coucher auprès du beau dormeur. Mais je n'ose pas.
Près de lui, un cahier à dessin. Je m'assois, tout doucement, je ne veux pas le réveiller, mais je suis curieuse, je veux voir ce que cet homme a dessiné. Lentement, sans faire de bruit, je tourne les pages. Des croquis au crayon, à la plume, au fusain. Quelques aquarelles. Des arbres, des roches, des nuages, des chevalements et encore des chevalements, squelettes sur des routes abandonnées. Et toujours de nouveau, le cimetière de Krugerdorf.
Je ferme le cahier. Antoine ouvre les yeux. « Bonjour, lui dis-je. C'est beau ce que tu fais. » Il me sourit. Il se lève, va chercher ses vêtements, s'habille.
— Et toi, qu'est-ce que tu es venue faire ici?
— Je cherchais le cimetière.
— Ah bon.
Antoine sort deux pommes de son sac à dos.
— C'est un peu le paradis ici, dit-il.
Il me tend une pomme.
— Et il n'y a pas de serpent et on ne va pas nous chasser.
Nous mangeons. Quand il a fini, Antoine prend son crayon.
— Tu permets?
Il ouvre son cahier, commence à dessiner. Je vois qu'il trace mon visage sur la page qui était ouverte lorsqu'il dormait. Sabine au cimetière israélite, Sabine au fond des bois.
Le mouvement de sa main, ses yeux qui me regardent, le bruit presque imperceptible du crayon sur le papier, tout cela est comme une douce caresse sur mes joues, ma bouche, mes tempes et mes cheveux. Je sais que pour une fois tout est vraiment pour le mieux dans le meilleur des mondes possibles, qu'ici, dans le refuge qu'ont trouvé des persécutés, je viens de trouver ce qui sera peut-être mon bonheur.
Ça vous irait, un petit retour en arrière? Oui? Merci... Avant de me donner la voiture, ma mère l'avait fait inspecter, avait fait refaire les freins. Le garagiste avait approuvé :
— Comme ça, votre fille sera sûre de pouvoir s'arrêter si jamais elle rencontre un orignal.
Maman avait ri de la remarque. Mais quand je lui ai parlé de ma rencontre avec Antoine, elle me l'a rappelée. « Sabine, m'a-t-elle dit, parfois il faut freiner un peu. »
Ne vous inquiétez donc pas pour moi, je ne suis pas folle. Je réfléchis à ce que je fais. _To fall in love,_ c'est beau, très beau. Mais parfois, ça se termine mal. Si Iroquois Falls n'a plus ses chutes, l'amour, vous le savez tous, en cache encore.
Pour commencer, j'avais besoin de faire quelque chose de raisonnable. J'ai décidé de changer de thèse. Il y a eu tout un échange de courriels avec mon prof, on s'est parlé au téléphone, puis elle est d'accord que j'abandonne les chiffons, que je me consacre à ce qui me fascine de plus en plus : l'histoire de ce groupe de Juifs du Nord si courageusement errants.
J'écrirai sur les Lévy, les Solomon, les Nosov, les Perlmutter, les Abramson et tous les autres venus de si loin! J'apprendrai l'hébreu avec Sarah, Max qui aime les vieux livres m'aidera dans mes recherches. Antoine, j'en suis sûre, me donnera bien quelques dessins pour illustrer le chapitre le plus important qui portera sur le cimetière de Krugerdorf.
Je resterai donc ici, au moins le temps de cette thèse. Après, on verra. Si tout va bien, vraiment bien, on s'achètera peut-être une petite maison, Antoine et moi. Il paraît qu'elles ne coûtent pas cher ici, les maisons. Et alors je demanderai à mon ami le curé de venir la bénir, pourquoi pas?
_Marguerite Andersen_
###### Incitatif pédant: Y'aura quoi
à Iroquois Falls?
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau, vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez écrit un rapport sur la communauté de Wawa, dans lequel vous préconisez l'ouverture de condominiums géants pour les chiens, rapport que vous avez intitulé _Tous les chiens font Wawa_.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
La rumeur populaire dit que vous avez vendu votre âme au diable, mais que celui-ci vous l'a retournée, prétextant un défaut de fabrication.
_Il va répondre. Elle enchaîne._
Vous imitez le pinson comme personne. Vous êtes somnambule...
###### Gregory Rotheingham
_surpris_
Ah! Vous me l'apprenez!
###### Annie-France Brazeau
On a des photos. Et je pourrais continuer longtemps comme ça, mais je pense que nous en savons déjà beaucoup trop! Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous venez nous présenter votre tout nouveau rapport, qui vise à donner un souffle nouveau à la ville d'Iroquois Falls. Comment s'appelle votre rapport, cette fois-ci, Gregory?
###### Gregory Rotheingham
Eh bien, mon rapport s'appelle : _Y aura qui? Y aura quand? Y aura quoi? À Iroquois Falls?_
###### Annie-France Brazeau
Et que suggérez-vous pour diversifier l'économie d'Iroquois Falls?
###### Gregory Rotheingham
Eh bien, Annie-France, ce que je propose aux gens d'Iroquois Falls, c'est de bâtir leur économie sur les ressources déjà disponibles dans la communauté.
###### Annie-France Brazeau
Les Iroquois?
###### Gregory Rotheingham
Ben non! Il n'y a pas d'Iroquois à Iroquois Falls.
###### Annie-France Brazeau
Les chutes?
###### Gregory Rotheingham
Non. Il n'y a pas de chutes non plus à Iroquois Falls.
###### Annie-France Brazeau
Ça promet! Alors sur quoi allez-vous vous baser, Gregory?
###### Gregory Rotheingham
Sur deux pôles : alors le premier, c'est le tourisme, mais surtout, il y a les sous-produits de la forêt.
###### Annie-France Brazeau
Décrivez-moi les sous-produits de la forêt, qu'est-ce que c'est?
###### Gregory Rotheingham
Ben, entre autres, il y aurait les queux de castor au miel. Ça pourrait ressembler à du _beef jerky_ , mais sucré.
###### Annie-France Brazeau
Il va falloir mâcher longtemps!
###### Gregory Rotheingham
Dans la région, il pousse des champignons un peu hallucinogènes, mais là, par exemple, ce serait un marché parallèle.
###### Annie-France Brazeau
Oui, ces champignons expliquent bien des choses dans votre cas, Gregory Rotheingham.
###### Gregory Rotheingham
Eh, eh, j'aime pas les champignons.
###### Annie-France Brazeau
Écoutez, revenons à notre rapport, Gregory. Décrivez-moi ce que vous avez dans ce rapport-là.
###### Gregory Rotheingham
_montrant le rapport_
Alors, on va relancer l'économie d'Iroquois Falls en utilisant l'écorce de bouleau pour faire des cornets.
###### Annie-France Brazeau
Ah oui, ces fameux cornets pour appeler l'orignal. J'imagine que vous allez les transformer en cornets de crème glacée?
###### Gregory Rotheingham
Absolument! Comment avez-vous deviné? Saviez-vous que d'après mes calculs, un parent dépensera en moyenne 28,35 $ en cornets pour la crème glacée dans la vie d'un seul enfant. Alors, j'ai pensé à un cornet réutilisable! un seul cornet dure toute la jeunesse de nos enfants.
###### Annie-France Brazeau
Fantastique! On pourra donc économiser un précieux 28,35 $.
###### Gregory Rotheingham
Absolument!
###### Annie-France Brazeau
Et combien se vendra ce cornet réutilisable?
###### Gregory Rotheingham
35 $.
###### Annie-France Brazeau
Bon, si je vous comprends bien, Gregory, vous allez rebâtir l'économie d'Iroquois Falls sur le dos des enfants.
###### Gregory Rotheingham
Ah, pas seulement sur le dos des enfants, quand même. Sur le dos des adultes aussi! On va faire une révolution dans le monde des consommations estivales avec la bière-popsicle santé.
###### Annie-France Brazeau
Holà! Un instant, Gregory! Bière, popsicle, santé, c'est trois choses qui ne vont pas ensemble. Essayons-les par deux, juste pour voir : Bière-santé... non, ça ne vas pas. Popsicle-santé... ça ne va pas non plus. Bière- _popsicle_... toujours pas.
###### Gregory Rotheingham
Je vous explique. L'idée, c'est qu'on prend les bières les plus populaires, on les vide dans un verre de plastique et on les congèle avec une belle branche de sapin au milieu. Puis, ce qui va arriver, c'est que les gens vont lécher, lécher et, rendu au centre, ils vont avoir une chique de gomme de sapin. Alors, ce sera rafraîchissant, santé, alcool et, en plus, la gomme masque l'haleine si on se fait prendre par la police.
###### Annie-France Brazeau
Gregory Rotheingham, je suis désolée, mais y a pas personne qui va acheter une bière congelée avec de la gomme de sapin au centre.
###### Gregory Rotheingham
Au contraire, Annie-France, tous vos amis montréalais vont en déguster sur les terrasses cet été.
###### Annie-France Brazeau
Un instant, Gregory. Montréal est sans doute ce qui est le plus loin au monde de la gomme de sapin. Ça marchera jamais, votre affaire.
###### Gregory Rotheingham
Pas de problème. On va leur dire que c'est populaire à Londres.
###### Annie-France Brazeau
Vraiment? Londres craque pour la gomme de sapin?
###### Gregory Rotheingham
Oui, oui!
###### Annie-France Brazeau
Ah, je peux comprendre dans le fond... Euh? Est-ce que Tony Blair aime la gomme de sapin?
###### Gregory Rotheingham
_prenant une voix de vendeur d'aspirateur_
Ah ben, oui, Tony Blair, ça fait déjà des années qu'il a compris que la gomme de sapin, c'est très bon pour la santé. Ça vient dans des petites capsules et ça se vend comme des petits pains chauds!
###### Annie-France Brazeau
Gregory, parmi vos projets, vous voulez également attirer des touristes à Iroquois Falls. Comment allez-vous faire?
###### Gregory Rotheingham
Ah ben, ça les touristes, on va en attirer avec notre carnaval : un carnaval extrême.
###### Annie-France Brazeau
Un carnaval extrême. Extrême, c'est très tendance. Ça pourrait réussir. Qu'est-ce que vous allez présenter dans votre carnaval?
###### Gregory Rotheingham
Il va y avoir, entre autres, une course de canot sur la rivière Abitibi.
###### Annie-France Brazeau
Gregory, tous les carnavals ont une course de canot. Entre vous et moi, on ne peut pas dire que c'est très extrême.
###### Gregory Rotheingham
Oui, mais en connaissez-vous des courses de canot où on saute le barrage électrique? C'est en hommage aux Iroquois qui ont donné leur nom à la ville en se tuant au bas des chutes.
###### Annie-France Brazeau
Mais, voyons Gregory, vous voulez tuer les participants!
###### Gregory Rotheingham
Absolument pas. La ville a déjà un nom, qu'est-ce qu'on ferait avec d'autres morts? Les participants seront équipés d'un deltaplane et, quand ils arriveront au barrage, ils vont s'envoler et le gagnant c'est celui qui va se rendre le plus loin.
###### Annie-France Brazeau
Étonnant! Parlez-moi d'une autre épreuve, monsieur _flyé_?
###### Gregory Rotheingham
Ben, on aura aussi la course de traîneau à chiens inversée.
###### Annie-France Brazeau
La course de traîneau à chiens inversée? Qu'est-ce que c'est?
###### Gregory Rotheingham
Voyez-vous, ce sont les participants qui sont attelés. Dans le traîneau, il y a un _pitbull_ enragé. Les participants vont avoir affaire à courir.
###### Annie-France Brazeau
Et j'ai lu dans votre rapport que vous voulez utiliser de nouvelles technologies pour attirer les touristes.
###### Gregory Rotheingham
Bien sûr! On va faire un rallye pour enfants. On va faire monter plein d'enfants dans le train à Porquis Junction. On en fera descendre deux ou trois à chaque kilomètre avec un GPS dans les mains, et ils doivent se rendre à Iroquois Falls. Ceux qui vont réussir vont gagner un cadeau.
###### Annie-France Brazeau
Ceux qui vont réussir? Mais c'est épouvantable! S'ils se perdent en forêt ou s'ils rencontrent un animal sauvage? Y avez-vous pensé seulement?
###### Gregory Rotheingham
Ben non, ben non, il ne faut pas s'inquiéter, c'est juste la moitié de l'activité. L'autre moitié, c'est les adultes qui partent en motoneige et ceux qui ramassent des enfants obtiennent le prix.
###### Annie-France Brazeau
En plus, les pauvres enfants vont perdre leurs cadeaux, Gregory? Mais c'est épouvantable!
###### Gregory Rotheingham
Oui, mais de toute façon, le cadeau, c'est une bière-santé-popsicle! Les enfants n'auraient pas pu la prendre et il aurait fallu qu'ils la donnent à leurs parents.
###### Annie-France Brazeau
Oui, bon ça va. Vous savez, Grégory, _Incitatif pédant_ est une émission très écoutée. On vient tout juste de recevoir un courrier électronique vous concernant. Alors je lis : « Bonjour madame Brazeau, vous êtes grande! Par contre, je me questionne au sujet de Gregory Rotheingham. Au mieux, c'est un lunatique, au pire c'est un hystérique. » Voilà, c'est un courrier électronique que nous venons de recevoir.
_Elle le lance sur la table. Gregory regarde le courriel et le saisit brusquement._
###### Gregory Rotheingham
Beuh... Mais c'est pas un courrier électronique, c'est écrit à la main!
###### Annie-France Brazeau
Oh... Euh...
###### Gregory Rotheingham
Et c'est VOTRE écriture, Annie-France!
###### Annie-France Brazeau
_mal à l'aise_
Ah, ça, c'est troublant, ça. Alors, j'aimerais demander à cet auditeur de bien vouloir cesser d'envoyer des courriers électroniques en imitant mon écriture.
###### Gregory Rotheingham
Mais c'est écrit au stylo!
###### Annie-France Brazeau
On n'a plus de temps, Gregory.
###### Gregory Rotheingham
Et il y a vos empreintes digitales!
###### Annie-France Brazeau
Plus de temps!
###### Gregory Rotheingham
Et c'est votre parfum sur la lettre!
###### Annie-France Brazeau
Je ne vous entends plus!
_Elle se bouche les oreilles et chante._
###### Entre ailleurs et autrefois
_Elle roule dans le brouillard,_
_Sur une petite route du Nord_
_Elle roule vers la petite ville_
_Où elle a grandi tranquille_
_Dans la lumière de ses phares_
_Trop de blanc dans la nuit noire_
_Comme si tout s'est confondu_
_Comme si elle s'est perdue_
Tout le monde trouve le chemin
Par où sortir de son enfance
À la recherche de son destin
À la poursuite de sa chance
Mais tout le monde ne trouve pas
Le petit sentier qui y ramène
On n'a pas tous cette chance-là
Trouver sa place au bout de sa peine
_Elle roule depuis des heures_
_Vers la maison de sa sœur_
_Où demeuraient ses vieux parents_
_Dans le temps qu'ils étaient vivants_
_Elle connaît la route par cœur_
_Mais ne la reconnaît pas_
_Est-ce qu'elle a fait une erreur_
_Suit-elle bien la route qu'elle croit_
Tout le monde trouve le chemin
Par où sortir de son enfance
À la recherche de son destin
À la poursuite de sa chance
Mais tout le monde ne trouve pas
Le petit sentier qui y ramène
On n'a pas tous cette chance-là
Trouver sa place au bout de sa peine
_Elle roule dans le brouillard_
_Sur une petite route du Nord_
_Ne voit pas très loin devant_
_Ne se retrouve pas vraiment_
_Elle a le sentiment fou_
_Qu'elle n'avance pas du tout_
_Entre ailleurs et autrefois_
_Peut-on retourner chez soi?_
_Peut-on retourner chez soi_
_Entre ailleurs et autrefois?_
###
### Salut Spanish!
Au bord de la route 17, là où la rivière Spanish se jette dans le lac Huron en formant un bel archipel et un bon havre pour une nouvelle marina — une des plus appréciées des Grands Lacs — se trouve un village franc et honnête, malgré son nom trompeur. Car, sachez-le, il n'y a rien d'espagnol à Spanish (non plus à Espanola, en amont sur la rivière).
Sur les vieilles cartes de la Nouvelle-France, cette rivière se nommait « l'Espagnole ». Ça on le sait. Mais pourquoi, ça on ne le sait pas. une dizaine d'hypothèses circulent. La plus colorée soutient que des explorateurs français furent étonnés d'entendre des mots espagnols dans le parler des Amérindiens de l'endroit. On leur aurait dit qu'une femme faite prisonnière lors d'un raid amérindien dans le Sud aurait appris sa langue à ses enfants.
La langue maternelle, c'est tenace, on le sait. C'est pour ça qu'à défaut d'espagnol, vous entendrez parler français ici. Environ le tiers des neuf cents habitants de Spanish sont Canadiens français. Il y en a par ici depuis le milieu du 19e siècle. Ils sont venus bûcher sur la Rive-Nord même avant l'arrivée du chemin de fer. Ils sont venus cultiver. Ils étaient nombreux dans les belles années forestières et minières.
Mais comme partout dans le Nord, le vent a tourné. En six ans, selon le dernier recensement, Spanish a perdu dix pour cent de ses habitants. S'il a bien fallu dire salut à ceux qui sont partis, aujourd'hui nous saluons ceux qui tiennent à y rester. Spanish a dans son passé un homme qui n'a laissé presqu'aucune trace dans l'histoire, sauf... ses histoires. Mais quelles histoires!
###### Jules Couvrette
Jules Couvrette était un des pires péteurs de broue, conteur de coups, fabricateurs d'histoires de fou que vous puissiez imaginer. Il en a tant et si bien conté, de ses aventures insensées, tirées par les cheveux, exagérées à qui mieux mieux, qu'enfin on s'est demandé : a-t-il vraiment pu exister, ce Jules Couvrette?
Qu'il ait vraiment existé, on en est sûr au musée de Massey, où j'ai trouvé le document _Gems of History_ , d'où j'ai tiré ce que je vais conter. Ce petit cahier dactylographié est le fruit d'une enquête orale locale que la Spanish River Women's Institute a menée dans les années soixante-dix. Ces enquêteuses ont découvert que plusieurs familles pionnières connaissaient et racontaient de curieuses histoires immanquablement attribuées à Jules Couvrette, sans pourtant savoir qui il était exactement. En tout cas, ces histoires ont fini par donner tout un chapitre dans le cahier. Moi, je ne fais que les répéter. Vous en jugerez, à les écouter, si Jules Couvrette a existé.
Un après-midi d'hiver, Jules Couvrette est entré dans le magasin général de Spanish, dont monsieur Sandowski était le propriétaire. Par hasard, il s'y trouvait un client qui n'était pas du coin. Jules Couvrette a dit:
— Sandowski, tu paies combien pour une corde de lièvres ces temps-ci?
— Ça dépend. T'en as combien?
— Deux cordes, peut-être même trois! » a répondu Jules Couvrette.
Puis il est sorti.
— C'est un fou, celui-là? a demandé l'étranger.
— Oui et non. S'il revient, tu lui demanderas comment il les attrape, ses lièvres.
Sur ces entrefaites, Jules Couvrette est revenu et l'étranger lui a posé sa question.
— C'est simple. La rivière est couverte de glace lisse ces temps-ci. Mes garçons et moi, on y place partout de petits tas d'éclisses de cèdre pas plus grosses que des crayons. À la brunante, on les allume. Les lièvres aiment la chaleur, donc ils s'approchent. une fois bien réchauffés, ils s'endorment. Mais quand le feu s'éteint, la glace se reforme et leurs derrières restent pris dedans. Donc le matin, mes garçons et moi, on a juste à les faucher puis à les corder.
— Vraiment? Mais qu'est-ce que vous faites l'été?
— L'été? Alors là, on va sur le chemin de la réserve. Le terrain, par là, c'est de la belle roche lisse. Là-dessus, on laisse partout des petits tas de poivre. Les lièvres qui passent là sentent ça. Quand ils éternuent — atchoum! — leur tête se lance par en avant contre le roc puis ils s'assomment. Avec le poivre et les éclisses, j'ai pas besoin de collets. Mais dis donc, Sandowski, tu la paies combien, la corde de lièvres aujourd'hui?
« Vas te promener! » a été la réponse de Sandowski. Cette réponse a soulagé l'étranger.
Une autre fois, Jules Couvrette se faisait voler ses patates la nuit, dans son champ au bord de la rivière. Donc une nuit, il s'y cache avec sa hache.
Vers minuit, il a entendu la vase clapoter, puis un corps se traîner entre deux rangs de patates. Soudain, c'est le face-à-face avec son voleur, qui était... une barbotte géante!
D'un grand coup de hache, Jules Couvrette lui a fendu la tête en deux moitiés. Mais la barbotte a quand même réussi à se sauver dans l'eau.
Le pire, c'est que la nuit suivante, la barbotte est quand même revenue voler des patates. Puis maintenant, elle déterrait deux rangs en même temps!
Comme tout cultivateur, Jules Couvrette bûchait en hiver. Mais, parfois, il façonnait ses harnais non pas avec du cuir, mais de la babiche.
La babiche, quand c'est mouillé, ça s'étire. une fois, son harnais s'est tellement étiré que le cheval était rendu à l'étable, mais la bille qu'il traînait était encore dans le bois.
Mais Jules Couvrette avait du savoir-faire. Il a juste dételé le cheval et accroché le harnais à une souche. Pendant la nuit, la babiche a rétréci. Donc, au matin, la bille était rendue près de la grange. Ça n'avait pas été forçant.
Cette histoire-ci remonte au temps où il y avait encore des pins géants sur la Rive-Nord. Jules Couvrette racontait qu'à son arrivée dans la région, il était marteleur près de la rivière aux Sables. Le marteleur, on l'appelle aussi le _fitteur_. C'est l'homme qui marque d'un coup de hache les bons arbres à abattre.
Une fois, Jules Couvrette a planté sa hache dans le pin le plus immense qu'il ait jamais vu. Puis il en a fait le tour comme il faut faire pour juger de quel côté il allait tomber. Quand il a eu fini d'en faire le tour et qu'il est revenu là où sa hache était plantée, il a trouvé juste la lame. Le manche, lui, avait disparu.
Jules Couvrette était perplexe, mais il a vite compris. Ça avait été si long, faire le tour de cet arbre, que le manche avait eu le temps de pourrir!
Les chiens de Jules Couvrette étaient les plus intelligents du pays. Il en avait un si fin qu'il lui suffisait de pointer du doigt ce qu'il voulait et ce chien-là s'en occupait.
Dans ce temps-là, Jules Couvrette défrichait. Il y avait de grosses souches sur sa terre. Donc, il s'est procuré de la dynamite. Prudent quand même, il a utilisé une longue mèche. Malheureusement, ça a donné le temps à sa meilleure vache de s'en approcher de trop près. Sans réfléchir, Jules Couvrette a pointé la souche du doigt. Son chien est parti en flèche. Et le pire est advenu. Juste comme le chien l'a atteint, la souche a explosé. Son pauvre chien a revolé trente pieds en l'air, et est retombé juste à ses pieds, déchiré en deux!
Sans hésiter, Jules Couvrette a vite recollé les deux bouts du chien ensemble et, miraculeusement, son chien semblait bien se porter. Mais il avait agi si vite que les deux pattes d'en arrière pointaient par en l'air!
Mais ça ne le dérangeait pas, au contraire. Ce chien-là courait sur deux pattes d'abord, puis il se roulait de l'autre côté et courait sur les deux autres. Il ne se fatiguait jamais. Il était deux fois meilleur qu'avant!
Les chiens de Jules Couvrette étaient aussi bons à la chasse. Il en avait un qui ne lâchait jamais sa proie, au grand jamais.
Une fois, ce chien s'est mis à pourchasser un renard derrière Blind River. Quand il les a revus les deux, le lendemain, ils étaient à Walford. Mais ils ne couraient plus, ils marchaient lentement, tellement ils étaient morts de fatigue.
Ce chien-là était aussi champion pour suivre une piste à l'odeur. Mais une fois, il s'est trompé. Il s'est mis à suivre une piste d'orignal pendant sept jours. Quand il s'est enfin arrêté, il n'y avait pas un animal autour. Jules Couvrette était perplexe, mais il a fini par comprendre. Son chien avait suivi la piste de reculons. C'était donc à cet endroit que l'orignal était né.
Jules Couvrette trappait aussi et son chien était fameux pour ça.
Tout ce qu'il avait à faire, c'était de sortir une des planches sur lesquelles il tendait les peaux à sécher. Le chien savait quelle prise faisait sur cette planche-là — castor, martre, pécan — puis il partait lui en chercher une.
Une fois, Jules Couvrette a sorti la planche à repasser. Son chien est parti. Mais il ne l'a plus jamais revu.
Même sans son chien, ce chasseur savait chasser. une fois, Jules Couvrette, en chaloupe, poursuivait à la rame un orignal à la nage. C'était épuisant. À force de ramer, il a réussi à s'approcher assez près de la bête pour lui passer une corde autour du panache. Là, il pouvait relaxer, en attendant que l'orignal atteigne la rive pour le tirer. Malheureusement, sa chaloupe s'est cognée contre un arbre submergé et Jules Couvrette est tombé dans le lac.
L'automne suivant, il est retourné au même lac et il a entendu un vacarme d'enfer dans le bois. C'est ce même orignal qui est apparu, traînant toujours la chaloupe! Mais le plus impressionnant, c'est que couché dedans, il y avait un bébé orignal, un veau de l'année. Puis la femelle suivait derrière!
La fois où Jules Couvrette a eu le plus peur, c'est une fois où il était parti trapper sans carabine. Six loups l'ont piégé dans un arbre où il avait grimpé pour leur échapper. Trois sont partis, mais trois sont restés là à le surveiller.
C'est sournois un loup. Mais jamais il n'aurait imaginé ce qui lui arrivait. Les loups sont revenus avec un castor, et le castor s'est mis à ronger l'arbre. Heureusement que la forêt était dense à cet endroit. Ça l'a sauvé, parce que l'arbre n'est pas tombé d'un coup, il s'est appuyé contre l'arbre voisin. Jules Couvrette a eu le temps de s'y agripper avant que son arbre ne s'écrase.
Mais ce n'était pas fini. Le castor a recommencé. Mais je vous ai dit que la forêt était dense : c'est la même chose qui est arrivée. Après le cinquième arbre, le castor s'est découragé, puis les loups sont partis.
Mais c'est pas tout ça qui explique que Jules Couvrette ait passé toute une nuit dans la forêt. C'est qu'il n'a pas osé descendre de l'arbre avant le lendemain matin.
Et tout ça, mes amis, ce n'est qu'un survol des histoires de Jules Couvrette que le musée de Massey conserve. Il en a conté beaucoup d'autres et prenait bien plus son temps que moi.
Ça se comprend qu'on ait pu douter de son existence, tant on peut douter de ses aventures. Ce grand conteur de peurs était si fort dans son art que ses fictions l'ont remplacé dans la réalité. Mais sachez qu'il a bel et bien vécu à Spanish, Jules Couvrette. Son nom figure sur une vieille carte du canton à titre de propriétaire d'une terre bordant la rivière, là où elle fait un coude, au sud du village. Son nom se trouve dans la liste des électeurs du secteur 26 du district d'Algoma Est. Et là on a une date. C'était en 1917. Ce sont de rares détails sur la vraie vie de Jules Couvrette qu'on peut découvrir en fouillant dans la bibliothèque publique de Spanish.
Mais ce qui compte, c'est son nom dans ses contes. Ce qui compte encore plus, c'est que tant d'autres aient répété ce nom en répétant ses contes, pour être complices de ce grand esprit blagueur à la hauteur d'un grand pays inspirateur. Il y a des discours pleins de bons sens qui déguisent la bêtise. Il faut peut-être tourner le dos au réalisme quand il nourrit le pessimisme. Nous en avons trop, dans le Nord de l'Ontario, de dures réalités à changer, de tristes réalités à corriger. C'est par la force de l'esprit qu'on y aura réussi. C'est en ayant du bagout qu'on en viendra à bout. Pour ces qualités-là, à Spanish, on a un fameux modèle: on a Jules Couvrette.
###### Raconte-moi des histoires
_J'ai besoin de croire que ça ira mieux_
_Je suis pris dans le courant_
Je suis comme un chien qui se chasse la queue
_C'est pas mal étourdissant_
_Ma petite vie normale, puis ma job banale_
_C'est pas valorisant_
_Je veux m'aligner, je voudrais gagner_
_Mais je sais pas comment_
Oh! s'il te plaît
Raconte-moi des histoires, fais-moi de beaux accroires
Fais-moi rêver un peu, dis-moi que ça ira mieux
Invente-moi un espoir, le bon sens j'y crois plus
Fais-moi sortir du noir
Je suis pas mal tanné d'être un ti-cul
_J'ai obéi à mes parents, j'ai écouté mes enseignants_
_J'ai fait pas mal ce qu'on m'a dit de faire_
_J'ai essayé de suivre la mode_
_Jai bu la Bleue, j'ai bu la Bud_
_Mon coat de cuir m'a coûté cher_
_Je fais de mon mieux pour être du monde_
_Pendant un bout j'avais une blonde_
_Avant je fumais, maintenant j'arrête_
_Mais quand je me lève le matin_
_Pour faire mon bon petit citoyen_
Je sens juste qu'on se paie ma tête
Oh! s'il te plaît
Dis-moi que ça va changer si j'y mets plus de cœur
Que tout va s'arranger, que c'est pas mieux ailleurs
Fais-moi rêver un peu, fais-moi de beaux accroires
Dis-moi que ça ira mieux
Envoye, là, raconte-moi des histoires
_Vas-tu falloir que je sacre mon camp_
_Vas-tu falloir que je pique une crise_
_Vas-tu falloir que je prenne mon trou_
_Que je brûle un lampion à l'église_
_J'ai l'impression de mériter mieux_
_Que je vaux quand même autant qu'un autre_
_Je sais, c'est pas beau d'être envieux_
_Dis-moi que tout ça, c'est pas de ma faute_
Oh! s'il te plaît, raconte-moi des histoires
Oh! s'il te plaît, raconte-moi des histoires
###### Tout feu tout flamme
L'histoire commence une nuit d'automne. une nuit de pleine lune _..._ pleine, ronde et orange comme si elle avait avalé le soleil. Quand la lune se pare de ses atours de feu comme ça, les Anglais la nomment _blood moon_ ; les Amérindiens et les poètes l'ont baptisé la « lune de feu » et Environnement Canada appelle ça une éclipse lunaire. Peu importe... cette nuit-là, la lueur flamboyante de la lune était telle qu'elle embrasait les arbres comme en plein jour et donnait aux eaux du lac Huron un aspect cuivré.
Quand je dis le lac Huron, évidemment je ne parle pas du lac au grand complet. Ça ferait trop pour ceux qui n'ont pas beaucoup d'imagination. Imaginez plutôt la pleine lune qui se mire dans le North Channel et imaginez que le lac la trouve tellement belle qu'il ne bouge pas d'une goutte dans l'espoir qu'elle va venir se coucher sur lui. Tout le monde sait que l'eau et le feu, ça ne se mélange pas _..._ sauf les vieux Ojibwés qui prétendent que le feu de la lune lui, peut venir se mêler à l'eau d'un lac le temps d'une caresse, voire d'un long baiser. Mais seulement si le lac reste d'un calme plat. Cette nuit-là, le lac Huron a ouvert ses flots si doucement que la lune est venue s'allonger sur lui. Le feu et l'eau se sont effleurés pendant un fragment d'éternité où la lune a eu le temps de s'abandonner tellement creux dans l'eau du lac qu'elle a touché le fond.
En cette nuit de poésie cosmique, Paulo, un jeune Gaspésien chercheur d'épaves _,_ était campé seul au creux de la baie de l'île Fox. Et cette nuit-là, précisément, il avait décidé de coucher à la belle étoile tellement le temps était doux. Comme Paulo n'était pas ronfleur, il n'avait pas attiré l'attention des incommensurables amoureux. Sauf qu'il fut réveillé au milieu de la nuit par la chaleur intense du clair de lune. Du moment où il a ouvert les yeux, il a vu le lac Huron faire la cour à la lune et la lune s'étendre sur lui de tout son rond. Au moment précis où la lune touchait le fond du lac, Paulo a étiré le cou pour mieux voir ce qui brillait dans l'abîme des eaux. C'est tout ce que ça prenait pour briser le charme. Surprise dans ses ébats les plus intimes — et les plus rares — la lune s'est sauvée tellement vite qu'elle a laissé une petite partie d'elle-même au fond du lac. Pour hurler sa peine de grand lac, le lac Huron s'est mis à éclabousser les étoiles, à battre les côtes rocheuses des îles, à écraser les roseaux et le riz sauvage de l'embouchure de la Spanish... Bref, le lac Huron faisait tous les temps!
En vrai homme de la mer, Paulo savait que tôt ou tard, le lac allait retrouver son calme et qu'à ce moment-là, il pourrait plonger jusque dans ses entrailles, s'il le fallait, pour aller cueillir le trésor mystérieux qu'il avait entrevu et croyait voir encore entre les soubresauts du lac. Finalement, ce n'est qu'aux petites heures du septième matin que Paulo, vêtu de son costume d'amphibien, a pu se glisser dans l'eau du lac à la manière d'une rumeur dans l'esprit d'une commère. Et là, enfoui dans les profondeurs ténébreuses, il a trouvé l'objet de son délire encore chaud au toucher. L'objet, c'était un croissant de lune en or sur un collier de la même couleur. Il n'arrivait pas à lire l'inscription sur le bijou parce que sa vision était devenue floue. Pire, ses oreilles bourdonnaient. Il manquait d'air! Il a voulu se projeter vers la surface, mais il s'était mis tellement de poids à la ceinture pour se rendre au fond qu'il nageait comme une roche. C'est à ce moment-là que les événements ont pris une tournure des plus inattendues.
À peine remonté de quelques brasses, il échappe sa lampe en essayant de détacher sa ceinture de plombs. Affolé, il serre la main droite pour éviter de perdre la lune d'or. Ses poumons brûlent, sa tête va éclater. Dans un dernier coup de pinceau lumineux de sa lampe qui tombe en vrille sous lui, il croit apercevoir une forme humaine, avec de longs cheveux noirs. une main tendue vers lui. une main de femme. Ensuite, dans le noir, quelque chose ou quelqu'un lui arrache son trésor.
Au bout d'un moment d'obscurité, Paulo n'éprouve plus aucune douleur ni aucune panique. Et aussi soudainement qu'elle est disparue, la trace du trésor qu'il convoite réapparaît. À dix ou quinze pieds de lui, un scintillement bref lui fait signe, comme un phare dans la nuit liquide. Là, sur le fond du lac, où il a posé les pieds, il se met à courir vers le scintillement intermittent. Rapidement il se rend assez près de l'objet brillant qu'il convoite pour entrevoir, à la lueur du bijou qu'on lui a volé, la silhouette noire d'une femme aux longs cheveux. Dans le rayonnement cuivré du bijou qui éclaire le beau visage de la jeune femme, il voit que ses yeux sont exactement de la même couleur de feu que la lune... que le bijou. « Hmm... pas laide », qu'il se dit, avant de se ressaisir et de continuer à la poursuivre jusqu'à la surface de l'eau.
Il émerge non pas à la surface du lac, mais dans un vieil hôtel. Vieux comme quelque chose qui ressemble à l'époque de son grand-père. une photo en noir et blanc derrière le grand comptoir attire son attention. Sur la photo, un hôtel ainsi qu'un jeune homme et sa jeune femme vêtus de leurs habits du dimanche et de leur fierté de propriétaires. Il se dit : « Ça doit être à eux-autres, l'hôtel » en lisant les lettres blanches dans le coin inférieur droit: Lake Huron Hotel, 1931. Une odeur de brûlé l'attire vers la porte au fond de la salle. Il aboutit dans la cuisine où il aperçoit sa voleuse, penchée sur la cuisinière au gaz, en train de brasser doucement quelque chose dans un chaudron. Elle ne semble plus avoir envie de se sauver. Mais elle n'a plus le bijou. Il s'approche un peu. La belle inconnue sourit. un petit sourire énigmatique de Mona Lisa. Ça intrigue notre homme plus que ça le rassure, mais il continue d'approcher. Après tout, ça le démange de savoir qu'est-ce qu'elle fait cuire. Rendu tout près d'elle, il ne peut s'empêcher de la reluquer encore avant de regarder dans le chaudron. Ses longs cheveux noirs comme la nuit ont un parfum céleste. Sauf qu'un coup d'œil dans le chaudron le fait sursauter. À sa grande stupéfaction, il s'aperçoit que sa belle étrangère est venue lui voler son croissant de lune jusqu'au fond du lac Huron juste pour le faire fondre.
Instinctivement, il plonge sa main droite dans le chaudron brûlant. Sa main fond instantanément, se transforme en eau. Pire que ça, sa main d'eau étant encore attachée à son bras, il reste comme soudé au chaudron. Aussitôt, l'étrange fondeuse de bijoux jette un caillou dans le chaudron et de l'autre main elle saisit le bijou. Immédiatement la main de Paulo réapparaît dans le fond du chaudron chaud. Il se dépêche de la retirer, ouvre le poing et y trouve le caillou. Le rire de la femme lui fait lever les yeux. En la regardant, il remarque aussitôt que ses yeux couleur de cuivre sont passés au doré — comme le bijou, d'ailleurs. De plus en plus dépassé, il regarde à nouveau le caillou. Puis il entend la voix de la femme.
— _Gracias_.
Quand il lève les yeux, il ne subsiste d'elle que le sourire qu'elle avait dans la voix.
— Non! Attends! crie-t-il en levant la main.
Du coup, il échappe son caillou. Mais avant de toucher le sol, le caillou se transforme en oiseau doré. L'écho lui revient : « _Adios_ ». L'oiseau part en direction de l'écho. Il suit l'oiseau. Au même moment le chaudron explose sur la cuisinière à gaz et embrase l'hôtel au complet.
Dehors, encore le feu! Spanish est ravagée par les flammes. Chaque maison que son oiseau-guide survole devient un énorme bûcher. Épouvanté par le spectacle, il se met à souffler sur chaque maison qu'incendie l'oiseau. Tous les bâtiments sur lesquels il souffle deviennent d'énormes bulles d'eau bouillonnantes s'évaporant sous l'effet de la chaleur. Leur vapeur monte au ciel en colonnes de fumée qui retombe sous forme de neige. Mais les énormes flocons sont en fait des feuilles de calendrier. Elles portent toutes la même date : 23 mars 1902. Paulo n'essaie même plus de comprendre. Et là, dans les vestiges d'une maison brûlée et à demi recouverte de la neige de calendrier, ses yeux captent un scintillement. En s'approchant, il reconnaît l'objet qu'il convoite. C'est bien le bijou, mais il est recouvert de glace. Et plus de traces de la mystérieuse femme. En saisissant l'objet, il est envahi par un froid glacial en même temps que sa propre chaleur se transfère au bijou. Il reste là, gelé, à genoux, le bijou dans les mains. Au moment où même sa pensée va se glacer à tout jamais, la femme émane du bijou, là devant lui. Ses yeux ont encore pâli. Ils sont devenus gris-argent de clair de lune. Entre les mains de Paulo, le bijou est lui aussi devenu encore plus scintillant dans son nouveau lustre gris-argent. Il peut maintenant y lire l'inscription : Consuela Isabella Cortés de Luna. Avec un dernier sourire — triste, celui-là —, la belle se penche sur son visage et l'embrasse. Et le feu de ce baiser torride n'a pas fini de se transmettre à Paulo que la femme de feu part en fumée.
Maintenant, tous les conteurs de contes vous diront que c'est ici que s'est arrêtée la destinée de Paulo, l'homme de la mer venu se noyer pour un trésor de fond de lac, une nuit d'automne, à Spanish, Ontario. C'est qu'ils ne connaissent pas aussi bien que moi toutes les vertus du feu, surtout lorsqu'il est allumé par une étincelle dans le fond d'un regard langoureux. En fait, c'est comme si le baiser lui avait dégelé l'âme. En tout cas, c'est la conclusion à laquelle il est arrivé puisqu'il se met à flotter doucement dans les airs, s'élevant au-dessus de son propre corps. Peu après, sa tête perce un plafond transparent, exactement comme lorsqu'il remonte à la surface après une plongée. Sauf que cette fois, il émerge... sur un nuage. un nuage qui forme un quai aérien menant à la lune immense qui a, elle aussi, retrouvé sa couleur habituelle. Assise sur le rebord de l'astre de la nuit, les pieds pendant juste au-dessus du nuage où il marche, la belle l'attend. Mais quelques pas infranchissables les séparent pourtant. Toutes choses étant revenues à leur place, l'eau et le feu ne peuvent plus se mélanger. Alors, avec les étoiles pour témoins, la belle donne à Paulo son bijou, comme une promesse. Celle de lui rendre visite à la prochaine lune de feu.
Aussitôt, tout disparaît! Et il est emporté dans une chute vertigineuse. Heureusement, le lac Huron s'en est mêlé. Se sentant sans doute concerné par toute cette histoire, qui après tout s'était ourdie dans le creux de ses entrailles, le lac Huron a élevé ses plus hautes vagues à la rencontre du plongeur et l'a attrapé à mi-ciel pour amortir sa chute.
Finalement c'est comme ça qu'on l'a retrouvé, flottant à la dérive sur le lac, le bras droit tendu vers le haut, le poing fermé comme une huître qui cache jalousement sa plus belle perle. Mais c'est seulement après un mois — d'une pleine lune à l'autre — que Paulo a vraiment refait surface. Son réveil a suscité bien de l'émoi à l'hôpital. On lui a d'abord raconté qu'il avait été le sujet du coma le plus mouvementé de l'histoire de l'humanité. Ensuite, on lui a rendu le bijou qu'on avait fait expertiser pendant son coma. Il était typique de l'orfèvrerie espagnole du XVIIe siècle. Tout le monde en a conclu que, si le collier s'était retrouvé là près de la rive nord du lac Huron, c'était sans doute que la vieille histoire amérindienne était vraie. Que les Ojibwés auraient ramené d'une razzia dans le sud des États-unis une femme espagnole qui aura transmis sa langue à sa descendance, puisqu'on a baptisé la ville et la rivière du nom de Spanish.
Par contre, pour Paulo, l'histoire était tout autre. Le collier de lune, c'était en fait l'incarnation d'une caresse furtive un soir d'automne quand la lune en feu s'est étendue sur le lac Huron. « Un rêve », qu'on lui a dit. « un rêve de comateux. » Désormais, à chaque fois qu'il doute, Paulo prend dans sa main le croissant de lune sur son collier et chaque fois ses yeux s'arrêtent sur le premier mot qui y est gravé : Consuela. Consolation.
_Richard J. Léger_
###### _Incitatif pédant_ : Spanish Fly
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau. Vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez publié un rapport sur la communauté de Deux-Rivières, dans lequel vous suggérez à Deux-Rivières de se tourner vers la production d'émissions pour enfants. Vous avez intitulé votre rapport _Cinq moutons, quatre moineaux, trois marmottes, Deux-Rivières et une souris verte_.
_Il va répondre. Elle enchaîne._
Vous êtes débrouillard, vous avez déjà organisé un lave-auto-bénéfice et, avec l'argent recueilli, vous avez fait laver votre auto.
_Il va répondre. Elle enchaîne._
Vous avez écrit des chansons, surtout des ballades misogynes, dont votre plus grand succès qui s'appelle « La crème de toutes les filles, j'lui donne 35 % _»_.
_Il va répondre. Elle enchaîne._
Vous appartenez à un ordre initiatique secret, dont les membres ne doivent révéler leur implication secrète sous aucun prétexte, sous peine de recevoir des châtiments corporels extrêmes...
###### Gregory Rotheingham
_sans réfléchir_
C'est vrai!
###### Annie-France Brazeau
Et je pourrais continuer longtemps comme ça, mais je ne suis pas certaine qu'on ait envie d'en savoir plus. Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous venez nous présenter un tout nouveau rapport sur le développement économique du pittoresque village de Spanish, sur les rives du lac Huron.
###### Gregory Rotheingham
Ben, sur « une » des rives.
###### Annie-France Brazeau
Oui... Et j'imagine que vous allez exploiter les beaux paysages, les abondantes ressources naturelles et la pêche miraculeuse.
###### Gregory Rotheingham
Non, Annie-France. La prospérité de Spanish est liée à sa gastronomie, à l'érotisme et, évidemment, à ses liens avec l'Espagne.
###### Annie-France Brazeau
Et comment s'intitule votre rapport?
###### Gregory Rotheingham
_Liche et relish pour s'mettre riche à Spanish._
###### Annie-France Brazeau
C'est presque sensuel. Mais l'idée de faire de Spanish l'Espagne d'Amérique est excellente. Le métissage culturel, c'est tellement tendance. J'imagine qu'on pourra désormais manger de la paella et boire une bonne sangria à Spanish.
###### Gregory Rotheingham
Pas tout à fait, Annie-France. Voyez-vous, notre stratégie est d'inspiration espagnole, mais à l'accent local.
###### Annie-France Brazeau
Non, non, Gregory. Vous voulez dire :
_avec un accent espagnol_
« notré stratégie est d'innspirationne espagnole, mais
_avec un accent populaire_
à l'ac-chant lôcal ».
###### Gregory Rotheingham
Une question Annie-France : une paella, c'est quoi au juste?
###### Annie-France Brazeau
Du riz avec des fruits de mer.
###### Gregory Rotheingham
Ah, pas à Spanish! Ici, à Spanish, ça devient du riz frit avec de la barbotte.
###### Annie-France Brazeau
Ah, vous transposez les recettes espagnoles pour les adapter au Nord de l'Ontario.
###### Gregory Rotheingham
Exactement!
###### Annie-France Brazeau
Votre recette de paella est très réussie. Quoi de plus nord-ontarien...
###### Gregory Rotheingham
Que la barbotte!
###### Annie-France Brazeau
Non, que la friture!
###### Gregory Rotheingham
Une autre question Annie-France : la sangria, qu'est-ce que c'est?
###### Annie-France Brazeau
Du vin rouge avec des fruits.
###### Gregory Rotheingham
Ben, imaginez qu'à Spanish, ce sera un pichet de bière avec des bleuets et des framboises sauvages qu'on écrase dedans.
###### Annie-France Brazeau
Vous croyez vraiment que les gens vont adopter ces créations gastronomiques, Gregory Rotheingham?
###### Gregory Rotheingham
Certainement, parce que, en plus de bien manger, ils vont pouvoir s'amuser grâce à une grande fiesta et une corrida en plein cœur de Spanish.
###### Annie-France Brazeau
Une corrida, voir un taureau se faire massacrer par un toréador, c'est pas un peu cruel comme attraction touristique?
###### Gregory Rotheingham
Non, non, non, Annie-France! On va en faire une adaptation nord-ontarienne.
###### Annie-France Brazeau
Ah, je pense que je vous vois venir. Adaptation nord-ontarienne...
###### Gregory Rotheingham
Annie-France, imaginez une belle grosse arène en plein air, des milliers de spectateurs scandant : « Olé! Olé! » Et sous les acclamations du public, le toréador fait son entrée, il enlève et secoue son manteau de chasse orange fluorescent : « Olé Olé! »
###### Annie-France Brazeau
Très élégant! Je l'imagine saluant la foule avec sa calotte de base-ball.
###### Gregory Rotheingham
Il faut sentir la dignité du toréador se tenant à l'extrémité de l'arène quand, tout à coup, entre un beau gros _buck_ avec un panache de 60 pouces. Le toréador le regarde droit dans les yeux et lâche le _call_ de la femelle.
_Il imite le cri de l'orignal appelant une femelle._
L'orignal gratte le sol et fonce! Olé Olé!
###### Annie-France Brazeau
Stupéfiant! Et, en plus, allez-vous installer un orignal en bronze à l'entrée du village?
###### Gregory Rotheingham
Mais non, ça déjà été fait à Smooth Rock Falls et à Hearst. Nous, notre mascotte, ça va être une grosse mouche, symbole de l'amour et de la passion.
###### Annie-France Brazeau
Une mouche?
###### Gregory Rotheingham
Oui! La _Spanish Fly_! Spanish va devenir le paradis de l'amour et de la passion espagnole. On va construire une usine du plus puissant des aphrodisiaques naturels, le _Spanish Fly_.
###### Annie-France Brazeau
Le _Spanish Fly_? Les bras m'en tombent — mais seulement les bras, par exemple. Vous allez construire une usine pour un aphrodisiaque dont l'ingrédient de base est une mouche espagnole, complètement absente de la région?
###### Gregory Rotheingham
Mouche espagnole, mouche espagnole, voyons, une mouche, c'est une mouche et, ici, à Spanish, de la mouche, on en a en masse! Vous connaissez sûrement les rouleaux de pellicules collantes qu'on utilise pour attraper les mouches dans nos maisons?
###### Annie-France Brazeau
Oui, je me suis déjà pris les cheveux dedans.
###### Gregory Rotheingham
On installe des pellicules géantes partout dans le village. Quand la pellicule est pleine de mouches, on la laisse sécher au soleil. On broie ensuite mouches et pellicule assaisonnée pour en faire une fine poudre qu'on va mettre en comprimés qu'on vendra partout sur la planète. Vous savez, la population vieillit et, trop souvent, la passion aussi. Nous allons donc exploiter ce marché de baby-boomers avec mon slogan : « Pour remettre le "boom" dans baby-boom, on se ravitaille en _Spanish Fly_! »
###### Annie-France Brazeau
Aye, aye, aye! Mais, dites-moi, avez-vous fait des tests cliniques, Gregory?
###### Gregory Rotheingham
À Elliot Lake, à moins d'une demi-heure d'ici, il y a une armée de retraités qui ne demandent qu'à raviver leur flamme d'antan. On y a testé notre produit et les résultats sont renversants.
###### Annie-France Brazeau
Que s'est-il donc passé?
###### Gregory Rotheingham
Des dames de plus de 50 ans se sont mises à poser nues pour un calendrier1!
###### Annie-France Brazeau
Étonnant. La passion comme moteur de développement économique, c'est original. Où donc s'arrêteront toutes vos élucubrations libidineuses?
###### Gregory Rotheingham
Sur l'île Aird, dans l'archipel de la rivière Spanish, ce paradis terrestre qu'on croyait perdu à jamais.
###### Annie-France Brazeau
Le paradis perdu, vous l'avez donc retrouvé?
###### Gregory Rotheingham
Eh oui! Et on va le transformer en terrain de camping naturiste. Le camping Adam et Ève, pour amoureux à la recherche de sensations fortes.
###### Annie-France Brazeau
Oh!
###### Gregory Rotheingham
Vous savez que des pommiers sauvages poussent sur cette île?
###### Annie-France Brazeau
Vraiment?
###### Gregory Rotheingham
Au pied de chaque pommier, les nudistes pourront planter leur tente. Ce sera comme un retour au Jardin d'Éden.
###### Annie-France Brazeau
Mais c'est fabuleusement romantique, Gregory. un paradis terrestre : la pomme sans le serpent.
###### Gregory Rotheingham
Non, non, les serpents, on va les prendre à Serpent River, le village voisin. Les gens pourront même choisir leur serpent, parmi des serpents à sonnettes, des cobras ou des boas constrictors.
###### Annie-France Brazeau
Ouiiii... et les gens pourront même brûler les serpents.
###### Gregory Rotheingham
Hein?
###### Annie-France Brazeau
Ça leur fera du « boa » de chauffage. Gregory Rotheingham, vous vous moquez de moi. Votre paradis terrestre, personne ne voudra s'en approcher!
###### Gregory Rotheingham
Au contraire, du camping nudiste évoquant le paradis perdu, les gens vont payer jusqu'à 800 $ pour une fin de semaine.
###### Annie-France Brazeau
Huit cents dollars pour un camping naturiste? On n'est pas des tout-nus!
###### Gregory Rotheingham
Bah...
###### Annie-France Brazeau
Mais, dites-moi Gregory, vous allez uniquement miser sur l'attrait de l'Espagne et du paradis terrestre pour attirer des touristes?
###### Gregory Rotheingham
Pas tout à fait, Annie-France. On va aussi miser sur le tourisme occulte. On a à Spanish une marmotte capable de prédire l'avenir.
###### Annie-France Brazeau
Ben là, Gregory Rotheingham, un peu de sérieux. une marmotte qui prédit l'avenir! Quoi d'autre? un porc-épic acuponcteur?
###### Gregory Rotheingham
Annie-France, il y a une marmotte qui prédit l'avenir à Spanish. Elle vivait à Ottawa jusqu'à ce qu'elle se fasse écraser par un autobus de Greenpeace.
###### Annie-France Brazeau
Écoutez Gregory. Si votre marmotte est capable de prédire l'avenir, comment ça se fait qu'elle ne savait pas qu'un autobus s'en venait?
###### Gregory Rotheingham
Annie-France, vous êtes de mauvaise foi! La marmotte a commencé à prédire l'avenir seulement après sa mort.
###### Annie-France Brazeau
De mauvaise foi? Parce que je ne crois pas qu'une marmotte morte,
_elle trébuche deux fois sur ces mots_
bon, qu'un siffleux trépassé puisse prédire l'avenir, je suis de mauvaise foi?
##### Gregory Rotheingham
Parfaitement! Spanish Joe...
###### Annie-France Brazeau
Spanish Joe?
###### Gregory Rotheingham
C'est son nom! Bref, la marmotte marmonne ses visions à son maître, Rick Story, qui les transmet ensuite à l'humanité2.
###### Annie-France Brazeau
Ah là, je comprends! Spanish Joe a rencontré un docteur Dolittle qui parle le langage de la marmotte. Tout s'explique!
_On entend une voix à l'accent espagnol appeler Gregory._
###### Gregory Rotheingham
Annie-France, je l'entends.
###### Annie-France Brazeau
Qui ça?
###### Gregory Rotheingham
C'est Spanish Joe. Il me « télépathise » un message. La transe m'envahit. Il veut communiquer.
###### Annie-France Brazeau
J'ai toujours su que vous aviez des voix dans la tête!
###### Gregory Rotheingham
Non! Spanish Joe me chuchote quelque chose... C'est en espagnol.
###### Annie-France Brazeau
Attendez! Moi aussi, j'entre en communication! Ah non! C'est un mauvais numéro.
1. Authentique. Ce projet visait à amasser des fonds pour l'achat d'équipement hospitalier.
2. Cette histoire paraît abracadabrante, mais Gregory Rotheingham ne l'a pas inventée : le jour de la marmotte, les prédictions de Spanish Joe sont même rapportés par certains médias nord-ontariens.
###### Petite fuite amoureuse
_Il descendait au lac, déboutonnait sa chemise_
_S'étendait sur un quai, se livrait à la brise_
_Qui calmait dans son cœur le si puissant émoi_
_Que lui avait donné l'été de deux mille trois_
_Elle n'y allait pas autant qu'elle voulait_
_Jusqu'à la marina où son chum l'attendait_
_Sur le sofa son père surveillait ses sauvettes_
_Irait-il se coucher finalement, tabarouette?_
Au bord de l'eau
Au bord de l'eau
On peut voir l'horizon
On respire
Et c'est beau
_En ces soirées chanceuses où le père dormait tôt_
_Elle prenait le chemin qui mène au bord de l'eau_
_Sur sa bicyclette rouge, dans sa jupe d'été_
_Ses humeurs mélangées, mi-angoisse, mi-gaieté_
_Elle traversait en flèche, sans le moindre coup d'œil_
_Ce village sans vie, sans lumière, comme en deuil_
_Elle savait que l'amour et l'avenir se joindraient_
_Loin d'ici, à la ville, pas ici, en forêt_
Au bord de l'eau
Au bord de l'eau
On peut voir l'horizon
On respire
Et c'est beau
_Il l'a vue s'approcher comme une voile vers un port_
_Il voulait l'emporter tellement plus loin encore_
_Ils sont partis depuis, comme ils l'avaient promis_
_Jeune amour, petit village, ne font pas bon ménage_
Au bord de l'eau
Au bord de l'eau
On peut voir l'horizon
On respire
Et c'est beau
###
### Salut Sturgeon Falls!
Nipissing-Ouest compte environ 13 000 habitants et 6 000 de ces habitants vivent à Sturgeon Falls. Au contraire de la tendance nord-ontarienne, le dépeuplement est ici moins prononcé qu'ailleurs. La présence française, elle, est bien prononcée. En ville, comme dans les environs, les trois quarts des habitants sont Canadiens français. Et franchement, ça devrait se savoir. Car, dans le passé, les Esturgeois ont montré à tout le pays comment se tenir debout pour le français en Ontario. Il y a eu non pas une, mais deux crises scolaires à Sturgeon Falls.
La plus récente, on s'en souvient bien. Dans les années soixante-dix, des Esturgeois résolus ont dû troubler la paix et braver les fanatiques pour amener les autorités scolaires locales à respecter la loi scolaire provinciale. C'est au prix de tels efforts que la communauté a pu obtenir l'école secondaire française Franco-Cité. Mais la première crise scolaire esturgeoise, on la rappelle moins souvent. Pourtant, c'est celle des deux qui éclaire mieux le présent. Vers 1910, les habitants anglophones et francophones de Sturgeon Falls avaient aidé à relever de la faillite l'usine de pâte et papier, en y investissant des fonds municipaux. (Ça préparait un déjà-vu, oui je sais!) Malgré cet élan de solidarité entre les deux communautés linguistiques, la Sturgeon Falls Pulp Company a hautainement décidé qu'elle ne verserait pas ses taxes municipales à la grande école séparée française, seulement à la petite école publique anglaise. Les Esturgeois ont protesté, dénonçant cette injustice. Ils ont reçu pour réponse une pluie de critiques orangistes à la une de tous les grands journaux de l'Ontario : le _Globe_ , le _Daily News_ , le _Telegram_ , le _Mercury_... Il a fallu des personnes de la trempe de Zotique Mageau, tour à tour maire et député provincial, pour manœuvrer au parlement ontarien et négocier localement de manière à gagner un compromis injuste, mais qui évitait le pire. En 1928, l'Abitibi Power and Paper a racheté l'usine. Eh bien, eux aussi ont aussitôt fait le même coup. Décidément, cette grosse usine n'a pas toujours été la plus respectable des voisines.
À l'automne 2004, l'usine qui était passée aux mains de Weyerhauser est disparue. Malgré des tentatives de rachat et sans grand énervement des gouvernements, la compagnie a démoli une usine rentable. une stratégie corporative exigeait ce sacrifice de Sturgeon Falls. Les affaires sont les affaires. Puisqu'on ne peut pas dire mieux, on ne dira pas adieu. On dira bon débarras. Le véritable esprit de Sturgeon Falls est tout le contraire de cette attitude déplorable. Certainement, le village survivra au départ de son usine. « Sans faute », c'est la devise officielle de la ville.
Jadis, le grand cœur communautaire qui bat toujours à Sturgeon Falls s'est incarné dans un homme qui avait le physique de l'emploi, un homme plus gros que nature. Il pesait, à ce qu'on dit, au moins trois cents livres. Il a été vicaire de la paroisse Sacré-Cœur à partir des années quarante jusqu'aux années soixante. Mais il a été surtout une institution ambulante. Il s'appelait Joseph Eugène Marchand. On l'appelait respectueusement « le gros père Marchand ».
On l'a apprécié, ça j'en suis sûr. Car dans ma petite enquête, la première réponse que tous m'ont donnée a été un rire sympathique et nostalgique. Mais avant de faire son portrait, une mise en garde : notre homme a dit des choses qui dans la bouche d'un autre seraient moins bien acceptées. Je vais pourtant devoir les répéter. Car c'est dans ses blagues au goût douteux qu'est le paradoxe du personnage. Fermez les yeux et imaginez un gros monsieur qui est parfois un peu grossier et un peu malpropre, mais qui est toujours très aimé de tous. Puis là, mettez-lui un col romain!
L'imposante présence du père Marchand était en soi un enseignement spirituel. Son humour enseignait mieux qu'un sermon la tolérance et la générosité. Ces vertus ne viennent qu'aux humbles. Elles appartiennent à ceux qui ne se prennent pas trop au sérieux.
###### Joseph Marchand
Le gros père Marchand, on l'appelait _Half-Tonne_. On l'appelait _Five-by-Five_ : cinq pieds de haut, cinq pieds de large. Il était gros, mais il ne s'en faisait pas pour ça. Au cinéma Odéon, le propriétaire lui avait fait une place spéciale, parce qu'un fauteuil ordinaire n'aurait jamais fait l'affaire. Il mangeait son pop-corn, buvait son Coke et passait ses commentaires. Il avait beau faire noir, on savait qu'il était là. Dans le noir du confessionnal aussi, on savait qu'il y était. Aux enfants qui avouaient un petit péché, il répondait d'une voix tonitruante que toute l'église entendait : « T'as pas fait ça toi! » Tu sortais de là rouge comme une tomate. Il n'épargnait pas les adultes non plus. une fois, un homme s'est confessé d'avoir conté des blagues salées. Le père Marchand a répondu : « Pas celles que je t'ai contées au moins? »
Mais pour le voir dans la lumière du jour, coiffé de son chapeau de paille blanc et débordant de sa chaise, il suffisait de passer par la pharmacie Maranda. C'est là qu'il trônait tous les avant-midi, en empestant la place de ses cigares Punch et en apostrophant les clients. L'été sa chaise était sur le trottoir. une fois, un jeune homme passait avec son bébé qui pleurait. Le père Marchand l'a taquiné en disant : « Tu as eu ton _fun_? Paie pour à c't'heure! » L'hiver, sa chaise était à l'intérieur, près du comptoir. Là, on a pu le voir décacheter ses cartes de Noël et tonner « Espèce de _cheap_! » si elle ne contenait pas de billets de banque.
Quand le hasard amenait un touriste américain, il était content, car il était d'origine franco-américaine. Certains étés, il a loué pour tout un mois le taxi de Gustave Leblanc, qui le conduisait jusqu'à Boston, où il visitait sa sœur et allait à l'opéra. Ses fameux cigares Punch, il en a tant fumé qu'un jour le représentant lui en a offert une boîte de cent, en guise d'hommage à un _gros_ client.
Quand il n'était pas à la pharmacie, il était là où il y avait du monde : aux tournois sportifs, aux noces, aux banquets. une fois, dans une noce, c'était une demoiselle Maisonneuve qui prenait époux. Le père Marchand a pris la parole pour dire : « Regardez-moi ce garçon chanceux. C'est rare, un nouveau marié qui prend possession d'une maison neuve! » En vérité, le père Marchand ne l'a pas dit exactement comme ça. Sa formule était plus grivoise que la mienne. Je vous la laisse imaginer.
En tout cas, c'était un rare prêtre qui ne se gênait pas d'entrer à l'hôtel. Bien sûr, il était secrétaire du club Kinsmen, ça lui donnait un prétexte. Il a aussi été un si bon Chevalier de Colomb qu'en 1962, ceux-ci ont honoré son dévouement. Quand même, on voyait le gros père Marchand bien moins souvent assis dans l'hôtel que debout dans l'église devant l'autel. Pour dire la messe, il avait sa manière à lui qui était sans longueurs. Trop gros pour faire ses génuflexions, il se contentait de tourner un peu le pied et de plier un peu le genou. Son sermon commençait par une phrase en latin, selon la coutume de l'époque. Mais la suite déboulait si vite qu'on aurait cru à... du latin! Il avait une entente avec Charley Goudreau, le maître de chant dans le jubé, qui levait son bras en montrant sa montre après huit minutes. Là, le père Marchand bouclait son sermon en laissant son idée en suspens. À la communion, il pouvait bien vous dire très fort, faussement impatient : « Envoye, sors-la, ta langue! »
C'était encore l'époque de la viande interdite le vendredi. Mais pour ça, il n'était pas de son époque. Ce n'était vraiment pas son genre de religion. Il n'était pas curé, mais simple vicaire et ça devait bien faire son affaire, parce qu'en fait d'administrateur, on en a vu des meilleurs. Il avait été curé à Blind River. Là-bas, on se souvient de l'avoir vu payer ses dépenses avec des enveloppes de la quête sorties de sa poche de veston. Sa comptabilité n'était pas très serrée. On pourrait facilement imaginer que pour des raisons de décorum, on n'ait pas voulu du père Marchand comme prêtre. Mais si c'était arrivé, il y aurait eu à Sturgeon Falls quelques âmes sauvées en moins. Il y en a qui n'ont jamais vu le dedans d'un confessionnal, mais qui ont jasé de choses qui comptent en d'autres lieux appropriés, avec un homme qui n'allait pas les intimider.
Bien sûr, le gros père Marchand aimait son boire et son manger. Plus d'un enfant de chœur se souvient de s'être fait dire en pleine messe : « Verse, verse, c'est pas toi qui paie! » ou bien : « Mets-en, c'est pas de l'onguent! ». un petit magasin du coin se trouvait près de l'église. Plus d'un enfant de chœur s'est fait dire : « Vas-y me chercher un _cône_ de crème à glace ». Ou mieux encore : « Cours chez monsieur Maranda emprunter 20 $ pour moi ». une autre de ses plaisanteries était sa petite formule anglaise qui annonçait que les restes allaient finir dans son assiette. Il disait : « _If it's going to go to waste, it might as well go to mine._ »
Un avant-midi, un séminariste a visité la chambre du père Marchand. Celui-ci s'est versé un verre sans en offrir à son visiteur. C'était du whisky de marque Teacher's. « Faut que tu comprennes, que le père a expliqué, Teacher's ça veut dire que c'est pas pour des enfants d'école. » une fois, le père Marchand mangeait un steak au restaurant, c'était chez Chink's ou chez St-Amand. Et le pauvre père s'est échappé... que voulez-vous, il a pété! Dans le silence gêné de la tablée, qu'est-ce qu'il a dit en regardant son steak? Il a dit : « Beugle tant que tu voudras, je vais te manger pareil! »
Voilà le genre de souvenirs qu'a laissés le gros père Marchand : des plaisanteries que les aînés se racontent encore cinquante ans plus tard. Cet homme que la nature a éprouvé quant au physique l'a compensé par l'esprit comique. On dirait que parce qu'il devait s'accepter, il savait se faire accepter des autres, quelles que soient leur langue ou leurs croyances. Dans une ville où, par le passé, l'intolérance a nourri des injustices, son souvenir n'est pas d'un innocent symbolisme. C'est comme s'il y était pour y montrer que la paix du cœur vient à ceux qui savent vraiment être plus présents aux autres qu'à soi-même.
Son entourage le lui a parfois bien rendu. un immense homme comme lui salissait vite ses vêtements, qui étaient durs à trouver et coûtaient bien plus cher qu'à l'ordinaire. Eh bien, ses amis se cotisaient pour lui acheter de temps en temps des habits neufs de _clergyman_ faits sur mesure pour lui à Hamilton. On dit qu'ils étaient surdimensionnés de 60 %! C'est le genre de faveur qu'on fait à un homme qu'on estime. Mais le paradoxe de cette estime mérite qu'on le médite. Comment un homme si gros pouvait-il être si habile à réduire le poids de vos problèmes? Comment son humour populacier d'homme de la rue pouvait-il inspirer un respect particulier pour l'homme de l'Église? En se moquant du savoir-vivre, il enseignait à vivre. C'est tout à l'honneur de Sturgeon Falls que cette communauté a si bien accepté et tant aimé ce gros homme pour qui l'entregent n'obligeait pas au raffinement, un homme pour qui la bonne humeur valait mieux que les bonnes apparences.
Le père Marchand a continué à plaisanter jusqu'après sa mort. Dans son testament, à l'endroit où la formule traditionnelle dit « _I leave all my wordly possessions_ », lui a fait écrire : « _I leave all my worthless possessions_. » Mais au contraire de ce qu'il a écrit, le père Marchand a légué à sa communauté le bien le plus précieux. Les plus belles pages des livres d'histoire locale sont généralement consacrées aux bâtisseurs d'institutions. Or l'institution du père Marchand a été sa relation terre-à-terre avec sa communauté. Il a été comme la pincée de sel dans la soupe, comme le petit chocolat qu'ont sort quand la visite est là. Il a mis dans la vie un peu plus de couleur locale et de chaleur humaine. Il a pu faire tout ça parce qu'il a aimé son monde et leur petit coin du monde. Ce n'est pas innocent comme symbolisme ça non plus. C'est pour ça qu'il aura une page dans le petit livre d'histoire du _Salut de l'arrière-pays_. Sa grosse chaleur humaine a fait du bien dans le passé. Aujourd'hui je la rappelle, pour qu'elle inspire aussi l'avenir.
###### L'été qui s'étend
_Tu me dis qu'il fera beau, qu'il fera très beau demain_
_Que le printemps appelle, que je m'y glisserai enfin_
_Je prépare le canot, nous partirons très tôt_
_Sur le lac, dans le froid crépuscule du matin_
Ici, dans le Nord, l'hiver prend tout son temps
Mais ma vie avec toi, c'est l'été qui s'étend
_À quoi tient ce mystère, que t'aies voulu de moi_
_Toi qui chasses l'hiver, toi qui casses le froid_
_C'est en toi que je me vois avancer dans ma vie_
_Tu le sais, t'en réjouis, me regardes et souris_
Ici, dans le Nord, l'hiver prend tout son temps
Mais ma vie avec toi, c'est l'été qui s'étend
_Ton regard m'avait dit : je serai ton pays_
_Ses forêts, ses rivières, ses menaces, ses mystères_
_Ta nature est en moi, tu es mon point de repère_
_Ma confiance est en toi, tu as vaincu l'hiver_
Ici, dans le Nord, l'hiver prend tout son temps
Mais ma vie avec toi, c'est l'été qui s'étend
###### Avance
Excusez-moi, est-ce que je suis en retard? J'étais au Ti Morton, à jaser avec les copains, et dans ce temps-là, on voit pas le temps passer...
Bon, fais-nous un petit conte, qu'ils m'ont dit. Les contes, ça commence généralement avec « Il était une fois ». Il y en a de toutes les sortes : les contes drôles, les contes héroïques, les contes d'aventure, les contes d'amour... Les contes grivois, aussi : ceux-là viennent de la France, je pense bien. Celui-ci, ben, je sais pas trop quelle sorte que c'est, vous me donnerez votre opinion tout à l'heure, si ça vous tente.
Bon. « Il était une fois », il y a très, très longtemps, dans un village loin, loin de Sturgeon Falls... disons, l'été passé, au lac des Cèdres — OK, j'ai menti pour « il y a très, très longtemps » et pour le village, faut qu'il y ait des menteries dans les contes, vous savez — il était une fois, donc, des copains qui se tenaient ensemble souvent, quasiment tout le temps. Ils s'appréciaient, dans le fond on pourrait dire qu'ils s'aimaient dur comme fer, même s'ils ne l'auraient jamais dit à haute voix. Voyez-vous, il y a des choses qui se disent pas, ou difficilement, puis pour les copains, les gestes comptent sans doute plus que les paroles. Non pas qu'ils aiment pas parler, jaser, rire aussi — rire souvent!
Début de l'été, il fait beau, l'hiver est comme un lointain souvenir d'enfance, quelque chose qui ne se répéterait plus jamais. Les copains organisent une petite partie de pêche, pis partent. Ils arrêtent presque tout de suite : ça leur prend des frites. Comme d'habitude, il y en a deux qui les prennent au P'tit Riv, les deux autres chez Larry's. Finalement ils ne partent pas si tôt que ça... Et en montant au lac, de chaque côté de la route, les feuilles sont de ce vert encore tendre et plein de promesses, alors que celui plus profond des conifères rappelle le drapeau franco-ontarien.
Le Grand, ainsi nommé parce qu'il mesure à peu près cinq pieds dix-huit pouces, dit : « Parlant du drapeau franco-ontarien, ça me fait penser à la crise d'octobre. »
Yvan lui dit : « Ça fait aucune espèce de bon sens. De quoi tu parles? »
Yvan n'est pas aussi grand que Le Grand — forcément —, mais c'est un grand solide pareil. Quand ils jouaient encore au basket, Yvan pouvait tenir le ballon d'une seule main — Le Grand, lui, tenait le panier! — puis quand Yvan travaille avec ses deux mains, c'est le genre capable : réparer un moteur, monter une maison, il peut tout faire. un gars en or.
Margot — ah, Margot! — Margot dit : « Le Grand fait toujours du sens... à sa façon. Il doit être dans ses souvenirs. »
Margot est ben fine, ben belle, souriante, intelligente, indépendante. Il y a beaucoup de gars autour qui ont essayé de la dompter, dans le temps, mais c'était aussi difficile que de faire marcher des poules en file indienne. Puis à chaque fois qu'un gars se faisait brave avec elle — ou tendre, ce qui arrivait moins souvent —, elle le regardait droit dans les yeux avec ses beaux grands yeux gris à elle, et lui disait, doucement: « Je sais pas ce que je ferais sans toi, mais j'aimerais essayer. » C'était jamais cruel ou méchant, puis elle était toujours une amie fidèle; les gars comprenaient à la longue. une fille... moderne, on peut dire.
« Je poigne pas », dit Avance.
C'est de famille, ça. Avance, de son vrai nom Jean-Pierre, comme son grand-père, un bûcheron et personnage haut en couleur venu ici de l'Abitibi pour faire de la terre en Ontario. C'était un homme fort, vraiment fort, qui avait un cheval, un gros percheron, qu'il appelait Hercule. Mais il avait de la misère avec, disait-il, lui qui avait du cœur au ventre, qui était travaillant pas à peu près, mais qui avait peu d'expérience comme cultivateur. « Je dessouchais, disait-il à qui voulait l'entendre, puis je lui fais "Avance, Hercule"! Et ce maudit cheval-là fait un pas en avant, puis un pas en arrière, pis il bouge pus, bonyeu, il comprend rien! » Et son auditoire s'esclaffe.
Il y en a toujours ben un qui fait, « As-tu remarqué s'il se grattait la tête? ». _Anyway_ , après ben du taquinage, le grand-père Jean-Pierre a fini par comprendre, pis le cheval aussi, après que son maître l'a rebaptisé... « Toé ».
Toujours est-il que le copain Jean-Pierre se fait appeler Avance depuis longtemps parce qu'il poigne une idée, « Ah, je vais faire ça » il dit, puis il commence à douter de lui-même, à se dire que c'est peut-être pas une bonne idée après tout, puis il branle dans le manche à n'en plus finir. Alors les copains ont fini par lui répéter « Avance, Hercule ». On l'a appelé Hercule un boutte, mais lui disait: « Je ne suis pas un cheval, voyons! ». Pis Margot lui disait : « Mais tu te grattes la tête comme le cheval de ton grand-père! » Ça l'a froissé un peu — quand il l'a poignée —, alors on a fini par lui dire, quasiment tout le temps, « Avance, Jay Pee », puis à un moment donné on a commencé à l'appeler Avance, tout simplement, qui était comme plus positif. On ne voulait plus qu'il recule. Certains l'ont même anglicisé, pour l'appeler Vance, mais ça a pas poigné, surtout avec Jay Pee qui semblait pas comprendre l'astuce, et on a laissé tomber. Surtout que, dans le temps, il y avait déjà ben assez d'anglais autour, même trop, au dire de certains.
Bon, j'en étais où, là? Ah oui. Alors là, ils sont rendus au lac. Installés dans le bateau d'Yvan, ils s'en vont taquiner le doré.
— Vous êtes _slows_ , ben petits, pis vous avez les yeux ben _weirds_ », dit Le Grand.
— De quoi tu parles encore?, demande Yvan.
— Je taquine le doré, répond Le Grand.
— Je poigne pas, lance Avance, comme d'habitude
— Pas grave, dit Margot, elle n'était pas si drôle que ça.
Mais à force de les taquiner, on finit par les trouver, puis par en prendre, et Avance, qui n'est pas le meilleur pêcheur de la gang, en rentre un beau de cinq livres, et Le Grand dit : « C'était le mien, ça! ». Et tous de rire, sauf Avance, qui poigne pas. « une chance que tu poignes du poisson, au moins », dit Margot, tout en sourire. Et à force de prendre du poisson sous le soleil, on finit par avoir soif et faim. Ils mangent du bon doré frais, et du riz, parce que Margot a dit : « Pas de patates deux fois aujourd'hui, OK? » et personne n'a voulu la contrarier, ça vaut pas la peine.
Confortablement installé, une petite bière à la main, Le Grand, grand philosophe à ses heures, s'étire et dit : « Ouais, c'est ça qui est ça. » Pis là, sur le même ton, il fait : « Avance, comment s'appelait la petite fille de Verner que tu sortais avec dans ce temps-là? »
Avance était tellement fou d'elle que les copains l'avaient baptisée sa « maladie vernerienne ».
— Oh, ben là... fait Avance, scandalisé. Pis toi et ta Sainte-Charlotte?
— Comment on appelle les gens de Cache Bay? poursuit Le Grand, visiblement sur sa lancée.
Margot le sait très bien, car ils se content les mêmes farces plates à chaque partie de pêche depuis la première.
— Je sais pas, elle répond, les Invisibles?
— Puis les gens de Field? enchaîne Yvan, embarquant dans le jeu.
— Euh... les Champêtres? fait Le Grand.
— Je poigne pas, dit Avance.
— T'oublies, c'est tout, fait Margot.
— Un vieux de Lavigne?
— Un raisin sec!
— Un pacifiste anglophone de Warren?
— En chœur, la réponse : « Warren Peace! »
— Pis un résident de North Bay?
— Un sacram...
— Bon, ça va faire, dit Margot, c'est pus drôle.
— Bon, dit Yvan, les gens de Sturgeon d'abord?
— Ben, nous-autres, grands comme on est, pis pas juste moi là, répond Le Grand, je crois que les Esturgeons ça suffit, c'est des maudits grands poissons qui ont donné leur nom à la rivière et au village.
Là, le soleil commence à se coucher en prenant tout son temps. Le Grand s'étire de nouveau et dit :
— C'est comme je disais en montant ce matin.
Pour une fois, les trois autres sont synchronisés : ils se tournent vers lui en même temps et disent en chœur :
— _What_?
— La crise d'octobre, la vraie, la nôtre, enchaîne Le Grand. 1971. Saint-Pierre de Porc-Épic! Y a toujours ben des _imites_!
— Ah non, pas encore! dit Yvan.
Vous savez comme moi, la loi provinciale était en place et toute. Pis la Nipissing Board of Ed., gros poisson qui avait avalé le Sturgeon Falls Board, avec 11 membres de North Bay pour nos trois crapets-soleils de Sturgeon, décide que non, on l'aura pas notre école. Mais notre monde était pas près de croire qu'eux autres savaient mieux que nous autres ce qui était bon pour nous autres. Les gens étaient écœurés de se faire écœurer. Oui, comme dit Le Grand, « il y a toujours ben des _imites_! »
— Tu parles, dit Margot.
Et cette décision, c'était en effet la limite : quatre jours après la décision, 90 % des élèves francos boycottaient les cours et manifestaient à l'extérieur de l'école avec leurs parents et des membres de la communauté. Ça prenait du courage pareil!
— Ouain, j'étais là, moi, dit Avance.
Monsieur V. avait fermé le magasin et dit à son employé, Avance, de l'accompagner. Jay Pee pensait qu'une école française, c'était une bonne chose, puis tout de suite après se demandait si ça valait la peine si ça allait déchirer la communauté. Mais il se faisait payer pareil, puis il en a appris de tout ce beau monde convaincu de la justesse de leur cause. Payé pour apprendre...
— C'est un petit peu la faute à Yvan aussi, dit Le Grand.
Dans ce temps-là, les élèves étaient vraiment embarqués, chauffés à bloc. Ils arrivaient pas à dormir, alors ils se ramassaient ensemble chez l'un ou l'autre, les parents les laissaient faire, les encourageaient même. Ça fait que, des fois, ils finissaient par s'endormir en classe, quand ils allaient encore en classe. À un moment donné, Yvan — ben, « Monsieur », comme ils appelaient le nouvel enseignant — aurait dit comme ça : « Vous me dormez en pleine face, vous feriez mieux d'aller vous coucher dans le corridor, c'est plus confortable et il y a plus de place. » Et là, la petite sœur de Margot regarde Le Grand d'un air espiègle...
Alors quand le Directeur de l'Éducation — le grand fromage — a fini par daigner venir, pensant tout régler par sa seule et unique éminente présence, il a eu plus de misère qu'il escomptait. Pensez-y comme il faut : il y avait dans ces années-là bien des _love-in_ , puis John Lennon et Yoko Ono ont fait leur _bed-in_ pour la paix à Montréal. Alors, les élèves ont fait un _sleep-in_ — comme ça ils pouvaient protester tout en se reposant un peu. Monsieur Fromage a dû enjamber une bonne centaine de jeunes corps franco-ontariens pour arriver jusqu'au bout du corridor. Comme acte de résistance non violente, c'était vraiment pas pire. Imaginez si Ghandi était venu à Sturgeon, sûrement qu'il l'aurait trouvé bonne. Puis quand le niaiseux de romancier de Montréal, bien des années plus tard, a dit que les Franco-Ontariens étaient « des cadavres encore chauds », les copains ont bien ri, pensant à cette belle gang de jeunes, pas du tout morts pour la cause! Pas endormis non plus, je vous le dis! Mais oui, « encore chauds », ben chauds et ben vivants!
— Ouais, enchaîne Margot. Ma petite sœur m'a dit que le bonhomme leur criait « _Get up! Get up!_ ».
Combien de fois a-t-elle déjà été contée celle-là? Il était juste devant elle. Alors, du plancher, elle le regarde avec un sourire malicieux et lui dit « Nos mamans nous réveillent EN FRANÇAIS, _Mister Cheese_. » Puis lui, choqué, de répondre, du haut de sa hauteur, « _My name is not Cheese_. » Puis la petite sœur, du tac au tac : « On le saurait pas à sentir vos pieds. » Des fois, Margot fait dire plutôt à sa petite sœur à ce moment-là : « _Maybe not, but it's soon going to be Mud_. »
Oui, à bien des niveaux, c'était un moment extraordinaire. Les gens commençaient à comprendre qu'ils avaient du pouvoir, il fallait seulement apprendre comment s'en servir, tenir bon, tous ensemble. J'ai commencé à le comprendre aussi.
— Et tu donnais pas ta place, toi, dit Yvan en faisant un clin d'œil à Margot.
Et bien sûr, il parle de la fameuse réunion du personnel. un petit air bête du Nipissing Board est venu dire qu'il était interdit que les enseignants prennent position parce qu'ils étaient des employés du Conseil. Et Margot, toute souriante — c'est de famille — de dire : « Employés ou pas, on veut notre école, et on va l'avoir. C'est pas parce qu'on est capable de parler anglais qu'on aura pas une école française. _We're bilingual, it makes us twice as wise. You should try it, you've really got nothing to lose_. » L'air bête a regardé Margot bouche bée, puis Margot est repartie de plus belle : « Avez-vous lu _Alice in Wonderland_ au moins? », et prend tout son temps, avant de lancer « _Off with his head!_ »
Ah ben là! Entre les fous rires et les cris et le grand branle-bas de combat, le gars finit par battre en retraite, au son de Margot qui ajoute, comme ça : « Il doit pas lire beaucoup, lui. »
— Vous savez, je n'ai jamais osé le dire, mais j'avais vraiment peur de perdre mon poste, dit Yvan. J'admirais tellement ton courage, Margot.
— Je pense pas que j'en avais tant que ça, dans le fond. Mais la communauté était une vraie inspiration, les jeunes comme ma petite sœur, les parents, les gens comme Monsieur V., ils y croyaient dur comme fer. Il fallait avoir l'air courageux.
— J'étais pas trop sûr quoi en penser, dit Avance. Mais Monsieur V. m'en parlait d'une façon bien tranquille et bien convaincante. Je pense bien qu'il m'a donné la confiance de prendre position pour la première fois de ma vie.
Le Grand grandissait tellement vite, à ce moment-là, qu'il faisait quasiment rien d'autre que manger, jouer au basket, et dormir en classe. Il dit : « Moi, ça me dérangeait pas de m'allonger dans le corridor. » Puis il s'étire encore une fois — quand Le Grand s'étire, il s'étire longtemps — et dit ce que tous savent qu'il va dire : « Ouais, c'est ça qui est ça. »
Et voilà comment les copains ont passé cette partie de pêche là, à rire beaucoup, comme d'habitude. Ils avaient assisté aux funérailles de Monsieur V. le matin même, avant de monter au lac. Ils avaient tous perdu un ami, une espèce d'ange gardien — surtout Avance — et c'est sûrement la grande émotion qu'ils éprouvaient qui a fait brasser la cage de leurs souvenirs.
Ben là, faut que je parte. Je vais retourner au Ti Morton retrouver les copains, pis je vais prendre un gros Morton — triple triple! — , on fait pas les choses en petit ici. Je vais retrouver Le Grand, et Yvan, et Margot. On va rire un peu.
_By the way_ , un jour, après bien des hésitations, j'ai fini par dire à Margot ce que je voulais lui dire depuis longtemps. C'était justement lors d'une partie de pêche, après notre « crise d'octobre », en se promenant un peu sur le bord du lac, après le souper. J'avais finalement osé me confier à Monsieur V., puis il m'a encouragé à suivre mon cœur. Et Margot m'a répondu, comme je le craignais, avec son drôle de sourire: « Jay Pee, je sais pas ce que je ferais sans toi, mais j'aimerais essayer ».
Mais devinez quoi? Elle a jamais essayé! Et on est ensemble depuis. Et maudit qu'elle est toujours aussi fine, belle, souriante, intelligente, et d'esprit indépendant. Même plus qu'avant. Et toujours un des copains, même si elle serait plutôt... ma copine. Voyez-vous, moi, Jay Pee, j'ai fini par... avancer. Et c'est beaucoup grâce à Monsieur V., aux gens de Sturgeon, aux copains, et à Margot. « Avance, Jay Pee, avance », qu'ils me disaient...
Là, mon conte est fini. Excusez-le.
_Robert Dickson_
###### _Incitatif pédant :_ Sturgeon Falls, la Ville-lumière des pommes de terre
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau, vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat. Vous avez notamment écrit _Liche et relish pour s'mettre riche à Spanish._ Vous avez également travaillé dans un cirque. Vous avaliez des couteaux de dix pouces, mais vous avez tout abandonné sous prétexte que ça vous coupait l'appétit.
_Gregory glisse un papier à Annie-France._
Vous avez également une piscine hors terre à vendre, elle n'a presque pas servi, elle est comme neuve et vous êtes prêt à considérer toute offre sérieuse, veuillez contacter Gregory au 705-67... Voyons, qu'est-ce que c'est que ça?
###### Gregory Rotheingham
674-4...
###### Annie-France Brazeau
Et je pourrais continuer longtemps comme ça, Gregory Rotheingham, mais je pense que nous en savons déjà beaucoup trop sur vous! Bonjour.
###### Gregory Rotheingham
Bonjour, Annie-France.
###### Annie-France Brazeau
Gregory, vous venez nous présenter un tout nouveau rapport sur le développement économique de Sturgeon Falls.
###### Gregory Rotheingham
Oui!
###### Annie-France Brazeau
Et j'imagine que vous allez détourner des autobus de touristes japonais de Niagara Falls pour leur faire visiter Sturgeon Falls.
###### Gregory Rotheingham
L'idée est bonne, mais il n'y a plus de chutes, elles ont été noyées par le barrage électrique y a longtemps.
###### Annie-France Brazeau
Mais c'est une manie chez vous, il n'y avait pas de chutes à Iroquois Falls, pas d'Espagnols à Spanish, pas de raisins à Lavigne. « Sud-bury » est au nord et, là encore, pas de chutes à Sturgeon Falls. Gregory, est-ce que le Nord de l'Ontario s'est bâti sur de la fausse représentation?
###### Gregory Rotheingham
Non, le Nord de l'Ontario est quand même Au nord ET en Ontario, alors on ne peut pas parler de fausse représentation.
###### Annie-France Brazeau
Alors, si vous n'utilisez pas les chutes comme moteur de développement économique, sur quoi allez-vous miser, Gregory?
###### Gregory Rotheingham
Annie-France, à Sturgeon Falls, on ne mise pas. On a essayé, et le gouvernement a quasiment dit oui, mais ils ont dit : « _No casino_. ».
###### Annie-France Brazeau
Je gage que vous avez trouvé autre chose.
###### Gregory Rotheingham
Je gage pas, mais j'ai trouvé le _jackpot_. On va se concentrer sur la chose qui a mis Sturgeon Falls sur la _mappe_ : les cabanes à _chips_.
###### Annie-France Brazeau
Des _chips?_ Si je comprends bien, vous allez ouvrir des kiosques pour vendre des croustilles.
###### Gregory Rotheingham
Pas des croustilles, Annie-France. Ici, ce qu'on appelle des _chips_ , chez vous on appelle ça des patates frites!
###### Annie-France Brazeau
Ah bon, je comprends. Et comment avez-vous intitulé votre rapport?
###### Gregory Rotheingham
_Imagine la cantine à poutine de Sturgeon._
###### Annie-France Brazeau
Hou, y a du travail là-dedans.
###### Gregory Rotheingham
Absolument!
###### Annie-France Brazeau
Mais je vous avertis, Gregory, le marché de la frite est très compétitif : il y a les frites belges et les _New-York fries_ et qui sont toutes deux très tendance. Pensez-vous pouvoir vous démarquer?
###### Gregory Rotheingham
Patate que oui, patate que non.
###### Annie-France Brazeau
Une autre comme ça et je vous embarque dans la friteuse. Ça va faire du Gregory frit.
###### Gregory Rotheingham
Sérieusement, on est confiant. Voyez-vous, la Belgique et New-York ne se sont pas diversifiées, elles ont une seule sorte de _chip_. Nous, on devient le « Village global de la patate ». Nos cabanes à _chips_ seront le reflet de toutes les tendances politiques et sociales.
###### Annie-France Brazeau
Des cabanes à frites politiques et sociales. Bravo Gregory, c'est la septième fois que vous venez à cette émission et vous réussissez toujours à faire pire.
###### Gregory Rotheingham
Ben non, regardez, pour commencer, on va ouvrir une cabane qui va s'appeler _La reine de la patate_.
###### Annie-France Brazeau
J'vous arrête tout de suite, Gregory, _La reine de la patate_ , y en a environ 382 au Québec. Votre concept n'est pas très original.
###### Gregory Rotheingham
Attendez. Pour entrer chez _La reine de la patate_ de Sturgeon, les clients passent sur un pont-levis, ils vont avancer sur un tapis rouge et toutes les _french fries_ sont coupées à la guillotine.
###### Annie-France Brazeau
Et j'imagine que les clients vont s'asseoir sur des trônes?
###### Gregory Rotheingham
Ah non, le trône, c'est comme partout ailleurs : une heure après le repas.
###### Annie-France Brazeau
Quel raffinement! Mais vous n'avez pas peur que les clients antimonarchistes boycottent Sturgeon Falls?
###### Gregory Rotheingham
Non, parce qu'ils peuvent aller dans une autre cabane concept : _La république démocratique de la patate_.
###### Annie-France Brazeau
Une autre cabane, stupéfiant!
###### Gregory Rotheingham
Ici, le spécial de la semaine est choisi démocratiquement par un vote des clients. Cette semaine, le spécial, c'est la poutine végétarienne.
###### Annie-France Brazeau
De la poutine végétarienne?
###### Gregory Rotheingham
Ben, oui... c'est de la salade César.
###### Annie-France Brazeau
Et j'imagine que de l'autre côté de la rue, il y aura une cabane à frite communiste.
###### Gregory Rotheingham
Bien sûr! Les clients pourront vivre une authentique expérience communiste : on vous promet une file d'attente in-ter-mi-na-ble, et ici la cabane appartient aux employés prolétaires qui sont tous barbus, même les femmes.
###### Annie-France Brazeau
Enfin, un nouveau débouché pour les femmes à barbe.
###### Gregory Rotheingham
Ici, on ne demande pas du ketchup, on crie : « Amenez-moi l'Armée rouge! »
###### Annie-France Brazeau
Hamburgers de tous les pays, unissez-vous!
###### Gregory Rotheingham
Et pour terminer, on aura _La dictature de la patate_.
###### Annie-France Brazeau
Arrêtez, je me rends!
###### Gregory Rotheingham
À l'entrée, il y aura une grosse statue de Saddam Hussein et, là, il n'y a pas de choix, la serveuse décide tout : « Toi, tu vas prendre deux hot-dogs au vinaigre extra moutarde. _Enwoye_ , donne-moi 7 $. »
###### Annie-France Brazeau
Mais c'est monstrueux!
###### Gregory Rotheingham
Non, c'est du marketing!
###### Annie-France Brazeau
Est-ce que vous avez d'autres projets?
###### Gregory Rotheingham
_montrant un sachet de sauce_
Ben oui. Vous savez c'est quoi, ça, Annie-France?
###### Annie-France Brazeau
C'est un sachet de sauce BBQ en poudre.
###### Gregory Rotheingham
Ben, l'été, on va vider deux « 18 roues » de poudre dans la rivière Sturgeon. Les touristes vont pouvoir se baigner dans le _gravy_!
###### Annie-France Brazeau
Oui! Qui n'a jamais rêvé de prendre la place d'une frite dans une poutine.
###### Gregory Rotheingham
_dynamique_
Exactement!
###### Annie-France Brazeau
Mais quelle sorte de drogue avez-vous consommée pour arriver avec ces plans de fous?
###### Gregory Rotheingham
Non, non, pas des plans de fous, des plants de _pot_. C'est justement le dernier élément de notre stratégie de marketing : la marijuana!
###### Annie-France Brazeau
Arrêtez. Vous m'effrayez, Gregory. Si on vous laisse faire, Sturgeon Falls deviendra un véritable enfer.
###### Gregory Rotheingham
Au contraire, Annie-France, on va inviter tout le monde à se convertir à la nouvelle Église de l'univers.
###### Annie-France Brazeau
L'Église de l'univers? C'est quoi, les gens communient avec des étoiles?
###### Gregory Rotheingham
Non, avec des joints.
###### Annie-France Brazeau
Quoi?
###### Gregory Rotheingham
C'est une religion à Sturgeon Falls. Le révérend Michel Éthier intègre la marijuana dans ses messes et il les célébrera tous les jours à 11 h et à 16 h, juste avant l'heure du repas. Après la messe, les gens vont avoir faim pis ils vont se garrocher sur nos cabanes à _chips_1.
###### Annie-France Brazeau
Machiavélique, Gregory! Mais, il n'y a pas assez de gens à Sturgeon Falls pour faire vivre tant de cabanes à frites. Vous aurez besoin de beaucoup de touristes et ils ne vont pas se déplacer uniquement pour vos « patates frites ».
###### Gregory Rotheingham
Mais les touristes, on va les attirer avec le musée Sturgeon River House, une de nos fiertés locales.
###### Annie-France Brazeau
Ah oui, je l'ai visité. C'est très mignon, ce petit poste de traite. Très tendance pour les touristes européens.
###### Gregory Rotheingham
Ben justement, des Européens, des Américains, même des Africains viendront admirer les améliorations que j'ai apportées à notre musée.
###### Annie-France Brazeau
Ah quel dommage! C'était tellement beau. Enfin, dites-moi donc comment vous allez saboter le musée?
###### Gregory Rotheingham
On va informer les gens sur le rôle joué par les Amérindiens dans l'histoire des cabanes à _chips_. On va recréer le premier wigwam qui servait des _chips_ aux voyageurs qui passaient sur la rivière. Et en plus des wigwams à frites, les Amérindiens ont inventé le ketchup.
###### Annie-France Brazeau
Ben voyons donc!
###### Gregory Rotheingham
C'est une recette ancestrale où on recueille de petites baies rouges qui poussent dans les marais près du musée. On les écrase pour en faire du ketchup.
###### Annie-France Brazeau
Ce que vous décrivez là, c'est pas du ketchup, c'est de la sauce aux atacas.
###### Gregory Rotheingham
Entéka.
###### Annie-France Brazeau
Personne va vouloir mettre ça sur ses frites et encore moins croire que c'est du ketchup amérindien.
###### Gregory Rotheingham
Vous l'avez dit, il y a juste les touristes qui vont venir, ils verront pas la différence! Et on va tellement les impressionner, parce que Sturgeon Falls c'est comme Paris, mais plus beau.
###### Annie-France Brazeau
Je ne veux pas dénigrer Sturgeon Falls, mais quand même, Paris, c'est la capitale de la mode, de la culture...
###### Gregory Rotheingham
Voyons, Annie-France, on a tout ça à Sturgeon : la « _mud_ », on en a. On a même des _mud bogs_ , et la culture, on a des cultures de soya, de canola pis de patates. Et à Paris, il y a un Moulin Rouge, ben, on en a un ici aussi.
###### Annie-France Brazeau
C'est rien de plus qu'un motel.
###### Gregory Rotheingham
_en s'animant_
À Paris, ils ont la rive gauche. Ici, le P'tit Riv est à gauche, puis Larry's est à droite. Nos deux cabanes à frites se font face. C'est nos Champs Élysées de la patate! Attendez de goûter la poutine bleu-blancrouge : fromage bleu, frites blanches, sauce au vin rouge! La cabane s'ouvre et on oublie le Louvre! Quand un visionnaire aura compris l'affaire, on fera de Sturgeon Falls la Ville lumière des pommes de terre! Notre réussite viendra des frites!
###### Annie-France Brazeau
J'entends déjà le chant des masses de touristes.
_Elle chante._
« Allons enfants de la...
###### Gregory Rotheingham
_se joignant à elle_
...patate, la pomme de terre est arrivée! »
###### Pour le moulin
_Les esturgeons sont disparus_
_Pour le moulin, pour le moulin_
_Plus de chutes sur la rivière non plus_
_Pour le moulin, pour le moulin_
_Des forêts entières abattues_
_Pour le moulin, pour le moulin_
_Ce qu'il lui fallait, ben il l'a eu_
_Mais regarde donc ça, son tour est venu_
_À ce moulin, à ce moulin_
_Il ne sert plus, on n'y croit plus_
_À ce moulin, à ce moulin_
_On le démolit, il est disparu_
_À ce moulin, à ce moulin_
_On a donné son dernier dû_
_Dans une ville dont le nom_
_Est un paysage passé_
_On sait bien que toute chose_
_A sa durée limitée_
_Une époque s'est achevée_
_Ben une autre va succéder_
_Y en a plus de moulin dans 'a place?_
_Pas sûr qu'il faille qu'on le remplace!_
_Il a fallu dompter la nature_
_Pour le moulin, pour le moulin_
_Se faire des muscles et de la couenne dure_
_Pour le moulin, pour le moulin_
_On s'y faisait de bons copains_
_Dans ce moulin, dans ce moulin_
_Et une bonne paie, ça c'est certain_
_Mais pour ça fallait endurer_
_Que le gros moulin, que le gros moulin_
_Impose en ville ses volontés_
_De gros moulin, de gros moulin_
_Arrange tout pour le bénéfice_
_Du gros moulin, du gros moulin_
_Puis qu'à la fin il nous trahisse_
_Dans une ville qui porte le nom_
_De vieilles beautés sacrifiées_
_On se souviendra du prix_
_Que le pain nous a coûté_
_Que l'usine claire la place_
_Pour faire place à du meilleur_
_Si on ne sait pas quoi tout de suite_
_Attends, on le saura tout à l'heure_
_Les esturgeons sont disparus_
_Pour le moulin, pour le moulin_
_Plus de chutes sur la rivière non plus_
_Pour le moulin, pour le moulin_
_Des forêts entières abattues_
_Pour le moulin, pour le moulin_
_Ce qu'il lui fallait, ben il l'a eu_
_Mais regarde donc ça, son tour est venu_
_À ce moulin, à ce moulin_
_Il ne sert plus, on n'y croit plus_
_À ce moulin, à ce moulin_
_Il est démoli, il est disparu_
_Sais-tu... je m'y fais mieux que j'aurais cru_
1. Le « révérend frère » Michel Éthier a défrayé les manchettes alors qu'il invoquait devant les tribunaux la Charte des droits et libertés pour contester une accusation de possession de marijuana, faisant valoir que sa consommation est un sacrement pratiqué dans l'Église de l'univers dont il est membre.
###
### Salut Fauquier!
À Fauquier, deux ponts haut perchés enjambent la rivière, côte à côte. Cette double silhouette témoigne de l'histoire des lieux : le chemin de fer et le chemin de terre ont ouvert la région à la colonisation et à l'industrie forestière. une statue de marmotte vous souhaite la bienvenue dans ce village sur le bord de la rivière Groundhog.
Fauquier est une petite patrie canadienne-française du Nord de l'Ontario sur la route 11. Sept cents personnes habitent ici et plus de 90 % sont francophones. Véritable pépinière de talents, ce petit village a vu grandir plusieurs écrivains, chanteurs et artistes, Pierre Albert, Robert Poisson, Carmen Laberge, et bien sûr, le grand sculpteur Maurice Gaudreault, dont les figurines de glaise de renommée nationale ont immortalisé la vie rurale.
Un autre personnage de Fauquier, peu connu, mériterait une place dans l'histoire nord-ontarienne. Mieux encore, il mériterait d'entrer dans la légende, tellement ses exploits paraissent surhumains. Il a d'ailleurs inspiré Maurice Gaudreault, qui lui a donné les traits d'un petit bonhomme portant chemise à carreaux, sciotte au dos et hache à la main. Au dire de ceux qui l'ont connu, il a été le meilleur bûcheron que le Nord de l'Ontario ait jamais vu. Pourtant, on en a vu en masse et puis des bons.
Vous vous imaginez sans doute un géant à la Jos Montferrand, sept pieds trois, trois cents livres. Eh bien, la réalité fait mieux que l'imagination. Il était le parfait contraire! Imaginez au cœur de la forêt boréale des épinettes qui tombent à la bascule comme des rangées de dominos. Imaginez que ça se passe comme ça toujours, jour après jour. Maintenant, au pied d'un de ces grands arbres, imaginez un homme tout court, à peine plus de cinq pieds, un homme tout maigre, à peine 130 livres. Cette petite carrure appartenait à un travailleur plus grand que nature. Même son sobriquet évoquait sa fière allure. On l'appelait Bucko... Bucko Beauchamp.
###### « Bucko » Beauchamp
Au dire de ceux qui l'ont connu, Bucko Beauchamp n'était pas du genre à lancer des défis ou à relever des paris. Il n'aimait pas se vanter. Il savait travailler. Ça suffisait. On raconte pourtant qu'une fois, à l'âge de douze ans, le jeune Bucko qui travaillait avec son père lui aurait dit: « Je m'excuse, son père, mais tu sais pas bûcher. »
À 14 ans, comme beaucoup de garçons de sa génération, il avait déjà quitté l'école pour travailler dans le bois. Ceux qui l'ont vu faire disent que sa force n'était pas dans l'effort musculaire; elle était dans son savoir-faire. Il fallait le voir aller pour comprendre. Il était vite comme un taon. Il était agile comme un suisse. Il était d'une efficacité mécanique. Il était un fonceur méthodique. Pas un de ses pas n'était perdu. Pas un de ses gestes n'était inutile. Il en était prodigieux. Pour bûcher, Bucko Beauchamp avait un don.
C'était dans les années quarante, dans le temps des _bucksaws_ , avant les scies à chaîne et bien avant les immenses machines. Les bûcherons travaillaient dans ce qu'on appelle une _strip_ , une bande boisée de 66 pieds que le contremaître leur assignait. En fin d'après-midi, vers quatre heures et demie, un homme ordinaire était fier d'avoir abattu, débité et empilé deux, parfois trois cordes de bois. Bucko Beauchamp lui, avait généralement quitté sa _strip_ bien plus tôt, vers deux heures. Il avait bu son thé et mangé son sandwich tout en travaillant, sans s'arrêter. Et il laissait derrière lui au moins quatre, parfois cinq, ou même six cordes d'épinette bien empilées. Il en faisait deux fois plus que les autres en prenant deux heures de moins. Ce n'était pas exceptionnel, c'était habituel. Ceux qui ont essayé de le suivre ont pu parfois soutenir son rythme un jour ou deux, mais pas plus. Le petit Bucko Beauchamp valait trois gros bûcherons à lui tout seul. Quand il bûchait, il n'était plus le même homme.
On raconte qu'une fois, il était rentré au camp vers deux heures de l'après-midi, comme d'habitude. un _foreman_ lui demande : « Qu'est-ce que tu fais là à cette heure-ci? C'est pas une maison de chambre icitte! » C'était mal parti, pour trois raisons. un : ce contremaître ne connaissait évidemment pas son homme. Deux : ça adonnait que ce jour-là, Bucko ressemblait à un petit gars de la ville perdu dans le bois. Après une longue soirée bien arrosée, il était entré au travail en portant son beau linge propre. Et trois : c'est que le pauvre Bucko bégayait. Alors si le petit homme drôlement habillé a répliqué quelque chose comme : « M-m-ma journée est faite. V-v-va mesurer mon bois p-p-pour voir », admettons que le _petit boss_ n'a pas dû être impressionné. En tout cas, le contremaître l'a congédié cet après-midi-là. Mais il l'a regretté ce soir-là. « Pauvre toé, qu'on lui a dit. Tu viens de _clairer_ le meilleur homme de la place. » On a demandé à Bucko de revenir. Bucko, lui, a juste demandé sa paie. Sans être vantard, il était quand même fier. Si on ne respectait pas sa réputation, il fallait respecter sa parole.
« Venez pas me tanner », qu'il disait en partant pour sa _strip_ avec son thermos de thé. Mais certains, parfois, le cherchaient et le trouvaient. une fois, deux gaillards vantards de Val Gagné nouvellement arrivés se sont moqués de lui. « On est là pour faire une piastre, nous autres », qu'ils ont dit en le regardant de haut. Eh bien, ces deux-là ont bientôt lâché le chantier en se plaignant que leurs longues journées leur donnaient de courtes paies. Bucko, lui, leur aurait dit : « C'est drôle, moi je bûche pourtant pas au fanal. »
Une autre fois, à ce qu'on raconte, c'est un gros bûcheron qu'on appelait l'Espagnol qui a lancé le défi :
— On verra qui aura vidé sa _strip_ le premier.
— C-c-c-correct, je vais t'attendre, qu'il aurait répliqué, Bucko.
On se trouvait dans une forêt d'épinettes qui avaient de 15 à 20 pouces à la souche. Il fallait d'abord l'abattre, puis l'ébrancher à la scie et à la hache. Ensuite, un partenaire traînait les troncs à l'aide d'un cheval puis les débitait en billes. Enfin, on les empilait en cordes à environ vingt pieds d'intervalle. une _strip_ pouvait donner entre 40 et 80 cordes. Dans les bandes de bois voisines, c'était facile de suivre à l'œil le progrès de chacun. En tout cas, cette semaine-là, plusieurs ont pris le temps de s'approcher pour constater ce qui se passait. C'était un carnage! Bucko était parti en peur. Il bûchait une corde à l'heure! Les épinettes tombaient comme des soldats au front. Et ils tombaient exactement là où le général Beauchamp le voulait. Car il avait l'adresse et la stratégie pour aligner ce qu'il abattait. Bien entendu qu'au bout du compte, il avait fait le ménage dans son boisé bien avant l'autre. L'Espagnol en a perdu son latin.
Des anecdotes de ce genre-là ne sont pas nombreuses, cependant, dans les mémoires. Bucko n'était pas tellement porté à se faire valoir. On raconte quand même qu'une fois, vers la fin des années cinquante, Bucko a participé à un concours dans les environs de Thunder Bay. On était rendu maintenant à l'époque des scies à chaîne, des tronçonneuses. Curieusement, Bucko n'avait pas trop d'intérêt pour l'entretien des machines. Si ça chaîne cassait, au lieu de la réparer, il pouvait tout aussi bien arrêter là en se disant qu'il reprendrait son erre d'aller le lendemain. En tout cas, c'est justement ce qui est arrivé : en plein concours, sa chaîne a brisé. un autre que lui aurait essayé de la remplacer au plus vite. Bucko, lui? Eh bien, il a vu qu'il avait une grande avance sur les autres, et il a jugé qu'il gagnerait même s'il s'arrêtait là. Le pire c'est qu'il a presque eu raison. En fin de compte, il a fini deuxième. Puis qu'est-ce qu'il a dit de ça? Il a dit : « Si j'avais su, je l'aurais peut-être changée, la chaîne! »
Quand il n'était pas dans le bois, Bucko Beauchamp n'en parlait pas. Mais il gardait son tempérament de travailleur pressé. Si tu voyageais avec lui, il t'attendrait une fois peut-être, mais sûrement pas deux. Ça ne l'empêchait pas pourtant de payer très souvent la tournée à ses amis. Et par la nuit étoilée après ses soirées bien arrosées, il rentrait en chantant son air préféré : « C'est écrit dans le ciel... » Écrit dans le ciel, peut-être, mais ailleurs j'ai cherché en vain. Dans les chantiers où il a travaillé, de Smooth Rock Falls à Thunder Bay, on dit que son rendement remarquable était parfois affiché. Mais aucun article de journal ni livre d'histoire n'a rapporté ses performances phénoménales.
Bucko Beauchamp aurait dit un jour : « Moi, je fais tout vite. Je vis vite. Je travaille vite. Je vais mourir vite. » Pensait-il vraiment dire une parole prophétique? En tout cas, il est mort jeune d'une manière trop familière. En 1963, à 38 ans, il a péri dans une collision frontale au pont de Gregoire's Mill. Sur sa pierre au cimetière de Fauquier, on voit sa photo. On apprend son prénom : Lucien. Et on lit l'inscription : « Parti, mais pas oublié ». C'est vrai sans doute pour sa famille, ses proches, ses anciens compagnons de travail qui m'en ont parlé. Mais il faudrait que ce soit vrai aussi dans la mémoire du Nord de l'Ontario.
Des milliers de travailleurs forestiers ont insufflé leur esprit à notre jeune pays qui a trop vite vieilli et qui s'essouffle déjà tant sa jeunesse s'en va. Il nous en faudrait des figures mythiques capables de symboliser leur esprit, pour léguer aux bâtisseurs de l'avenir la force des bâtisseurs du passé. Le cœur à l'ouvrage, la fierté du travail bien fait, le respect de la forêt boréale, voilà des qualités morales que nous garderons si nous gardons bien vivant le souvenir d'un petit bûcheron plus grand que nature de Fauquier, Ontario. Qu'il grimpe au rang de Jos Montferrand. Qu'il fasse la _runne_ à Paul Bunyan. Chez nous on en a eu un vrai et on se rappelle qui il était. Il n'avait pas besoin d'être grand, notre géant. Il était Bucko Beauchamp.
###### C'est pas toujours les meilleurs
qui l'emportent
_Qui sait comment ça va aller_
_Où nos vies vont finir par mener_
_De nos efforts, va-t-on goûter les fruits_
_On n'a pas tous la récompense_
_Que l'on mérite pour la dépense_
_De nos espoirs et de nos énergies_
_Au bout du jour, les traces qu'on a laissées_
_Dépendent pas juste des bottes qu'on a chaussées_
C'est pas toujours les meilleurs qui l'emportent
On donne quand même notre meilleur, peu importe
Ce coin de pays mérite mieux qu'on y a mis
On connaît notre valeur, et on attend notre heure
_Quand on verra le bout de la route_
_La chose qui comptera parmi toutes_
_C'est pas de gagner, mais de ne pas s'être perdu_
_On fait ce qu'on peut, on fait ce qu'on doit_
_On remet mieux que ce qu'on reçoit_
_On s'est donné, mais on s'est pas vendu_
_Au bout du jour, le progrès accompli_
_Dépend pas juste du cœur qu'on y a mis_
C'est pas toujours les meilleurs qui l'emportent
On donne quand même notre meilleur, peu importe
Ce coin de pays mérite mieux qu'on y a mis
On connaît notre valeur, et on attend notre heure
###### Tamara de Fauquier
Bonjour Grand-père Gagnon! Sais-tu que j'ai fait un long voyage pour revenir ici à Fauquier, te rendre visite, à ta dernière demeure? De la rue Grzela, je sens les regards étonnés des gens qui me voient plantée là, au beau milieu du cimetière, sous la pluie froide d'automne. Ils se demandent probablement pourquoi je n'ai pas attendu qu'il fasse beau pour venir prier sur ta tombe. Cher Grand-père, je ne suis pas ici pour te vénérer comme un saint, ou même pour prier pour le salut de ton âme. Si ma démarche ressemble à un geste religieux, il s'agit du plus primitif des rites. Je suis venue répondre à l'appel de la terre, des ancêtres, du sang. J'obéis à l'instinct qui me dit que les morts ne sont pas morts.
Grand-papa, toi qui es décédé avant ma naissance, tu m'as toujours manqué! Ta femme, Tamara, m'a si souvent parlé de toi, les larmes aux yeux, que c'est comme si je te connaissais. Tu sais, elle était inconsolable. Elle a porté ton deuil jusqu'à la fin de sa vie. Sur ta tombe, le jour des funérailles, elle aurait juré qu'elle ne se remarierait pas. Elle a tenu parole! Aujourd'hui, à quarante-cinq ans, je me demande comment, entre Tamara et toi, l'amour a pu être si fort.
Quand tu as tiré ta révérence, tu n'avais pas mon âge. À cette époque, la vie était brutale et souvent courte. J'aurais aimé te voir heureux, aux côtés de Tamara, cette femme si belle que tous les hommes t'enviaient. Mais la connaissais-tu vraiment, ta femme? Je sais que la réponse est non. Si je suis revenue à Fauquier, aujourd'hui, c'est pour te raconter tout ce que Tamara t'a caché, par sagesse sûrement, car si tu l'avais su, tu te serais cru indigne d'elle. Tu aurais été incapable de reconnaître l'amour.
Quand je regarde en arrière, dans mes souvenirs les plus lointains, je vois toujours en premier lieu la figure de cette femme qui a marqué mon enfance : Tamara Gagnon, la mère de ma mère. De ton vivant comme après ta mort, on disait de Tamara « la Finlandaise » qu'elle était « différente ». Les gens du village s'amusaient de son accent inimitable. Tous pensaient qu'il provenait du finnois. Même moi. Maintenant, je sais que c'est plus compliqué que ça! Te souviens-tu que son français était d'une correction grammaticale si impeccable que le maire du village lui avait demandé de s'occuper de l'école du rang, celle qui a existé longtemps avant la construction de l'école Sainte-Jeanne-d'Arc? Elle n'enseignait plus quand je suis née, mais elle avait gardé sa déformation professionnelle. Elle corrigeait, sans jamais se fatiguer, ce qu'elle appelait mon « mauvais français ». Et elle exigeait une diction précise. Pas de marmonnements! « Articule et fais des phrases, s'il te plaît! Les phrases, ça sert à se faire comprendre... »
Je ne tenais pas vraiment compte de ses remarques, car personne d'autre ne trouvait que je m'exprimais mal. Tout le monde me comprenait! Penses-tu vraiment que j'allais me faire remarquer en imitant sa manie de parler comme un livre? Jamais! Mais je dois admettre qu'il y a peut-être un lien entre son obsession grammaticale et le choix que j'ai fait d'entreprendre des études de littérature. Eh oui, c'est d'elle que je tiens l'amour des mots.
Tamara me racontait qu'elle était née en Finlande. Pourtant, elle ne ressemblait pas aux autres Finlandais connus dans la région : des bûcherons sans famille, vivant dans des camps au cœur de la forêt boréale. Je ne les voyais qu'à la fin de la saison de la coupe, quand il m'arrivait d'aller à Kap ou à Hearst. Ce qu'ils pouvaient dépenser en quelques jours! Tout ce qu'ils avaient gagné y passait, enrichissant les vendeurs d'alcool et les tenanciers pratiquant la traite des Blanches. Tamara-aux-mœurs-irréprochables était pourtant de leur race. J'ai même entendu dire qu'elle avait deux cousins parmi eux : Minima et Pitkanen.
Tamara insistait pour que nous l'appelions Moumi, ce qui signifie « grand-mère » en finnois. Elle, qui était incapable de formuler une seule phrase en anglais, paniquait quand un étranger lui adressait la parole en cette langue. À part la bizarrerie de son intonation, son attachement obstiné au français correct et sa phobie de l'anglais, nous, ses petits-enfants, ne voyions en elle qu'une ménagère pratique, qui nous préparait des mets délicieux et nous forçait à trop manger. Oh, comme elle nous gâtait!
Puis, un jour, elle a été atteinte d'Alzheimer et d'une folie douce qui nous a révélé que nous ne savions rien de ses origines. Beaucoup de choses qu'elle avait réussi à cacher pendant toute sa vie d'adulte ont alors émergé par bribes, quand lui revenaient en désordre des souvenirs jusqu'alors enfouis. Ainsi, elle nous a parlé du courage de sa mère qui avait dû élever toute seule trois enfants en travaillant comme femme de ménage. Nous ne savions pas que Tamara avait deux frères. Dans son délire, elle pleurait en parlant d'eux, se culpabilisant parfois de les avoir abandonnés quand leur mère est morte d'épuisement. De temps en temps, elle se trouvait des excuses, se réconfortait du fait qu'ils n'étaient plus des gamins. Ils avaient quinze et dix-sept ans à l'époque; ils pouvaient se débrouiller tout seuls; et puis, ils n'admettaient déjà plus l'autorité de leur sœur aînée alors âgée de dix-neuf ans... Ils l'auraient forcée à vendre la maison familiale et à partager l'argent en trois parts égales. Dégoûtée de leur comportement, Tamara aurait décidé de recommencer sa vie ailleurs et se serait embarquée sur un navire en partance pour le Canada.
Et là, tout d'un coup, Tamara garda le silence pendant plusieurs semaines. Ce mutisme commençait à nous inquiéter quand elle se mit soudain à parler des dialectes que nous ne connaissions pas. Ma mère, qui avait appris un peu de finnois, nous assura qu'il ne s'agissait pas de cette langue. Comme je me doutais bien que Moumi n'avait pas régressé vers un babillage pareil à celui des bébés, j'ai décidé d'enregistrer ses discours insaisissables. J'étais alors étudiante en lettres françaises à l'Université Laurentienne. Je suis allée voir des professeurs du département des langues modernes et je les ai suppliés d'écouter mes enregistrements. À mon grand étonnement, l'un d'entre eux m'a appris que Moumi parlait parfaitement le russe et l'allemand. Il m'a dit qu'elle s'adressait à deux personnages, qui étaient probablement son père et sa mère, qu'elle appelait Alfred et Paula Khun. Elle leur répétait inlassablement : « C'est moi, Tamara Khun. Vous ne me reconnaissez pas? Vous ne vous souvenez pas de moi? Je suis rentrée du Canada. Vous m'avez tellement manqué... » Nous savions que la mère de Tamara, mon arrière-grand-mère, s'appelait Paula. Mais, dans la famille, personne n'avait jamais entendu parler d'Alfred Khun. Nous avions toujours supposé que le mari de Paula s'appelait Malkki, puisque Tamara portait ce nom avant de prendre celui de Gagnon.
Ces révélations m'ont étonnée. Mais j'ai renoncé pendant longtemps à percer le mystère des deux noms de jeune fille de Tamara. un jour, le hasard m'a conduit à Paris, où je représente le Canada dans l'équipe du Secrétariat général de la Francophonie. Ce poste m'a donné l'occasion de retracer les origines de mon aïeule. J'ai consacré systématiquement mes vacances à cette recherche. J'ai ainsi pu effectuer maints séjours en Finlande, en Russie et en Allemagne, séjours au cours desquels j'ai pu rassembler les pièces du casse-tête. une fois celles-ci réunies, il ne me restait plus qu'à résoudre cette question qui me tourmentait : Robert Gagnon, cet humble bûcheron de Fauquier, a-t-il jamais su ce que j'ai mis tant d'années à découvrir?
Grand-père, laisse-moi te raconter pourquoi j'ai finalement conclu que Tamara Khun-Malkki-Gagnon ne t'avait jamais révélé l'histoire de son enfance. Si elle l'avait fait, même en obtenant sous serment ton engagement à ne jamais le répéter à personne, je suis certaine que tu aurais à ton tour passé le secret à un parent, ou à un ami, en lui intimant l'ordre de se taire — après tout, c'est la nature humaine! Et ton confident aurait imité ton indiscrétion. Et, de fil en aiguille, la rumeur se serait propagée. Or, je n'ai rien découvert à Fauquier, même pas une rumeur permettant d'entrevoir un début de solution du mystère des noms de Moumi. Je suis donc certaine qu'au Canada, Tamara n'a révélé à personne la tragédie de sa jeunesse. Même pas à toi, Grand-père Gagnon, sûrement parce qu'elle ne voulait pas ta pitié, mais ton amour. Elle a gagné son pari : tu l'as aimée au-delà de toute mesure. Aujourd'hui, j'estime que tu as le droit de tout savoir. J'irai à l'essentiel.
Paula Malkki, la mère de Tamara, devint très tôt orpheline de ses deux parents. À quatorze ans, elle quittait Helsinki et s'exilait en Russie, où elle fut engagée comme femme de ménage chez les Khun, une riche famille juive qui avait fui l'Allemagne pour des raisons faciles à deviner. L'antisémitisme ne date pas d'hier. La Russie a longtemps été une terre de refuge pour les Juifs. Dès le premier regard, Paula sentit qu'Alfred, le fils aîné des Khun, la dévorait des yeux. Il avait alors seize ans. L'attirance était réciproque, mais Paula redoutait un amour sans issue, vu la différence de leurs conditions. Pendant trois longues années, la vierge prudente est parvenue à éviter ce jeune homme, et rien de grave ne s'est produit. Mais un jour, Alfred Khun déclara à Paula qu'il était amoureux fou d'elle et prêt à tout pour la conquérir. Bien que terrorisée, elle en fut si flattée qu'elle répondit par son plus beau sourire au lieu de s'éclipser en faisant semblant de n'avoir rien entendu. L'engrenage était enclenché! Ils devinrent la proie des flammes et l'inévitable se produisit.
À dix-sept ans, Paula tomba enceinte. Et ce qui devait arriver arriva : les maîtres trouvèrent inacceptable que leur fils s'entiche d'une servante. Pour étouffer le scandale, ils décidèrent de chasser la fille. Mais si fort était l'amour d'Alfred pour Paula qu'il déserta le toit familial pour la suivre. Il interrompit ses études en médecine, se trouva de petits emplois pour assurer la subsistance de sa bien-aimée qui, à terme, lui donna une adorable petite fille : Tamara Khun. Deux autres enfants, David et Jacob, naîtront de cette « union libre », car Alfred et Paula ne se sont jamais mariés officiellement. J'ai retrouvé en Finlande des cousins, descendants de ces grands-oncles. Ils m'ont beaucoup aidée dans ma recherche.
Alfred parvint à reprendre ses études sans cesser de travailler pour subvenir aux besoins des siens. Devenu médecin, il gagnait si bien sa vie qu'il offrit à sa fille Tamara, alors âgée de sept ans, une préceptrice pour lui apprendre le français et l'allemand. Née en Russie, Tamara avait déjà le russe comme première langue et, à la maison, sa mère lui parlait en finnois. À dix ans, elle pouvait se faire comprendre en quatre langues : le russe, le finnois, le français et l'allemand. Mais c'était peut-être trop pour une petite fille, fût-elle surdouée! Je me demande si ce ne serait pas la raison pour laquelle, bien plus tard, Tamara se révéla imperméable à l'anglais. Je crois que, dans l'apprentissage des langues, elle avait atteint le point de saturation. Adulte, elle sentait qu'elle ne maîtriserait jamais une nouvelle langue comme celles qu'elle avait apprises dans son enfance.
Une tragédie devait mettre fin à cette jeunesse riche et heureuse. Tamara avait alors douze ans. Le Dr Alfred Khun, qui adorait l'équitation, se tua en tombant de cheval. Il eut la nuque brisée. Sans lui, Paula savait que la vie en Russie deviendrait infernale. Elle décida alors de retourner en Finlande avec ses enfants, auxquels elle donna son nom de jeune fille. Il valait mieux qu'on les considère comme des bâtards plutôt que comme des Juifs, russes par surcroît! Car, à l'époque, l'hostilité entre Russes et Finlandais, tout comme les préjugés contre les Juifs dans la plupart des pays d'Europe, étaient propres à faire trembler. Tamara cessa donc d'être Khun pour s'appeler Malkki.
Grand-père Gagnon, j'imagine que sans la comprendre, tu devais voir l'infinie tristesse dans les beaux yeux de ta femme. Maintenant, tu le sais : c'était la tristesse d'être de nulle part. Née en Russie d'un père juif d'origine allemande, expatriée en Finlande puis au Canada, cette femme que tu as amenée vivre à Fauquier était citoyenne du monde, par la force des choses. Si elle ne t'a jamais tout dit, ce n'est pas parce qu'elle était cachottière. Je crois plutôt qu'elle ne voulait pas t'effrayer ou te faire fuir. Toi qui savais à peine lire et écrire, tu étais pour elle comme le rocher qui tient. Elle a porté trois noms : Khun, Malkki, Gagnon. Elle a parlé quatre langues. Elle a pratiqué trois religions : juive en Russie, car Alfred n'avait jamais renié la foi de ses pères; luthérienne en Finlande, car Paula et ses enfants n'auraient pu être autre chose sans se couper de tout le monde; et catholique au Canada français, parce que sans cela, votre mariage aurait été impossible.
Comme tu vois, dans la famille, je ne suis pas la première à mener une existence cosmopolite. Moumi Tamara m'a précédée. Peut-être même m'a-t-elle montré le chemin. Le monde était sa patrie longtemps avant ma naissance. Sur ses traces, j'ai compris que je suis juive, allemande, russe, finlandaise, canadienne-française... Et je suis aussi, fondamentalement, ce que Tamara est devenue au bout de son errance : citoyenne de Fauquier.
_Melchior Mbonimpa_
###### _Incitatif pédant_ : Fauquier de l'avenir
###### Annie-France Brazeau
Bonjour, ici Annie-France Brazeau, vous écoutez _Incitatif pédant_.
_On entend le thème musical de l'émission._
Nous recevons aujourd'hui Gregory Rotheingham, du ministère des Opérations improbables et des Études interminables.
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Gregory Rotheingham, auteur de rapports incontournables dans le monde du fonctionnariat, vous avez notamment écrit un rapport sur la communauté de Mattice dans lequel vous préconisez l'ouverture d'une gigantesque buanderie pour laver le linge sale de toute la région, un rapport que vous avez intitulé _Les chemises de l'archiduchesse sont-elles sèches à Mattice?_
###### Gregory Rotheingham
Bonj...
###### Annie-France Brazeau
Vous êtes déjà monté sur scène à Las Vegas. Vous avez d'ailleurs passé la nuit en prison pour avoir interrompu Tom Jones.
_Il va répondre. Elle enchaîne._
Vous êtes également un homme de science réputé : vous vous êtes vous-même greffé un estomac de plomb.
###### Gregory Rotheingham
Un estomac de plomb?
###### Annie-France Brazeau
Une jambe de bois!
###### Gregory Rotheingham
Une jambe de bois!
###### Annie-France Brazeau
Un pied mariton!
###### Gregory Rotheingham
Un pied mariton!
###### Annie-France Brazeau et Gregory Rotheingham
_chantant_
« Un pied mariton Madeleine, un pied mariton Madelon... »
###### Annie-France Brazeau
Et je pourrais continuer longtemps comme ça, mais je ne suis pas sûre que nous avons envie d'en savoir plus! Gregory Rotheingham, bonjour.
###### Gregory Rotheingham
Bonjour Annie-France.
###### Annie-France Brazeau
Gregory, vous venez nous présenter votre tout nouveau rapport qui vise à donner un second souffle à l'économie de Fauquier.
###### Gregory Rotheingham
Absolument. Fauquier a le potentiel des plus grands villages.
###### Annie-France Brazeau
Et quel est le titre de votre rapport?
###### Gregory Rotheingham
_Pour venir à Fauquier, fauqu'i' ait d'l'avenir à Fauquier._
###### Annie-France Brazeau
Faut qu'on l'enferme! Dites-moi, qu'est-ce qui incarne le mieux l'avenir économique de Fauquier?
###### Gregory Rotheingham
La marmotte!
###### Annie-France Brazeau
Une marmotte dans un rapport de développement économique, c'est du sabotage!
###### Gregory Rotheingham
Non, c'est la mascotte de Fauquier. Vous avez remarqué qu'on a une grande marmotte sur le bord de la route.
###### Annie-France Brazeau
Oui, je voyage beaucoup, mais c'est la première fois que j'arrive dans un village pour être accueillie par une grosse marmotte. Ah non, c'est vrai, une fois à North Bay, j'ai été accueillie par Mike Harris.
###### Gregory Rotheingham
Fauquier a la plus grosse marmotte au monde! C'est l'inspiration des premiers Jeux internationaux de la marmotte.
###### Annie-France Brazeau
Après le Jour de la marmotte, ce sera les Jeux de la marmotte? Ça peut durer deux ou six semaines, tout dépend si le maire voit son ombre.
###### Gregory Rotheingham
Justement, les Jeux de la marmotte débuteront le Jour de la marmotte. La marmotte de Fauquier fera, elle aussi, ses prédictions pour annoncer la fin de l'hiver.
###### Annie-France Brazeau
Gregory, je dois vous avouer une chose : je ne suis pas une marmotte, mais je fais moi aussi des prédictions et je vous dis que votre projet est voué à l'échec.
###### Gregory Rotheingham
Pourquoi?
###### Annie-France Brazeau
Parce que je n'ai pas vu l'ombre du rapport.
###### Gregory Rotheingham
...
###### Annie-France Brazeau
Écoutez, Gregory Rotheingham, il y a déjà trop de marmottes qui font des prédictions le 2 février.
###### Gregory Rotheingham
Mais vous me prenez pour un imbécile?
###### Annie-France Brazeau
C'est vrai!
###### Gregory Rotheingham
Bien sûr qu'il y a trop de marmottes qui font des prédictions le 2 février! Notre marmotte va devancer toutes ses concurrentes.
###### Annie-France Brazeau
Comment?
###### Gregory Rotheingham
Elle va faire ses prédictions le 1er février.
###### Annie-France Brazeau
Gregory Rotheingham, vous voulez dénaturer le Jour de la marmotte qui est, je vous le rappelle, le 2 février.
###### Gregory Rotheingham
Mais non! Le 1er février, notre marmotte va seulement prédire si elle va voir son ombre le lendemain.
###### Annie-France Brazeau
Et quoi encore! J'imagine que vos Jeux de la marmotte auront une épreuve de creusage de tunnel.
###### Gregory Rotheingham
Absolument! À chaque année, on va amener les dix plus grands criminels au monde. On les enferme chacun dans une cellule de prison avec une cuillère à thé. Le premier à sortir gagne sa liberté.
###### Annie-France Brazeau
Gregory, un peu de sérieux. une cuillère à thé, ça n'existe pas.
_Elle sort une cuillère._
Tiens, prenons celle-ci et interrogeons-la : « Dis-moi, cuillère, crois-tu qu'il existe un dieu suprême et ultime? »
_Prenant la voix de la cuillère._
Oh non, je ne crois pas.
_Reprenant sa voix._
Mais tu es une cuillère athée! une cuillère athée! une cuillère athée...
###### Gregory Rotheingham
Calmez-vous, Annie-France!
###### Annie-France Brazeau
Mais Gregory, pensez-y un instant, regardez des gens creuser un tunnel pour s'évader de prison, ça risque d'être endormant.
###### Gregory Rotheingham
Et justement! Qui dit marmotte dit « dormir comme une marmotte ». On va organiser un dormothon.
###### Annie-France Brazeau
Un dormothon?
###### Gregory Rotheingham
Oui! Les gens vont venir hiberner à Fauquier.
###### Annie-France Brazeau
Hiberner?
###### Gregory Rotheingham
On vend 200 places par année. Fin octobre, on enferme ce monde-là dans le gymnase de l'école Sainte-Jeanne-d'Arc et on ouvre les portes au mois d'avril.
###### Annie-France Brazeau
Ouh... Six mois à hiberner dans le village de la marmotte, ça va finir par sentir l'ours.
###### Gregory Rotheingham
L'objectif secret de l'activité, c'est d'assurer l'avenir démographique de Fauquier.
###### Annie-France Brazeau
En hibernant? Comment allez-vous faire?
###### Gregory Rotheingham
Chaque participant aura signé une autorisation pour prendre part à l'activité... et, en tout petits caractères sur la feuille, on aura inscrit que tous les enfants conçus pendant l'hiver sont la propriété exclusive de Fauquier.
###### Annie-France Brazeau
C'est monstrueux!
###### Gregory Rotheingham
Exactement! C'est la revanche des berceaux.
###### Annie-France Brazeau
Mais on ne peut pas baser l'économie de toute une communauté sur des gens qui creusent et des gens qui dorment.
###### Gregory Rotheingham
Oh, mais la marmotte a plus d'un tour dans son sac. On va organiser des thérapies pour les gens qui ont peur de leur ombre.
###### Annie-France Brazeau
Bravo! Le public cible à l'échelle de la planète est de 42 personnes.
###### Gregory Rotheingham
Et qui dit marmotte dit mauvaise vue. À chaque année, on va organiser un grand congrès d'oculistes.
###### Annie-France Brazeau
Ah oui, je vois.
###### Gregory Rotheingham
Et les gens pourront venir passer gratuitement un test de la vue à Fauquier.
###### Annie-France Brazeau
Oui. Je vois.
###### Gregory Rotheingham
Et pour mieux représenter la nouvelle réalité, on va changer la pancarte sur la route à l'entrée de la ville.
_Il montre la nouvelle pancarte._
F
A U
Q U I
E R P O
P U L A T
I O N 7 5 0
###### Annie-France Brazeau
Mais Gregory, c'est assez aride tout ça. Vous n'avez pas un petit côté culturel à vos projets?
###### Gregory Rotheingham
Vous lisez dans mes pensées, Annie-France! Tous les mois de juillet, les gens vont venir de partout au monde pour assister à notre concours d'imitation de Roger Whittaker.
###### Annie-France Brazeau
Roger Whittaker?
###### Gregory Rotheingham
C'est ça!
###### Annie-France Brazeau
Je comprends qu'il existe des concours d'imitation d'Elvis, mais... pourquoi Roger Whittaker?
###### Gregory Rotheingham
C'est un grand siffleux!
###### Annie-France Brazeau
Vous êtes complètement tordu, Gregory. Et l'histoire des ovnis qu'utilise Moonbeam, le village voisin de Fauquier, pour attirer des touristes, j'imagine que c'est aussi vous qui êtes derrière ça?
###### Gregory Rotheingham
Non, ce contrat-là m'a échappé. Mais j'ai une grande admiration pour leur projet et j'ai décidé de détourner leur clientèle vers Fauquier.
###### Annie-France Brazeau
Pensez-vous vraiment qu'ils vont délaisser les ovnis pour la marmotte?
###### Gregory Rotheingham
Oui, parce qu'on a de grands mystères ici aussi.
###### Annie-France Brazeau
Des mystères qui peuvent faire compétition aux objets volants non identifiés de Moonbeam?
###### Gregory Rotheingham
Ah oui! Sur la rivière Groundhog, on a des « ofnis » : des objets flottants non identifiés.
###### Annie-France Brazeau
Et qu'est-ce qui flotte sur la rivière qui pourrait intéresser les touristes?
###### Gregory Rotheingham
On ne sait pas, on les a pas identifiés.
###### Annie-France Brazeau
Et ils vont arrêter à Fauquier pour ça?
###### Gregory Rotheingham
Pas seulement pour ça. Il y a, madame Brazeau, un mystère qui règne autour de la rivière Groundhog.
###### Annie-France Brazeau
Vraiment?
###### Gregory Rotheingham
Il y a, dans le fond de la rivière, une accumulation de billes de bois qui remontent à l'époque où il y avait de la drave sur la rivière.
###### Annie-France Brazeau
Attendez, je veux être certaine de comprendre. Il y a des billots de bois au fond de la rivière?
###### Gregory Rotheingham
Exactement!
###### Annie-France Brazeau
C'est parce que du bois, c'est censé flotter...
###### Gregory Rotheingham
Justement! On va récupérer ces mystérieuses billes qui coulent pour construire... des sous-marins!
###### Annie-France Brazeau
Des sous-marins en bois, oui! Ça va être des sous-marins bio! Imaginez la panique de l'équipage quand ils vont croiser un poisson-scie!
###### Gregory Rotheingham
Ah ben, les gros poissons, on connaît ça à Fauquier. Sur la rivière Groundhog on pêche l'esturgeon en aval pis de « l'ouesturgeon » en amont.
###### Annie-France Brazeau
Étonnant!
###### Gregory Rotheingham
Et on a une île à Poisson sur la rivière qui a un potentiel touristique.
###### Annie-France Brazeau
L'île à poissons maintenant. Vous allez me faire croire que vous avez des poissons qui vivent sur une île, des poissons terrestres?
###### Gregory Rotheingham
Non, ça s'appelle l'île à Poisson parce que l'ancien propriétaire s'appelait Poisson. Voyons, des poissons terrestres, ça n'existe pas.
###### Annie-France Brazeau
Et qu'est-ce qu'on y retrouve sur l'île à Poisson?
###### Gregory Rotheingham
Des ormes, des frênes, des fougères géantes et des sirènes!
###### Annie-France Brazeau
Gregory, vous avez reconnu que les poissons terrestres n'existent pas et là, vous me dites que sur l'île à Poisson, il y a des sirènes.
###### Gregory Rotheingham
Il y en aura. On va greffer une queue d'esturgeon à la reine du carnaval, et on va l'installer sur l'île à Poisson. Ça, ça va attirer les touristes!
###### Annie-France Brazeau
Vous vivez sur une autre planète Gregory. Qu'est-ce que vous allez faire avec les pieds de la reine et la tête de l'esturgeon?
###### Gregory Rotheingham
Euh... On les greffera aussi pis il se présentera aux élections.
###### Annie-France Brazeau
On me fait signe qu'il nous reste malheureusement assez de temps pour un dernier de vos projets pour Fauquier.
###### Gregory Rotheingham
J'en ai plus pour Fauquier, mais j'en ai pour sa banlieue : Strickland.
###### Annie-France Brazeau
Qu'allez-vous faire à Strickland?
###### Gregory Rotheingham
On va y établir un camp d'été de discipline pour les enfants. Strickland sera comme Disneyland, mais en plus strict.
###### Annie-France Brazeau
Et vous avez imaginé un slogan.
###### Gregory Rotheingham
« Strickland : vos enfants ne s'amuseront pas... ou vous serez remboursé. »
###### Annie-France Brazeau
Wow! Je vais prendre la deuxième option.
###### Gregory Rotheingham
Ce n'est pas un choix! Parmi les activités du camp de discipline de Strickland, il y aura courir dans des pneus ou faire la file pendant deux heures sous la pluie avec une tranche de pain dans les mains.
###### Annie-France Brazeau
Du non-plaisir assuré! En terminant, Gregory, je suis surprise que vous n'ayez rien conçu à partir du symbole qui représente le mieux Fauquier : les deux ponts, le pont pour les voitures et celui du train.
###### Gregory Rotheingham
Ah, mais au contraire. Si des comptines ont pu mettre le pont d'Avignon sur la mappe, il n'y a pas de raison que Fauquier, qui a deux ponts, ne soit pas deux fois plus connue.
###### Annie-France Brazeau
Bien sûr, j'imagine une foule faisant une immense ronde en chantant : « Sur les ponts / De Fauquier / On y danse / On y danse / Sur les ponts / De Fauquier... » Ah! Le train s'en vient! Aaaah aaaaaaaaah aaaaah garglgglglglg!
###### La onze, la onze
_La route d'asphalte et le chemin de fer_
_Et les deux ponts sur la rivière_
_Je les ai dans le coin de l'œil, toute la journée_
_À longueur de semaine, à longueur d'année_
_Et dans le creux de l'oreille, j'ai, jour et nuit_
_Les dix-huit roues, les trains qui crient_
_Ça fait longtemps qu'on n'arrête plus ici_
_Qu'il arrive plus de monde venu se faire une vie_
_Et je me rappelle les milliers de chargements_
_Tout le bois que nous envoyons depuis longtemps_
La Onze, la Onze
T'as beau avoir deux voies
Tu emportes toujours plus que tu renvoies
La Onze, la Onze
Tu peux mener partout
Comment ça se fait que tu mènes pas chez nous?
_J'ai une petite job, qui rapporte juste assez_
_Pour payer le pick-up que je viens d'acheter_
_Ma blonde vient d'Ottawa, c'est rare, c'est icitte qu'elle veut rester_
_Mais c'est inquiétant, quand on entend que tout le Nord va diminuant_
_Faudrait qu'un jour, quelqu'un puisse m'expliquer_
_Comment ça se fait qu'on peut pas développer_
_Des industries ici comme celles d'ailleurs loin sur la route_
_Et puis comment ça se peut qu'ils pensent vivre mieux_
_Tassés dans le Sud dans leurs banlieues_
La Onze, la Onze
T'as beau avoir deux voies
Tu emportes toujours plus que tu renvoies
La Onze, la Onze
Tu peux mener partout
Comment ça se fait que tu mènes pas chez nous?
La Onze, la Onze
Où sont partis nos milliers de chargements?
La Onze, la Onze
Vas-tu falloir que je te prenne finalement?
### Table des matières
Note de l'éditeur
Préface
Collaborations
Bienvenue au Salut!
Salut Gogama!
Salut Verner!
Salut Chapleau!
Salut Earlton!
Salut Iroquois Falls!
Salut Spanish!
Salut Sturgeon Falls!
Salut Fauquier!
Achevé d'imprimer
en novembre deux mille dix sur les presses
de l'Imprimerie Gauvin, à Gatineau (Québec).
|
{
"redpajama_set_name": "RedPajamaBook"
}
| 3,128
|
package org.elasticsearch.xpack.transform.rest.action;
import org.elasticsearch.client.node.NodeClient;
import org.elasticsearch.common.xcontent.XContentParser;
import org.elasticsearch.rest.BaseRestHandler;
import org.elasticsearch.rest.RestRequest;
import org.elasticsearch.rest.action.RestToXContentListener;
import org.elasticsearch.xpack.core.transform.TransformField;
import org.elasticsearch.xpack.core.transform.action.PreviewTransformAction;
import java.io.IOException;
import java.util.List;
import static org.elasticsearch.rest.RestRequest.Method.POST;
public class RestPreviewTransformAction extends BaseRestHandler {
@Override
public List<Route> routes() {
return List.of(new Route(POST, TransformField.REST_BASE_PATH_TRANSFORMS + "_preview"));
}
@Override
public String getName() {
return "transform_preview_transform_action";
}
@Override
protected RestChannelConsumer prepareRequest(RestRequest restRequest, NodeClient client) throws IOException {
XContentParser parser = restRequest.contentParser();
PreviewTransformAction.Request request = PreviewTransformAction.Request.fromXContent(parser);
return channel -> client.execute(PreviewTransformAction.INSTANCE, request, new RestToXContentListener<>(channel));
}
}
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 8,365
|
Cessna 172N Skyhawk, N75558, Brenda Lee Houston: Fatal accident occurred July 27, 2008 in McMurray, Washington
Location: McMurray, WA
Accident Number: LAX08FA246
Aircraft: CESSNA 172N
Defining Event: Controlled flight into terr/obj (CFIT)
The instrument rated commercial pilot departed under visual flight rules (VFR) for a planned cross-country flight. Shortly after departure, the pilot was receiving radar flight following for about 15 minutes. During this time, the controller advised the pilot that no one had successfully proceeded south VFR and that low ceilings and reduced visibility existed between her destination and her point of departure. Radar data revealed that the flight was initially on a southeasterly heading at an altitude of about 2,400 feet. About 18 minutes after departure, the flight initiated a descent to 900 feet while remaining on the southeasterly heading. As the flight approached an uncontrolled airport, radar data showed a slight climb to 1,400 feet, followed by a climbing 180-degree turn to the northwest, which was opposite the direction of her intended destination. Two minutes later the flight initiated a right turn to a northerly course, while continuing to climb. As the flight proceeded northward, radar data depicted a series of turns to the left and right with the altitude fluctuating between 1,500 feet and 2,900 feet before radar contact was lost. Wreckage and impact signatures were consistent with the airplane impacting trees and mountainous terrain at 2,250 feet msl within a heavily wooded area on an easterly heading. All major components of the airframe were located at the accident site. Review of recorded weather data revealed that instrument meteorological conditions prevailed in the area at the time of the accident, with restricted visibilities in rain showers, and overcast clouds with bases and tops at 1,900 and 10,000 feet, respectively. Examination of the recovered airframe revealed no mechanical anomalies that would have precluded normal operation. Examination of the engine revealed that the number four cylinder exhaust valve was stuck in the open position and was bent. The camshaft was intact and each of the camshaft cam lobes exhibited severe wear and spalling signatures. The corresponding tappets exhibited severe spalling on their respective camshaft contact surfaces. The engine exhibited signatures consistent with a high time engine.
The pilot's improper decision to continue VFR flight into instrument meteorological weather conditions. Contributing to the accident were low ceilings, reduced visibility, and mountainous terrain.
Flight planning/navigation - Pilot (Cause)
Weather planning - Pilot (Cause)
Decision making/judgment - Pilot (Cause)
Mountainous/hilly terrain - Contributed to outcome (Factor)
Clouds - Effect on operation (Factor)
Low ceiling - Effect on operation (Factor)
Below VFR minima - Decision related to condition (Cause)
On July 27, 2008, at 1439 Pacific daylight time, a Cessna 172N airplane, N75558, impacted trees and terrain while maneuvering near McMurray, Washington. The airplane was registered to Crest Airpark, Inc., Kent, Washington, and operated by the pilot, under the provisions of Title 14 Code of Federal Regulations (CFR) Part 91. The commercial pilot and two passengers were killed. The airplane was substantially damaged. Visual meteorological conditions prevailed at the time of departure and instrument meteorological conditions prevailed in the area at the time of the accident. No flight plan was filed for the personal flight. The cross-country flight originated from the Roche Harbor Airport (WA09), Roche Harbor, Washington, about 1402, with an intended destination of Auburn, Washington.
A family member of the pilot reported the airplane missing to the Federal Aviation Administration (FAA) on July 27, 2008, after becoming concerned when the airplane had not arrived at its intended destination. The FAA subsequently issued an Alert Notification (ALNOT). The airplane was located by air units on July 27, 2008, approximately 2120, in a heavily wooded mountainous area about 6 miles east of McMurray.
A witness who was flying in a Cessna 170 reported she intended to fly from Blakely, Washington, to Auburn. During the flight, they elected to return to Blakely due to lowering ceilings and decreasing visibility south of Port Townsend, Washington. During the return flight to Blakely, she was receiving traffic advisories from Whidbey Approach Control. The controller advised her about opposite direction traffic. The witness stated that she observed a Cessna that matched the accident airplane passing in front of her position on an eastbound heading. She added that during this time, the visibility to the east of her current position appeared to be "very poor."
There were no known witnesses to the accident sequence.
According to a family member of the pilot, the flight was destined for Auburn to drop off two passengers, pickup one passenger, return to WA09 to pickup two additional passengers, and return to Auburn.
Review of radar data provided by the FAA revealed that a primary radar contact was initially obtained at 1404, about 5.2 miles southeast of WA09 at an altitude of 2,200 feet mean sea level (msl). The flight path of the airplane was generally on a southeasterly heading with at an altitude between 2,400 and 2,500 feet msl until 1420, where a gradual descent to 900 feet msl was observed. At 1427, a slight climb was initiated until approaching the vicinity of Arlington Airport, Arlington, Washington, at 1429. While in the vicinity of Arlington, radar data depicted the airplane initiated a left turn from a southeasterly course to a northwesterly course while climbing to an altitude of 1,700 feet msl. At 1431, the airplane initiated a right turn to northerly course until 1434, where the airplane performed a series of left and right turns with the altitude fluctuating between 1,500 feet msl and 2,900 feet msl. The last three radar returns were consistent with the airplane being in a descending right-hand turn. The last recorded radar return was at 1439, located 0.17 miles southwest of the accident site at an altitude of 2,200 feet msl.
PERSONNEL INFORMATION
The pilot, age 47, held an airline transport pilot certificate with airplane multiengine rating, and a commercial pilot certificate with airplane single-engine land and airplane single-engine sea ratings. In addition, the pilot held a turbojet powered rating and type ratings in various transport category aircraft. A first-class airman medical certificate was issued on November 26, 2007, with the limitations stated, "must wear corrective lenses, not valid for any class after." The pilot reported on her most recent medical certificate application that she had accumulated 14,200 total flight hours.
Review of the pilot's rental records provided by Crest Airpark revealed the pilot had rented a Cessna 172 on seven separate occasions, totaling 12.3 hours of usage between July 5, 2007 and July 17, 2008. Of the seven rentals, six were conducted in a Cessna 172SP and one in a Cessna 172N. According to the rental record, the pilot had rented a Cessna 172SP and Cessna 172N within the previous 90 days to the accident. The total flight time for both rentals was 2.5 hours.
The four-seat, high-wing, fixed-gear airplane, serial number (S/N) 17267807, was manufactured in 1977. It was powered by a Lycoming O-320-H2AD engine, S/N L-4858-76, rated at 160 horsepower and equipped with a McCauley two bladed fixed-pitch propeller. Review of copies of maintenance logbook records showed an annual inspection was completed on April 1, 2008, at a recorded tachometer reading of 2,609.8 hours; airframe total time of 9,935.4 hours; and engine time since major overhaul of 2,559.8 hours. At the time of the accident, the airplane had accumulated 105.2 hours since the time of the inspection.
The engine was overhauled on June 19, 2001, at a total time of 4805.0 hours. According to an entry in the engine logbook, the number four cylinder and number four piston were replaced on August 30, 2007, due to low compression. According to the FAA inspector who interviewed the mechanic that performed the maintenance, the replacement cylinder was removed from another O-360-H2AD engine. Logbooks for the engine in which the cylinder was removed revealed that metal was found in the oil filter and oil suction screen in 2005.
Review of Lycoming Service Instruction Number 1009AT, the recommended time between overhauls is 2,000 hours. The Service Instruction states for the O-320 engine, "if an engine is being used in a 'frequent' type service and accumulates 40 hours or more per month, and has been so operated consistently since being placed in service, add 200 hours to the TBO time."
METEOROLOGICAL INFORMATION
A review of recorded data from the automated weather observation station at Arlington Municipal Airport, Arlington, Washington, located about 11 miles southwest of the accident site reported at 1435; wind from 320 degrees at 7 knots; visibility 10 statute miles; scattered cloud layer at 500 feet above ground level (agl), broken cloud layer at 2,000 feet agl, and an overcast ceiling at 2,700 feet agl; temperature 15 degrees Celsius; dew point 13 degrees Celsius; and an altimeter setting of 30.05 inches of Mercury.
The National Transportation Safety Board's staff meteorologist reviewed weather data and reported that the northwest portion of the National Weather Service (NWS) Weather Depiction Charts for 1200 depicted an area of instrument flight rules (IFR) conditions over Washington in the vicinity of the accident site. Surrounding that area was an area of marginal visual flight rules (MVFR) conditions over portions of western Washington, Oregon, and northern California. The closest visual flight rules (VFR) conditions were depicted over eastern Washington. MVFR conditions were depicted over the route of flight and the accident site, with IFR conditions in the immediate vicinity. The closest station model to the south of the accident site depicted MVFR conditions with a ceiling overcast at 1,900 feet agl. The northwest section of the NWS hourly Radar Summary Chart for 1419 PDT (2119Z) depicted a band of "light" intensity echoes associated with rain showers over the Cascade Range and Puget Sound area, and in the immediate vicinity of the accident site.
The northwest section of the NWS hourly Radar Summary Chart for 1419 depicted a band of "light" intensity echoes associated with rain showers over the Cascade Range and Puget Sound area, and in the immediate vicinity of the accident site.
According to the Skagit County Sheriff Deputy, who was located just north of the accident site, "spotty rain showers" and a "typical marine layer" were present around the time of the accident.
Review of communication transcripts provided by Whidbey Approach Control revealed that the pilot initially contacted Whidbey Approach Control at 1408. At 1410, the controller notified the pilot of the accident airplane that he didn't have "anybody go south of bush point in the last hour VFR." The pilot responded to the controller that it "looks pretty good so far." At 1413, the controller informed the pilot to maintain VFR at or above 2,500 feet msl due to other traffic in the vicinity. At 1414, the controller advised the pilot to contact Whidbey Approach Control on 120.70, and the pilot acknowledged. The pilot contacted Whidbey Approach Control at 1415, reporting they were at an altitude of 2,500 feet msl. The controller canceled the altitude restriction for the accident airplane at 1417, followed by an acknowledgment from the pilot.
At 1419, the controller reported to another aircraft in the vicinity that the weather around Seattle extending north 15 to 20 miles was "pretty limited visibility with low ceilings," and subsequently asked if the pilot of the accident airplane copied that radio transmission. The pilot of the accident airplane acknowledged that they heard the controller's transmission and reported that they were going to initiate a descent to 2,000 feet msl. The controller responded, "maintain VFR altitudes your discretion." At 1423, the controller terminated radar service with the accident airplane and informed the pilot to squawk VFR and frequency change approved. The pilot subsequently responded acknowledging the controller's request. No further radio communications were received from the accident airplane.
The airplane impacted trees and terrain within heavily wooded mountainous area about 6 miles east of McMurray at an elevation of 2,214 feet msl. The airplane came to rest on its left side on a heading of about 085 degrees magnetic. The wreckage energy path was measured about 175 feet in length on an approximate heading of 085 degrees magnetic from the first identified point of contact (FIPC) with trees to the main wreckage and engine. Located within the wreckage debris path were various broken trees and wreckage debris. All major components of the airframe, primary flight controls, engine, and propeller were observed at the accident site.
The left and right wings were separated from the fuselage. The right wing was located near the FIPC and exhibited a semicircular impact about 2 feet outboard of the wing root. The left wing and fuselage were located about 100 feet beyond the right wing. The left wing was separated from the fuselage and exhibited leading edge compression throughout its span. The engine was separated from the fuselage and was located about 30 feet beyond the main wreckage.
The Skagit County Coroner's office conducted an autopsy on the pilot on July 29, 2008. The medical examiner determined that the cause of death was "blunt trauma injuries."
The FAA's Civil Aeromedical Institute (CAMI) in Oklahoma City, Oklahoma, performed toxicology tests on the pilot. According to CAMI's report, carbon monoxide, cyanide, volatiles, and drugs were tested with negative results for carbon monoxide, cyanide, and volatiles. An unspecified amount of Azacyclonol was detected in the urine, however, it was not detected in the blood.
On August 7, 2008, at the facilities of Precision Airmotive, Marysville, Washington, the Precision MA-4SPA carburetor, serial number 10-5217, was examined by representatives of Precision Airmotive under the supervision of the Safety Board investigator-in-charge (IIC). The examination revealed that the carburetor exhibited impact damage. The mixture and throttle levers were separated. The inlet fuel fitting was finger tight. The fuel inlet screen was removed and found to be free of debris. The right pontoon of the delrin type float was found completely filled with a blue liquid that appeared to be consistent with 100 low-lead aviation fuel. No visible leaks were noted with the pontoon. A blue stain was observed on the inner edge seam at the top of the right pontoon.
On August 26, 2008, at the facilities of AvTech Services, Maple Valley, Washington, the engine and airframe were examined.
Examination of the recovered airframe revealed no mechanical anomalies that would have precluded normal operation. Flight control continuity was established throughout the airframe to all primary flight controls.
Examination of the engine revealed that all of the cylinders remained attached to the engine crankcase. All of the engine accessories were separated from the engine. The top spark plugs and rocker box covers were removed. Mechanical continuity was established throughout the engine and valve train when crankshaft was rotated by hand. Thumb compression was obtained in proper firing order on all cylinders except for number four. The engine was disassembled and visually examined.
No anomalies were noted with the number one, two, and three cylinders.
Examination of the number four cylinder revealed that the exhaust valve was stuck in the open position. Tooling marks were observed on the rocker box casting internal surface next to the exhaust valve spring. The exhaust valve springs and exhaust valve were removed from the cylinder. During removal of the exhaust valve, significant binding was noted. The valve springs visually appeared to be distorted and were intact. The exhaust valve stem was found bent. The exhaust valve guide was found loose within its respective bore within the casting of the cylinder head. The exhaust pushrod was intact and straight. A wear pattern was present around the circumference of the pushrod, consistent with contact against the pushrod tube, which was dented.
The crankshaft was intact and undamaged. All four connecting rods remained attached to the crankshaft and were undamaged. The camshaft was intact and each of the camshaft cam lobes exhibited severe wear and spalling signatures. The corresponding tappets exhibited severe spalling on their respective camshaft contact surfaces.
The single drive dual magneto exhibited damage consistent with impact. When the magneto drive shaft was rotated by hand, spark was produced on each post with impulse coupling engagement. According to the Lycoming investigator, the "e-gap timing was not within specifications" and the points were "opening slightly."
The Lycoming investigator stated that the engine exhibited significant internal wear characteristics consistent with a "high time engine."
No additional anomalies were noted with the engine.
The two bladed fixed pitch propeller remained attached to the crankshaft propeller flange. One of the propeller blades exhibited a slight aft bend about 6 inches from the propeller blade tip. The opposing propeller blade was bent aft and slightly curled.
VFR encounter with IMC
Loss of visual reference
Controlled flight into terr/obj (CFIT) (Defining event)
Certificate: Airline Transport; Commercial
Age: 47, Female
Airplane Rating(s): Multi-engine Land; Single-engine Land; Single-engine Sea
Restraint Used: Seatbelt, Shoulder harness
Last Flight Review or Equivalent:
Flight Time: 14200 hours (Total, all aircraft)
Aircraft Manufacturer: CESSNA
Model/Series: 172N
Time Since Last Inspection: 105.2 Hours
Airframe Total Time: 9935.4 Hours as of last inspection
ELT: C91 installed, activated, aided in locating accident
Engine Model/Series: O-320
Registered Owner: CREST AIRPARK, INC.
Operator: Brenda Lee Houston
Conditions at Accident Site: Instrument Conditions
Observation Facility, Elevation: ARO, 142 ft msl
Observation Time: 1435 PDT
Lowest Cloud Condition: Scattered / 500 ft agl
Wind Speed/Gusts, Direction: 7 knots, 320°
Visibility (RVV):
Departure Point: Roche Harbor, WA (WA09)
Destination: Auburn, WA (S50)
Type of Airspace: Unknown
Passenger Injuries: 2 Fatal
SEATTLE — A defective carburetor is blamed for the 2008 crash of a small plane north of Arlington that killed three people.
A King County Superior Court jury awarded two families $26.1 million for the deaths of Dr. Tory Becker, an Auburn spine surgeon, Enumclaw airline pilot Brenda Houston, and her 10-year-old daughter, Elizabeth Crews.
The jury awarded both compensatory and punitive damages against the engine manufacturer.
The single-engine Cessna was flying from San Juan Island to Auburn when its engine failed, causing the airplane to crash into a heavily forested area north of Arlington. Investigators at the scene discovered that the carburetor float, an accessory which supplies fuel to the engine, had leaked and was full of fuel.
"Once we analyzed the defects in the carburetor, our investigation focused on the carburetor design, manufacturing process, and failure history," said Robert Hedrick, an attorney with Aviation Law Group in Seattle who represents the Becker family. "Sure enough there was a significant history of similar failures for years before this accident," Hedrick said.
Attorneys said the manufacturer implemented a fix for the carburetor problem more than two years before the crash but the fix was not implemented for thousands of aircraft already operating the field, including the Cessna that crashed.
The trial took place in February and March in Seattle. After hearing testimony and arguments from both sides, the jury awarded compensatory damages to the families. In a second phase of the trial, the jury awarded $6 million in punitive damages.
Becker was in private practice in Auburn and was a staff surgeon at Auburn Regional Medical Center.
Brenda Houston was an experienced airline pilot with United Airlines.
Source: http://www.kirotv.com
Yeager Airport unveils veteran memorial project
CHARLESTON, W.Va. - Twenty Air National Guard members who died 62 years ago in one of the worst air tragedies in southern West Virginia are being honored in a living memorial at Yeager Airport.
The special project, which aims to recognize contributions of West Virginia veterans, features photographs in the baggage claim area for the Wall of Honor. The wall was unveiled Monday, with the 130th Air Guard Honor Guard displaying post and retire colors.
The focal point of the first display is a tragic chapter of Yeager Airport's own history.
In 1951, an Air National Guard transport plane clipped the top of a hill and crashed about 10 miles north of Charleston. Nineteen servicemen immediately died; two others died later of their severe burns. Twenty of the 21 were from West Virginia.
The servicemen were flying home from Godman Air Base in Fort Knox, Ky., for the funeral of Major Woodford "Jock" Sutherland, who had been piloting an airplane in Florida that had gone the wrong direction on a taxiway and ended in a ground collision between two F-51 fighter planes.
"This event has big ties to the community," said Tim Murnahan, assistant director of the Yeager Airport. "It's altogether fitting and proper for this flight group to be our first showcase of photos on the 62nd anniversary of the crash."
Story and Photos: http://www.wvgazette.com
Asheville Regional Airport (KAVL), North Carolina: Plane Landing Gear Malfunction
A landing gear wheel on a twin-engine Cessna gave way upon landing this afternoon at Asheville Regional Airport, causing a fuel tank to rupture.
No one was injured in the 1:27 p.m. incident, according to airport spokeswoman Tina Kinsey. The Cessna 310 has three landing gear, and the wheel on the right side of the plane is the one that collapsed.
"The plane had to be removed from the runway," Kinsey said. "The runway reopened at 2:26 p.m. A few flights were impacted with slight delays."
A pilot and one passenger were on board. A fuel tank on the wing tip ruptured, but no fire resulted, Kinsey said.
No further information was available.
Man charged with murder after Welsh pilot found dead
Steven Barrett
A man has been charged with murder after a Welsh pilot was found dead in his Edinburgh flat over the weekend.
Steven Barrett, 27, originally from Kenfig Hill, near Bridgend, was found dead in Lochend Butterfly Way in the early hours of Saturday.
Mr Barrett had recently left the RAF and is understood to have been working as a first officer for budget airline Flybe out of Edinburgh Airport.
Police in Scotland said a 23-year-old man had been charged and is due to appear at Edinburgh Sheriff Court on Monday.
Mr Barrett attended Cynffig Comprehensive School before earning a Bachelor of Science degree at Swansea University.
While at university he became an active member of the University of Wales Squadron.
He was then selected to go to Cranwell to begin training to be a pilot with the RAF, graduating as a flying officer in 2007.
From Cranwell he went to RAF Wynton and later to Linton-on-Ouse, successfully completing the Basic Jet Training Course.
He was presented with his wings, achieving the title Flight Lieutenant Stephen Barrett Bsc (Hons) RAF in 2009.
Following the news of his death, friends posted tributes on the internet.
A message from the Rest Bay Lifeguards Club – which Mr Barrett used to a member of – read: "Steven was a member in our junior section for several years.
"He went on to achieve his dream and worked as a pilot.
"Our thoughts are with his family at this very difficult time."
While Hywel Davies wrote: "This is really sad news.
"I have a happy memory during one of the junior training sessions before we did risk assessments of Steven breast stroking out through 8ft surf much faster than I could get out front crawl (I was trying to catch up to tell him to return to shore).
"Gareth Law was doing backstroke at the same speed.
"I remember Steven as an awesome athlete and him and his brother Stuart were most valued club members.
"My thoughts go out to his family.
On Twitter Sara Orwin said: "Can't understand why you were taken from us so soon and so young.
"You were a great guy, Steven Barrett #restinpeace."
And Georgiana Browne wrote: "RIP Baz! You will be sorely missed by us all."
Story and Photo: http://www.walesonline.co.uk
Labels: Courts and Crime, Military, Pilots, Police Reports
Texas Turbine Conversions putting the brakes on expansion at North Texas Regional Airport (KGYI)
The development of North Texas Regional Airport-Perrin Field's west side has hit a snag after the first prospective tenant for undeveloped portion of NTRA has decided to stay where they are for the time being.
Texas Turbine Conversions had planned to move to a built-to-suit hangar along Highway 289. On Thursday the company will ask the Grayson County Regional Mobility Authority Board to release them from the lease agreement for the new hangar.
"There's been a change in circumstance with them and they've requested (the Board) to … release them from their obligation," said NTRA Director Mike Shahan.
The lease agreement between the airport and Texas Turbine called for the airport to construct the terminal shell and the company to finish the building. The company was to repay the airport for the cost of the hangar plus 5 percent interest over 20 years of a 40-year lease. Shahan said Texas Turbine has already made one down payment, slightly over $50,000, which will be forfeited with the termination of the agreement.
He stressed that the company, which has been modifying propeller aircraft since the 1990s, according to its website, still plans to operate out of the hangar it currently occupies at NTRA.
"They're not moving or anything," Shahan said. "They just can't commit to such a big project right now."
He said the airport still plans to complete construction of the 22,500-square-foot hangar. "We're going to continue to work on the hangar and go ahead and finish it out," Shahan said.
The airport director said NTRA has two prospects for the hangar. "We've got at least one company very seriously considering it," he said.
"It's a setback, but we're moving on," Shahan said. "We'll get it built. We'll get a tenant in there. Hopefully we'll get a tenant in there before we even have it built."
Source: http://heralddemocrat.com
GRAYSON COUNTY, TEXAS -- What was a big economic announcement for North Texas Regional Airport this summer is now being put on hold.
This past July, Texas Turbine Conversions announced their plan to lease to purchase a hangar out at NTRA.
Grayson County Commissioners approved spending up to $1.8 million dollars to build a hangar for the company to lease.
Today, Airport Director Mike Shahan says Texas Turbines is "holding off" on their lease because of a "change of circumstance."
Shahan would not give further comment. We'll bring you more information on this story as it becomes available.
Source: http://www.kxii.com
Labels: Airplanes, Airports, Hangars
Peru Perenco Helicopter Crash Leaves Nine Dead: Caída de helicóptero en Loreto deja al menos nueve muertos
A helicopter charted by French oil company Perenco SA in Peru crashed today in the northern Amazon jungle, killing all nine people on board, daily El Comercio reported in its online edition.
The Mil Mi-8PS carrying four crew members and five oil workers operated by Helicopteros del Pacifico fell to the ground near the Arabela River near where Perenco is drilling in Block 67, the Lima-based newspaper said.
It was the deadliest helicopter accident in the country since June 2010, when a chopper chartered by Samsung C&T Corp. and Korea Engineering Consultants Corp. crashed in the southern jungle, killing 14, El Comercio said.
Telephone numbers listed for Perenco's Lima office weren't answered outside business hours.
La nave se precipitó cerca del río Arabela, en el distrito de Napo, provincia de Maynas. Viajaban trabajadores de petrolera Perenco
Un helicóptero de la compañía Helipac (Helicópteros del Pacífico) cayó por causas que aún se desconocen cerca del río Arabela, en una zona con vegetación conocida como Curaray, correspondiente al distrito de Napo, provincia de Maynas, en la región Loreto.
Según pudo conocer El Comercio, la nave con cuatro tripulantes y cinco trabajadores de la empresa petrolera Perenco se dirigía al lote 67 de la base petrolera de esta compañía. Todos habrían muerto. Otras versiones señalan que serían 13 los fallecidos.
El helicóptero modelo MI-8 se precipitó aproximadamente a las 11:10 a.m. y aún continuaría ardiendo en llamas.
Un equipo de contingencia de la empresa Perenco y personal de Defensa Nacional del gobierno regional se dirigen a la zona. Aún se desconoce el nombre de los accidentados.
En los exteriores de la empresa petrolera, en Iquitos, se encuentran amigos y familiares de los trabajadores buscando información. Una mujer que dijo tener a un amigo y a su hermano entre las víctimas, comentó esta tarde a RPP que en representantes de la compañía le habían dicho que todos los ocupantes han fallecido.
Según reportó Canal N, tres de los fallecidos serían Hernán Cervantes, Máximo Rolando Cuello y Frider Pasmiño.
Source: http://elcomercio.pe
A helicopter Helipac Company (Pacific Helicopters) dropped for reasons still unknown about Arabela River, in an area known as Curaray vegetation, for the district of Napo province Maynas, in the Loreto region.
According to El Comercio could meet, the vessel with four crew and five employees of the oil company Perenco headed to lot 67 of the base oil from this company. All have died. Other versions say that the deceased would be 13.
The model helicopter MI-8 crashed at about 11:10 am and still continue to burn in flames.
A contingency team Perenco and Defence personnel regional government target area. Even the name is unknown to the injured.
In the foreign oil company in Iquitos, friends and family are workers looking information. A woman claiming to be a friend and brother among the victims, said this afternoon RPP that company representatives had been told that all passengers have died.
According to Canal N reported, three of the deceased would Hernán Cervantes, Rolando Maximum Neck and Frider Pasmiño.
Beech A36TC, N36SG: Accident occurred April 06, 2013 in Big Bear City, California
NTSB Identification: WPR13LA184
14 CFR Part 91: General Aviation
Accident occurred Saturday, April 06, 2013 in Big Bear City, CA
Aircraft: BEECH A36TC, registration: N36SG
Injuries: 1 Serious,3 Minor.
This is preliminary information, subject to change, and may contain errors. Any errors in this report will be corrected when the final report has been completed. NTSB investigators may not have traveled in support of this investigation and used data provided by various sources to prepare this aircraft accident report.
On April 6, 2013, about 1035 Pacific daylight time, a Beech A36TC airplane, N36SG, was substantially damaged during a forced landing in Big Bear Lake, Big Bear City, California. The pilot received serious injuries, and his three passengers received minor injuries. The personal flight was conducted under the provisions of Title 14 Code of Federal Regulations Part 91. Visual meteorological conditions prevailed at the accident location. The airplane had departed on an instrument flight rules (IFR) flight plan from McClellan-Palomar Airport (CRQ), Carlsbad, California, and was destined for Big Bear City Airport (L35).
According to the pilot, his passengers include his niece, her husband, and their 5-year-old son. En route, the pilot cancelled his IFR flight plan, and continued, using flight following services from SoCal Approach. The pilot planned to follow his normal route, which included a slow let down to L35 after passing ARRAN intersection. Shortly after the airplane passed ARRAN, at an estimated altitude of between 9,500 and 9,000 feet, the airplane experienced a "very sudden" onset, high-frequency vibration. The pilot reduced the power setting, and notified the air traffic controller of an engine problem. Initially the manifold pressure gauge provided very erratic indications, but the pilot was unable to discern whether this was a valid indication, or the instrument responding to the vibration. The airplane began to descend, and the pilot adjusted the engine controls to yield normal manifold pressure and rpm cruise values. However, the airplane did not respond commensurately with the indicated power setting, and continued its descent. When the airplane was approximately abeam the "observatory" (a visual landmark) at an altitude of about 6,900 feet, the pilot switched to the airport advisory frequency, and notified them of his engine problem. The pilot observed a causeway that crossed the lake between his position and the airport, and began a turn to his right (south) to avoid the causeway. At some point, the pilot recognized and broadcast that he was unable to make the airport, and planned to land in a field west of the airport. The pilot banked to the right to turn for the field, but the right wingtip contacted the lake. The airplane "flipped," and came to rest inverted in water about 1 to 3 feet deep. The impact location was about 0.3 miles southwest of the causeway, and about 1.25 miles southwest of the L35 runway 8 threshold. L35 elevation was 6,752 feet.
An eyewitness on his property on the lake was attracted to the airplane by its sound. He estimated that at that time he noticed the airplane, it was about 30 to 35 feet above the lake, and headed south, towards him. He then saw it bank right, and impact the lake. He summoned 911 assistance, and then ran into the lake to render assistance.
According to Federal Aviation Administration (FAA) records, the pilot held a private pilot certificate with airplane single engine land and instrument airplane ratings. He reported that he had a total flight experience of approximately 1,300 hours, including about 1,250 hours in the accident airplane make and model. The airplane was manufactured in 1979, and was equipped with a Teledyne Continental TSIO-520 engine.
The L35 1035 automated weather observation included winds from 260 degrees at 8 knots, variable between 224 and 284 degrees; visibility 10 miles, clear skies; temperature 15 degrees C; dew point minus 8 degrees C; and an altimeter setting of 30.14 inches of mercury.
Two people were injured Saturday morning, April 6, when an airplane crashed in shallow water on the south shoreline of Big Bear Lake.
San Bernardino County sheriff's deputies described the crash as "more of a hard landing." The injuries were minor.
The airplane, a Beech A36 Bonanza, lost engine power and ditched in shallow water at 10:41 a.m., according to Federal Aviation Administration spokesman Allen Kenitzer.
Three people were aboard, according to the San Bernardino County Sheriff's Department, although Kenitzer said there may have been as many as four. The aircraft had substantial damage.
Details about the aircraft involved in the crash were not available, but specifications found on the Internet indicate it is normally a six-seat, single-engine aircraft.
No information was available on where the plane took off from or where it was headed.
The FAA and the National Transportation Safety Board are investigating, Kenitzer said in an email, noting that it could take months before a preliminary report is prepared.
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 5,955
|
Home | University hubs | Queen's University Belfast | Reviews students | Pros
Uni overview
What are the best things about studying at your university?
"The good lecture content and good project options."
"All of the lectures are readily available online to review."
"The library is very accessible and everything is within a five-minute walk of the university."
"The university provides a range of opportunities for students to enhance their student experiences at university to the maximum. Aside from the availability of broad ranges of clubs and societies, the students' union runs a student council that forms the most powerful student body of the university and all students are welcomed to run and be elected for the positions. Staff members, lecturers and tutors alike are very supportive of their students in general (at least for the school of law)."
"The drama society at my university is probably my favourite part of my experience. I love most of my lecturers too and find them to be really passionate about what they're teaching, which makes it easier to be engaged in learning."
"I loved everything about my experience. We have top-of-the-range facilities and I've never had a bad experience with a staff member, either lecturer or clerical. The social scene is incredible and there are so many societies and clubs to choose from, along with the speakeasy/Mandela being everyone's favourite night-out spots."
"I like the fact that my accommodation is within walking distance of my classes."
"My degree is of excellent quality and I've received great help from the university staff. Belfast is a pretty city and has a lot of things to do."
"The lecturers at QUB law school are phenomenal and so supportive. I love Belfast, it's a great city that's small enough to have friendly people and a sense of community while, at the same time, a really cool artsy side to it. Belfast is definitely a burgeoning city in terms of culture, art and hipster cafes. You just need to do a little walking and you'll find some fantastic bars, cafes and shops."
"The course is very well laid out and the teachers are well tuned in and willing to give support."
"The university provides a creche for my children and I've received a lot of support from my department."
"Paying a reasonable amount instead of the stupid fees charged in Britain is a big plus."
"It has a good library and gym. Rent in the city is cheaper than elsewhere in the UK."
"I like the social activities, course options and the library."
"The area is great with many local things including pubs, clubs, shops and cinemas etc. My course is great and the learning environment is excellent."
Sign up to access to use your dashboard and receive extra advice in your inbox
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 8,187
|
Cesty '85, Panton na Portě - Live (1985) je sampler živých nahrávek z národního finále Porty 1985 vydaný Pantonem na LP desce. LP bylo vyrobeno velice rychle, v podstatě přes noc, obal byl již vytištěn předem a seznam písní je pouze na přiloženém listě. Panton se tímto LP dostal mezi samplery z Porty, které pravidelně vydával Supraphon.
Písně
Padá láska (Wabi Ryvola), Stopa
Starodávná (Luboš Marek / Karel Filip), Sem tam
Máme své chvíle, Milena Zíchová
Tragédie filmového štábu (Miroslav Paleček / Michael Janík), Miroslav Paleček a Michael Janík
Sůl v očích (Marek Eben, písnička ze hry Matěj Poctivý Studia Y), Bratři Ebenové
Na rozloučenou (Pavel Lohonka), Žalman & spol.
Signály (Jan Nedvěd), Brontosauři
Už se nám čas krátí (tradicionál, úprava Jiří Tichota / Jiří Tichota), Spirituál kvintet
Píseň nová o strašlivé s ocasem kometě (Miroslav Janoušek), Miroslav Janoušek a Luboš Vondrák
L'avor azur, Dagmar Andrtová-Voňková
Nehrálo se o ceny (Jaroslav Samson Lenk, Jaromír Vondra, Ladislav Kučera / Ladislav Kučera), Hop trop
Balada o pozdním hasebním zásahu požárního sboru v Kostelní Řečici, Ivo Jahelka
Vítr (Jaroslav Samson Lenk), Máci
Vím co smím (Pavel Zajíc), Nezmaři
Česká hudební alba
Folková hudební alba
Samplery
Koncertní alba z roku 1985
Kompilační alba Pantonu
Koncertní alba Pantonu
Kompilační alba z roku 1985
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 4,715
|
Karel Javůrek (30. července 1815 Praha – 23. března 1909 Praha) byl český malíř, autor mnoha pláten s výjevy z českých dějin. Studoval u Christiana Rubena na pražské malířské akademii, absolvoval i stáže v Drážďanech, Antverpách a Paříži. Upozornil na sebe dvěma obrazy s husitskou tematikou – Loučení Husa s krajany (1846) a Černá země (1852; původně nazýván Spáleniště kostnické). V následujících desetiletích pak vytvořil stovky maleb zpodobňujících téměř všechna období české historie a také několik církevních obrazů s biblickými motivy. Společně s Jaroslavem Čermákem je považovaný za průkopníka české historické školy. Každoročně vystavoval, jeho díla se dobře prodávala. Jeho umělecký styl ustrnul v mladém věku, nevyvíjel se. Všechny obrazy jsou poplatné sentimentálnímu romantismu 40. let 19. století, ke konci jeho života již velmi zastaralému. Současníci oceňovali jeho lásku k dějinám, vlastenectví a pěknou kompozici obrazů.
Život
Byl synem pražského továrníka, výrobce pentlí, otce ale ztratil již v mladém věku. Jako dvacetiletý studoval nejprve rok ve Vídni u Josefa Danhausera, poté se vrátil do Prahy na malířskou akademii. Z finančních důvodů si musel vedle studia přivydělávat kreslením. Jeho učitelem byl nejprve František Tkadlík, kterého později vystřídal nový ředitel Christian Ruben. Ten se stal Javůrkovi největším vzorem, pod jeho vedením maloval na akademii historické obrazy až do roku 1850.
Jako student se s Rubenem také zúčastnil revolučních událostí roku 1848; dvě pikantní historky z té doby zaznamenal Rudolf Jaroslav Kronbauer.
Roku 1850 odešel na dva roky do Antverp, kde v té době působili Gustaaf Wappers, Teigmans a Versagern. V roce 1855 pak odjel společně s Jaroslavem Čermákem na roční stáž do dílny Thomase Coutura v Paříži; pobytu v tomto městě využil i ke zdokonalování techniky kopírováním starých mistrů v Louvru a Palais du Luxembourg. Během padesátých let také pravidelně trávil letní měsíce v Drážďanech.
V následujících desetiletích žil v Praze a až do pozdního věku tvořil historické malby, jejichž výběr každoročně prezentoval v Rudolfinu na výstavě Krasoumné jednoty. Na počátku 20. století byl považován za patriarchu české malířské obce a jednoho z nejpopulárnějších umělců. Současníci oceňovali jeho rozsáhlou tvorbu i vlastenectví a lásku k dějinám, neopomíjeli ale na druhou stranu ani to, že jeho styl ustrnul a po dlouhých šedesát let zůstával věrný původnímu rubenovskému romantickému pojetí.
V souvislosti s historickou tematikou své tvorby byl Javůrek sběratelem starožitností, zejména historických zbraní. Jeho sbírku dále rozšiřoval syn Jaromír a další potomci. Díky účasti na revolučních událostech roku 1848 a umělecké podpoře významných témat českých dějin vnímáme Javůrka jako charakteristickou osobnost národního obrození. Pohřben byl na Olšanských hřbitovech po obřadu v kostele sv. Ignáce z Loyoly na Karlově náměstí, za účasti mnoha umělců.
Rodinný život
Karel Javůrek se dne 23. 5. 1852 oženil v pražském kostele Panny Marie Sněžné s Marií Ungerovou (1818 – 1. 1. 1871). Manželé Javůrkovi měli syna Jaromíra (1859–??) a dceru Zdenku (1861–??). K jeho potomkům patří také bioložka Olga Javůrková (Olga Fassatiová) a její syn Tomáš Fassati.
Dílo
Karel Javůrek patřil k velmi plodným malířům. Ve svých dílech zachytil téměř všechna období českých dějin. Podobně jako jeho současníci — Jaroslav Čermák, Antonín Lhota a Josef Vojtěch Hellich — se v mládí postavil proti dosavadním tradicím pěstovaným na pražské akademii. Jeho práce se vyznačují láskou k dějinám a pěknou kompozicí. Věhlasu a významu Jaroslava Čermáka ale nedosáhl.
Proslavily ho především obrazy Loučení Husa s krajany (1846) a Černá země (1852). Na druhém z nich, původně nazvaném Spáleniště kostnické, zachycuje Jana z Chlumu a Václava z Dubé, jak sbírají zemi na místě, kde byl upálen Jan Hus. Inspirací byla báseň J. E. Vocela. Obraz byl oceňovaný v českých vlasteneckých kruzích, ale naopak si výběrem tématu vysloužil kritiku německého listu Bohemia.
K dalším historickým obrazům patří: Smrt císaře Albrechta, Zatčení císaře Sigmunda, Zavraždění Svatopluka, Jan Lucemburský, Konec posledního Přemyslovce, Ladislav Pohrobek, Svatý Vojtěch, Vršovci, Bedřich Falcký aj. Maloval také oltářní obrazy a portréty.
Po celou svou šedesátiletou uměleckou dráhu zůstával věrný romantickému rubenovskému pojetí, jeho styl zůstal neměnný. Námětem mu byly především historické studie, básně a divadelní hry, a zpracovával je v sentimentálním romantickém stylu. Postavy vyzařují spíš šablonovitý herecký patos než reálné životní emoce. Kompozice všech obrazů je podobná, většinou vypadají, jakoby zachycovaly poslední scény dramat na jevišti. Uměleckým cílem je připomínat dávné události. Lpěním na osvědčeném stylu a neochotou přijímat nové směry se sám Javůrek počátkem 20. století stal symbolem a ilustrací minulých dob. Až do konce života byl jedním z nejpopulárnějších umělců, jeho obrazy snadno nacházely ochotné kupce a byly reprodukovány v časopisech (např. Světozor nebo Zlatá Praha).
Retrospektivní autorskou výstavu připravilo Javůrkovi ve spolupráci s Národní galerií v Praze Muzeum umění a designu Benešov (Karel Javůrek, malíř české historie. Benešov 9. 8. – 7. 9. 1994, kurátor Tomáš Fassati), které současně vydalo životopisnou publikaci.
Zastoupení ve sbírkách (výběr)
Javůrkovo dílo je zastoupeno ve sbírce Národní galerie v Praze, Národního muzea v Praze, Galerie hlavního města Prahy a lze jej vidět také v historických objektech (zámcích a kostelích), pro které je vytvářel.
Galerie
Reference
Externí odkazy
Soupis pražského obyvatelstva 1830-1910 (1920), Jawurek Karl (1815)
Pobytové přihlášky pražského policejního ředitelství (konskripce) 1850-1914, Jawurek Karl (1816)
České malířství 19. století
Čeští malíři 19. století
Narození 30. července
Narození v roce 1815
Pohřbení na Olšanech
Narození v Praze
Úmrtí 23. března
Úmrtí v roce 1909
Úmrtí v Praze
Muži
Čeští malíři 20. století
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 1,683
|
How One Globe-Trotting Grandmother Faced Her Breast Cancer
From Diagnosis to Reconstruction
Sometimes society holds stereotypes, imagining the quiet, meek woman who slowly falls ill with frightening diseases like cancer. We rarely expect the active, fit, and healthy to show up as sick. Yet, Vicki Lucas reminds us that breast cancer is not limited by lifestyles or stereotypes.
Living on the outskirts of Salt Lake City, Utah, Vicki is a busy woman. She has mothered six grown children and thoroughly enjoys her 26 grandchildren. However, she might not fit the usual "granny" stereotype. Skiing, running, hiking, and rockhounding are all part of Vicki's list of hobbies. In addition she continues to teach nursing classes at a local college and enjoys being actively involved in her church and community. Ironically, an accident during one of her adventures began Vicki's journey into unfamiliar territory.
An Injury and Two Diagnoses
Vicki had always maintained regular mammograms. However, this time she had been out of the country for two years and had missed her yearly checkup by about six months when she finally visited her doctor and had her usual screening mammogram.
Being as active as ever, in the days that followed her mammogram, Vicki found herself exploring an unusually rocky area. Unfortunately, she slipped and severely injured her knee. During medical follow-up, she learned she had an anterior cruciate ligament (ACL) injury, putting the usually active Vicki in an awkward metal brace from her hip to her ankle.
The ACL injury required surgery. Thus it was that as she was leaving the pre-op for her upcoming ACL surgery, only focused on getting her knee well, she received a call from her primary physician with the results from her mammogram. Her doctor asked if she wanted to come into the office to discuss the findings, but Vicki asked the doctor to tell her over the phone. When the doctor explained there was a problem Vicki understood, but all she could think about was taking care of her knee before she could handle anything else. She told her doctor that she would follow up on the issue after her knee surgery.
Thankfully, her doctor did not agree and told Vicki she must find an oncologist immediately.
Despite dealing with two critical medical issues simultaneously, Vicki believes the entire series of events was a miracle. She considers the injury to her leg a blessing in disguise, as it forced her to slow down and take life one day at a time. Her knee surgery was scheduled for two weeks later. In the meanwhile she was able to consult with an oncologist.
She was diagnosed with Stage 1 breast cancer.
Treatment: A Best-Case Scenario
Her oncologist recommended a lumpectomy, which was performed ten days after her knee surgery. Thankfully, her surgeries occurred during a holiday period, making life easier since she was already on break from her teaching duties, and family and friends were available to help.
She has kept her good nature and sense of humor, recalling a visit to the imaging center following her biopsy. The women in the waiting area were watching her maneuvering across the room with her crutches. She looked at them and smiled, "What can I say? Bum knee and bum breast."
Incidentally, two years before Vicki's diagnosis, she had received breast implants. When they did the lumpectomy, the cancer was removed without agitating the implant and it was not necessary to remove it. More importantly, however, when Vicki's sentinel node was removed and biopsied, the surgeon was able to declare her lymphatic system free of cancer. This meant a less aggressive treatment plan.
Vicki had a nine-week course of radiation after the lumpectomy, pinpointed to the specific area of her tumor. However, the last few treatments of radiation were more intensive "bursts" of radiation which affected the surrounding area and increased the scar tissue that formed in her breast. Vicki said the radiation wasn't as bad as some women have to endure, but it wasn't pleasant.
As radiation came to an end, Vicki couldn't help but feel her diagnosis and treatment weren't as bad as they could have been. Today she explains, she came to the conclusion, "I'm going to enjoy what I've got as long as I've got it!"
A Growing Discomfort and a Choice
Vicki really didn't think about reconstruction for a long time though. After all, the deformed area from the lumpectomy was pretty small, and her oncologist never even discussed the possibility with her. Most importantly, she was alive.
Over time, some of her friends did talk to her about plastic surgery, but still Vicki says she didn't think about it too much until she began to be bothered by an area that had started to harden. When she brought the hardened area to the doctor's attention, she was told she was experiencing a common side effect of breast implants: capsular contraction. It occurs when the body, sensing the "intrusion" of a foreign object (breast implants, in this case) forms scar tissue around the object to essentially seal it off from the rest of the body. However, over time that scar tissue (especially if exacerbated by radiation) can thicken and become extremely hard. In some cases, this can be quite a painful process. In Vicki's case, the hardening wasn't particularly painful, but it was extremely annoying. It bothered her that the original implant had begun to change shape, and her areola was more ovalular than round.
Not surprisingly, Vicki felt mixed emotions–gratitude that her cancer hadn't been worse, but disappointed that "my breast wasn't normal."
Vicki made a choice for herself. She decided that she didn't need to live with this uncomfortable condition and that she wanted to have her breast look normal again. She was referred to a well-informed plastic surgeon. He told her that some doctors claim that the scar tissue can be reduced by squeezing or massaged, but this does not work. He recommended removing the volume of scar tissue, then trading the implant with a new type of implant that is texturized. It does not adhere as easily to existing scar tissue and therefore has demonstrated a lower rate of encapsulation.
Today: Permission to be Normal Again
On the day of the interview for I'm Taking Charge, about four weeks have passed since Vicki had the procedure to trade out the implant. She says she still has some swelling, but she's glad that she chose to undergo the reconstruction, despite being unsure of how much of the process will be covered by her insurance. Most insurance policies will cover mastectomy reconstruction but not repair from a lumpectomy. Everyone's insurance varies, however, and may or may not include reconstruction following a lumpectomy. Vicki is content that she made the right decision for her, regardless of the cost.
When asked if she could go back and change anything that happened, Vicki says she would broaden the circle of people whom she allowed to support and encourage her. When she was diagnosed she decided to keep her situation relatively private, only sharing it with a handful of close friends and family. She felt it would be easier to make decisions with less input from too many people. Today she feels differently. She wants to share her experience with her co-workers, community, and church family. She wants her journey to be a success story, especially for her daughters and granddaughters, to let others know what she went through and how to take charge of their lives if faced with cancer.
At the end of the day, Vicki considers herself extremely blessed that her cancer was caught early and she did not need a mastectomy. Typical of many survivors, she said she often feels that she doesn't deserve to have had such a positive result from her experience with cancer, when many women don't receive a prompt diagnosis, treatment, or proper care. Vicki humbly shares that her prognosis is "wonderful" and happily admits, "I've given myself permission to be normal again."
She still recalls how everything unfolded. The scary thoughts don't go away, she explains, but she feels like she's learned so much from what she went through. She has been given time and more energy to pursue the things that bring her happiness – friends, church, career, community, nature, and recreational activities. Perhaps the most heartfelt message she has gleaned is that "life is precious and I know how much my family truly cares." Vicki is grateful that her damaged breast is now as normal as it can be, and life can move on–back to the busy and active lifestyle that Vicki loves so much.
Susan J.Z. Barnes 2018-01-11T04:28:04+00:00
Kimberley Hamilton October 8, 2016 at 4:39 pm - Reply
completely reminds me of my grandmothers story. However, she said that she had been on so many journeys, and this was just another, with a bit of a bumpy detour. You can make it through this if you look back at everything else you have already made it through.
#VoicesofLove
Sally Casey October 10, 2016 at 1:17 pm - Reply
Thanks, Kimberly! I really like this perspective! Awesome words for our #Voicesoflove!
About the Author: Susan J.Z. Barnes
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 5,114
|
Q: Functionality of void operator()() I am confused about the functionality of void operator()().
Could you tell me about that, for instance:
class background_task
{
public:
void operator()() const
{
do_something();
do_something_else();
}
};
background_task f;
std::thread my_thread(f);
Here, why we need operator()()? What is the meaning of the first and second ()? Actually, I know the operation of normal operator, but this operator is confusing.
A: The first part operator() is the way to declare the function that is called when an instance of the class is invoked as a function. The second pair of parentheses would contain the actual arguments.
With a return value and arguments this might make a bit more sense:
class Adder{
public:
int operator()(int a, int b){
//operator() -- this is the "name" of the operator
// in this case, it takes two integer arguments.
return a+b;
}
};
Adder a;
assert( 5==a(2,3) );
In this context, the std::thread will internally invoke f() inside the thread, i.e. whatever is inside the body of operator() is what gets done inside that thread.
A: You can overload the () operator to call your object as if it was a function:
class A {
public:
void operator()(int x, int y) {
// Do something
}
};
A x;
x(5, 3); // at this point operator () gets called
So the first parentheses are always empty: this is the name of the function: operator(), the second parentheses might have parameters (as in my example), but they don't have to (as in your example).
So to call this operator in your particular case you would do something like task().
A: The first () is the name of the operator - it's the operator that is invoked when you use () on the object. The second () is for the parameters, of which there are none.
Here's an example of how you would use it:
background_task task;
task(); // calls background_task::operator()
A: All the hints that were contributed above are correct to sequential programs, I mean, programs without threads. Using threads things change.
First of all, by default parameters to std::thread are functions and functions parameters. Probably you were studying the book "C++ concurrency in action", and the author shows an interesting example:
void do_some_work();
thread my_thread(do_some_work); //thread receives the function address
Suppose this function:
void do_other_job(int k);
In the body of the code, you should do:
k=3;
thread my_thread2(do_other_job, k);
in order to spawn another thread.
So, using threads the compiler interprets f ( in std::thread my_thread(f);) by default as a function instead of a class. To change that you have to initiate an operator() to warn the compiler you are working with a class.
An alternative code could be:
class background_task{
public:
background_task(){
do_sth();
do_sth_else();
}
void operator()(){}
};
background_task f;
thread mythread10(f);
Eventually, it's not correct, using threads, feeding the operator, so this code doesn't work:
void operator()(int x){
do_sth();
cout<<"x = "<<x<<endl;
}
It happens because all the code inside the brackets are read-only, and cannot be changed during the run-time. If you aim to insert a variable in the constructor, it must be put in the thread initialization. So:
class backg{
public:
backg(int i){
do_sth(i);
}
void operator()(){}
};
int main(){
thread mythread{ backg(12) }; //using c++11
return 0;
}
Will run without mistakes, and will perform the function do_sth(12) in the thread spawned.
I hope I have helped.
|
{
"redpajama_set_name": "RedPajamaStackExchange"
}
| 3,383
|
Downtown Hanover's newest space has 'character'
The space above Something Wicked Brewery, known as the Blackrose Ballroom, is now available for rent for events.
Downtown Hanover's newest space has 'character' The space above Something Wicked Brewery, known as the Blackrose Ballroom, is now available for rent for events. Check out this story on eveningsun.com: http://evesun.co/2svfXSi
Lindsey Welling, The Evening Sun Published 7:00 a.m. ET June 5, 2017 | Updated 12:09 p.m. ET June 8, 2017
The Black Rose Ballroom sits above Something Wicked Brewing Company at 34 Broadway. It is a recently opened rental space for private events. Lindsey Welling, The Evening Sun.
Steve Buenzow stands in front of one of two big murals that standout in the new Black Rose Ballroom, Thursday, May 25, 2017. Black Rose Ballroom, above Something Wicked, is ready to be used, after four years of renovation.(Photo: Ty Lohr, The Evening Sun)Buy Photo
The building at 34 Broadway in Hanover is filled with character.
Upstairs, in the Black Rose Ballroom, retro mixes with modern. The space, which recently opened, serves as a private event space and is available for rent.
Two murals, one depicting the signing of the Declaration of Independence, span floor to ceiling on the back wall, a giant wall sculpture of a black rose hangs on a side wall and two long rows of hanging orb-shaped lights reflect on the chocolate wood floor.
More: New brew: Aldus teams with Miscreation, Something Wicked
"I wanted to keep this kind of classic character that it has," said Black Rose Ballroom owner Steve Buenzow.
The idea for the venue started four years ago when Buenzow bought the building in an auction.
"I was looking for an investment property," Buenzow said.
Buenzow has a background in electrical engineering and said he grew up doing construction work and was looking for a new challenge. The building was vacant for about seven years prior and needed a lot of work, Buenzow said.
He got to work renovating the downstairs, which opened in 2015 as Something Wicked Brewing Company. When that was finished, he turned his focus to the upstairs.
With the help of some friends, Buenzow did major renovations to the plumbing and kitchen but only some minor atheistic renovations, including a new coat of paint and new lighting.
"I didn't want to fundamentally change the look and feel because when people see that it brings back memories," Buenzow said.
The history of the building dates back 130 years.
The name for the Black Rose Ballroom originated from the giant black rose that was already in the building when it was purchased, Thursday, May 25, 2017. Black Rose Ballroom, above Something Wicked, is ready to be used, after four years of renovation. (Photo: Ty Lohr, The Evening Sun)
Many remember the building from when it was the Moose Lodge from 1910 to 2002.
More: Borough may accept bids on oldest house in Hanover
Some of the original pieces in the room still stand from when it was the Moose Lodge, including the two wall murals and rose sculpture.
The high ceiling ballroom might bring back some memories for anyone who remembers sitting on Santa's lap during Christmas time or going to dinner events at the Moose Lodge.
The public had the chance to peak inside the venue on June 3 at Main Street Hanover's Sip & Stroll.
Buenzow said some other events are already lined up, including a fundraiser. Another idea is an Oktoberfest event with Something Wicked and other local breweries.
More: Hanoverians celebrate Oktoberfest in the square
Read or Share this story: http://evesun.co/2svfXSi
Police: Man exposed himself, assaulted girl
Nine people displaced after fire in Littlestown
8,300 Met-Ed customers lose power in Hanover area
Unique summer desserts now at 3 Hogs
Eisenhower Drive: 53 properties could be taken under one plan
Adams County boy pays it forward
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 4,127
|
{"url":"https:\/\/math.stackexchange.com\/questions\/1706438\/relation-between-exterior-derivative-and-lie-bracket","text":"# Relation between exterior derivative and Lie bracket\n\nThere is a formula connecting the exterior derivative and the Lie bracket $$d\\omega (X,Y) = X \\omega(Y) - Y \\omega(X) - \\omega([X,Y]).$$\n\nWhat is a good way to remember this? By which I mean, what structure does this reveal? (Or, what essentially is going on here?) In his book, John Lee says, \"In a sense, the Lie bracket is dual to the exterior derivative.\" But that's not really satisfying to me.\n\n\u2022 I'm still wondering in what sense the Lie bracket is dual to the exterior derivative. \u2013\u00a0Andrews Apr 17 at 13:54\n\u2022 You should not remember this because this is not the formula followed at all places. For example in Kobayashi and Nomizu, what you said does not hold but $d\\omega(X,Y)=\\frac{1}{2}(X(\\omega(Y))-Y(\\omega(X))-\\omega([X,Y]))$.. This is because their wedge product is defined differently... \u2013\u00a0user537667 Aug 14 at 19:15\n\u2022 \u201cIn particular, it shows that if we know all the Lie brackets of basis vector fields in a smooth local frame, we can compute the exterior derivatives of the dual covector f ields, and vice versa.\u201d Page 369, 2nd edition of Lee\u2019s smooth manifold book \u2013\u00a0user537667 Aug 14 at 20:28\n\nEric, if you take $X=\\partial\/\\partial x^i$, $Y=\\partial\/\\partial x^j$, and $\\omega = f dx^k$, then of course the formula gives you what you expect from the first two terms, and the bracket term disappears because $[X,Y]=0$ in this case. Why is the bracket term there in general? It's because $d\\omega$ is a tensor field, and therefore must be linear over the $C^\\infty$ functions. However, $X(\\omega(Y)) - Y(\\omega(X))$ is not linear over the $C^\\infty$ functions, and it is the bracket term that precisely corrects for that. In general, formulas such as this that one can guess by applying to coordinate vector fields have bracket terms appearing exactly for that reason.\nFix a local frame $$(E_a)$$ and let $$(\\theta^a)$$ denote its dual coframe, so that $$\\theta^a (E_b) = \\delta^a{}_b$$; in particular each such contraction is constant. Then, for frame and coframe elements the exterior derivative formula simplifies to $$\\phantom{(ast)} \\qquad d\\theta^a(E_b, E_c) = -\\theta^a([E_b, E_c]) . \\qquad (\\ast)$$ But the right-hand side is (up to sign) exactly the structure function $$C^a{}_{bc}$$ of the frame, and the structure functions tell us everything there is to know about the geometry of the frame $$(E_a)$$, or equivalently the coframe $$(\\theta^a)$$. Put another way:\nThe exterior derivatives $$d\\theta^a$$ contain the same information encoded by the Lie bracket but instead expressed in the (dual) language of the dual coframe.","date":"2019-09-23 01:08:17","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\": 0, \"img_math\": 0, \"codecogs_latex\": 0, \"wp_latex\": 0, \"mimetex.cgi\": 0, \"\/images\/math\/codecogs\": 0, \"mathtex.cgi\": 0, \"katex\": 0, \"math-container\": 8, \"wp-katex-eq\": 0, \"align\": 0, \"equation\": 0, \"x-ck12\": 0, \"texerror\": 0, \"math_score\": 0.9365297555923462, \"perplexity\": 223.3942603585502}, \"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-2019-39\/segments\/1568514575844.94\/warc\/CC-MAIN-20190923002147-20190923024147-00107.warc.gz\"}"}
| null | null |
{"url":"https:\/\/www.gradesaver.com\/textbooks\/math\/algebra\/algebra-and-trigonometry-10th-edition\/chapter-11-11-6-counting-principles-11-6-exercises-page-823\/23d","text":"## Algebra and Trigonometry 10th Edition\n\n$600$ three-digit numbers are possible.\nFirst digit: 6 choices: $4,5,6,7,8,9$ Second digit: 10 choices Third digit: 10 choices Using the Fundamental Counting Principle: $6\\times10\\times10=600$","date":"2019-10-19 05:49:31","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\": 0.9027107954025269, \"perplexity\": 7297.743711271104}, \"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-2019-43\/segments\/1570986688826.38\/warc\/CC-MAIN-20191019040458-20191019063958-00533.warc.gz\"}"}
| null | null |
Augsburg took 13 points from its last five German Bundesliga games before the competition's winter break but one does not agree with the relatively short odds available about Markus Weinzierl's team winning at high-flying Hertha on Saturday.
The trip to Hertha will be Augsburg's toughest assignment in any tournament since it lost 0-2 at home to Dortmund in the German DFB-Pokal in the middle of December and one cannot get carried away with the Bavarian side's German Bundesliga results from late November onwards in spite of its points tally. During that period Augsburg did not face any of the German Bundesliga's top five teams and it met its two toughest opponents – Schalke and Wolfsburg – at home.
Hertha was the surprise packet of the first half of the German Bundesliga season, posting a record of 10-2-5 to occupy third position on the ladder – six points adrift of Dortmund but three points ahead of Monchengladbach. The straightforward nature of the German Bundesliga draw means that every side played each other before the teams had four weeks off and, according to one's collateral form analysis, Hertha enjoys a 11-3-3 lead over Augsburg, including a 1-0 away win in the reverse fixture – one notes, though, that Pal Dardai's outfit scored from the penalty spot shortly after Markus Weinzierl's charges went down to 10 men.
William Hill are listing odds of 5/4 2.25 +125 1.25 1.25 -0.80 about Hertha beating Augsburg and that does not make sense. Odds of 5/4 2.25 +125 1.25 1.25 -0.80 about a German Bundesliga home side implies that they are slightly worse than its guest yet Hertha is third on the ladder, nine rungs and 13 points in front of Augsburg. One's German Bundesliga collateral form analysis supports the ladder and it does not hurt Hertha's case that it took out the reverse fixture as well. Hertha should be odds on to defeat Augsburg at home so the German Bundesliga winter bronze medallist (who we tipped to finish in the top 4) rates as one's best bet on the card as the competition gets back under way.
Sunday's German Bundesliga match between Monchengladbach and Dortmund is the highlight of Round 18 and one likes a couple of bets in the clash between fourth and second – over two and a half goals and the Foals not to lose to their guests.
Monchengladbach and Dortmund were the kings of over two and a half goals in the first 17 rounds of the German Bundesliga tournament – 14 Foals games went over the popular line and 15 BVB matches exceeded it, including their reverse fixture. Marathonbet is quoting odds of 57/100 1.57 -175 0.57 -1.75 0.57 about two and a half goals and one cannot see that price lasting until kick-off time.
Nor can one see Monchengladbach remaining odds against to avoid defeat in its German Bundesliga blockbuster versus Dortmund – 888Sport and Unibet are offering 5/4 2.25 +125 1.25 1.25 -0.80 , which is remarkable even allowing for the fact that BVB thrashed the Foals 4-0 in August. According to one's German Bundesliga collateral form analysis, Dortmund boasts a 8-4-5 edge over Monchengladbach so, with BVB hitting the road, one would argue that the game should be very close to pick 'em.
Finally, here are one's German Bundesliga collateral form analysis results for every one of Round 18's nine matches. The asterisk signifies the direct head-to-head winner.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 9,521
|
Q: WeakEventManager not handling interfaces correctly? I try to use the WeakEventManager to - well - use weak events. As source type I am using an interface and that is causing a problem. See code sample below.
When I start the program as is, it errors when calling WeakEventManager (last statement, see comment "here's the error"). It says, there is no PropertyChanged event on IMyEventSource. I then thought to add event PropertyChangedEventHandler PropertyChanged to IMyEventSource. That indeed fixes the issue but leaves me with the warning that my event declaration in the interface would hide the inherited member. Which makes sense as IMyEventSource says it would implement INotifyPropertyChanged. It's feels like catch-22.
Obviously I can accept the warning and add new to the event declaration. But I hope someone can tell me how to do better.
class Program
{
public static void Main(string[] args)
{
MyEventSource source = new MyEventSource();
WeakEventListener listener = new WeakEventListener(source);
source.Name = "John Dow";
Console.ReadLine();
}
}
public interface IMyEventSource : INotifyPropertyChanged
{
// event PropertyChangedEventHandler PropertyChanged; // fixes the issue
string Name { get; set; }
}
public class MyEventSource : IMyEventSource
{
private string name;
public string Name
{
get => name;
set { name = value; PropertyChanged?.Invoke(this, new PropertyChangedEventArgs("Name")); ; }
}
public event PropertyChangedEventHandler PropertyChanged;
}
public class WeakEventListener
{
private void OnEvent(object source, PropertyChangedEventArgs args) => Console.WriteLine("WeakEventListener received event.");
public WeakEventListener(MyEventSource source)
{
WeakEventManager<IMyEventSource, PropertyChangedEventArgs>.AddHandler(source, "PropertyChanged", OnEvent); // here's the error
}
}
|
{
"redpajama_set_name": "RedPajamaStackExchange"
}
| 2,794
|
title: quota Config
overview: Generated documentation for Mixer's Template Configuration Schema
order: 1190
layout: docs
type: markdown
---
<a name="rpcIstio.mixer.v1.config.descriptorIndex"></a>
### Index
* [ValueType](#istio.mixer.v1.config.descriptor.ValueType)
(enum)
<a name="istio.mixer.v1.config.descriptor.ValueType"></a>
### ValueType
ValueType describes the types that values in the Istio system can take. These
are used to describe the type of Attributes at run time, describe the type of
the result of evaluating an expression, and to describe the runtime type of
fields of other descriptors.
<table>
<tr>
<th>Value</th>
<th>Description</th>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.VALUETYPEUNSPECIFIED"></a>
<tr>
<td>VALUETYPEUNSPECIFIED</td>
<td>Invalid, default value.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.STRING"></a>
<tr>
<td>STRING</td>
<td>An undiscriminated variable-length string.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.INT64"></a>
<tr>
<td>INT64</td>
<td>An undiscriminated 64-bit signed integer.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.DOUBLE"></a>
<tr>
<td>DOUBLE</td>
<td>An undiscriminated 64-bit floating-point value.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.BOOL"></a>
<tr>
<td>BOOL</td>
<td>An undiscriminated boolean value.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.TIMESTAMP"></a>
<tr>
<td>TIMESTAMP</td>
<td>A point in time.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.IPADDRESS"></a>
<tr>
<td>IPADDRESS</td>
<td>An IP address.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.EMAILADDRESS"></a>
<tr>
<td>EMAILADDRESS</td>
<td>An email address.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.URI"></a>
<tr>
<td>URI</td>
<td>A URI.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.DNSNAME"></a>
<tr>
<td>DNSNAME</td>
<td>A DNS name.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.DURATION"></a>
<tr>
<td>DURATION</td>
<td>A span between two points in time.</td>
</tr>
<a name="istio.mixer.v1.config.descriptor.ValueType.STRINGMAP"></a>
<tr>
<td>STRINGMAP</td>
<td>A map string -> string, typically used by headers.</td>
</tr>
</table>
<a name="rpcQuota"></a>
## Package quota
<a name="rpcQuotaIndex"></a>
### Index
* [Template](#quota.Template)
(message)
<a name="quota.Template"></a>
### Template
<table>
<tr>
<th>Field</th>
<th>Type</th>
<th>Description</th>
</tr>
<a name="quota.Template.dimensions"></a>
<tr>
<td><code>dimensions</code></td>
<td>repeated map<string, <a href="#istio.mixer.v1.config.descriptor.ValueType">ValueType</a>></td>
<td>The unique identity of the particular quota to manipulate.</td>
</tr>
</table>
|
{
"redpajama_set_name": "RedPajamaGithub"
}
| 9,665
|
Brachiaria albicoma är en gräsart som först beskrevs av Jason Richard Swallen och García-barr., och fick sitt nu gällande namn av Fernando Omar Zuloaga och Thomas Robert Soderstrom. Brachiaria albicoma ingår i släktet Brachiaria och familjen gräs.
Artens utbredningsområde är Colombia. Inga underarter finns listade i Catalogue of Life.
Källor
Gräs
albicoma
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 4,828
|
\section{Introduction}
Weakly Interacting Massive Particles (WIMPs) are the candidates of dark matter in our universe.
Up to now any direct interaction of WIMPs with nuclei has not been observed yet. Since 2009, the
CDEX Collaboration has been working on searching for low mass WIMPs with HPGe
detector\cite{CDEX_introduction1,CDEX_introduction2}. Several results have
been produced by the first phase CDEX-1\cite{CDEX_result1,CDEX_result2,CDEX_result3}. In
the second phase CDEX-10, the HPGe detector is designed to be immersed in liquid argon which
serves as an anti-Compton detector. In this design, the LAr veto detector also serves as both
the passive shielding detector and the low temperature medium for the HPGe detector.
Liquefied noble gases as detection medium for dark matter are widely used by a number of existing and
proposed experiments\cite{XENON,Darkside}. Liquid argon has high scintillation yield ($\sim$40 photons/keV\cite{scinYield})
and is easily purified and scalable to large mass with relatively low cost, which makes it promising
for this application.
The sensitivity of the LAr veto detector is strongly dependent on the number of detected
scintillation photons. In order to develop and optimize the light collection of the
liquid-argon-based detector, a prototype detector was designed and manufactured in Tsinghua
University and then tested in Sichuan University.
In the following sections we will describe the prototype, the single-photoelectron calibration,
and we report the photoelectron yield measured with $^{137}$Cs 662 keV lines at different
collimated positions.
\section{The prototype of the liquid argon detector}
In this section, we will describe the device in detail, especially the light collection system
of the prototype detector.
\subsection{The detector system}
Fig.~\ref{fig1} shows the schematic diagram of the LAr detector system. The detector system is
composed of the following four parts: LAr Dewar, purification system, cooling system, vacuum pump.
The LAr Dewar is the main part of the prototype detector system. As shown in Fig.~\ref{fig2}, the
vacuum-insulated stainless steel Dewar contains a 7-liter active volume of LAr viewed by a
206-mm-diameter 8" ETL 9357 photomultiplier tube (PMT) with a quantum efficiency (QE)
of $\sim$17\%\cite{PMT}. The active LAr region is defined by a Teflon
cylinder which is 150 mm in diameter and 400 mm high. The PMT is held in place by two Teflon rings
below and above. To detect the 128 nm argon scintillation light, we
\begin{center}
\includegraphics[width=7cm]{LAr_system.eps}
\figcaption{\label{fig1} Schematic diagram of the prototype detector system.}
\end{center}
use the tetraphenyl butadiene (TPB) as the wavelength
shifter (WLS), with a peak emission wavelength of 420 nm\cite{lab1}. TPB is deposited by
vacuum evaporation onto the inner surface of the Teflon cylinder and the PMT window (see section 4.2 for
more discussion) with the thickness of $\sim$0.3 mg/cm$^2$. In addition, by placing a piece of stainless
steel above the PMT, the heat convection and the evaporation rate can be reduced.
\begin{center}
\includegraphics[width=5.5cm]{PMT_Teflon.eps}
\figcaption{\label{fig2} The layout of LAr container and PMT.}
\end{center}
The LAr Dewar is evacuated by the vacuum pump so that the gas impurities can be removed before running the
detector. The cooling system, composed of a cooling chamber and liquid nitrogen supplied from outside,
is employed to maintain the temperature of LAr during run time. Argon gas enters the cooling chamber and is
then liquefied by the heat exchanger (made of copper). Liquefied fluid flow back into the Dewar through a pipe.
Previous studies indicate that the contamination in LAr can seriously reduce the scintillation light intensity
through collisional de-excitation of Ar$^*_2$ excimers\cite{lab2}.
The LAr used in the experiment was commercially bought and the contamination level is lower than 10 ppm.
Impurities in the detector can build up over time via outgassing, and reduced LAr purity.
Hence the purification system is proposed in the next stage of experiment. A SAES getter will be used to
ensure high purity of LAr.
More details about the detector structure and the TPB coating process can be found in Ref.\cite{previousPaper}.
\subsection{The data acquisition system}
Each anode signal is fan out as two identical signals: one is fed into the discriminator to provide
the trigger signal and the other is sent to a 8 bit, 50 MS/s digitizer (CAEN V1721) to digitize the signal
pulse. The data are saved into hard disk for offline analysis.
The trigger threshold is set to about 1/2 of the single photoelectron mean amplitude. To
measure the DAQ dead time, the discriminator output of the
random trigger events (10 Hz) from a pulse generator served as a trigger and were digitized
as well. When the digitizer is triggered by an event, the data stream is then downloaded
and stored on a local hard disk. The time window of each event is 20 \textmu s (5 \textmu s before the
trigger and 15 \textmu s after).
\section{Single-photoelectron calibration}
For a measurement of the photoelectron yield of the prototype detector, we need a gain calibration of
the PMT. An LED calibration procedure was used to evaluate the charge response of the PMT to single
photoelectron. The PMT was immersed in liquid argon and the light pulses of $\sim$10 ns duration
at $\sim$440 nm wavelength from an LED were injected in through an optical fiber that
terminates on the PMT window. The intensity of the blue light can be adjusted by setting
the voltage of the LED.
The number of photoelectrons (p.e.) induced by each incident light pulse follows
a Poisson distribution and the probability of having $n$ photoelectrons is expressed as\cite{lab3}:
\begin{equation}
\label{eq1}
P(n)=\mu^n\cdot\frac{{\rm e}^{-\mu}}{n!}
\end{equation}
Where the mean value $\mu$ is determined by the light intensity. To evaluate the charge response
of the PMT to a single photoelectron, it is necessary to ensure
that most of the signals come from single electron events, which can be achieved by requiring
the number of signals with two photoelectrons being below, for instance, 10\% of that of single
photoelectron. According to Eq.~(\ref{eq1}), we can write:
\begin{equation}
\label{eq2}
\frac{P(2)}{P(1)}=\frac{\mu}{2}=0.1
\end{equation}
Eq.~(\ref{eq2}) can be translated into $\mu$ = 0.2 and $P(0)$ = 81.9\%. Hence, the number of 2
photoelectrons signals will be 10\% lower than that of 1 photoelectron if the light
intensity is adjusted such that the number of empty triggers is 81.9\%. Under this condition,
for any other number of photoelectrons the probability will be negligible.
The PMT response to photoelectrons can be roughly treated as a series of Gaussian functions
of different weight. Taking account of the electronic noise, an exponential functions should
be added to the PMT response function\cite{lab4}. For our purpose, however, a simplified fit
function $f(x)$ is sufficient:
\begin{eqnarray}
\label{eq3}
f(x)=a\cdot{\rm e}^{-bx}+\sum^3_{n=1}c_n\cdot G(n\mu_1,\sqrt{n}\sigma_1)
\end{eqnarray}
The exponential function with fit parameters $a$ and $b$ represent the pedestal which is
dominated by electronic noise. The Gaussian functions $G(n\mu_1,\sqrt{n}\sigma_1)$, with mean value
$n\mu_1$, standard deviation $\sqrt{n}\sigma_1$, and a normalization constant $c_n$, are the sum of
the single photoelectron response ($n$=1) and multi-photoelectron responses ($n>$1).
A function generator pulses the LED at a certain frequency and triggers the DAQ system simultaneously.
Each signal of the PMT output was collected. After subtraction of the baseline, the integral of the recorded
waveform is evaluated within a fixed 100-ns window around the arrival time of the light pulse. The charge
spectrum of the PMT was then obtained and fitted with Eq.~(\ref{eq3}), as shown in Fig.~\ref{fig3}.
The parameter $\mu_1$, with a value of 4.625 pC, is the absolute gain of the PMT for a single photoelectron. More
test on the performance of the PMT operating at both room and cryogenic temperatures can be
found in Ref.\cite{dys}.
\begin{center}
\includegraphics[width=9cm]{single_pe_spectrum.eps}
\figcaption{\label{fig3} The single photoelectron response spectrum of the ETL 9357FLB PMT at
1600V at liquid argon temperature. The dotted red curve is the exponential part of the fit function. The three
dashed curves (green, blue, pink) are the gaussian parts of the fit function and represent 1-3 p.e. response
respectively.}
\end{center}
\section{The Gamma ray response of the prototype detector}
In this section, the photoelectron yield of the prototype detector is estimated by exposing the
detector to an external $\gamma$ source collimated at different positions. The difference in
photoelectron yield between coating the PMT window or not is figured out.
The Gamma source we used is $^{137}$Cs with an intensity of 10 mCi. The Gamma lines are collimated
by a 50-mm-thick lead collimator with a 10 mm diameter hole. Three positions of collimated Gamma source along the
vertical direction are used: A, B and C, corresponding to 100 mm, 200 mm and 300 mm from the bottom
of the active volume (shown in Fig.~\ref{position}).
\begin{center}
\includegraphics[width=3.1cm]{position5.eps}
\figcaption{\label{position} Schematic diagram of the three positions of collimated Gamma source.}
\end{center}
\subsection{The photoelectron yield for 662 keV Gamma lines}
The Gamma-induced scintillation spectra for the three positions of collimated Gamma source, shown in Fig.~\ref{fig4}-\ref{fig6},
were obtained after subtraction of a background spectrum acquired with no source present. The two peaks
in each spectrum are the compton edge (left) and the full absorption peak (right), respectively. They
are consistent quite well with the spectra from a GEANT4-based Monte Carlo simulation.
\begin{center}
\includegraphics[width=8.5cm]{Cs137_spectrum_top2.eps}
\figcaption{\label{fig4} Scintillation spectrum of $^{137}$Cs collimated at position A (solid black
histogram), as well as the Monte Carlo simulation result (dashed blue histogram). The full absorption
peak is fitted with the sum of a Gaussian and a linear function(red line).}
\end{center}
\begin{center}
\includegraphics[width=8.5cm]{Cs137_spectrum_central2.eps}
\figcaption{\label{fig5} Scintillation spectrum of $^{137}$Cs collimated at position B (solid black
histogram), as well as the Monte Carlo simulation result (dashed blue histogram). The full absorption
peak is fitted with the sum of a Gaussian and a linear function(red line).}
\end{center}
\begin{center}
\includegraphics[width=8.5cm]{Cs137_spectrum_bottom2.eps}
\figcaption{\label{fig6} Scintillation spectrum of $^{137}$Cs collimated at position C (solid black
histogram), as well as the Monte Carlo simulation result (dashed blue histogram). The full absorption
peak is fitted with the sum of a Gaussian and a linear function(red line).}
\end{center}
Each full absorption peak has been fitted with the sum of a Gaussian and a linear
function. The best-fit function, superimposed on the histogram, shows the mean value of
the full-absorption. According to the absolute gain of the PMT for single photoelectron of
4.625 pC, the photoelectron yield (${\rm PY}_\gamma$) for the 662 keV line, defined for each fit
as $\mu_p/E_\gamma$, can then be given (shown in Table~\ref{tab1}).
\begin{center}
\tabcaption{ \label{tab1} photoelectron yield for each position of collimated Gamma source.}
\footnotesize
\begin{tabular*}{86mm}{ccccc}
\toprule
\shortstack{Gamma source\\position} & $\mu_p$/pC & p.e. & PY$_\gamma$/(p.e./keV) & Error/\% \\
\hline
A & 243.4 & 52.6 & 0.079 & 0.16 \\
B & 191.3 & 41.4 & 0.062 & 0.26 \\
C & 155.0 & 33.5 & 0.051 & 0.13 \\
\bottomrule
\end{tabular*}
\end{center}
We found that the photoelectron yield decreased with the increase of the distance from PMT window to
the position that the $^{137}$Cs source was collimated. It is certain that the distance from PMT
window to the location of the Gamma energy deposition significantly affected the number of
photons collected by the PMT. It is due to the depletion of the argon scintillation light
(via absorption mechanism) during transmission\cite{lab7,lab8}. Note that LAr is highly transparent
to visible light, so the depletion of visible light is negligible during transmission.
Meanwhile, because the reflectivity of Teflon is not strictly equal to 100\%, loss of photons occurs
in the process of reflection.
Some parameters of the prototype detector, such as the Surviving Fraction of the Ar$^*_2$ excimers ($S_{{\rm Ar}^*_2}$)
(defined as the ratio of the scintillation light yield in the experiment
to the yield recorded in the case of pure liquid Argon),
the absorption length for 128 nm photons in liquid argon ($L_{\rm abs}$) and the reflectivity of Teflon ($R_{\rm Teflon}$)
to UV and visible light (considered to be the same\cite{Teflon_Ref}), are affected by the LAr purity or Teflon polishing method and
therefore remain unknown in our experiment. By adjusting their values in the simulation, it is found that
the photoelectron yield is mainly determined by $S_{{\rm Ar}^*_2}$ while the ratio of the photoelectron yields
of different Gamma source positions is mainly determined by $L_{\rm abs}$ and $R_{\rm Teflon}$.
By ensuring that the simulation spectra are in good agreement with the experiment results, the parameters were
finally determined (listed in Table~\ref{tab2}).
\begin{center}
\tabcaption{ \label{tab2} The values of the parameters determined in the simulation.}
\footnotesize
\begin{tabular*}{30mm}{cc}
\toprule
parameter & value \\
\hline
$S_{{\rm Ar}^*_2}$ & 4.0\% \\
$L_{\rm abs}$ & 1.0 m \\
$R_{\rm Teflon}$ & 84\% \\
\bottomrule
\end{tabular*}
\end{center}
The value of $S_{{\rm Ar}^*_2}$ (4\%) indicates that the contamination level of LAr was $\sim10^3$ ppm of
Nitrogen equivalent concentration according to the results of Ref.~\cite{lab8}. According to the previous
studies, Teflon, or PTFE,
has more than 90\% reflectivity for UV and visible light\cite{Teflon_Ref}. The Teflon reflectivity still has enough
space for improvement in our experiment. To obtain higher photoelectron and better detector resolution, it is
necessary to improve the LAr purity and the reflectivity of Teflon. In the next stage of experiment, the
purification system is proposed and new methods of Teflon polishing will be researched.
\subsection{WLS coating on the PMT window}
\label{WLS}
The PMT in the experiments above was coated with TPB by vacuum evaporation, with the thickness of
$\sim$0.3 mg/cm$^2$. The TPB coating shifts VUV light to visible light so that it can pass through
the PMT window and be detected by the PMT. But on the other hand the transparency of the PMT window
to visible light would be significantly reduced by the TPB coating. Therefore attention should be attached to
the TPB coating on the PMT window. A contrast experiment was carried out to investigate the influence
of the PMT window coating. In the contrast experiment, the PMT window was uncoated while other aspects
remain the same. The $^{137}$Cs source was collimated at position A and the spectrum is shown
in Fig.~\ref{fig7}.
\begin{center}
\includegraphics[width=9cm]{Cs137_spectrum_top_without_TPB.eps}
\figcaption{\label{fig7} Scintillation spectrum of $^{137}$Cs collimated at position A under
the condition that the PMT window was uncoated.}
\end{center}
The mean value of the full-absorption peak is 185.7 pC, which corresponds to about 40.2 photoelectrons.
The photoelectron yield can then be given as ${\rm PY}_\gamma = 0.061$ p.e./keV. We can draw a conclusion
that the photoelectron yield increases of about 30\% with the presence of the PMT window coating. The result is
consistent with the one in Ref.~\cite{lab9}.
\section{Summary}
The China Dark Matter Experiment is designed to directly detect the dark matter with the high-purity
germanium detector. In the second phase CDEX-10, a LAr veto detector is employed to cool down and further
decrease the background of the HPGe detector. In order to develop
and optimize the light collection of the liquid-argon-based detector, a prototype was built and tested.
A gain calibration of the PMT has been done by using an LED calibration procedure. The absolute
gain of the PMT for single photoelectron is 4.625 pC under the liquid argon temperature.
We measured the photoelectron yield for 662 keV Gamma lines. The photoelectron yield turns out to be 0.079 p.e./keV
when the Gamma lines were collimated to position A. But the photoelectron yield decreased with
the increase of the distance from PMT window to the position that the $^{137}$Cs source was collimated.
It is probably due to the depletion of the argon scintillation light
(via absorption mechanism) during transmission and the photons loss in the process of reflection.
The good agreement between the experimental and simulation results has been reached and some important parameters
of the prototype detector have been determined by simulation.
A contrast experiment was carried out to investigate the influence of the PMT window coating. It shows
that PMT window coating improves the total light collection by about 30\%.
\section{Outlook}
As the veto detector of the low energy threshold HPGe detector in CDEX-10, LAr detector is expected
to have a good performance of energy threshold and veto efficiency which are mainly decided by the
photoelectron yield of the LAr detector. By reference to the LAr parameters obtained above, we can
simulate and predict the energy threshold and the veto efficiency of the CDEX-10 LAr veto detector under the
similar experimental conditions. Some preliminary results have been obtained.
To increase the photoelectron yield of the CDEX-10 LAr veto detector, the following methods will be considered:\\
1) Purifying LAr with a getter.\\
2) Optimizing the TPB coating process to improve the coating uniformity.\\
3) Researching a new polishing method of Teflon.\\
4) Replacing the PMTs with higher QE ones at reasonable cost.\\
5) Optimizing the layout of the PMTs to maximize the photocathode coverage.
\end{multicols}
\vspace{-1mm}
\centerline{\rule{80mm}{0.1pt}}
\vspace{2mm}
\begin{multicols}{2}
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 4,707
|
{"url":"https:\/\/openseespydoc.readthedocs.io\/en\/latest\/src\/loadConst.html","text":"# 7.5. loadConst command\u00b6\n\nloadConst('-time', pseudoTime)\n\nThis command is used to set the loads constant in the domain and to also set the time in the domain. When setting the loads constant, the procedure will invoke setLoadConst() on all LoadPattern objects which exist in the domain at the time the command is called.\n\n pseudoTime (float) Time domain is to be set to (optional)\n\nNote\n\nLoad Patterns added afer this command is invoked are not set to constant.","date":"2021-05-08 19:51:48","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\": 0.8056310415267944, \"perplexity\": 2065.4623463833072}, \"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-2021-21\/segments\/1620243988923.22\/warc\/CC-MAIN-20210508181551-20210508211551-00385.warc.gz\"}"}
| null | null |
{"url":"https:\/\/brilliant.org\/problems\/there-must-be-odd\/","text":"# There Must Be Odd\n\nLet $$T$$ be the number of nonempty subsets of $$\\{1,2,\\ldots,100\\}$$ containing at least one odd integer. Determine the remainder when $$T$$ is divided by $$1000$$.\n\n\u00d7\n\nProblem Loading...\n\nNote Loading...\n\nSet Loading...","date":"2017-10-22 03:14:38","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\": 0.8169299960136414, \"perplexity\": 336.0075745315491}, \"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-2017-43\/segments\/1508187825057.91\/warc\/CC-MAIN-20171022022540-20171022042540-00144.warc.gz\"}"}
| null | null |
{"url":"http:\/\/blog.invibe.net\/posts\/2012-03-22-how-to-leave-iphoto.html","text":"\u2022 iphoto.app is certainly a nice tool, but it is also\n\n1. slow, unresponsive and locks you in some ugly closed-source format.\n2. also, try to look in forums when you want to share pictures on different {computers \/ OSs \/ iphoto versions \/ places \/ users} = nightmare!\n3. on top of that, the *cloud stuff is intellectually just very corrupted...\n4. what decided me to drop it entirely was a sudden corruption of the library. It took 2 days to recover my files and re-rotate correctly all pictures...\n5. last nail in the coffin was the fact that libraries are not backward compatible : you have to upgrade to the new product.\n\u2022 so time to take back possession of your pictures!\n\n\u2022 I found some supporting arguments here: http:\/\/accretiondisc.com\/blog\/2011\/10\/26\/leaving-iphoto\/\n\n\u2022 solution: put the files in folders according to a date organization of the type 2011\/P0747444.jpg, other tools for tagging, etc... follow naturally (and you share as you share folders on dropbox or whatever)\n\n\u2022 first, let's set the default application for upcoming pictures from cameras:\n\n1. With your camera connected, open Image Capture\n2. You will see your camera under devices. Click the icon.\n3. Down at the bottom, you will see \u201cConnecting this iPhone opens\u2026\u201d with a pull-down menu under it.\n4. select \"autoimport\" (I choose \"delete after importation\")\n5. another solution is to use a feature from dropbox...\n\u2022 then, let's export the data from iPhoto: File...\/Export.... I kept the original file names but with the modifications (mainly rotations that were applied to my old camera without a gravity sensor). you should end up with lots of files in one directory. but it ... does not work as iphoto raises a completly unmeaningful error ('could not create file').\n\n\u2022 so, let's explore the library:\n\n1. it is package, and we can right-click in the finder to 'show package content'\n2. click the Masters folder to open it, and you'll see folders organized by year; inside are more folders organized by date and album, and inside them are photos in JPG format and video clips.\n\u2022 then I found https:\/\/github.com\/BMorearty\/exportiphoto and you just have to type\n\npython exportiphoto.py Pictures\/iPhoto\\ Library Pictures\/Photos\n\n\nto achieve the export\n\n\u2022 I used the following commands to post-process all\n\ncounting if everything was moved (there was some corruption in the library file with MOVies)\nfind ~\/Pictures\/iPhoto\\ Library\/Masters\/2012 -iname *MOV |wc -l\nfind ~\/Pictures\/iPhoto\\ Library\/Masters -iname *MOV |wc -l\nfind ~\/Pictures\/Photos -iname *MOV |wc -l\n\n# brute force to compare the content of the trees (brute force for the computer, you can get a coffee)\nmkdir ~\/Pictures\/Photos\/2011\nchmod -R u+rwX ~\/Pictures\/Photos\/2011\nfind ~\/Pictures\/iPhoto\\ Library\/Masters\/2011 -iname *MOV -exec cp {} ~\/Pictures\/Photos\/2011\/ \\;\nmkdir ~\/Pictures\/Photos\/2012\nfind ~\/Pictures\/iPhoto\\ Library\/Masters\/2012 -iname *MOV -exec cp {} ~\/Pictures\/Photos\/2012\/ \\;\n\n\n(for the last exampls, see http:\/\/en.wikipedia.org\/wiki\/Xargs )\n\n\u2022 at the end, the great relief is to remove all this iphoto cr*p:\n\ncd ~\/Pictures\/\nrm -rf iPhoto*\ncd ~\/Library\/Caches\nrm -rf com.apple.iPhoto\n\n\u2022 I'll be glad to help anybody with similar problem","date":"2018-04-24 10:52:29","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\": 0.26986196637153625, \"perplexity\": 9734.73114059502}, \"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-2018-17\/segments\/1524125946597.90\/warc\/CC-MAIN-20180424100356-20180424120356-00208.warc.gz\"}"}
| null | null |
Баяндай () — село в Енбекшиказахском районе Алматинской области Казахстана. Входит в состав Ташкенсазского сельского округа. Код КАТО — 194077200.
Население
В 1999 году население села составляло 805 человек (388 мужчин и 417 женщин). По данным переписи 2009 года, в селе проживали 824 человека (412 мужчин и 412 женщин).
Примечания
Населённые пункты Енбекшиказахского района
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 5,332
|
Sports News Thu, 22 Nov 2018
Essien bids Drogba farewell
Former Ghana international Michael Essien has wished Didier Drogba farewell after his illustrious career.
The pair had an illustrious career at Chelsea wining several trophies including the UEFA Champions League in 2012.
Drogba retired from football at the end of this season at the age of 40 and his former teammate had some great words for him.
"Mon frere @didierbrogba. It was a pleasure playing with you. You inspired many of us to go the extra mile and you have been a great ambassador for African Football and @chelseafc. Thank you brother and good luck in your future career.??????" Essien tweeted.
Drogba featured for Le Mans, Guingamp, Marseille, Chelsea, Shanghai Shenhua, Fenerbahce, Montreal Impact and FC Phoenix Rising Stars.
Source: Footballmadeinghana.com
Ghanaian players on the move in the ongoing transfer window
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 1,785
|
\section{Introduction}
Graphene offers unique electronic properties~\cite{katsnelson2012} that make it a promising material for future nanoelectronics and quantum information technology.
Its low nuclear spin density (natural carbon consists of 99\% $^{12}$C, with no nuclear spin), leading to nearly negligible hyperfine interaction, and its weak spin-orbit interaction~\cite{huertas2006,min2006} make graphene also highly interesting for spintronics~\cite{han2014,roche2015} and spin-based applications.
In particular, the promise of long spin-relaxation times~\cite{recher2009} makes graphene an attractive candidate for hosting spin-based quantum bits~\cite{loss1998,trauzettel2007}.
Motivated by this, graphene quantum dots (QDs) have been extensively investigated in recent years, both experimentally and theoretically.
While theoretical studies so far focused on, among others, level statistics~\cite{deraedt2008,libisch2009}, magnetic edge states~\cite{cheng2015}, and spin relaxation~\cite{struck2010}, the major experimental challenge lies in the confinement of charge carriers in graphene.
In fact, the gapless electronic band structure and Klein tunneling effects~\cite{katsnelson2006,beenakker2008} make the purely electrostatic confinement of charge carriers in graphene a difficult task.
Instead, the most viable way to confine charge carriers in graphene nowadays is based on etching nanostructures into graphene flakes~\cite{stampfer2011}.
So far, Coulomb blockade~\cite{stampfer2008,ponomarenko2008,liu2009,puddy2013}, excited states~\cite{schnez2009,volk2011,guttinger2009}, charge sensing~\cite{wang2010,guttinger2011}, and spin-filling sequences~\cite{guttinger2010} have been studied in detail in etched graphene quantum dot devices.
More recently, also charge pumps~\cite{connolly2013} and charge relaxation times of excited states~\cite{volk2013} in graphene quantum dots have been investigated experimentally.
The extracted lifetimes are a factor of 5 -- 10 larger, with an extracted lower bound of around 60 -- 100~ns compared to III/V quantum dots~\cite{hanson2007,fujisawa2002,jang2008}.
It has been suggested~\cite{volk2013} that the increased lifetimes of excited states in graphene can be attributed to weak electron-phonon interaction due to the absence of piezo-electric phonon modes in graphene.
However, a detailed theoretical estimate of electron-phonon interaction-induced charge relaxation in graphene has been missing so far.
Here we present and discuss analytical calculations of charge relaxation times in graphene quantum dots.
In particular, we compare two simple models of QDs that have recently been discussed in the literature~\cite{recher2009,schnez2008,struck2010}.
These studies focused on spin-relaxation processes and assumed a ratio of QD radius-to-phonon wavelength much smaller than the one reported in Ref.~\onlinecite{volk2013}.
These results are thus not applicable to the charge relaxation processes in large QDs.
In this paper, after briefly reviewing both models, we derive analytical expressions for charge relaxation times due to electron-phonon interaction, going beyond the previously used dipole approximation for the electron-phonon coupling.
We find that the quantitative results significantly differ between the two models, which can be traced back to the different way in which confinement is handled.
In particular, we find that, in a model where electrons are confined electrostatically, charge relaxation times are much larger than in other models, suggesting electrostatic confinement as a preferred way to define quantum dots in graphene.
\section{Graphene Quantum Dot Models}
We start by briefly reviewing the two models used for the calculation of charge relaxation times.
In both models the quantum dot is taken to be circular, while the method of confinement differs between the two (see Figures~1a and 1b).
\begin{figure}[t]
\includegraphics[width=1.0\linewidth]{Figure1}
\caption{(a,b) Schematic illustrations of the two different graphene quantum dot models describing a circular dot with radius $R$:
(a) Infinite mass boundary model and (b) Electrostatic confinement model.
(c,d) The corresponding total potential for both models. The quantities appearing in the labels are defined in the text.
\label{fig:potential}
}
\end{figure}
In the first model, which we will refer to as the ``infinite mass boundary'' (IM) model, the electron behaves like a free electron inside the dot, while being confined by an infinitely large mass-like potential~\cite{schnez2008} at the boundary that might be realized by etching the dot.
The second model, which will be referred to as the ``electrostatic confinement'' (EC) model, uses a finite electrostatic potential to achieve confinement~\cite{recher2009}.
To prevent Klein tunneling, we account for a finite mass term that might be induced by an underlying substrate.
Within the nearest-neighbor tight-binding scheme in the low-energy approximation, both models can be described by a similar Hamiltonian in the valley isotropic form~\cite{schnez2008,recher2009}:
\begin{equation}
\mathbf{H}^{\tau_{\mathrm{v}}} = v_{\mathrm{F}} \vec{p} \cdot \vec{\sigma} + \tau_{\mathrm{v}} V(r) \sigma_z + U(r).
\label{eq:Hamiltonian}
\end{equation}
Here, $v_{\mathrm{F}} \approx 10^6$~m/s is the Fermi velocity, $\vec{p}$ is the momentum operator, $\vec{\sigma} = (\sigma_x,\sigma_y)^{\mathrm{T}}$ is a two-dimensional vector of Pauli matrices acting in sublattice space, and $\tau_{\mathrm{v}} = \pm 1$ distinguishes between the $K$ and $K'$ valleys, respectively.
$V(r)$ specifies the mass term and only depends on the radial coordinate for a circularly symmetric quantum dot.
For the IM~model, $V(r)$ is given by $V_{\mathrm{IM}}(r) = 0$ for $r < R$ and $V_{\mathrm{IM}}(r) = \infty$ for $r > R$, while we use $V_{\mathrm{EC}}(r) = \Delta_V = \mathrm{const}$ for $V(r)$ in the EC~model.
Finally, $U(r)$ denotes the electrostatic confinement potential, which is zero in the IM~model.
For the EC~model, we use $U_{\mathrm{EC}}(r) = 0$ for $r < R$ and $U_{\mathrm{EC}}(r) = U_0 = \mathrm{const}$ for $r > R$.
The different potential landscapes are illustrated in Figures~1c and 1d.
The Hamiltonian given in Eq.~(\ref{eq:Hamiltonian}) commutes with the operator of total angular momentum $\mathbf{J}_z = L_z + (\hbar/2)\sigma_z$, where $L_z = (\vec{r} \times \vec{p})_z$ denotes the $z$~component of the orbital angular momentum operator and the second term corresponds to the contribution of the pseudospin (or sublattice spin) in graphene.
The eigenfunctions of the Hamiltonian can thus be chosen to also be eigenfunctions of $\mathbf{J}_z$ and hence we make the ansatz
\begin{equation}
\psi(r,\varphi) = \left(
\begin{array}{c}
\psi_{m}^{(1)}(r,\varphi) \\
\psi_{m}^{(2)}(r,\varphi)
\end{array}
\right) = \left(
\begin{array}{c}
\chi_{m}^{(1)}(r) \\
\chi_{m}^{(2)}(r) \mathrm{e}^{i \varphi}
\end{array}
\right) \mathrm{e}^{i m \varphi},
\end{equation}
with $m \in \mathbb{Z}$ being the angular momentum quantum number.
The Dirac equation $(\mathbf{H}^{\tau_{\mathrm{v}}}-E)\psi(r, \varphi) = 0$ has been solved for both models in previous works~\cite{schnez2008,recher2009}.
In both cases, the above ansatz results in a simple, Bessel-type differential equation for $\chi_{m}^{(1,2)}(r)$.
The boundary conditions, however, are different in both models.
In the IM~model, the wave functions are restricted by the condition that the radial component of the probability current is zero, so that the electron remains within the dot at all times.
This results in the condition $\psi_{m}^{\tau_{\mathrm{v}},(2)}(R,\varphi) / \psi_{m}^{\tau_{\mathrm{v}},(1)}(R,\varphi) = \tau_{\mathrm{v}} i\mathrm{e}^{i \varphi}$~\cite{berry1987}.
One then finds that the possible wave functions in the IM~model are given by
\begin{equation}
\psi_{m,n}^{\tau_{\mathrm{v}}}(r,\varphi) =
\begin{cases}
\alpha \left(
\begin{array}{c}
J_m(k r) \\
i J_{m+1}(k r) \mathrm{e}^{i\varphi}
\end{array}
\right) \mathrm{e}^{i m \varphi}, & r < R \\
0, & r \geq R,
\end{cases}
\end{equation}
where $k = E / (\hbar v_{\mathrm{F}})$, $J_m$ denotes the Bessel function of the first kind of order $m$, and $\alpha > 0$ is a normalization constant.
The boundary condition then translates to the transcendental equation
\begin{equation}
J_{m+1}(k R) = \tau_{\mathrm{v}} J_m(k R),
\label{eq:IM-boundary}
\end{equation}
which has solutions for discrete values of $k$ that we will label by an additional, positive integer quantum number $n$, sorted by increasing energy.
Due to the oscillatory nature of Bessel functions, this equation has an infinite number of solutions.
Note also that the wave function is discontinuous at $r = R$, as the mass term features an infinitely high step there, so that the first derivative of the wave function has a $\delta$-function-like behavior at $r = R$.
This leads to a discontinuity in the wave function itself, in contrast to the behavior of the wave function for a non-relativistic particle, which is continuous even for potentials containing a $\delta$-like singularity since the equation of motion features the second spatial derivative of the wave function.
This behavior is illustrated in Figure~2a, where one example of such a wave function is plotted.
\begin{figure}[t]
\includegraphics[width=1.0\linewidth]{Figure2}
\caption{
Examples of typical wave functions in the two models.
(a) Infinite mass boundary model.
(b) Electrostatic confinement model.
The quantum numbers belonging to the shown wave functions are $\tau_{\mathrm{v}} = +1$ ($K$~valley), $m = 3$ and $n = 3$ for both cases.
In case of the EC~model, the parameters were set to $\Delta_V = U_0 = 10\hbar v_{\mathrm{F}} / R \approx 120$~meV, following Ref.~\onlinecite{struck2010}.
\label{fig:wave-function}
}
\end{figure}
By contrast, all potentials in the EC~model only possess finite steps and hence the allowed wave functions must be continuous at $r = R$.
The general solution of the Dirac equation in this case is given by
\begin{widetext}
\begin{equation}
\psi_{m,n}^{\tau_{\mathrm{v}}}(r,\varphi) =
\begin{cases}
\alpha \left(
\begin{array}{c}
J_m\left(\sqrt{k_<^2} r\right) \\
i \sqrt{\frac{\kappa_< - \lambda_{\tau_{\mathrm{v}}}}{\kappa_< + \lambda_{\tau_{\mathrm{v}}}}} J_{m+1}\left(\sqrt{k_<^2} r\right) \mathrm{e}^{i \varphi}
\end{array}
\right) \mathrm{e}^{i m \varphi}, & r < R \\
\alpha \eta \left(
\begin{array}{c}
K_m\left(\sqrt{-k_>^2} r\right) \\
i \sqrt{\frac{\lambda_{\tau_{\mathrm{v}}} - \kappa_>}{\lambda_{\tau_{\mathrm{v}}} + \kappa_>}} K_{m+1}\left(\sqrt{-k_>^2} r\right) \mathrm{e}^{i \varphi}
\end{array}
\right) \mathrm{e}^{i m \varphi}, & r \geq R.
\end{cases}
\end{equation}
\end{widetext}
Here we introduced $\lambda_{\tau_{\mathrm{v}}} = \tau_{\mathrm{v}}\Delta_V / (\hbar v_{\mathrm{F}})$, $\kappa_< = E / (\hbar v_{\mathrm{F}})$, $\kappa_> = (E-U_0) / (\hbar v_{\mathrm{F}})$, and $k_{<,>}^2 = \kappa_{<,>}^2 - \lambda_{\tau_{\mathrm{v}}}^2$.
The function $K_m$ is the modified Bessel function of the first kind of order $m$, while the constant $\alpha > 0$ is again a normalization constant.
Finally, the constant $\eta$ is needed to ensure that both components of the spinor are continuous at $r = R$.
Explicitly, it is given by $\eta = J_m(\sqrt{k_<^2} R)/K_m(\sqrt{-k_>^2} R)$.
By equating the lower components of the wave functions at $r = R$, one obtains an equation that determines the possible energy values:
\begin{equation}
\begin{split}
& \sqrt{\frac{\kappa_< - \lambda_{\tau_{\mathrm{v}}}}{\kappa_< + \lambda_{\tau_{\mathrm{v}}}}} J_{m+1}\left(\sqrt{k_<^2} R\right) K_m\left(\sqrt{-k_>^2} R\right) \\
= & \sqrt{\frac{\lambda_{\tau_{\mathrm{v}}} - \kappa_>}{\lambda_{\tau_{\mathrm{v}}} + \kappa_>}} J_m\left(\sqrt{k_<^2} R\right) K_{m+1}\left(\sqrt{-k_>^2} R\right).
\end{split}
\end{equation}
In contrast to the characteristic equation for the IM~model [Eq.~(\ref{eq:IM-boundary})], the above equation only has a finite number of roots, as the modified Bessel functions are exponentially damped.
An example of a typical wave function is shown in Figure~2b.
Compared to the wave functions in the IM~model, the EC~wave functions feature a leaking tail due to the finite confinement potential.
Having reviewed the two models and the resulting wave functions, we can now move on to calculate charge relaxation times due to electron-phonon interaction.
\section{Electron-Phonon Interaction and Charge Relaxation Times}
Electron-phonon interaction plays an important role in many graphene experiments.
For example, electron-phonon scattering has been shown to limit the charge-carrier mobility in graphene samples~\cite{hwang2008} and has also been argued to be the main mechanism for charge relaxation in QDs~\cite{volk2013}.
To gain an estimate of the charge relaxation times in graphene quantum dots, we thus focus on relaxation processes due to electron-phonon interaction.
Theoretically, electron-phonon interaction in graphene has already been studied in the literature~\cite{suzuura2002,ando2005,manes2007,mariani2008} in depth.
We use a perturbative approach and look only at the dominant, lowest-order, i.e.. one-phonon, processes.
Since we want to provide an estimate for charge relaxation times observed in an experiment at low temperature~\cite{volk2013}, we can neglect phonon absorption.
Thus, the only process we consider here is the decay of an excited QD state to a lower state by emission of one phonon.
Furthermore, the energy differences between the electronic states observed in experiment are rather low ($\Delta_E \lesssim 10$~meV~\cite{volk2013}) and thus we only take into account low-energy, acoustic phonons.
Moreover, since out-of-plane lattice vibrations couple only via two-phonon processes~\cite{manes2007}, we can further restrict ourselves to in-plane phonon modes.
The latter couple to the electronic system via two mechanisms: (i) a deformation potential (DEF) caused by a change of the area of the unit cell, and (ii) a change of the carbon-atom bond length (BLC) resulting in a modified hopping parameter.
Since we are interested in one-phonon processes involving acoustic, low-energy (i.e., low-momentum) phonons, inter-valley scattering is not possible and we can thus focus on only one valley, choosing the one at $K$.
Following the approach used in Ref.~\onlinecite{mariani2008}, we find that to lowest order in the lattice deformation field, the electron-phonon interaction Hamiltonian can be written as
\begin{equation}
\begin{split}
H_{\mathrm{el-ph}} = \sum_{\substack{\mathrm{a} = \mathrm{DEF},\\\mathrm{BLC}}} g_{\mathrm{a}} \sum_{\substack{m',n'\\ m,n}} & \sum_{\vec{q},\mu} f_{\mu}(q) (b_{\vec{q},\mu} + b_{-\vec{q},\mu}^{\dagger}) \\
& \times c_{m',n'}^{\dagger} c_{m,n} \mathcal{M}_{m',n',m,n}^{\mathrm{a},\mu}(\vec{q}).
\end{split}
\end{equation}
Here, $g_{\mathrm{a}}$ is the electron-phonon coupling constant related to mechanism $\mathrm{a} = \mathrm{DEF,BLC}$.
Numerically, we use $g_{\mathrm{DEF}} = 30$~eV and $g_{\mathrm{BLC}} = 1.4$~eV~\cite{suzuura2002,struck2010}.
We also defined $f_{\mu}(q) = \sqrt{\hbar / (A \rho \omega_{\mu}(q))} q$, wherein $A$ is the area of the graphene sheet, $\rho$ is its mass density, and $\omega_{\mu}(q)$ is the frequency of an in-plane phonon with wave vector $\vec{q} = q (\cos \varphi_{\vec{q}},\sin \varphi_{\vec{q}})^{\mathrm{T}}$ and longitudinal or transverse polarization $\mu = \mathrm{l}, \mathrm{t}$.
The operators $b_{\vec{q},\mu}$($b_{\vec{q},\mu}^{\dagger}$) and $c_{m,n}$ ($c_{m,n}^{\dagger}$) are the phononic and electronic annihilation (creation) operators, which destroy (create) a phonon or electron with the indicated quantum numbers, respectively.
Finally, $\mathcal{M}_{m',n',m,n}^{\mathrm{a},\mu}(\vec{q})$ denotes the matrix element for the involved states.
It can be written in the form
\begin{equation}
\begin{split}
\mathcal{M}_{m',n',m,n}^{\mathrm{a},\mu}(\vec{q}) = \int \mathrm{d}^2 r \, & \psi_{m',n'}^*(\vec{r}) \left(
\begin{array}{cc}
\alpha_{\mathrm{a},\mu} & \beta_{\mathrm{a},\mu} \\
\beta_{\mathrm{a},\mu}^* & \alpha_{\mathrm{a},\mu}
\end{array}
\right) \psi_{m,n}(\vec{r}) \\
& \times\mathrm{e}^{i \vec{q} \cdot\ \vec{r}},
\end{split}
\label{eq:matrix-element}
\end{equation}
where the constants $\alpha_{\mathrm{a},\mu}$ and $\beta_{\mathrm{a},\mu}$ are given by $\alpha_{\mathrm{a} = \mathrm{DEF}, \mu = \mathrm{l}} = -1$, $\beta_{\mathrm{BLC},\mathrm{l}} = \exp(2 i \varphi_{\vec{q}})$, $\beta_{\mathrm{BLC},\mathrm{t}} = i \exp(2 i \varphi_{\vec{q}})$ and are zero for all other cases~\cite{suzuura2002}.
To facilitate the evaluation of these matrix elements, the exponential in Eq.~(\ref{eq:matrix-element}) has so far been approximated by the dipole approximation in the recent literature~\cite{struck2010}.
This is possible if the wavelength of the phonon is much larger than the QD radius.
In the experiment which we want to compare our results to, however, the QD radius is given by $R = 55$~nm\cite{volk2013}, which is larger than the typical wavelength of the involved phonons ($\lambda \approx 5-10$~nm, for a phonon of energy $\hbar \omega = \Delta_E \lesssim 10$~meV).
Thus, the dipole approximation for $\exp(i \vec{q} \cdot \vec{r})$ is no longer applicable.
Instead, one can make use of the Jacobi-Anger identity $\exp( i \vec{q} \cdot \vec{r}) = \exp(i q r \cos(\varphi-\varphi_{\vec{q}})) = \sum_{l = -\infty}^{+\infty} i^l J_l(q r) \mathrm{e}^{i l (\varphi-\varphi_{\vec{q}})}$ to expand the exponential of Eq.~(\ref{eq:matrix-element}) into a Fourier series.
After evaluating the angular part of the integral, the matrix elements reduce to:
\begin{widetext}
\begin{equation}
\mathcal{M}_{m',n',m,n}^{\mathrm{a},\mu}(\vec{q}) = -2 \pi i^{m'-m} \mathrm{e}^{-i (m'-m) \varphi_{\vec{q}}} \times
\begin{cases}
M_0^{(1,1)}(q) + M_0^{(2,2)}(q), & \mathrm{a} = \mathrm{DEF}, \mu = \mathrm{l} \\
0, & \mathrm{a} = \mathrm{DEF}, \mu = \mathrm{t} \\
i M_{-1}^{(1,2)}(q) \mathrm{e}^{+3 i \varphi_{\vec{q}}} - i M_{+1}^{(2,1)}(q)\mathrm{e}^{-3 i \varphi_{\vec{q}}}, & \mathrm{a} = \mathrm{BLC,\mu = \mathrm{l}} \\
-M_{-1}^{(1,2)}(q) \mathrm{e}^{+3 i \varphi_{\vec{q}}} - M_{+1}^{(2,1)}(q)\mathrm{e}^{-3 i \varphi_{\vec{q}}}, & \mathrm{a} = \mathrm{BLC},\mu = \mathrm{t},
\end{cases}
\end{equation}
\end{widetext}
where we introduced $M_{\zeta}^{(i,j)}(q) = \int_0^{\infty} \mathrm{d} r \, r \chi_{m',n'}^{(i),*}(r) \chi_{m,n}^{(j)}(r) J_{m'-m+\zeta}(q r)$ as a short-hand notation that hides the dependence of the matrix element on $m$, $n$, $m'$, and $n'$, but avoids a cluttering of the final expressions due to too many indices.
Finally, we can make use of Fermi's golden rule,
\begin{widetext}
\begin{equation}
\tau^{-1}((m,n) \to (m',n')) = \frac{2 \pi}{\hbar} \sum_{\vec{q},\mu} | \langle m',n';\mu,\vec{q}| H_{\mathrm{el-ph,ac}} |m,n\rangle |^{2} \times \delta(E_{m,n} - E_{m'n'} - \hbar \omega_{\mu}(q)),
\label{eq:relax-time}
\end{equation}
\end{widetext}
to get an analytic expression for the charge relaxation time $\tau$.
For low-energy acoustic phonons, we use a linear approximation of the phonon dispersion, $\omega_{\mu}(q) = v_{\mu} q$, where $v_{\mathrm{l}} \approx 19.5 \times 10^3$~m/s and $v_{\mathrm{t}} \approx 12.2 \times 10^3$~m/s is the speed of sound for longitudinal and transverse phonons, respectively \cite{struck2010}.
After approximating the sum over $\vec{q}$ in Eq.~(\ref{eq:relax-time}) with an integral $\sum_{\vec{q}} \to \frac{A}{(2 \pi)^2} \int \mathrm{d}^2 q$, any interference terms in the squared matrix elements disappear as they contain oscillating exponentials of $\varphi_{\vec{q}}$.
The final result for the relaxation times reduces to a sum of three contributions $\tau^{-1}((m,n) \to (m',n')) = \tau^{-1}|_{\mathrm{DEF,l}} + \tau^{-1}|_{\mathrm{BLC,l}} + \tau^{-1}|_{\mathrm{BLC,t}}$, given by:
\begin{widetext}
\begin{equation}
\tau^{-1}|_{\mathrm{a},\mu} =
\begin{cases}
\theta(\Delta_E) \frac{4 \pi^2 \Delta_E^2}{\hbar^3 v_{\mathrm{l}}^4 \rho} \left|M_0^{(1,1)}\left(\frac{\Delta_E}{\hbar v_{\mathrm{l}}}\right) + M_0^{(2,2)}\left(\frac{\Delta_E}{\hbar v_{\mathrm{l}}}\right)\right|^2, & \mathrm{a} = \mathrm{DEF}, \mu = \mathrm{l} \\
\theta(\Delta_E) \frac{4 \pi^2 \Delta_E^2}{\hbar^3 v_{\mathrm{l}}^4 \rho} \left[\left|M_{-1}^{(1,2)}\left(\frac{\Delta_E}{\hbar v_{\mathrm{l}}}\right)\right|^2 + \left|M_{+1}^{(2,1)}\left(\frac{\Delta_E}{\hbar v_{\mathrm{l}}}\right)\right|^2\right], & \mathrm{a} = \mathrm{BLC},\mu = \mathrm{l} \\
\theta(\Delta_E) \frac{4 \pi^2 \Delta_E^2}{\hbar^3 v_{\mathrm{t}}^4 \rho} \left[\left|M_{-1}^{(1,2)}\left(\frac{\Delta_E}{\hbar v_{\mathrm{t}}}\right)\right|^2 + \left|M_{+1}^{(2,1)}\left(\frac{\Delta_E}{\hbar v_{\mathrm{t}}}\right)\right|^2\right], & \mathrm{a} = \mathrm{BLC},\mu = \mathrm{t},
\end{cases}
\end{equation}
\end{widetext}
where $\Delta_E = E_{m,n} - E_{m'n'}$ denotes the energy difference between the initial and final electron state and $\theta(\Delta_E)$ is the step function
\section{Numerical Results and Discussion}
\begin{figure*}[!]
\includegraphics[width=1.0\linewidth]{Figure3}
\caption{
Numerical results for the charge relaxation times for the finite sets of states described in the text.
(a) Infinite mass boundary model.
(b) Electrostatic confinement model.
(c) Electrostatic confinement model, with the wave function artificially set to zero outside the quantum dot (green histogram).
For comparison, the distributions (histograms) shown in panels (a) and (b) are also shown.
\label{fig:Relax-Histo}
}
\end{figure*}
To obtain an estimate of the order of magnitude of charge relaxation times in graphene quantum dots, we apply the expressions obtained in the previous section to the electronic states of the two quantum dot models.
For both models, we use a dot radius of $R = 55$~nm and, in the case of the electrostatic confinement model, we follow Ref.~\onlinecite{struck2010} and use $\Delta_V = U_0 = 10\hbar v_{\mathrm{F}} / R \approx 120$~meV.
To obtain numerical results for the relaxation times, we need to limit the number of states to a finite number.
For the EC~model, the number of states is naturally limited by the finite size of the well and hence we include all of them in the numerical analysis.
For the IM~model, which features an infinitely high mass barrier, no such natural limitation exists and the number of states has to be chosen manually.
To allow for a comparison with experiment~\cite{volk2013}, we focus on low-lying excited states in the conduction band.
Furthermore, in order to facilitate the comparison between the two models, we choose an energy cutoff $E^{\mathrm{IM}}_{\mathrm{max}}$ for the states in the IM~model.
This energy cutoff can be chosen to either match the maximum possible energy in the EC~model or to result in roughly the same number of included states as in the EC~model.
However, the former possibility, i.e., $E^{\mathrm{IM}}_{\mathrm{max}} = \Delta_V \approx 120$~meV, results in an unfeasibly small number of states due to a comparatively large level spacing in an infinitely deep well.
We therefore limit the energy of the states such that the number of states in the conduction band is roughly the same in both models, which results in a chosen maximum energy of $E^{\mathrm{IM}}_{\mathrm{max}} = 250$~meV.
The highest possible angular momentum quantum numbers turn out to be $m_{\mathrm{max,IM}} = 18$ and $m_{\mathrm{max,EC}} = 13$.
For each model, we compute the relaxation times between all possible pairs of states in the conduction band and collect the results in a histogram shown in Figures~3a and 3b.
In both models, the calculated times range over several orders of magnitude.
However, the range of $\tau$ values that occur most often differs significantly between the two models.
In the IM~model, one mostly finds relaxation times between 1~\textmu s and 0.1~ns, whereas the EC~model results in a wide spread of times.
In the latter, the majority of times fall into the interval between 10~ms and 10~ns, thus predicting the existence of long-lived excited states.
This behavior has its origin in the finite probability to find the electron outside the dot, since it can be traced back to the presence of a leaking tail of the wave function.
To demonstrate this, we set the leaking tail of the wave function in the EC~model artificially to zero, renormalize the wave function, and recalculate the relaxation times.
In Figure~3c, we show the resulting distribution of times as a green histogram.
When compared to the two distributions of the IM and EC~model (half-transparent blue and red histograms, respectively), we find that the resulting distribution is much closer to the one found in the IM~model, where no leaking tail is present.
Further analysis of the numerical results shows that the contribution of the region $r \geq R$ is roughly of the same order of magnitude as the one of the region $r \leq R$.
However, the signs of the two contributions are opposite, and hence the average electron-phonon matrix element in the EC~model is rather small.
This leads to the (on average) small decay widths and correspondingly large life times in the EC~model.
Unfortunately, the physical origin of this effect is not easy to pinpoint, especially since the sign and magnitude of the contributions of the two different regions to the matrix element strongly depend on the concrete quantum numbers of the involved states.
While the total relaxation times differ significantly between the two models, the relative contributions of the two different electron-phonon coupling mechanisms are roughly the same in both cases.
If one calculates the individual contributions of both the deformation potential and the bond length change mechanism to the decay width, one finds that the former is on average a factor of $10^6$ larger than the latter.
This factor is three orders of magnitudes larger than what would be expected from the ratio of the squared coupling constants alone and indicates that the form of the electronic wave functions and their interplay with the phonon play a significant role.
In other words, the change of the on-site energies (which corresponds to the deformation potential mechanism) is much more important than the change of the hopping parameters for the localized states of the quantum dot.
\begin{figure*}[!]
\includegraphics[width=1.0\linewidth]{Figure4}
\caption{
Level spacing versus charge relaxation times in the two quantum dot models.
(a) Infinite mass boundary model.
(b) Electrostatic confinement model.
The dashed lines serve as a guide to the eye for the asymptotic $\tau \sim \Delta_E^2$ behavior.
(c) and (d) Zoom-in into the region $\Delta_E < 10$~meV of panels (a) and (b), respectively.
The dashed line represents the experimentally observed lower bound as discussed in Ref.~\onlinecite{volk2013}.
\label{fig:Delta-vs-Relax}
}
\end{figure*}
For further analysis and comparison to experiment, we plot the relaxation times against the level spacing as shown in Figures~4a and 4b.
Here, one can observe the general trend that the calculated relaxation times as a whole grow monotonically with the level spacing.
For large values of the level spacing $\Delta_E$, one finds a lower bound obeying a $\tau \sim \Delta_E^2$ scaling law (compare dashed lines in Figures~4a and 4b).
This behavior can be directly extracted from the analytical calculation, since for large $\Delta_E$, the integrals $M_{\zeta}^{(i,j)}(\Delta_E / \hbar v_{\mathrm{F}})$ scale like $\Delta_E^{-2}$ and hence the decay rate $\tau^{-1}$ scales like $\Delta_E^2 \times [\Delta_E^{-2}]^2 = \Delta_E^{-2}$.
Also notable is the fact that even for small level spacings ($ \Delta_E < 10$~meV), i.e., in the experimentally relevant region~\cite{volk2013}, the two models lead to rather different predictions as seen more clearly in the zoom-in shown in Figures~4c and 4d.
In this region, only the EC~model consistently features excited states with life times above the lower bound observed in experiment (black, dashed line in Figures~4c and 4d).
At this point it is important to note that the experimental data from Ref.~\onlinecite{volk2013} were obtained on etched graphene nanostructures.
The termination of the nanostructure corresponds to the appearance of an infinitely high mass term and thus, out of the two investigated models, the IM~model is expected to correspond best to the experimental situation.
However, the fact that the average relaxation times in the IM~model are much lower than the lower bound observed in the cited experimental work seems to suggest that this simple model does not capture all of the relevant physics, even though it should be noted that it does feature transitions with relaxation times above the lower bound in the region $\Delta_E < 2$~meV, which partly contains the excited states seen in experiment (with energies in the range 1.7 -- 2.5~meV~\cite{volk2013}).
It is also possible that near the edges of the graphene nanostructure the sublattice symmetry is broken (e.g., due to the badly controlled etching procedure).
This would lead to a finite mass term toward the boundary of the dot, which would in turn lead to a leaking tail of the wave function.
As observed in the EC~model, this leaking tail could then be responsible for the large increase of the relaxation times.
Independently of this, the larger relaxation times present in the EC~model suggest that the presence of a leaking tail is beneficial for the longevity of the excited states of the quantum dot and hence electrostatic confinement might be the better way to define quantum dots with larger charge relaxation times.
In summary, we compared two models of graphene quantum dots and studied charge relaxation times due to emission of acoustic phonons.
We found that the two models lead to significantly different predictions that could be traced back to the different boundary conditions and the corresponding existence of a leaking tail of the wave functions.
Furthermore, we demonstrated that the relaxation times in the electrostatic confinement model are much larger than the times found in the infinite mass boundary model, exceeding the experimentally established lower boundary of 60 -- 100~ns.
Thus, for the realization of graphene quantum dots with long-living excited states, the confinement with electrostatic potentials as featured in the model, is to be preferred.
\section*{Acknowledgments}
The authors would like to thank G. Burkard, F. Haupt, and S.V. Rotkin for useful discussions.
Financial support by the Deutsche Forschungsgesellschaft (Grant No. SPP-1459) and the European Research Council (Grant No. GA-Nr. 280140) is gratefully acknowledged.
S. R. acknowledges financial support by the National Research Fund (FNR) Luxembourg.
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 8,147
|
Collines (frz. Hügel) sind die subnationalen Verwaltungseinheiten auf der untersten Verwaltungsebene in Burundi. Das Staatsgebiet Burundis ist in Provinzen, diese in Kommunen und diese wiederum in Collines untergliedert.
Siehe auch
Liste subnationaler Verwaltungseinheiten
Öffentliche Verwaltung (Burundi)
Geographie (Burundi)
|
{
"redpajama_set_name": "RedPajamaWikipedia"
}
| 9,168
|
\section{Introduction}
\begin{figure}[t]
\centering
\includegraphics[width=0.55\textwidth]{./introduction} \caption{A unified detector is expected to detect all categories (e.g., both the classes in {\color{red} COCO} and {\color{green} WIDER FACE}) while only separately labeled training sets are provided (``faces" are not labeled in COCO; simultaneously, instances of the 80 COCO classes are not annotated in FACE). Typically, two simple ways can achieve this goal: relabeling or plain joint training. However, relabeling is extremely costly (which needs to browse each image to discover instances belonging to the missing categories) and plain joint training leads to severe biases and a poor detector (which fails to detect ``face" in COCO evaluation set and ``person" in FACE evaluation set). Best seen in color.}
\label{fig:introduction}
\end{figure}
Object detection is a fundamental and essential computer vision problem in various application areas, such as self-driving~\cite{WEI2020107195}, medicine~\cite{wang2020focalmix} and security~\cite{AKCAY2022108245}, etc. Owing to the development of deep neural networks, more and more powerful detectors are proposed~\cite{ren2015faster,lin2017focal,QIN2020107404}.
A standard object detector is typically trained on a pre-prepared fully annotated training dataset with fixed categories. However, in many real-world applications, we cannot know all possible categories of interest beforehand, so it is often desired to add new object classes progressively. For instance, we have a detector in use, which was trained on an original dataset (e.g., COCO~\cite{lin2014microsoft} with 80 classes). Now, it is required to detect a new class ``human face" (abbr.: face) simultaneously while only some new training data labeled with ``face'' is additionally provided (e.g., WIDER FACE~\cite{yang2016wider}). As shown in Fig.~\ref{fig:introduction}, many faces in the original training data are not labeled, similarly, the instances belonging to the original categories (e.g., ``person") are not annotated in the new dataset.
As new categories are incrementally added, training data is collected and labeled specifically for them. {\color{highlight}Commonly, the models are expected to be efficient, namely smaller, faster and better~\cite{redmon2016you,yang2020distilling,yang2020factorizable,jing2020dynamic,jing2021amalgamating}}. An important question is how to train a unified detector that can manage all categories based on the limited datasets, because a unified detector introduces merely negligible extra computation and memory overheads with respect to the number of categories of interest. As shown in Fig.~\ref{fig:introduction}, there are two straightforward ways to achieve a unified detector. The one is relabeling, which transforms the problem into a standard object detection task. However, it's prohibitively expensive to relabel all missing instances in the original and new datasets. For massive amounts of data and ever-increasing categories of interest, browsing images from all datasets to discover instances belonging to the missing categories is impractical. The other one is plain joint training. While, in one dataset (e.g., COCO), samples related to unlabeled instances (e.g., ``face") will be regarded as negative (background) samples, resulting in conflicts and false gradients during training. Due to overfitting the false negative samples, it is difficult for the detector trained in the plain way to discover categories not labeled in corresponding training data. For example, {\color{highlight} as shown in Fig.~\ref{fig:introduction}}, the model trained on the combined training set of COCO and WIDER FACE cannot detect ``face" in the COCO evaluation set or ``person" in the WIDER FACE evaluation set.
Can we obtain a unified detector that can detect all categories well while avoiding relabeling manually and alleviating conflicts during training? We propose a general solution for this purpose in the work. (i) A conflict-free loss is carefully designed to avoid two possible conflicts during joint learning (details in Sec.~\ref{sec:cf_loss}): (a) Samples assigned as negative in one dataset may be actually positive instances belonging to classes from the other datasets. (b) Samples assigned as positive may be more suitable to be labeled as categories of the other datasets. Experimental results demonstrate that conflict-free loss can lead to an acceptable unified detector. (ii) To further improve performance, we propose a retraining phase (details in Sec.~\ref{sec:gt_mining} and Sec.~\ref{sec:retraining}). (a) We attempt to mine high-quality unlabeled ground-truth with the trained detector. The Monte Carlo Dropout is employed to obtain localization confidence, which is combined with the classification confidence to unearth pseudo annotations with more accurate bounding boxes. (b) Regarding the retraining strategies with the mined pseudo annotations, we realize that although a lot of unlabeled instances are mined, there still hide many latent ground-truth. Therefore, the conflict-free idea cannot be removed in the retraining stage. However, simply copying the training strategy from the first phase results in insufficient negative information, as only positive samples are added by pseudo annotations during retraining. Consequently, we propose an overlap-based negative samples weighting strategy to utilize negative samples modestly in the retraining process. In the end, we achieve a strong and unified detector.
{\color{highlight} Object detection, semi-supervised object detection, open-set object detection, and class incremental object detection are all related to the study of category-extended object detector. A significant amount of research has been conducted in these areas. Sec.~\ref{sec:related_work} provides an abridged overview of the relevant background.}
To sum up, the main contributions of this paper are listed as follows.
\textbf{(i)} We present a general solution for training the unified detector with only the original datasets and incrementally labeled new datasets, which is urgently required in realistic applications.
\textbf{(ii)} A conflict-free loss is proposed to avoid possible label ambiguity. Localization confidence is designed for mining more accurate pseudo annotations. An overlap-weighted loss is employed to deal with uncertain negative samples for retraining a stronger detector with pseudo annotations.
\textbf{(iii)} The entire pipeline does not introduce any additional manual labeling or modification of network structures and does not alter inference speed.
\textbf{(iv)} Extensive experiments are conducted to demonstrate the effectiveness of our proposed solution as compared with state-of-the-art approaches.
\section{Related Work}
\label{sec:related_work}
{\color{highlight} \paragraph{Object Detection} Object detection has made significant progress in the past decade. Detectors (R-CNN serials~\cite{ren2015faster,girshick2014rich,girshick2015fast}, RetinaNet~\cite{lin2017focal}, FCOS~\cite{tian2019fcos}, etc.) mainly focus on efficiency or performance enhancement based on sufficient and complete training data. Recently, object detection in more complex and realistic scenarios has also attracted much attention~\cite{wang2018geometry,yang2021training,gao2022discrepant}}.
\paragraph{Semi-Supervised Object Detection} SSOD aims to learn detectors based on a few labeled images and large amounts of unlabeled images. CSD~\cite{jeong2019consistency} and FocalMix~\cite{wang2020focalmix} utilize consistency regularization to make full use of the unlabeled data. Self-training and strong data augmentations are employed in STAC~\cite{sohn2020simple} to enhance the detector for SSOD. Although the issue of incomplete training data is also involved in this work, the main difference is that no extra images are introduced during training except the available partially-labeled datasets in our problem.
\paragraph{Open-Set Object Detection} OSOD~\cite{hall2020probabilistic,miller2018dropout} refers to the issue that objects from a new domain that are not seen in the training data may be encountered in the test data, which bring about more false positive predictions compared to the closed-set. In this work, there probably exists a non-ignorable domain gap between the new annotated training set and the original training set. Methods for OSOD inspire us to mine higher-quality pseudo annotations with fewer false positive predictions based on uncertainty estimation.
{\color{highlight}\paragraph{Class Incremental Object Detection} In recent years, class incremental learning has attracted much attention in diverse tasks, such as classification~\cite{rebuffi2017icarl,wu2019large,zhao2020maintaining,zhao2022energy}, segmentation~\cite{michieli2021knowledge} and also object detection. CIOD aims at developing a lifelong-learning detection system. In~\cite{shmelkov2017incremental,hao2019end,perez2020incremental,kj2021incremental}, efforts are made to learn detectors incrementally based on only new training images with new classes, i.e., the original training data is not available when new classes arrive. Specifically, knowledge distillation is employed to prevent the forgetting phenomenon in~\cite{shmelkov2017incremental} and the method in~\cite{hao2019end} further extend it to an end-to-end training manner. A meta-learned gradient preconditioning is proposed in~\cite{kj2021incremental} to not only minimize forgetting but also maximize knowledge transfer. In CIOD, the primary challenge is the catastrophic forgetting~\cite{mccloskey1989catastrophic,lao2021focl} for old classes due to a lack of original datasets. While the original data is still available in this work, we focus on a more immediate and realistic situation. Although it relaxes the constraints, we show that it is still a challenging problem (as shown in Fig.~\ref{fig:introduction}) which is frequently encountered in real-world applications and must be addressed urgently.}
\paragraph{Multi-Dataset Object Detection} MDOD tries to train a single detector on multiple datasets, which is the same purpose as in this work. In~\cite{zhao2020object} and~\cite{rame2018omnia}, a pseudo labeling approach is exploited. Dataset-aware focal loss is proposed in~\cite{yao2020cross} for the multi-dataset training. Compared to these methods, we present a more powerful pipeline to deal with this problem, which concentrates on three main questions: how to avoid label conflicts effectively and comprehensively (answered in Sec.~\ref{sec:cf_loss}), how to mine more accurate pseudo annotations (answered in Sec.~\ref{sec:gt_mining}), and how to make better use of pseudo labels (answered in Sec.~\ref{sec:retraining}).
\section{Category-extended Object Detector}
In this section, we introduce our solution for training category-extended detectors. We analyze the possible label conflicts hidden in the plain joint training and propose a general loss formula (conflict-free loss) for this problem (Sec.~\ref{sec:cf_loss}). It attempts to take full advantage of the exact information and avoid ambiguous one, leading to an acceptable detector in one training round. Then, to further improve the performance, we design the retraining phase. It consists of two components: (a) Unlabeled ground-truth mining with classification and localization confidence (CLC) (Sec.~\ref{sec:gt_mining}). The CLC-based unlabeled ground-truth mining method helps to obtain more accurate pseudo annotations. (b) Retraining with overlap-weighted negative samples (Sec.~\ref{sec:retraining}). The overlap-weighted negative samples in retraining help to make full use of the pseudo annotations and obtain a better detector.
\subsection{Notations}
For brevity of presentation and without loss of generality, we assume that there are original dataset $\mathcal{D}_o$ (denoted by categories $\mathcal{C}_o$, images $\mathcal{I}_o$ and ground-truth annotations $\mathcal{G}_o$) and newly-added dataset $\mathcal{D}_n$ (denoted by categories $\mathcal{C}_n$, images $\mathcal{I}_n$ and ground-truth annotations $\mathcal{G}_n$) with different label spaces\footnote{This technique is also applicable to multiple datasets.} ($\mathcal{C}_o$ and $\mathcal{C}_n$ do not include the special category ``background"). We aim to train a unified object detector on $\mathcal{D}_o$ and $\mathcal{D}_n$. The overall loss function for training can be formulated as a weighted sum of the classification loss ($\mathcal{L}_{cls}$) and the localization loss ($\mathcal{L}_{loc}$):
\begin{equation}
\mathcal{L}(\{\mathbf{p}_i\}, \{\mathbf{t}_i\};
\{\mathbf{p}^*_i\}, \{\mathbf{t}^*_i\}) =
\mathcal{L}_{cls}(\{\mathbf{p}_i\}, \{\mathbf{p}^*_i\})
+
\mathcal{L}_{loc}(\{\mathbf{t}_i\}, \{\mathbf{t}^*_i\}),
\label{eq:overall_loss}
\end{equation}
where $i$ is the index of an anchor, $\mathbf{t}_i$ is a $4$-dimensional vector representing the parameterized coordinates of the predicted bounding box, and $\mathbf{t}_i^*$ is that of the ground-truth box associated with a positive anchor. $\mathbf{p}_i^*$ and $\mathbf{p}_i$ are the ground-truth label and the predicted probability vector of anchor $i$, respectively. The localization loss $\mathcal{L}_{loc}$ usually can be Smooth $L_1$ loss~\cite{he2016deep}. The classification loss $\mathcal{L}_{cls}$ can be a binary cross-entropy (BCE) based loss. It views the classification task as a series of independent binary classification tasks, which is formulated as
\begin{equation}
\mathcal{L}_{cls} \coloneqq \mathcal{L}_{bce}(\{ \mathbf{p}_i \}, \{ \mathbf{p}^*_i \}) =
\frac{1}{N} \sum_{i,c}
l({p^c_i, p^*_i}^c)
,
\label{eq:bce_loss}
\end{equation}
where $
l({p^c_i, p^*_i}^c) = -[{p^*_i}^c log(p^c_i) +
(1 - {p^*_i}^c) log(1 - p^c_i)]
$. Under the BCE-based loss, $\mathbf{p}_i^*$ is a $|\mathcal{C}_o \cup \mathcal{C}_n|$-dimensional vector. If anchor $i$ matches with a ground-truth box in $\mathcal{G}_o$ or $\mathcal{G}_n$ of category $c$, $\mathbf{p}_i^*$ is denoted as a one-hot vector with only ${p_i^*}^c=1$. If it does not match any box in $\mathcal{G}_o \cup \mathcal{G}_n$, $\mathbf{p}_i^*$ will be set to $\mathbf{0}$.
As an example shown in Fig.~\ref{fig:introduction}, the detector trained with plain joint training on the combined dataset of COCO (dose not labeled ``face") and WIDER FACE (dose not labeled ``person") cannot detect ``face" in the COCO evaluation set or ``person" in the WIDER FACE evaluation set. The detector is biased due to the datasets and training method: in one dataset (e.g., COCO), samples related to unlabeled instances (e.g., "face") will be regarded as negative (background) samples, resulting in conflicts and false gradients during training. To relieve this problem, in the next subsection, we propose conflict-free loss, which attempts to make full use of the correct information and avoid using ambiguous information.
\subsection{Conflict-Free Loss}
\label{sec:cf_loss}
\begin{figure}[t]
\centering
\includegraphics[width=0.8\textwidth]{./cf_loss}
\caption{Illustration of conflict-free loss. Two possible conflicts are relieved, details in Sec.~\ref{sec:cf_loss}. {\color{red} Red} for original classes and {\color{green} green} for new classes. Best seen in color.}
\label{fig:cf_loss}
\end{figure}
In Eq.~\eqref{eq:overall_loss}, the localization loss is activated only for positive anchors ($i \in \text{Pos}$) and disabled otherwise ($i \in \text{Neg}$), so that there are few possible conflicts hidden in it and we keep the localization loss unchanged. However, the plain BCE-based loss will be affected by false signals during joint training on multiple datasets. Thus, the classification loss should be carefully designed. There are two possible conflicts in plain BCE-based loss. (i) Anchors assigned as negative samples in one dataset may be positive samples belonging to unlabeled instances with classes from the other datasets. (ii) The assigned positive samples may belong to categories of other datasets. Commonly, in one dataset, an anchor is assigned as positive for one specific category when the Intersection over Union (IoU) with the ground-truth is greater than a threshold (say, 0.5 typically). However, such loose restriction may ignore the possibility that the anchor is more suitable to be labeled as another category that is only annotated in other datasets.
Consequently, we propose the Conflict-Free (CF) loss, which is formulated as
\begin{equation}
\mathcal{L}_{cls} \coloneqq \mathcal{L}_{cf}(\{ \mathbf{p}_i \}, \{ \mathbf{p}^*_i \}) =
\frac{1}{N} \sum_{i,c} w(i, c) \cdot
l({p^c_i, p^*_i}^c)
,
\label{eq:cf_loss}
\end{equation}
where, $w(i, c)$ is given by Alg.~\ref{algo:omega_training}, $\star$ represents $o$ or $n$, $f(i)$ represents the maximum IoU of anchor $i$ with ground-truth, $\tau_{s}$ is a strict threshold (0.9, conservatively). An illustration of Eq.~\eqref{eq:cf_loss} and Alg.~\ref{algo:omega_training} is shown in Fig.~\ref{fig:cf_loss}. In conflict-free loss, the two possible conflict origins are removed. First, a negative sample will not contribute to the loss of classes that are not labeled in its dataset $(i \in \mathcal{I}_{\star} \ \& \ c \notin \mathcal{C}_{\star} \ \& \ i \in \text{Neg})$. Second, the conflict-free loss does not concern the positive anchors with insufficient large overlaps with ground-truth ($\text{IoU} < \tau_{s}$) to provide negative information to classes from the other datasets $(i \in \mathcal{I}_{\star} \ \& \ c \notin \mathcal{C}_{\star} \ \& \ i \in \text{Pos} \ \& \ f(i) < \tau_{s})$. Note that the name ``conflict-free" is used to represent a general loss formula for avoiding conflicts, which can be used directly or integrated with any other BCE-based loss (e.g., Focal loss~\cite{lin2017focal}) by replacing $l$.
\begin{algorithm}[tp]
\centering
\caption{Obtaining $\{\omega (i,c)\}$ for Eq.~\eqref{eq:cf_loss}.}
\begin{algorithmic}[1]
\For{ anchor $i \in 1, \cdots, N $; class $c \in 1, \cdots, |\mathcal{C}_o \cup \mathcal{C}_n|$}
\If{$i \in \mathcal{I}_{\star} \ \& \ c \notin \mathcal{C}_{\star}$}
\If{$i \in \text{Neg} \ | \ (i \in \text{Pos} \ \& \ f(i) < \tau_{s})$}
\State $\omega(i,c)=0$
\Else
\State $\omega(i,c)=1$
\EndIf
\Else
\State $\omega(i,c)=1$
\EndIf
\EndFor
\State \textbf{Output: $\{\omega (i,c)\}$}
\end{algorithmic}
\label{algo:omega_training}
\end{algorithm}
\begin{figure*}[t]
\centering
\includegraphics[width=0.65\textwidth]{./retraining}
\caption{Illustration of the retraining method. (top) Unlabeled ground-truth mining with both classification and localization confidence. (bottom) Retraining on the supplemented datasets with overlap-based weighting schedule for negative samples. The remaining legend is the same as in Fig.~\ref{fig:cf_loss}. Best seen in color.}
\label{fig:retraining}
\end{figure*}
It is demonstrated in the later experiments that an acceptable unified detector can be learned with conflict-free loss in one training round. While, there still exist many underutilized unlabeled objects. Inspired by self-training, a retraining phase is employed to further improve performance, which consists of two components: unlabeled ground-truth mining with classification and localization confidence (Sec.~\ref{sec:gt_mining}) and retraining with overlap-weighted negative samples (Sec.~\ref{sec:retraining}).
\subsection{Unlabeled Ground-Truth Mining}
\label{sec:gt_mining}
Unlike classification tasks, detectors need to predict the localization of objects, so that accurate localization annotations usually lead to better performance. Previous studies~\cite{jiang2018acquisition,he2019bounding} show that classification score cannot be interpreted as localization confidence: a predicted bounding box with high classification confidence may be localized inaccurately. If the inaccurate predicted results are adopted as pseudo annotations, they will prevent detectors from getting better performance. In pursuit of high-quality pseudo annotations, we propose the Classification and Localization Confidence (CLC) based unlabeled ground-truth mining method.
Monte Carlo Dropout-based Bayesian neural networks have shown great potential in measuring model uncertainty~\cite{gal2016dropout,kendall2017uncertainties,miller2018dropout,miller2019evaluating}. Inspired by this concept, during the inference phase, we perform $T$ forward passes of the base detector learned in Sec.~\ref{sec:cf_loss} with activated dropout layers as depicted in Fig.~\ref{fig:retraining}. Notice that since we introduce dropout operation only on the last layer of the classification and localization head while all the previous layers are identical and the computations are shared, the extra inference time during $T$ forward passes is ignorable. Then, for each image, we perform clustering on the bounding boxes of each class, resulting in a set of clusters. Each cluster $\mathcal{O}_i$ is made up of several bounding boxes with their respective classification score, defined as
\begin{equation}
\begin{aligned}
\mathcal{O}_i &= \{(\mathbf{t}_{ij}, \mathbf{p}_{ij}) \ | \ j=1,2,\cdots,|\mathcal{O}_i| \leq T \}, \\
\ \text{s.t.} &\ \text{IoU}(\mathbf{t}_{ij_1}, \mathbf{t}_{ij_2}) \geq \tau_{nms}, \forall \ \mathbf{t}_{ij_1}, \mathbf{t}_{ij_2} \in \mathcal{O}_i,\\
&\argmax_{r \in \mathcal{C}_o \cup \mathcal{C}_n} \ p^r_{ij_1} = \argmax_{r \in \mathcal{C}_o \cup \mathcal{C}_n} \ p^r_{ij_2}, \forall \ \mathbf{p}_{ij_1}, \mathbf{p}_{ij_2} \in \mathcal{O}_i,
\end{aligned}
\end{equation}
where the threshold for clustering is the same as the NMS IoU threshold $\tau_{nms}$ (0.5 typically). Then the cluster $\mathcal{O}_i$ is represented by a triplet $(\overline{\mathbf{t}}_i, \overline{p}_i, \overline{e}_i)$, where the integrated bounding box $\overline{\mathbf{t}}_i$ and classification confidence $\overline{p}_i$ can be calculated by
\begin{equation}
\overline{\mathbf{t}}_i = \frac{1}{|\mathcal{O}_i|} \sum_{j} \mathbf{t}_{ij}, \
\overline{p}_i = \frac{1}{|\mathcal{O}_i|} \sum_{j} \max(p^1_{ij}, \cdots, p^{|\mathcal{C}_o \cup \mathcal{C}_n|}_{ij}),
\end{equation}
More importantly, the localization confidence is given by
\begin{equation}
\overline{e}_i =
\frac{ 1 + \mathbbm{1}_{|\mathcal{O}_i| \geq \frac{T}{2}} }{2|\mathcal{O}_i|^2}
\sum_{j_1, j_2} \text{IoU}(\mathbf{t}_{ij_1}, \mathbf{t}_{ij_2}).
\label{eq:loc_conf}
\end{equation}
Then, we obtain the detection confidence $\overline{d}_i = \overline{p}_i \times \overline{e}_i$, which considers both classification and localization confidence, and can better reflect the quality of predictions. Finally, we accept an integrated bounding box as pseudo ground-truth for class $c$ only if it satisfies: (i) class $c$ is not labeled in the dataset originally; (ii) $\overline{d}_i$ is larger than a pre-defined threshold $\eta$. Generally speaking, only the predictions with enough high detection confidence will be selected as pseudo annotations for the unlabeled classes. Fig.~\ref{fig:retraining} (top) illustrates the mining method.
\subsection{Retraining with Pseudo Annotations}
\label{sec:retraining}
After acquiring the pseudo annotations, we try to train a more powerful unified detector with the ground-truth and pseudo annotations together. So how to make better use of pseudo annotations? The most straightforward way is viewing the combined dataset supplemented with pseudo annotations as fully labeled~\cite{sohn2020simple}. Then the detector can be trained in a normal way with common classification losses (Eq.~\eqref{eq:bce_loss}). However, there may still exist considerable instances that are difficult to be mined. Hence, there is a risk that the inaccurate supervisory information may be introduced. Instead, one can retrain the detector with the conflict-free loss again (Eq.~\eqref{eq:cf_loss} and Alg.~\ref{algo:omega_training}). Compared to the first phase, the negative samples still do not contribute to the loss of classes from different datasets, only positive samples are supplemented by the pseudo annotations. So that the imbalance between positive samples and negative samples may be intensified in this way. A solution to alleviate the imbalance problem mentioned above is to introduce some ``safe" negative samples~\cite{zhao2020object}, which can contribute to the loss of classes from different datasets while ``unsafe" negative samples are still discarded.
\begin{algorithm}[t]
\centering
\caption{Obtaining $\{\omega (i,c)\}$ for Eq.~\eqref{eq:cf_loss} during retraining.}
\begin{algorithmic}[1]
\For{ anchor $i \in 1, \cdots, N $; class $c \in 1, \cdots, |\mathcal{C}_o \cup \mathcal{C}_n|$}
\If{$i \in \mathcal{I}_{\star} \ \& \ c \notin \mathcal{C}_{\star} \ \& \ i \in \text{Neg}$}
\State $\omega(i,c)= Gom(f(i))$
\Else
\State $\omega(i,c)=1$
\EndIf
\EndFor
\State \textbf{Output: $\{\omega (i,c)\}$}
\end{algorithmic}
\label{algo:omega_retraining}
\end{algorithm}
\begin{figure}[t]
\centering
\includegraphics[width=0.25\textwidth]{./gom_func}
\caption{Overlap-based weighting schedule ($\tilde{a}=1$, $\tilde{b}=10,000$ and $\tilde{c}=25$). The negative samples' contribution to the loss of classes from different datasets is assigned with high weight only when they have a large overlap with annotated boxes.}
\label{fig:gom_func}
\end{figure}
Instead, we introduce an overlap-based weighting schedule~\cite{wu2018soft} here to re-weight the negative samples' contribution to the loss of classes from different datasets. The main hypothesis is that negative samples which have relatively large overlaps (like 0.3 IoU, 0.4 IoU) with the existing ground-truth (or mined pseudo annotations) are probably not related to any unlabeled instances. On the contrary, negative samples with small overlaps have a higher risk of being unlabeled instances, i.e., false background. Based on this hypothesis, the Gompertz function
$$
Gom(x) = \tilde{a} e^{-\tilde{b} e^{-\tilde{c}x}}
$$
is employed to describe the relationship between overlap and the weight of loss, where $\tilde{a}$ is an asymptote, $\tilde{b}$ sets the displacement along the x-axis, $\tilde{c}$ sets the growth rate. We plot this function in Fig.~\ref{fig:gom_func} with $\tilde{a}=1$, $\tilde{b}=10,000$ and $\tilde{c}=25$. The negative samples' contribution to the loss of classes from different datasets is assigned with high weight only when they have a large overlap with annotated boxes. Other negative samples are still given very low weights.
Our overlap-weighted retraining method can be formulated as training with Eq.~\eqref{eq:cf_loss} and $\{\omega (i,c)\}$ calculated from Alg.~\ref{algo:omega_retraining}. Considering many unlabeled instances are mined and the main concern in retraining is balancing the extra positive samples brought by pseudo annotations, we no longer consider the conflicts in positive samples in the retraining phase. In later experiments, we found that this method can take advantage of the mined pseudo annotations more fully.
\section{Experiments}
\subsection{Experimental Settings}
\label{sec:exp_set}
\paragraph{Datasets} We conduct experiments on several widely used datasets with different settings. MS COCO~\cite{lin2014microsoft} (2017) consists of 118,287 training images (COCO-train), 5,000 validation images (COCO-val) and 40,670 test images with 80 categories. PASCAL VOC~\cite{everingham2010pascal} (2007) has 9,963 images with 20 categories, 50\% for training/validation (VOC-train) and 50\% for testing (VOC-test). WIDER FACE~\cite{yang2016wider} is comprised of 32,203 images with one category ``face'', training 40\% (FACE-train), validation 10\% (FACE-val) and test 50\%. To evaluate the unified models, we also adopt the re-labeled evaluation sets \cite{zhao2020object}: VOC-subtest (a subset of VOC-test) contains 500 images with annotations for the 80 categories in COCO; FACE-subval (a subset of FACE-val) consists of 500 images with annotations for the ``face'' in FACE and the ``person'' in COCO; COCO-subval (a subset of COCO-val) has 500 images with annotations for the 80 categories in COCO and the ``face'' in FACE.
\begin{table}[t]
\centering
\caption{The statistics of the training sets used in our experiments.}
\resizebox{\textwidth}{!}{
\begin{tabular}{llllll}
\toprule
setup & training set & \multicolumn{1}{l}{\#categories} & \multicolumn{1}{l}{\#images} & \multicolumn{1}{l}{\#annotations} & \multicolumn{1}{l}{\#missing annotations} \\
\midrule
A & COCO75-train & 75 & 100,543 & 601,410 & 129,089 \\
& COCO5-train & 5 & 17,744 & 22,976 & 106,526 \\
B & COCO79-train & 79 & 100,543 & 506,916 & 222,696 \\
& COCO1-train & 1 & 17,744 & 39,769 & 90,620 \\
C & COCO60-train & 60 & 118,287 & 367,189 & 492,812 \\
& VOC-train & 20 & 5,011 & 15,662 & unknown \\
D & COCO-train & 80 & 118,287 & 860,001 & unknown \\
& FACE-train & 1 & 12,880 & 159,424 & unknown \\
E & COCO60-train & 60 & 118,287 & 367,189 & unknown \\
& VOC-train & 20 & 5,011 & 15,662 & unknown \\
& FACE-train & 1 & 12,880 & 159,424 & unknown \\
\bottomrule
\end{tabular}%
}
\label{tab:data_stat}%
\end{table}%
\paragraph{Setups} To analyze the effect of our pipeline, we design five experimental setups. The statistics of the training sets used in these setups are summarized in Tab.~\ref{tab:data_stat}.
\textbf{Setup A}: COCO75-train and COCO5-train. We split COCO-train into two sets. One set is named COCO75-train, in which only annotations of 75 categories are retained, while annotations related to the other 5 categories are removed; another set, named COCO5-train, contains only annotations of the 5 categories. COCO75-train and COCO5-train are regarded as original dataset $\mathcal{D}_o$ and new dataset $\mathcal{D}_n$, respectively. In COCO75-train, there exist considerable unlabeled objects of the remaining 5 categories, similar in COCO5-train. We report the performance of the 75 classes and the 5 classes on COCO-val. Since there are more missing annotations belonging to the new 5 classes in the combined training set, we should pay more attention to the results of these 5 classes.
\textbf{Setup B}: COCO79-train and COCO1-train. We perform another split on COCO-train: the original dataset COCO79-train contains 79 classes, and the new dataset COCO1-train has only one category. We report the performance of the 79 classes and the one class on COCO-val. Similarly, the results of the new class should be paid more attention.
\textbf{Setup C}: COCO60-train and VOC-train. In these experiments, we regard VOC-train as the new dataset and COCO60-train as the original dataset, which is generated by removing the annotations of the 20 VOC categories from COCO-train. We report the performance of all 80 classes on VOC-subtest and COCO-subval, and of the 20 VOC classes on COCO-val.
\textbf{Setup D}: COCO-train and FACE-train. We also train unified detectors on COCO-train (as original dataset) and FACE-train (as new dataset). The ``face'' is not annotated in COCO-train and the 80 COCO classes (``person'' mainly) are also not labeled in FACE-train. The performance of ``face" and ``person" on COCO-subval and FACE-subval is reported, in which the results of ``face" on COCO-subval and ``person" on FACE-subval should be paid more attention.
\textbf{Setup E}, which involves three training sets labeled with different classes: COCO60-train, VOC-train, and FACE-train. The whole label space has 81 classes. We report the performance of all 81 classes on VOC-subtest and COCO-subval, and of ``person" on FACE-subval.
\paragraph{Metrics} We use the standard metrics for object detection: AP, $\text{AP}_{50}$, and $\text{AP}_{75}$.
\paragraph{Implementations Details}
For fair comparisons, all experiments are conducted on RetinaNet~\cite{lin2017focal} with ImageNet-pretrained ResNet-50~\cite{he2016deep} as the backbone. The models are implemented with PyTorch~\cite{paszke2019pytorch} and MMDetection~\cite{chen2019mmdetection}. In all experiments, input images are resized to $1333\times 800$. The models are trained using SGD over 8 GPUs with 2 images per GPU with 0.9 momentum, 0.0001 weight decay. We train 12 epochs in total with an initial learning rate of 0.02, and decrease the learning rate by 0.1 at epoch 8 and 11. For the conflict-free loss, the more strict threshold $\tau_{s}$ is set to 0.9 conservatively as described in Sec.~3.2 in all experimental settings. For the CC-based unlabeled ground-truth mining method, the confidence threshold $\eta$ is set to 0.5 in all experimental settings for best overall performance. For the CLC-based mining method, the number of forward passes $T$ is set to 20 for all experimental settings; the confidence threshold $\eta$ is set to 0.525 in the experiment on COCO75-train and COCO5-train, and 0.5 in other experimental settings for best overall performance. For the retraining strategy ``safe negatives", the lower confidence threshold $\eta'$ is set to 0.1 to obtain high-recall pseudo annotations in all experimental settings. For the retraining strategy ``overlap-weighted", we set $\tilde{a}=1$, $\tilde{b}=10,000$ and $\tilde{c}=25$ as shown in Sec.~3.4 in all experimental settings. There is almost no change in hyper-parameters in different experimental settings, which shows that our pipeline is robust.
\subsection{Overall Performance}
\begin{figure}[t]
\centering
\subfloat[]{\includegraphics[height=0.16\textwidth]{./c_b1}},
\subfloat[]{\includegraphics[height=0.16\textwidth]{./c_b2}},
\subfloat[]{\includegraphics[height=0.16\textwidth]{./c_b3}},
\subfloat[]{\includegraphics[height=0.16\textwidth]{./c_b4}}
\caption{Overall Performance. Subfigs (a), (b), (c) and (d) show the results of setup A, B, C and D, respectively. The performance is evaluated on specific sets and categories for each setup, details in Sec.~\ref{sec:exp_set}. Best seen in color.}
\label{fig:compare_baselines}
\end{figure}
\begin{figure}[t]
\centering
\includegraphics[height=0.2\textheight]{./vis1.pdf}
\caption{Detection results of the plain joint training (top) and our method (bottom) on COCO-val (COCO evaluation set, the first two columns) and FACE-val (WIDER-FACE evaluation set, the last two columns). Red bounding boxes indicate objects belonging to the 80 categories in COCO, while green ones correspond to the category ``face'' in WIDER FACE. Best seen in color.}
\label{fig:vis1}
\end{figure}
We first show the performance of unified detectors trained by the proposed solution and the plain joint training on setup A-D. As shown in Fig.~\ref{fig:compare_baselines}, the performance of the detector obtained by plain joint training is poor, especially for categories that are not labeled in datasets. In setup C, due to the label ambiguity and domain gap, the detector trained in the plain way is heavily biased, which only gets 4.1\% AP$_{50}$ for the 20 VOC categories on COCO-val. In setup D, the performances of the detector trained in the plain way are only 3.4\% and 6.7\% in terms of AP$_{50}$ for ``face" on COCO-subval and ``person" on FACE-subval, respectively. We show that the proposed method achieves the performances of 42.6\%, 53.9\% and 58.5\% with improvements of 38.5\%, 50.5\% and 51.8\%, respectively. For setup A and B, we also provide the results based on fully labeled datasets, which can be seen as upper-bounds. These results demonstrate that our method can reduce the gap with fully labeled datasets.
Fig.~\ref{fig:vis1} illustrates some visualization results of the detectors trained on COCO-train and FACE-train. Due to the conflicts during plain joint training and the domain gap between the two sets, the detector trained in the vanilla way only can detect ``person'' in COCO-val and ``face'' in FACE-val (i.e., categories which are labeled in corresponding training data). Assisted by our approach, the detector can detect all categories on two evaluation sets. Both qualitative and quantitative results demonstrate the effectiveness of our method.
\subsection{Comparison to Other Methods}
In this section, we compare our conflict-free loss, overlap-weighted retraining method and CLC-based unlabeled ground-truth mining method with corresponding counterparts in Sec.~\ref{sec:compare_1}, Sec.~\ref{sec:compare_2} and Sec.~\ref{sec:compare_3}, respectively.
\begin{table}[t]
\centering
\caption{Compare conflict-free loss with other methods on COCO75-train ($\mathcal{D}_o$) and COCO5-train ($\mathcal{D}_n$). The results of 75 original classes and 5 new classes on COCO-val are reported. `\textbf{*}' indicates the categories that we should pay more attention to. Best results for these categories are in \textbf{bold}.}
\resizebox{1.\textwidth}{!}{
\begin{tabular}{llllllll}
\toprule
Data & Method & \multicolumn{3}{l}{75 original classes} & \multicolumn{3}{l}{5 new classes~\textbf{*}} \\
\cmidrule(lr){3-5} \cmidrule(lr){6-8}
& & AP & $\text{AP}_{50}$ & $\text{AP}_{75}$ & AP & AP$_{50}$ & AP$_{75}$ \\
\midrule
$\mathcal{D}_o$ & plain~\cite{lin2017focal} & 36.3 & 55.4 & 38.6 & N/A & N/A & N/A \\
$\mathcal{D}_n$ & plain~\cite{lin2017focal} & N/A & N/A & N/A & 20.2 & 36.6 & 19.4 \\
$\mathcal{D}_o$+$\mathcal{D}_n$ & plain~\cite{lin2017focal} & 35.8 & 54.9 & 38.1 & 20.6 & 36.3 & 20.3 \\
& dataset-aware~\cite{yao2020cross} & 36.4 & 55.3 & \textbf{38.8} & 22.7 & 39.4 & 22.5 \\
& conflict-free (ours) & \textbf{36.5} & \textbf{55.5} & 38.7 & \textbf{23.5} & \textbf{40.4} & \textbf{23.4} \\
{\color{highlight} fully labeled} & {\color{highlight}plain~\cite{lin2017focal}} & {\color{highlight}37.1} & {\color{highlight}56.3} & {\color{highlight}39.6} & {\color{highlight}30.6} & {\color{highlight}49.6} & {\color{highlight}31.5} \\
\bottomrule
\end{tabular}%
}
\label{tab:compare_loss}%
\end{table}%
\subsubsection{Compare conflict-free loss with others} \label{sec:compare_1}
We compare the proposed conflict-free loss with other methods under setup A. The dataset-aware loss in~\cite{yao2020cross} is a special case of conflict-free loss without considering the second conflict origin discussed in Sec.~\ref{sec:cf_loss}. We also provide the results of multiple detectors trained on separate datasets.
As shown in Tab.~\ref{tab:compare_loss}, the performance of the detectors trained separately on COCO75-train and COCO5-train are 36.3\% AP and 20.2\% AP respectively. The performance of the detector trained on the combined training set in a vanilla way drops to 35.8\% AP for the 75 classes, which is mainly caused by the conflicts during joint training; and improves to 20.6\% AP for the 5 classes, which reveals that more training data may lead to better feature representation. The detector trained with dataset-aware loss~\cite{yao2020cross} outperforms the plain method. Furthermore, our conflict-free loss achieves better results (23.5\% AP for the 5 classes), which is mainly due to that the conflict-free loss can avoid two possible conflicts simultaneously as discussed in Sec.~\ref{sec:cf_loss}. It is worth noting that the unified detector trained with conflict-free loss outperforms separate detectors, especially for new classes (23.5AP vs 20.2AP) owing to the abundant training samples provided by other datasets. It demonstrates that the conflict-free loss leads to a unified detector with acceptable results in one training round, outperforms other unified detectors, and is not worse than separate detectors.
\begin{table}[t]
\centering
\caption{Compare the unlabeled ground-truth mining method and the retraining method with other approaches under setup A ($\mathcal{D}_o$: COCO75-train, $\mathcal{D}_n$: COCO5-train). $\mathcal{\hat{G}}_{CC}$ and $\mathcal{\hat{G}}_{CLC}$ represent the mined pseudo annotations by CC-based and CLC-based methods, respectively. We report the results of the 75 original classes and the 5 new classes on COCO-val, respectively.}
\resizebox{1\textwidth}{!}{
\begin{tabular}{llllllll}
\toprule
Data & Method & \multicolumn{3}{l}{75 original classes} & \multicolumn{3}{l}{5 new classes~\textbf{*}} \\
\cmidrule(lr){3-5} \cmidrule(lr){6-8}
& & AP & $\text{AP}_{50}$ & $\text{AP}_{75}$ & AP & AP$_{50}$ & AP$_{75}$ \\
\midrule
$\mathcal{D}_o$+$\mathcal{D}_n$ & plain~\cite{lin2017focal} & 35.8 & 54.9 & 38.1 & 20.6 & 36.3 & 20.3 \\
& conflict-free (ours) & 36.5 & 55.5 & 38.7 & 23.5 & 40.4 & 23.4 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CC}$ & as fully labeled~\cite{sohn2020simple} & 36.6 & 55.4 & 38.9 & 24.0 & 39.9 & 24.6 \\
& conflict-free (ours) & 36.3 & 55.2 & 38.5 & 24.2 & 40.8 & 24.6 \\
& safe negatives~\cite{zhao2020object} & 36.6 & \textbf{55.7} & 38.7 & 24.5 & 41.1 & 25.0 \\
& overlap-weighted (ours) & 36.6 & 55.4 & \textbf{39.0} & 24.7 & 41.7 & \textbf{25.1} \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CLC}$ & overlap-weighted (ours) & \textbf{36.7} & 55.6 & \textbf{39.0} & \textbf{24.9} & \textbf{42.0} & \textbf{25.1} \\
{\color{highlight} fully labeled} & {\color{highlight}plain~\cite{lin2017focal}} & {\color{highlight}37.1} & {\color{highlight}56.3} & {\color{highlight}39.6} & {\color{highlight}30.6} & {\color{highlight}49.6} & {\color{highlight}31.5} \\
\bottomrule
\end{tabular}%
}
\label{tab:coco75+coco5}%
\end{table}%
\begin{table}[t]
\centering
\caption{Compare the unlabeled ground-truth mining method and the retraining method with other approaches under setup B ($\mathcal{D}_o$: COCO79-train, $\mathcal{D}_n$: COCO1-train). We report the results of the 79 original classes and the 1 new class on COCO-val, respectively.}
\resizebox{1.\textwidth}{!}{
\begin{tabular}{llllllll}
\toprule
Data & Method & \multicolumn{3}{l}{79 original classes} & \multicolumn{3}{c}{1 new class~\textbf{*}} \\
\cmidrule(lr){3-5} \cmidrule(lr){6-8}
& & AP & AP$_{50}$ & AP$_{75}$ & AP & AP$_{50}$ & AP$_{75}$ \\
\midrule
$\mathcal{D}_o$+$\mathcal{D}_n$ & plain~\cite{lin2017focal} & 35.4 & 54.4 & 37.7 & 37.9 & 69.2 & 36.5 \\
& conflict-free (ours) & \textbf{36.5} & \textbf{55.5} & \textbf{39.0} & 41.6 & 72.4 & 41.4 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CC}$ & as fully labeled~\cite{sohn2020simple} & 36.2 & 55.1 & 38.4 & 42.9 & 71.1 & 44.4 \\
& conflict-free (ours) & 36.1 & 54.8 & 38.3 & 44.4 & 74.4 & 45.6 \\
& safe negatives~\cite{zhao2020object} & 36.1 & 55.0 & 38.2 & 43.7 & 72.9 & 45.0 \\
& overlap-weighted (ours) & 36.0 & 54.8 & 38.1 & 44.6 & 74.2 & 46.0 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CLC}$ & overlap-weighted (ours) & 36.2 & 55.0 & 38.5 & \textbf{44.7} & \textbf{74.6} & \textbf{46.3} \\
{\color{highlight} fully labeled} & {\color{highlight}plain~\cite{lin2017focal}} & {\color{highlight}36.5} & {\color{highlight}55.5} & {\color{highlight}38.9} & {\color{highlight}51.2} & {\color{highlight}79.6} & {\color{highlight}54.2} \\
\bottomrule
\end{tabular}%
}
\label{tab:coco79+coco1}
\end{table}%
\subsubsection{Compare the overlap-weighted retraining method with others}
\label{sec:compare_2}
We describe other retraining strategies as follows. (i) As described in Sec.~\ref{sec:retraining}, we can view the combined dataset supplemented with pseudo annotations \textbf{as fully labeled}~\cite{sohn2020simple}, and train detectors with common classification loss. (ii) We can retrain the detector with the \textbf{conflict-free} loss again. Compared to the first phase, only positive samples are supplemented by the pseudo annotations. (iii) We can introduce some \textbf{safe negatives}~\cite{zhao2020object}. In addition to the threshold $\eta$ used to mine high-quality pseudo annotations, a lower threshold $\eta'$ is introduced to get high-recall pseudo annotations. Under the combination of ground-truth and high-quality pseudo annotations, some negative samples are actually unsafe negative samples which can match boxes from high-recall pseudo annotations. Anchors that do not match any ground-truth or high-recall pseudo annotations are considered to be safe negative samples. Safe negative samples can contribute to the loss of classes from different datasets while unsafe negative samples are still discarded.
For fair, we compare our overlap-weighted retraining method with these strategies with the same pseudo annotations (training detectors with conflict-free loss and mining pseudo annotations with classification confidence, which is described in the next paragraph). We provide the results of setup A, B, C and D in Tab.~\ref{tab:coco75+coco5},~\ref{tab:coco79+coco1},~\ref{tab:coco60+voc20} and~\ref{tab:coco80+face1} respectively ($\mathcal{D}_o$ + $\mathcal{D}_n$ + $\mathcal{\hat{G}}_{CC}$). In setup A, retraining with ``as fully labeled" leads to improvement compared to the results of one training round (0.5\%AP gains for the 5 new classes), which demonstrates the effectiveness of retraining. The detector retrained with conflict-free loss achieves better performance for the 5 classes. Relying on ``safe negatives" during retraining, the detector reaches 24.5\% AP for the 5 classes. Finally, our retraining method achieves the best results (24.7\% AP for 5 classes). Similar results are shown in setup B, C and D. These results imply that there are still many unlabeled ground-truth in the training data, thus the idea of avoiding conflicts cannot be ignored in retraining process. Whereas, if we use the same method as in the first training round, only positive samples would be added, causing insufficient negative information in retraining. Thus, it is important to employ suitable negative samples to achieve a balance between positive and negative samples. Our empirical experiments show that using the overlap-weighted negative samples is a better choice for the purpose.
\begin{table}[t]
\centering
\caption{Compare the unlabeled ground-truth mining method and the retraining method with other approaches under setup C ($\mathcal{D}_o$: COCO60-train, $\mathcal{D}_n$: VOC-train). We report the results of all 80 classes on VOC-subtest and COCO-subval, and the 20 VOC classes on COCO-val.
}
\resizebox{1.\textwidth}{!}{
\begin{tabular}{llllllll}
\toprule
Data & Method & \multicolumn{2}{l}{VOC-subtest} & \multicolumn{2}{l}{COCO-subval} & \multicolumn{2}{l}{COCO-val} \\
\cmidrule(lr){3-4} \cmidrule(lr){5-6} \cmidrule(lr){7-8}
& & \multicolumn{2}{l}{all 80 classes} & \multicolumn{2}{l}{all 80 classes} & \multicolumn{2}{l}{20 new classes} \\
\cmidrule(lr){3-4} \cmidrule(lr){5-6} \cmidrule(lr){7-8}
& & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$ \\
\midrule
$\mathcal{D}_o$+$\mathcal{D}_n$ & plain~\cite{lin2017focal} & 43.0 & 28.0 & 42.6 & 27.2 & 4.1 & 2.7 \\
& conflict-free (ours) & 46.8 & 30.1 & 50.3 & 30.8 & 32.3 & 18.6 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CC}$ & as fully labeled~\cite{sohn2020simple} & 48.0 & 31.5 & 48.7 & 29.9 & 27.5 & 18.6 \\
& conflict-free (ours) & \textbf{49.4} & 31.7 & 49.9 & 30.8 & 34.1 & 20.4 \\
& safe negatives~\cite{zhao2020object} & 49.0 & \textbf{33.3} & 49.9 & 31.5 & 34.3 & 21.7 \\
& overlap-weighted (ours) & 47.5 & 31.4 & 51.6 & 31.6 & \textbf{42.7} & 24.9 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CLC}$ & overlap-weighted (ours) & 49.0 & 33.0 & \textbf{52.2} & \textbf{31.7} & 42.6 & \textbf{25.4} \\
\bottomrule
\end{tabular}%
}
\label{tab:coco60+voc20}%
\end{table}%
\begin{table}[t]
\centering
\caption{Compare the unlabeled ground-truth mining method and the retraining method with other approaches on setup D ($\mathcal{D}_o$: COCO-train, $\mathcal{D}_n$: FACE-train). We report the results of categories ``person'' and ``face''. `\textbf{*}' indicates the categories that we should pay more attention to.}
\resizebox{1.\textwidth}{!}{
\begin{tabular}{llllllllll}
\toprule
Data & Method & \multicolumn{4}{l}{COCO-subval} & \multicolumn{4}{l}{FACE-subval} \\
\cmidrule(lr){3-6} \cmidrule(lr){7-10}
& & \multicolumn{2}{l}{person} & \multicolumn{2}{l}{face~\textbf{*}} & \multicolumn{2}{l}{person~\textbf{*}} & \multicolumn{2}{l}{face} \\
\cmidrule(lr){3-4} \cmidrule(lr){5-6} \cmidrule(lr){7-8} \cmidrule(lr){9-10}
& & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$ \\
\midrule
$\mathcal{D}_o$+$\mathcal{D}_n$ & plain~\cite{lin2017focal} & 71.1 & 40.8 & 3.4 & 0.9 & 6.7 & 3.7 & 54.3 & 28.0 \\
& conflict-free (ours) & 71.9 & \textbf{43.2} & 49.9 & 20.9 & 55.6 & 33.9 & \textbf{53.8} & 28.8 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CC}$ & as fully labeled~\cite{sohn2020simple} & \textbf{72.7} & 42.6 & 49.0 & 22.6 & 51.5 & 31.8 & \textbf{53.8} & 28.9 \\
& conflict-free (ours) & 71.2 & 42.1 & 53.9 & 22.8 & 56.4 & \textbf{34.7} & 53.7 & 28.8 \\
& safe negatives~\cite{zhao2020object} & 72.4 & 42.3 & 50.7 & 22.4 & 52.9 & 31.7 & 54.2 & \textbf{29.0} \\
& overlap-weighted (ours) & 71.4 & 42.8 & \textbf{54.0} & 22.5 & 58.3 & \textbf{34.7} & 53.1 & 28.7 \\
$\mathcal{D}_o$+$\mathcal{D}_n$+$\mathcal{\hat{G}}_{CLC}$ & overlap-weighted (ours) & 72.3 & 43.1 & 53.9 & \textbf{23.7} & \textbf{58.5} & \textbf{34.7} & 53.2 & \textbf{29.0} \\
\bottomrule
\end{tabular}%
}
\label{tab:coco80+face1}%
\end{table}%
\begin{figure}[t]
\centering
\includegraphics[height=0.12\textheight]{./tpfp_hist}
\includegraphics[height=0.12\textheight]{./tpfp_curve}
\caption{Distribution of localization confidence across true and false positive predictions (left). The curve of true positive predictions versus fasle negative predictions (right). Best seen in color.}
\label{fig:tpfp}
\end{figure}
\subsubsection{Compare CLC-based unlabeled ground-truth mining method with others}
\label{sec:compare_3}
We compare the CLC-based mining method with the approach that only relies on classification confidence (CC-based) similar to some work for classification tasks~\cite{berthelot2019mixmatch,huang2021behavior,lai2019improving}. The generating process also relies on a score threshold strategy, which only exploits the classification score to represent the detector's confidence in the predicted results. The predictions with high classification confidence will be selected as pseudo annotations to complete ground-truth for the unlabeled classes.
We investigate the performance of the CC-based and the CLC-based mining methods. We expect more accurate bounding boxes can be mined by our CLC-based mining method. To check the localization quality of bounding boxes, we count the true positive predictions and the false positive predictions with high overlap metric ($\text{IoU} \geq 0.75$) in the predicted results whose classification confidence is larger than 0.5. The distribution of localization confidence (Eq.~\eqref{eq:loc_conf}) across true positive predictions and false positive predictions are shown in Fig.~\ref{fig:tpfp} (left). We find the overwhelming majority of bounding boxes with low localization confidence are false negative predictions, which shows the localization confidence can reflect the quality of bounding boxes to some extent. Besides, as shown in Fig.~\ref{fig:tpfp} (right), the number of false positive predictions can be suppressed with the collaboration of localization confidence. The results in Table~\ref{tab:coco75+coco5},~\ref{tab:coco79+coco1},~\ref{tab:coco60+voc20} and~\ref{tab:coco80+face1} ($\mathcal{D}_o$ + $\mathcal{D}_n$ + $\mathcal{\hat{G}}_{CLC}$) also demonstrate that the pseudo annotations mined by the CLC-based method bring an improvement compared to the CC-based method.
\subsection{Extend to multiple training sets}
The above analysis and evaluations are conducted on two datasets, while, our method is also fully capable of handling multiple datasets. To verify it, we perform setup E, which involves three training sets labeled with different classes: COCO60-train ($\mathcal{D}_o$), VOC-train ($\mathcal{D}_{n_1}$), and FACE-train ($\mathcal{D}_{n_2}$). The results are listed in Table~\ref{tab:setup_e}, demonstrating the effectiveness of our solution again.
\begin{table}[tp]
\centering
\caption{Results on setup E. AP$_{50}$ (\%) / AP$_{75}$ (\%) are reported for the 81 classes on VOC-subtest, the 81 classes on COCO-subval, and the ``person" category on FACE-subval, respectively.}
\resizebox{1.\textwidth}{!}{
\begin{tabular}{llllllll}
\toprule
Data & Method & \multicolumn{2}{l}{VOC-subtest} & \multicolumn{2}{l}{COCO-subval} & \multicolumn{2}{l}{FACE-subval} \\
\cmidrule(lr){3-4} \cmidrule(lr){5-6} \cmidrule(lr){7-8}
& & \multicolumn{2}{l}{all 81 classes} & \multicolumn{2}{l}{ all 81 classes} & \multicolumn{2}{l}{person} \\
\cmidrule(lr){3-4} \cmidrule(lr){5-6} \cmidrule(lr){7-8}
& & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$ & AP$_{50}$ & AP$_{75}$\\
\midrule
$\mathcal{D}_o$+$\mathcal{D}_{n_1}$+$\mathcal{D}_{n_2}$ & plain~\cite{lin2017focal} & 40.0 & 27.0 & 42.1 & 25.7 & 0.2 & 0.0 \\
& conflict-free (ours) & 46.5 & 30.1 & 49.1 & 29.7 & 41.9 & 13.5 \\
$\mathcal{D}_o$+$\mathcal{D}_{n_1}$+$\mathcal{D}_{n_2}$+$\mathcal{\hat{G}}_{CLC}$ & as fully labeled~\cite{sohn2020simple} & 47.4 & \textbf{32.0} & 48.2 & 30.0 & 39.3 & 15.7 \\
& overlap-weighted (ours) & \textbf{49.6} & 31.4 & \textbf{51.2} & \textbf{30.4} & \textbf{48.5} & \textbf{18.5} \\
\bottomrule
\end{tabular}%
}
\label{tab:setup_e
\end{table}%
\subsection{BCE-based v.s. CE-based}
Similar to BCE loss, cross-entropy (CE) loss is also widely used in object detection. With CE loss, label $\mathbf{p}^*$ and prediction $\mathbf{p}$ should have $|\mathcal{C}_o \cup \mathcal{C}_n|+1$ categories (plus the ``background"), and the prediction $\mathbf{p}$ is activated by softmax instead of sigmoid. BCE-based and CE-based loss are both commonly used. There is usually no obvious difference in performance between these two losses in standard object detection~\cite{redmon2018yolov3}. However, in CE loss, all categories would affect each other (due to ``softmax"). Thus, we argue that BCE-based loss is a better choice for the task handled in this work, as it helps to avoid conflicts conveniently.
\begin{table}[t]
\centering
\caption{{\color{highlight}Performance with pseudo annotations mined by our unified detector in Sec.~\ref{sec:cf_loss} ($\mathcal{\hat{G}}_{CC}$, $\mathcal{\hat{G}}_{CLC}$) and the Separate Detectors ($\mathcal{\hat{G}}_{CC,SDs}$, $\mathcal{\hat{G}}_{CLC,SDs}$) under setup C.}}
\resizebox{1.\textwidth}{!}{
\begin{tabular}{lllllll}
\toprule
\multicolumn{1}{l}{\color{highlight}{Method}} & \color{highlight}{$\mathcal{\hat{G}}$} & \multicolumn{2}{l}{\color{highlight}{VOC-subtest}} & \color{highlight}{} & \multicolumn{2}{l}{\color{highlight}{COCO-subval}} \\
\cmidrule{3-4}\cmidrule{6-7}\color{highlight}{} & & \color{highlight}{AP$_{50}$} & \color{highlight}{AP$_{75}$} & \color{highlight}{} & \color{highlight}{AP$_{50}$} & \color{highlight}{AP$_{75}$} \\
\midrule
\multicolumn{1}{l}{\color{highlight}{as fully labeled~\cite{sohn2020simple}}} & \color{highlight}{$\mathcal{\hat{G}}_{CC,SDs}$} & \color{highlight}{44.3 } & \color{highlight}{28.3 } & \color{highlight}{} & \color{highlight}{48.0 } & \color{highlight}{27.2 } \\
\color{highlight}{} & \color{highlight}{$\mathcal{\hat{G}}_{CC}$} & \color{highlight}{48.0 (+3.7) } & \color{highlight}{31.5 (+3.2)} & \color{highlight}{} & \color{highlight}{48.7 (+0.7)} & \color{highlight}{29.9 (+2.7)} \\
\multicolumn{1}{l}{\color{highlight}{confict-free}} & \color{highlight}{$\mathcal{\hat{G}}_{CC,SDs}$} & \color{highlight}{46.2 } & \color{highlight}{26.8 } & \color{highlight}{} & \color{highlight}{50.1 } & \color{highlight}{29.9 } \\
\color{highlight}{} & \color{highlight}{$\mathcal{\hat{G}}_{CC}$} & \color{highlight}{49.4 (+3.2)} & \color{highlight}{31.7 (+4.9)} & \color{highlight}{} & \color{highlight}{49.9 (-0.2)} & \color{highlight}{30.8 (+0.9)} \\
\multicolumn{1}{l}{\color{highlight}{safe negatives~\cite{zhao2020object}}} & \color{highlight}{$\mathcal{\hat{G}}_{CC,SDs}$} & \color{highlight}{45.3 } & \color{highlight}{28.2 } & \color{highlight}{} & \color{highlight}{49.7 } & \color{highlight}{28.2 } \\
\color{highlight}{} & \color{highlight}{$\mathcal{\hat{G}}_{CC}$} & \color{highlight}{49.0 (+3.7)} & \color{highlight}{33.3 (+5.1)} & \color{highlight}{} & \color{highlight}{49.9 (+0.2)} & \color{highlight}{31.5 (+3.3)} \\
\multicolumn{1}{l}{\color{highlight}{overlap-weighted}} & \color{highlight}{$\mathcal{\hat{G}}_{CC,SDs}$} & \color{highlight}{46.3 } & \color{highlight}{28.7 } & \color{highlight}{} & \color{highlight}{50.0 } & \color{highlight}{29.4 } \\
\color{highlight}{} & \color{highlight}{$\mathcal{\hat{G}}_{CC}$} & \color{highlight}{47.5 (+1.2)} & \color{highlight}{31.4 (+2.7)} & \color{highlight}{} & \color{highlight}{51.6 (+1.6)} & \color{highlight}{31.6 (+2.2)} \\
\multicolumn{1}{l}{\color{highlight}{overlap-weighted}} & \color{highlight}{$\mathcal{\hat{G}}_{CLC,SDs}$} & \color{highlight}{49.2 } & \color{highlight}{30.8 } & \color{highlight}{} & \color{highlight}{50.9 } & \color{highlight}{29.9 } \\
\color{highlight}{} & \color{highlight}{$\mathcal{\hat{G}}_{CLC}$} & \color{highlight}{49.0 (-0.2)} & \color{highlight}{33.0 (+2.2)} & \color{highlight}{} & \color{highlight}{52.2 (+1.3)} & \color{highlight}{31.7 (+1.8)} \\
\bottomrule
\end{tabular}%
}
\label{tab:sds}%
\end{table}%
{\color{highlight}
\subsection{Separate detectors for unlabeled ground-truth mining}
\label{sec:SDs}
We have shown that the conflict-free loss can lead to an acceptable detector in one training round, and that the retraining phase can significantly improve performance. Conflict-free loss is recommended if you prefer a simple one-round training approach. If one adopts the two-rounds training, what about using multiple detectors trained on separate datasets for unlabeled ground-truth mining?
We compare the performance of using pseudo annotations mined by the unified detector and the separate detectors under setup C. The results in Tab.~\ref{tab:sds} demonstrate the pseudo annotations mined by the unified model are superior to those mined by multiple separate detectors. Considering that training multiple separate detectors is more cumbersome, we also recommend the unified detector as an initial detector in two-rounds training method.
}
\subsection{More visualization results.}
Fig.~\ref{fig:vis2} illustrates more visualization results of the detectors on setup C, showing that our method achieves powerful and robust detectors with limited data.
\section{Conclusion}
In this paper, we aim at training a category-extended object detector with limited data. To achieve this goal, a powerful pipeline is proposed. The conflict-free loss is designed to eliminate possible conflicts during joint learning. To further improve performance, a retraining phase is employed. We adopt a Monte Carlo Dropout-based method to estimate localization confidence to unearth more accurate pseudo annotations. We explore several strategies for retraining with the pseudo annotations, and empirically exhibit that employing overlap-weighted negative samples during retraining leads to better performance. The experimental results on multiple settings demonstrate that the proposed method achieves a feasible category-extended object detector and outperforms other approaches.
\begin{figure}[tp]
\centering
\caption{Detection results on COCO-val and VOC-test. Red bounding boxes indicate objects of the 60 categories in COCO60-train, while green ones correspond to the 20 VOC categories.}
\includegraphics[height=0.6\textheight]{./vis2.pdf}
\label{fig:vis2}%
\end{figure}%
{\color{highlight} Despite the promising results, there is still a significant gap between models trained on incrementally labeled datasets and models trained on fully labeled datasets. Exploring unsupervised domain adaptation algorithms would benefit unlabeled ground-truth mining by mitigating the effects of domain gap; designing class-distribution aware retraining methods would be beneficial in alleviating the class imbalance problem, which may be exacerbated by pseudo annotations. For other tasks, the incrementally labeled datasets can also be encountered in multi-label classification, semantic / instance segmentation, and other modalities, which are worth exploring further.}
\section*{Acknowledgements}
This work is supported in part by the National Natural Science Foundation of China (62171248, 61972219), the R\&D Program of Shenzhen (JCYJ2018050815\\2204044, JCYJ20190813174403598, SGDX20190918101201696), the PCNL KEY project (PCL2021A07), and the Overseas Research Cooperation Fund of Tsinghua Shenzhen International Graduate School (HW2021013).
\begin{small}
\bibliographystyle{elsarticle-num}
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 5,763
|
StoryCorps 536: Little Kids, Big Problems
In a grown-up world, kids often have no choice but to go along with the course charted by the adults around them. This means that they're subject to grown-up problems, often in ways they can't understand until much later in their lives. In this episode, we hear what it's like to be a kid caught up in a tangle created by adults.
Released on June 22, 2018.
Like the music in this episode? Support the artists:
"Scalloped" by Blue Dot Sessions from the album Crab Shack
"Zak's Bedtime / More Wonderful Books" by Christopher Bowen from the album $9.99 (Original Motion Picture Soundtrack)
"Opus 26" by Dustin O'Halloran from the album Piano Solos, Vol 2
"Thinking Of You" by Gillicuddy from the album …Plays Guitar
"The Donut Song" by Burl Ives from the album Little White Duck and Other Children's Favorites
Newer Podcast Older Podcast
Michael Garofalo (MG): When you're a kid, the people you depend on most, the people you look to for safety and support, are adults. It's a grown-up world and most of the time, kids really have no choice but to go along with the course charted by the adults around them.
This means that kids are subject to grown-up problems often in ways they can't understand until much later in their lives.
Parents, teachers, caretakers of all kinds — they do their best to protect kids from the darker parts of the grown-up world, but sometimes the problems are so big that kids get swept up in them.
Roy Ebihara (RE): We saw the men were holding the oil torches coming across to where we lived. They were going to burn down everything.
James Hanover Thompson (JHT): I was nine years old; David was seven. The police car pulled up and they said, "We're taking y'all to jail."
Ruben Aguilar (RA): I grew up when that happened. From six years old, all of a sudden I felt like I was 15.
In this episode of the StoryCorps podcast from NPR, we'll hear what it's like to be a kid caught up in a tangle created by adults. What does it look like from the kid's point of view and how do they carry the things that happen to them into adulthood? I'm Michael Garofalo. We'll be back after a short break.
MG: Welcome back. We've got three StoryCorps interviews for you in this episode and they all come from people who as kids got caught up in big adult problems.
We'll start with Roy and Aiko [EYE–CO] Ebihara [EBB–IH–HARRA]
In 1942, President Roosevelt issued Executive Order 9066 which authorized the internment of Japanese-Americans, like Roy and Aiko.
Roy was 8 years old then and he begins by describing the atmosphere in his hometown of Clovis, New Mexico in the weeks before his family was interned.
RE: A sheriff and two plainclothesmen barged into the house and searched for what they called contrabands. They took out my brother's box camera, my father's shortwave radio, and they took the ax to it and chopped everything up. My father never protested, never said a word. He just stood there. And one night, vigilante groups formed in town. We saw the men were holding the oil torches coming across to where we lived. They were going to burn down everything.
The state patrol came roaring in and told us quickly to gather up what we can in pillowcases and whatever can fit into the trunk of the car. I remember my sister Kathy my sister Mary and my brother Bill and I, the four of us were squeezed into the back seat of the sedan, and we left in the darkness of the night. We were all crying, and we couldn't stop. It was just terrifying. And some months later, we were put into camps. Do you remember those times?
Aiko Ebihara (AE): I do remember getting on these Army trucks. I was probably 5 or 6 years old. I thought I was going on a vacation. Did you know that you were not going to return to your house again?
RE: No. I was 8 years old, and I really didn't understand what this all meant and how it would affect our family. I guess I felt we were guilty of something but what, I didn't know. When we were released, if we talked about some of these things, even at dinner table, we were silenced, especially by my father who said I don't want to hear that. So you and I maintain that silence of our parents.
AE: Yeah, we regret it very much, yeah.
RE: You know, I just feel that I want to go back and accept that pride, that pride of who we are.
MG: That's Roy and Aiko Ebihara. In 2014, Roy and his family received a formal public apology from the mayor of Clovis, NM.
Next — we'll hear from a Mexican-American who was deported in 1933.
It happened as part of a Mexican Repatriation Program, run by the U-S Government.
During the Great Depression, hundreds of thousands of people of Mexican descent were forcibly deported to Mexico without due process including many American citizens.
Ruben Aguilar (AG-ee-lahr) was born in the U-S, but was deported with his parents, who were undocumented. At the time, he was 6 years old.
Here, he tells his story to a friend, Bill Luna.
Bill Luna (BL): So you were born in Chicago?
RA: In Chicago.
BL: Now what happened when you were deported to Mexico?
RA: Well, when I was deported what I remember is the way that the agents crashed into the house. Okay people. Line up against the wall. We were put into the trucks, taken to the train station, and then shipped out.
I grew up when that happened. From six years old, all of a sudden I felt like I was 15.
BL: You hadn't been to Mexico before then?
RA: Never. I could speak fluent English, but not Spanish.
BL: Now, how did you return to the U.S.? How did you come back here?
RA: I was an American citizen. So in 1945, I was drafted into the Army.
My father explained to me, he says, "You got a little card from Chicago to join the United States Army. You're going back to your country."
So I took the bus to the United States. It stopped in Laredo before we take off for Chicago. And I asked the bus driver, "Where is the washroom, sir?" And he said, "Right around the corner." So I go around the corner and I see a big sign, No Mexicans or dogs are allowed. And I said, Welcome back.
You know, it's a funny thing, because when I talk about it, you know, it looks like yesterday. Those things, you never get rid of that.
BL: How do you want to be remembered Ruben?
RA: I want to be remembered as somebody got hurt by his country, came back to this country, and is going to die in his country.
MG: Ruben Aguilar and Bill Luna in Chicago.
We'll end with a story of how a kiss made civil rights history.
It happened in Monroe, North Carolina in 1958.
Two African-American children — James Hanover Thompson and his friend David Simpson were accused of kissing a girl, who was white.
They were arrested, and charged with rape and it became known around the world as ""The Kissing Case."
Over time it was largely forgotten and the Thompson family rarely talked about it, but they did record this StoryCorps interview.
James Hanover Thompson sat down with his younger brother, Dwight, to talk about what happened.
JHT: We were playing with some friends over in the white neighborhood, chasing spiders and wrestling and stuff like that. One of the little kids suggested that one of the little white girls give us a kiss on the jaw. The little girl gave me a peck on the cheek and then she kissed David on the cheek. So, we didn't think nothing of it. We were just little kids.
Dwight Thompson (DT): How old was you then?
JHT: I was nine years old; David was seven. Really didn't hardly know what a kiss was. And, so, we went on home like nothing happened, you know?
And the next day, the police car pulled up and they said, "We're taking y'all to jail." I didn't know what was going on. But when we got down to the police station we understand that they said that we had raped a little white girl.
They uh took us down in the bottom of the police station to a cell. And they had us handcuffed — they started beating us. They was beating us to our body, you know? They didn't beat us to the face where nobody could see it; they just punched us all in the stomach and back and legs. We was hollering and screaming. We thought they was gonna kill us.
And so we stayed in jail for six or seven days before our parents ever got a chance to see us. And they sent me to a psychologist. And I would go see that man. Every week, I had to go see him. He'd tell me, "They should have castrated y'all." I mean, it were just something.
DT: So how that make you feel now? I mean, how old are you now?
JHT: I'm 62 now.
DT: 62.
JHT: I still feel the hurt and the pain from it. And nobody never said, "Hey, look, I'm sorry what happened to y'all. It was wrong." And I always sit around and wonder if this hadn't happened to me, you know, what could I have turned out to be? Could I have been a doctor? Could I have went off to some college or some great school? It just destroyed our life.
MG: James Hanover Thompson with his younger brother, Dwight. Both James and the other child, David Simpson, were eventually pardoned by the governor of North Carolina.
That's all for this episode. These stories were produced by John White, Jasmyn Belcher Morris, and myself. This episode of the StoryCorps podcast was produced by David Herman, and edited by me. As always you can find out what music we used on our website, StoryCorps-dot-org. If you ever want to leave a voicemail for someone you hear on the show, we have a number for that: it's (301) 744 "TALK." That's 3-0-1 7-4-4 T-A-L-K.
Until next time, this has been the StoryCorps Podcast, and I'm Michael Garofalo. Thanks for listening.
The Echoes of Gunfire
This week, difficult conversations between people whose lives have been greatly impacted by one thing: guns.
Across The Great Divide
This week, conversations between strangers who don't necessarily agree on things politically, but come together to find moments of connection during their StoryCorps interviews.
All I Had Was Hope And You
This week, we'll hear from veterans and their families as they navigate difficult conversations around deployment and the transition back to civilian life.
On The Brink Of Separation
In this week's episode, you'll hear the difficult conversations that one family is having as they deal with the recurring threat of deportation.
|
{
"redpajama_set_name": "RedPajamaCommonCrawl"
}
| 6,094
|
\section{Introduction}
The so-called ``solar neutrino problem'' has been solved by the
results of the Sudbury Neutrino Observatory
(SNO)~\cite{Ahm01,SNO05}. The SNO experiment has shown strong
evidence that the neutrino-flux deficit measured in
charged-current interactions is a result of neutrino flavor
oscillations between electron-neutrino production in the Sun and
their detection on Earth. The flux measured in neutral-current
interactions of high-energy solar neutrinos is in general
agreement with the flux predicted by the standard solar model
(SSM, Refs.~\cite{BP04,Tur04}). The current slight discrepancy
between the flux predicted by the SSM and the neutral-current flux
measured by SNO~\cite{SNO05} may be significant or not depending
on the uncertainty of the flux prediction: a small but significant
deficit could, e.g., be evidence for oscillations into sterile
neutrinos. To that end it is essential to further reduce the
uncertainty of nuclear inputs to the SSM in order to refine its
predictions. One critical quantity is the $^7$Be(p,$\gamma$)$^8$B
cross section at solar energies since it is linearly related with
the high-energy solar neutrino flux stemming from $^8$B
$\beta$-decay.
In recent years, many attempts have been undertaken to measure
this cross section with high-precision in direct-proton-capture
measurements using radioactive $^7$Be targets
~\cite{Ham01,Str01,Bab03,Jun03}. Unfortunately, these results do
not yield a completely consistent picture: The earlier
measurements (Refs.~\cite{Ham01,Str01}) yield lower zero-energy
astrophysical $S$ factors, $S_{17}(0)$, around 19 eV b, whereas
the two more recent ones (Refs.~\cite{Bab03,Jun03}) obtain results
which are about 15\% higher. All (p,$\gamma$) data sets, however,
were found to be consistent with an energy dependence of $S_{17}$
as given by the cluster model of Descouvemont and Baye
~\cite{Des94}.
In view of their importance for astro- and elementary-particle
physics, it is desirable to cross-check these results by other,
indirect measurements that have different systematic errors. One
possibility is Coulomb dissociation (CD) of $^8$B in the
electromagnetic field of a high-$Z$ nucleus. Such measurements
have been performed at intermediate \cite{Kik98,Dav01} and high
energies \cite{Iwa99}. The present paper reports on a CD
experiment similar to that of Iwasa \textit{et al.} (GSI-1, Ref.
\cite{Iwa99}), but with an improved experimental technique: In
GSI-1, the incident $^8$B beam could not be tracked before the
target, whereas in the present run we could measure the angles
before and after the target with good precision. Preliminary
results of the present study have been published
earlier~\cite{Sch03}. For the present publication, the data were
re-analyzed, leading to slightly different results for the
lowest-energy data points.
Another indirect method to deduce $S_{17}(0)$ is to determine the
Asymptotic Normalization Coefficients (ANC) of the proton wave
functions bound in the $^7$Be potential. This method makes use of
the fact that due to the very low proton binding energy radiative
proton capture is extremely peripheral and $S_{17}(0)$ can be
calculated directly from the ANC. These ANC are determined from
low-energy proton-transfer or from proton-removal cross sections
\cite{Muk01,Tra04,Cor03}. A recent re-examination of ANC-results
for $^7$Be(p,$\gamma$) by Trache \textit{et al.} yielded a
relatively small central value of $S_{17}(0)= 18.7 \pm 1.9$ eV
b~\cite{Tra04}. Still, this value is in line with all published
values of $S_{17}(0)$ except for Ref.\cite{Jun03}.
It is important to compare the results from direct and indirect
methods to determine the astrophysical $S$-factors with each other
since the indirect methods could also be used to study
astrophysically interesting reactions between {\em unstable}
nuclei where direct-capture reactions cannot be applied. The
reaction $^7$Be(p,$\gamma$)$^8$B could be an ideal test case,
provided that the remaining inconsistencies in $S_{17}$(0) from
the different, direct and indirect, methods can be resolved.
In the present paper, we give a comprehensive report of the CD
experiment performed at GSI~\cite{Sch03} where we impinged a
secondary $^8$B beam with an incident energy of 254 $A$ MeV on a
$^{208}$Pb break up target. As already mentioned in our earlier
publication, we focus on a crucial question that must be answered
if one wants to use the CD method to derive a precise value for
$S_{17}$(0): the contribution of E2 multipolarity to CD of $^8$B.
One can calculate that E1 is the dominant multipolarity in CD as
well as in direct proton capture, but it is obvious that the
equivalent photon field from a high-$Z$ target nucleus seen by a
projectile at a few hundred MeV per nucleon contains also a strong
E2 component. Experimental limits for a possible E2 contribution
were extracted in the work of Kikuchi \textit{et al.}~\cite{Kik98}
and Iwasa \textit{et al.}~\cite{Iwa99}; both papers found
negligible E2 contributions. Recently, Davids \textit{et al.} have
reported positive experimental evidence for a finite E2
contribution in CD of $^8$B, mainly from the analysis of {\em
inclusive} longitudinal momentum ($p_{||}$) spectra of $^7$Be
fragments measured at 44 and 81 $A$ MeV \cite{Dav01,Dav03}; they
therefore subtracted a calculated E2 contribution from their
$S_{17}$ data. In order to resolve these discrepancies, we have
analyzed observables that should be particularly sensitive to
contributions from E2 multipolarity, namely the angular
correlations of the $^8$B-breakup particles, proton and $^7$Be.
\section{Model calculations}
Accurate model calculations of the CD of $^8$B are essential for
several reasons. From a practical point of view, the relatively
bad energy resolution of the CD method requires to simulate e.g.
the effects of cross talk between neighboring energy bins, of the
finite size and resolution of the tracking detectors etc. These
simulations require a CD event generator that is reasonably close
to reality so that the remaining differences between the measured
and simulated cross-section distributions can be attributed to the
$S_{17}$ factor. For this purpose we have used a simple potential
model of $^8$B.
Since the current experiment (like most other direct and indirect
studies of the $^7$Be(p,$\gamma$)$^8$B reaction) does not allow to
measure at solar energies, the data set has to be extrapolated
towards $E_{rel}=0$. For this purpose we have to use the most
sophisticated model available. We will show below that a cluster
model of $^8$B~\cite{Des04} seems to be suited best for a reliable
extrapolation.
\subsection{Nuclear-structure of $^8$B}
The isotope $^8$B has one of the lowest proton binding energies of
all particle-stable nuclei known in the chart of nuclides ($B_p=
137.5$ keV, \cite{Til04}). The relevant parts of the $^7$Be and
$^8$B level schemes are depicted in Fig.\ref{level_scheme}.
\begin{figure}[bt]
\includegraphics[width=7cm]{fig01_level_scheme.eps}
\caption{Level schemes of $^7$Be and $^8$B relevant for direct
proton capture into the ground state of $^8$B and for resonant
capture via the M1 resonances at 0.770 MeV and 2.32 MeV excitation
energy.} \label{level_scheme}
\end{figure}
The simplest model for $^8$B is that of a p-wave proton coupled to
an inert $^7$Be core with $I^{\pi}=3/2^{-}$ to form the $^8$B
ground state with $I^{\pi}=2^{+}$. We have adopted this simplified
single-particle model of $^8$B to calculate cross sections within
the simulations to be described below. Details of the model are
described in Ref.~\cite{Dav03}. The proton is bound in a
Woods-Saxon potential with radius parameter $r_0 = 1.25$ fm and
diffuseness $a = 0.65$ fm. As usual, the p-wave potential depth
has been adjusted to match the $^8$B proton binding energy, this
yields a depth of 43.183 MeV. The s-, d- and f-wave potentials
have been adjusted to reproduce the s-wave scattering lengths of
the mirror $^7$Li+n reaction \cite{Koe83}, this yields $V_1 =
43.857$ MeV for channel spin $S$=1 and $V_2 = 52.597$ MeV for
channel spin $S$=2. We note that we obtain for the dominant
channel spin $S$=2 an s-wave scattering length for $^7$Be+p of
$a_{02}^{theo} = -8$ fm which agrees well with the recently
measured value of $a_{02}^{exp} = -7 \pm 3$ fm (Angulo \textit{et
al.}~\cite{Ang03}).
With this model we obtain astrophysical $S$-factors as a function
of the proton-$^7$Be relative energy, $E_{rel}$, as shown in
Fig.\ref{s17_pg_theo}. The non-resonant direct capture into the
$^8$B ground state proceeds mainly via s- and d-wave captures and
E1 $\gamma$-emission. Capture of p- and f-waves followed by E2
emission plays an insignificant role, in particular at solar
energies. The resonant component proceeds through the $1^+$
resonance at 770 kev (632 keV above threshold) which decays mainly
by M1 emission and is limited essentially to a narrow region
around the resonance energy, with minor but finite contributions
at relative energies above the resonance. The M1-resonance cross
section has been obtained from experimental data~\cite{Fil83}, it
is not contained in the model of Ref.~\cite{Dav03}. We have
ignored entirely the high-lying M1 resonance at 2.32 MeV since it
cannot be seen in our high-energy CD experiment, due to both, the
small cross section in CD and its large width.
The potential model of $^8$B sketched above ignores the well known
cluster structure of $^7$Be (see e.g. Ref.~\cite{Nun97}).
Descouvemont and Baye~\cite{Des94} have therefore applied a model
where $^8$B is assumed to consist of e.g. p+($^7$Be=$^3$He+$^4$He)
or $^3$He+($^5$Li=p+$^4$He) three-cluster structures, including
excited states of the clusters. Later, this model was slightly
refined~\cite{Des04} by allowing also for variations in the
cluster separation and by using different effective
nucleon-nucleon interactions; the results were found to be largely
unchanged. Our potential model can be viewed as a simplification
of the cluster model where the cluster distance is artificially
set to zero; the spectroscopic information available for $^7$Be
points, however, to a finite cluster distance of $\approx$3.5
fm~\cite{Des04}.
For computational simplicity, we will use the potential model to
simulate the differential observables in our experiment and come
back to the cluster model at the end of this article where we
discuss the extrapolation of $S_{17}$ towards zero relative
energy.
\begin{figure}[bt]
\hspace{-1.0cm}
\includegraphics[width=9.0cm]{fig02_s17_theo.eps}
\caption{(Color online)
Theoretical $S_{17}$ factors from a simple potential model of $^8$B
and their decomposition
into contributions from various partial waves.
}
\label{s17_pg_theo}
\end{figure}
\subsection{Coulomb dissociation of $^8$B}
As proposed by Baur \textit{et al.} \cite{Bau86}, CD can be used
favorably to measure radiative-capture cross sections by making
use of the strong flux of equivalent photons originating from a
heavy target nucleus as seen by a fast-moving projectile, which
replaces the presently insufficient intensity of available
real-photon sources. Assuming first-order perturbation theory for
the electromagnetic excitation process, CD cross sections can be
converted directly to photo-dissociation cross sections. The
latter are related to the astrophysically relevant
radiative-capture cross sections by the principle of detailed
balance. It is obvious that the indirect method of CD needs
theoretical input in the conversion process.
There are several sources for complications. They can be
identified and minimized by selecting appropriate observables and
kinematical conditions in the experiment.
\begin{enumerate}
\item Several multipolarities (E1, E2, M1, \dots) with different
weights contribute in radiative capture reactions and Coulomb
breakup. In principle, they can be disentangled by studying
angular distributions in CD, preferably in the center-of-mass
system of the excited nucleus. Their relative strengths depend on
projectile energy and scattering angle. \item An exchange of more
than one photon (``higher-order effects'') destroys the direct
relation between the CD cross section and the photo-dissociation
cross section. High projectile energies and large impact
parameters reduce this effect. \item The nuclear interaction
between projectile and target induces nuclear breakup and
absorption. It becomes relevant for small impact parameters.
\end{enumerate}
These are general features of CD that can be included in the
theoretical description of the breakup mechanism in various
approximations which lead to corrections of the simple
pure-Coulomb first-order approach.
\subsubsection{Semiclassical calculations}
The Coulomb-breakup mechanism can be described both in fully
quantal approaches and semi-classical models \cite{Ald56}. In the
latter case the projectile is assumed to move on a classical
trajectory with respect to the target. In our case, we use the
semi-classical model in first-order perturbation theory (PT) to
describe the CD of $^8$B in the Coulomb field of $^{208}$Pb, as
described in more detail elsewhere \cite{Ber94,Typ02,Dav03}. The
excitation amplitude is calculated in the relativistic approach
assuming a straight-line trajectory but correcting the excitation
functions for the deflection in the Coulomb field of the target
\cite{Win79}. This is appropriate at the high incident energy used
in the present experiment (254 $A$ MeV) and justified {\it a
posteriori} by the good agreement with the measured angular
distributions.
In addition to CD, nuclear overlap of $^8$B and $^{208}$Pb has to
be considered. This will mainly take flux out of the $^7$Be+p exit
channel; feeding this channel by nuclear interaction has been
calculated to be of minor importance by Bertulani and Gai
\cite{Ber98}. In order to take nuclear absorption into account we
modified the relativistic Coulomb-excitation functions by
multiplying them with a correction factor as described in Ref.
\cite{Typ02}. This factor is derived from an eikonal approximation
of the excitation functions taking both Coulomb and nuclear
potentials into account. In the present case we assume a diffuse
absorptive nuclear potential with a depth of 20 MeV and a radius
of 9.91 fm, i.e.\ the sum of the projectile and target radii. As
we will see below, this choice reproduces well the integral
scattering-angle distribution.
\subsubsection{Dynamical calculations}
Higher-order effects from the exchange of more than one photon can
be considered in semi-classical calculations that study the
time-evolution of the projectile system during the scattering. As
compared to a first-order calculation, the momenta of the outgoing
particles are modified in the Coulomb field of the target leading
to a distortion of relative-energy and angular-momentum
distributions. Esbensen \textit{et al.}~\cite{Esb04,Esb05} have
proposed that discrepancies between the results from
radiative-capture and from CD studies of the
$^7$Be(p,$\gamma$)$^8$B reaction are due to deficiencies of the
method how to evaluate $S_{17}$ from CD cross sections by using
first-order PT. They point out that a full dynamical calculation
of CD, if compared to a first-order PT calculation, will lead to
an increased $S_{17}$ factor at low $E_{rel}$ and a reduced one at
high $E_{rel}$, thus producing a smaller slope of $S_{17}$ {\it
vs.} $E_{rel}$ and a better agreement between the results from the
two methods. However, the amount of this modification depends on
the assumed E2 strength and thus is model dependent. Recently,
fully quantal calculations became available that consider the
post-acceleration of the fragments in the Coulomb field of the
target. In contrast to dynamical calculations in the semiclassical
approach, they predict an {\em increase} of the cross section at
low relative energies \cite{Alt05}. More theoretical work is
required in order to obtain a consistent picture of higher-order
effects.
To follow the suggestions of Esbensen \textit{et al.}, we have
also performed dynamical calculations of the CD of $^8$B at 254
$A$ MeV following the approach as described in \cite{Dav03} for
lower projectile energies assuming the simple potential model for
${}^{8}$B. The ${}^{8}$B nucleus moves on a Coulomb trajectory
taking the deflection into account. E1 and E2 multipoles were
considered in the standard far-field approximation with the full
strength as predicted by the model.
In both theoretical approaches, triple-differential cross sections
for the CD of ${}^{8}$B are obtained. These distributions of
observables cannot be compared directly to the measured data, but
have to be folded with the respective experimental resolutions. To
this end, the cross sections were converted to statistically
distributed ``event'' distributions from both (PT and dynamical)
calculations and run through our experimental filter as will be
described in more detail below.
\section{Experimental procedures}
Several other CD studies of $^8$B breakup \cite{Kik98,Dav01} have
used intermediate energies between 46 and 83 $A$ MeV as available
from cyclotron-based radioactive-beam facilities. At GSI, the 18
Tm SIS-18 synchrotron allows to go much higher in incident energy.
We have chosen a $^8$B incident energy of 254 $A$ MeV limited by
the maximum bending power of the KaoS spectrometer used for
determining the momenta of the break-up particles, p and $^7$Be.
In the following we will describe in detail the preparation and
identification of the secondary beam as well as the experimental
equipment used to measure the breakup.
\subsection{Preparation and properties of the $^8$B beam}
The $^8$B secondary beam was produced at the SIS/FRS
radioactive-beam facility at GSI~\cite{Gei92} by fragmenting a 350
$A$ MeV $^{12}$C beam in a 8 g/cm$^2$ Be target and separating it
from contaminant ions in a 1.4 g/cm$^2$ wedge-shaped Al degrader
placed in the FRS intermediate focal plane.
Typical $^8$B beam intensities in front of KaoS were $5 \times
10^4$ per 4 sec spill; the only contaminant consisted of about
20\% $^7$Be ions which could be identified event by event with the
help of a time-of-flight measurement. For this purpose a 3 mm
thick plastic scintillator detector was installed in the transfer
line between FRS and KaoS, about 85 m upstream from the breakup
target, to serve as a time-of-flight (ToF) start detector.
Positions and angles of the secondary beam incident on the Pb
breakup target were measured with the help of two parallel-plate
avalanche counters (PPAC) located at 308.5 cm and 71 cm upstream
from the target, respectively. The detectors, which were designed
and built at RIKEN \cite{Kum01}, had areas of $10 \times
10\,\text{cm}^2$ and allowed to track the incident $^8$B beam with
about 99\% efficiency and with position and angular resolutions of
1.3 mm and 1 mrad, respectively. In addition, they provided a ToF
stop signal with a resolution of 1.2 ns (FWHM). Fig.\ref{de_tof}
shows a two-dimensional plot of the ToF between the scintillator
detector and the second PPAC detector in front of the target.
\begin{figure}[bt]
\includegraphics[width=9cm]{fig03_de_tof.eps} \vspace*{-1.0cm}
\caption{(Color online)
Identification of primary and secondary fragments by
energy loss and time of flight. } \label{de_tof}
\end{figure}
One can see that a ToF measurement alone is sufficient to separate
the $^8$B beam from contaminants on an event-by-event basis.
\subsection{Detection of break-up fragments}
A schematic view of the experimental setup to detect the breakup
of $^8$B in semi-complete kinematics (i.e. without detecting
coincident $\gamma$-rays) at the KaoS spectrometer at GSI is shown
in Fig.\ref{kaos_setup}. Apart from the PPAC tracking detectors
mentioned above, it consisted of (i) the $^{208}$Pb breakup
target; (ii) two pairs of Si strip detectors; (iii) the magnets of
the KaoS spectrometer; (iv) two large-area multi-wire proportional
chambers (MWPC); (v) a ToF wall serving as a trigger detector. The
individual components will be discussed in detail below.
\begin{figure*}[bt]
\includegraphics[width=17.0cm]{fig04_kaos_setup.eps}
\caption{(Color online)
Artist's view of the experimental setup. Shown
schematically are the beam-tracking detectors (PPAC) in front of
and the fragment-tracking Si strip detectors (SSD) behind the
Coulomb-breakup target. Proton and $^7$Be positions in the focal
plane of the KaoS magnetic spectrometer are determined by
large-area multi-wire chambers (MWPC) followed by a
scintillator-paddle wall for trigger purposes.} \label{kaos_setup}
\end{figure*}
\subsubsection{Fragment tracking: Si strip detectors}
Downstream from the Pb target (which consisted of 52 mg/cm$^2$
$^{208}$Pb enriched to 99.0$\pm$0.1\% and had an area of 24 mm in
height times 36 mm in width), the angles and positions as well as
the energy losses of the outgoing particles were measured with two
pairs of single-sided Si strip detectors (SSD). These detectors
(300 $\mu$m thick, 100 $\mu$m pitch) were located at distances of
about 15 cm and 30 cm downstream from the target.
Fig.\ref{ssd_layout} shows schematically the layout of the SSD
detector array.
\begin{figure}[bt]
\vspace{0.8cm}
\includegraphics[width=7.5cm]{fig05_ssd_layout.eps}
\caption{(Color online)
Schematic view of the geometrical arrangment of the four
layers of single-sided Si strip detectors yielding the breakup
particles' trajectories directly after the target. }
\label{ssd_layout}
\end{figure}
The vacuum of the beamline housing the PPAC's, the target and the
SSD detectors was separated downstream from ambient air by a
stainless-steel window of 50 $\mu$m thickness.
\subsubsection{The KaoS magnetic spectrometer}
The KaoS magnetic spectrometer \cite{Sen93} consisted of a
large-aperture quadrupole and a horizontally-focussing dipole
magnet. The ratio between the smallest and largest momentum
accepted by KaoS amounted to about two, making KaoS an ideal
instrument to detect breakup of neutron-deficient nuclei into a
proton with $A/Z$=1 and an ion with $A/Z \approx 2$. Prior to our
measurement, the magnetic field of the KaoS dipole had been mapped
in three dimensions to obtain an empirical field map; this map was
then used to simulate the passage of charged particles through the
magnet using the code GEANT-3 \cite{geant}. To avoid multiple
scattering of the fragments in air, the chamber inside the
quadrupole and dipole magnets was filled with He gas at 1 bar
pressure, separated from the ambient air by thin He-tight foils.
\subsubsection{Fragment tracking: Multi-wire chambers and trigger detectors}
Behind the magnets, two large-area MWPC were installed as close to
the focal plane as possible. One chamber, with horizontal and
vertical dimensions of 60 cm and 40 cm, respectively, detected the
positions of protons behind KaoS. The other one, 120 cm wide and
60 cm high, was set to detect the $^8$B non-interacting beam and
the $^7$Be fragments. The separation of the position measurements
of protons and the heavy ions allowed to optimize each detector
voltage for optimum detection efficiency.
Behind the focal plane and parallel to it, a plastic-scintillator
wall with 30 elements (each 7 cm wide and 2 cm thick) was
installed and used for trigger purposes. The wall was subdivided
into two sections covering the respective MWPC in front of them.
Coincident signals in the left hand (proton) part and in the right
hand (ion) part of the wall indicated a breakup event (``breakup''
trigger). Singles hits in the right hand section were interpreted
as ``beam'' triggers and recorded with a down-scale factor of
1000.
\subsection{Monte-Carlo simulations}
Monte-Carlo simulations of the Coulomb breakup of $^8$B and the
detection of the breakup products are an indispensable part of the
present experiment. They were essential in planning the
experiment, helped to estimate the energy resolution and detection
efficiencies, and were instrumental in determining the proton and
$^7$Be momenta from the measured positions. As a tool for these
Monte-Carlo simulations the program package GEANT-3~\cite{geant}
was used.
The Monte-Carlo simulations started with an event generator that
simulated CD of $^8$B on $^{208}$Pb in first-order perturbation
theory or via a fully dynamical calculation by the theoretical
approaches mentioned above (subsect.II.B). Technically, the event
generator produced statistically-distributed ensembles of 500,000
CD-``events'' each that were used as input to a GEANT simulation
of the passage of each breakup particle through the Pb target, the
SSD detectors, the beamline exit window, the He-filled interior of
the magnets and the air behind KaoS before hitting the MWPC
volumes. At the target, the emittance of the $^8$B as measured
with the PPAC's was imposed, the momentum spread was assumed to be
the nominal FRS momentum acceptance, $\Delta p/p = \pm 1 \%$.
Momenta of each particle type (p,$^7$Be,$^8$B) were obtained from
two position measurements in the SSD and one position measurement
in the respective MWPC. To calculate each particle type's
momentum, a 36-term polynomial expression was derived; its
parameters were obtained in a GEANT simulation by sending
particles with known momenta (covering evenly the range of
relevant momenta) through the setup and fitting the momenta as a
function of the positions by varying the 36 polynomial parameters.
In a similar way, the invariant-mass resolution of the experiment
could be obtained by simulating breakup events of known invariant
mass and reconstructing this quantity from the simulated
positions. The top panel in Fig.\ref{minv_effic} shows the
$E_{rel}$ resolution ($1 \sigma$ width) as a function of the
p-$^7$Be relative energy, $E_{rel}$, as determined from the
simulation.
\begin{figure}[bt]
\includegraphics[width=7.5cm]{fig06_minv_effic.eps}
\vspace*{-1.0cm}
\caption{(Color online)
Calculated properties of the experimental setup from
GEANT simulations. Top: $E_{rel}$ resolution (squares). The upper
set of points (circles) has been determined from the experimental
data. \\
Bottom: Geometrical efficiency (upper set of data points).
In addition, we show in the lower set of data points also the
total efficiency including all analysis conditions. }
\label{minv_effic}
\end{figure}
The efficiency of our setup at high $E_{rel}$ is mainly given by
the finite sizes of the SSD and MWPC detectors. Below the maximum
around 0.5 to 1 MeV, the efficiency drops due to overlap of the
proton- and $^7$Be hit patterns in the SSD leading to apparent
multiplicity 1 instead of 2. Numerical values of the efficiency
could be obtained by simulating the full set of 500,000 CD events
with and without the above conditions and plotting the ratios of
these numbers for different, evenly spaced $E_{rel}$ bins. This
distribution is shown in the lower panel of Fig.\ref{minv_effic}.
The upper set of data points (circles) is obtained by requiring
two separated p-Be hits inside all detector volumes. The lower set
of data points (squares) is obtained by taking into account the
intrinsic detector and trigger efficiencies and applying all
analysis conditions, see subsect.~IV.B below. It can be seen that
the major part of the $E_{rel}$ distribution is covered with high
total efficiency (about 30-40\%). It should be noted that this
curve is insufficient to correct measured data for efficiency: the
total efficiency is a multi-dimensional function of both the
original and the smeared-out (by the experimental resolution)
angles and momenta of both particles. Therefore, we pass the
theoretical ``events'' through the experimental filter and compare
the results to the same quantities derived from the data.
\section{Data reduction and results}
The experiment described in the present paper recorded events
from three different sources:\\
(i) breakup events originating in the Pb target;\\
(ii) down-scaled beam particles;\\
(iii) background from a variety of sources (e.g. cosmic rays).\\
Though event classes (i) and (ii) are mainly correlated with a
corresponding trigger type (``breakup'' trigger for class (i),
``beam'' trigger for class (ii)) we have checked if by chance the
trigger types and event classes were mixed in rare cases, and have
corrected for that.
In the following, we first show how the total number of incident
$^8$B projectiles is obtained from the ``beam'' trigger events. We
then explain how the breakup events originating in the Pb target
were identified.
\subsection{Total number of $^8$B projectiles}
The absolute number of $^8$B ions impinging on the $^{208}$Pb
breakup target needs to be known to determine absolute cross
sections. To this end, ``beam'' trigger events were analyzed to
select those that correspond to $^8$B in the $\Delta E$-ToF plot,
Fig.~\ref{de_tof}. A 3$\sigma$ window around the $^8$B energy-loss
peak in each SSD was chosen. To convert the integrated number in
this spectrum to the total number of incident $^8$B ions, the
down-scale factor of the ``beam'' trigger (10$^3$) and the
efficiencies for detecting $^8$B ions in the ToF detectors (3 mm
scintillator, PPAC detectors) as well as in the $\Delta E$ (SSD)
detectors have to be taken into account. These numbers were
derived from sets of linear equations containing the coincidence
count rates and the respective efficiencies. A small number
(0.48\%) of $^8$B ions was found for the ``breakup'' trigger
condition due to random-noise coincidences with the left
(``proton'') part of the plastic-scintillator wall. Together with
the total from the ``beam'' trigger condition we obtain a total of
$(4.15 \pm 0.03) 10^9$ $^8$B ions impinging on the breakup target.
\begin{figure}[bt]
\vspace{-0.5cm}
\includegraphics[width=9.0cm]{fig07_de_ssd1.eps}
\caption{(Color online)
Energy-deposition of the incident $^8$B ions and of the
breakup fragments in the third Si strip detector. The outermost
contour corresponds to all events. The intermediate thin contour
depicts events where an incident $^8$B particle is correlated with
multiplicity $m = 2$ in each SSD. The innermost filled histograms
are obtained by requiring a $\Delta E$-cut on $^7$Be in each SSD
plus a p-$^7$Be vertex inside the target volume (see text). }
\label{de_ssd}
\end{figure}
\subsection{Identification and tracking of breakup products}
The coincident p and $^7$Be signals resulting from breakup in the
$^{208}$Pb target were identified among the class (i) events
(``breakup'' trigger) in several successive steps:
\begin{enumerate}
\item The $\Delta E$-ToF condition was applied to select only
incident $^8$B ions (see above). \item A multiplicity of $m \geq
2$ in each SSD was required. That meant that at least one empty
strip was found between two adjacent hit clusters. \item A
3$\sigma$-window around the $\Delta E$ peak corresponding to the
energy loss of $^7$Be in each SSD selected $^7$Be as one of the
reaction products. \item The coincident protons were found among
all events with $\approx 50$ keV $ < \Delta E < 500$ keV in each
SSD where the trajectories had a closest distance to the
coincident $^7$Be trajectory inside a volume given in x and y by
the size of the target ($\pm$18 mm in x- and $\pm$12 mm in
y-direction) and in z-direction (along the beam axis) of $\pm$25
mm around the target (located at $z=0$). The low-energy cutoff was
chosen individually for each detector; the number of protons below
this cutoff was estimated by fitting a Gaussian to the low-energy
tail of the Landau distribution.
\end{enumerate}
The inclusive $\Delta E$ spectra resulting from conditions 1 and 2
above are shown by the intermediate thin histogram in
Fig.\ref{de_ssd}, whereas conditions 3 and 4 lead to the innermost
(filled) histograms in Fig.\ref{de_ssd}. This procedure removed
all breakup events in layers of matter other than the target and
led to a practically background-free measurement.
The breakup protons loose only about 200 keV in the 300 $\mu$m
thick SSD. Nevertheless, after imposing the vertex condition, the
energy-deposition signals of protons in the SSD are clearly
resolved from noise.
\subsection{Invariant-mass reconstruction}
\subsubsection{Proton-ion opening angles}
The p-$^7$Be relative energy, $E_{rel}$, is derived from the total
energies, $E$, of the particles, their 3-momenta, $p$, and the
p-Be opening angle, $\theta_{17}$ according to
\begin{equation}
E_{rel} = \sqrt{(E_{Be}+E_p)^2 -p_{Be}^2 -p_p^2 -2 p_{Be} p_p
\cos(\theta_{17})}. \label{eq1}
\end{equation}
Whereas the proton and $^7$Be momenta can be obtained only from
the rather complicated momentum reconstruction procedure described
in the next subsection, the p-$^7$Be opening angle, $\theta_{17}$,
can be determined directly from constructing the vectors
connecting the breakup vertex in the target with the corresponding
hit positions in the SSD. Since protons fire only a single strip
in each SSD, their positions are given by the strip centroid, and
the variances of these positions - assuming that the hits are
evenly distributed over the strip width - by the the strip pitch,
100 $\mu$m, divided by $\sqrt{12}$. In contrast, the larger energy
deposits by the $^7$Be ions produce broader hit patterns in our
setup, with rather large fluctuations of the widths.
To reconstruct a breakup event, the p and $^7$Be hits in each SSD
have to be separated by at least one empty strip. Since this
affects the efficiency for identifying a breakup event for low
$E_{rel}$, we have to make sure that the GEANT simulation
accurately reproduces this efficiency. This has been achieved by
introducing a weighting function in GEANT that gradually increases
the efficiency for detecting two separated hits from zero to one
over the appropriate distance for each detector so that
experimental and simulated distance distributions look alike. In
Fig.\ref{p_Be_dist} we plot the horizontal-distance distribution
between
\begin{figure}[bt]
\vspace{-0.5cm}
\includegraphics[width=9cm]{fig08_p_be_dist.eps}
\vspace*{-1.2cm}
\caption{(Color online)
Horizontal (x-) distances between proton and $^7$Be hits
in the first Si strip detector. The thin (red) histogram shows the
distribution of experimental distances, the thick (blue) one those
from the present GEANT simulation. The dashed (green) histogram
shows the GEANT simulation that was used to evaluate our previous
results \cite{Sch03}. } \label{p_Be_dist}
\end{figure}
proton and $^7$Be hits in the first SSD. One can observe that
experiment and simulation yield very similar distributions. It
should be emphasized that in our earlier data analysis a step
function was assumed for this efficiency that jumped from zero to
full efficiency at a fixed distance of 0.4 mm in each SSD. This is
visualized by the dashed histogram in Fig.\ref{p_Be_dist}; it
clearly shows that we overestimated the GEANT detection efficiency
for small $E_{rel}$ in our previous paper~\cite{Sch03}. As we will
show below, this leads to slightly larger cross sections at low
$E_{rel}$ compared to those of Ref.~\cite{Sch03}.
The validity of this procedure can be checked immediately
by inspecting the
integral distribution of p-$^7$Be opening angles, $\theta_{17}$,
both in experiment and
in simulation. These distributions are shown in Fig.\ref{theta17}.
The agreement is excellent.
\begin{figure}[hb]
\vspace{-1.0cm}
\includegraphics[width=6cm]{fig09_theta17.eps}
\vspace*{-0.5cm}
\caption{(Color online)
Distribution of opening angles between proton and
$^7$Be, $\theta_{17}$. Data points are shown by the symbols, the
histogram indicates the GEANT simulation. } \label{theta17}
\end{figure}
\subsubsection{Momentum reconstruction of the fragments}
As mentioned already above, momenta for each particle
(p,$^7$Be,$^8$B) where calculated from two position determinations
in front of KaoS (in the SSD) and from another position
determination behind KaoS (in the MWPC). These 6 coordinates were
converted to momenta using three sets of 36-term polynomial
expressions, one for each ion. By combining event by event the
longitudinal momenta of p and $^7$Be, one can check the accuracy
of the momentum reconstruction; its width is a measure how well
angular straggling effects are treated in the GEANT simulation.
The comparison shows that the simulated momentum widths are more
narrow by 20\% compared to experiment. Therefore, the simulated
$\sigma_{Erel}$ values shown in Fig.\ref{minv_effic} (squares)
where uniformly increased by 20\% (circles).
\subsection{Angular distributions}
In the following we will present some angular distributions that
can be shown to be sensitive to an E2 amplitude in CD.
Fig.~\ref{vector_diagram} shows the coordinate systems used. With
$^8$B$^*$ we denote the momentum vector of the (excited) $^8$B
prior to breakup, as it is reconstructed from the measured proton
and $^7$Be momentum vectors. The angle $\theta_8$ is the
laboratory scattering angle of $^8$B$^*$ relative to the incoming
$^8$B beam. The polar angles, $\theta_{cm}$, and the azimuthal
angles, $\phi_{cm}$, of the breakup protons are measured in the
rest frame of the $^8$B* system. In the same way, one can
calculate the transverse proton momentum vector in the reaction
plane ($p_t^{in}$).
\begin{figure}[bt]
\vspace{0.8cm}
\includegraphics[height=4cm]{fig10_vector.eps}
\caption{(Color online)
Vector diagram showing the definitions of the angles
$\theta_{cm}$ and $\phi_{cm}$ as well as the proton in-plane
transverse momentum, $p_t^{in}$, in the frame of the $^{8}$B$^*$
system.} \label{vector_diagram}
\end{figure}
\subsubsection{Comparison to perturbation-theory calculations}
Fig.\ref{theta8} shows the $\theta_8$ distribution, in comparison
to two model calculations using first-order PT as discussed in
subsect.~II.B. The full histogram denoting pure E1 multipolarity
follows the data points very well, even to very large angles. The
dashed histogram, where both E1 and E2 with their full theoretical
strengths were assumed, deviates from the data points markedly for
$\theta_8$ values above about 0.7 degrees. Note that the
theoretical histograms were folded with the experimental response.
We conclude that even the $\theta_8$ distribution already
indicates E1 dominance, in a similar way as demonstrated in the
paper by Kikuchi \textit{et al.} \cite{Kik98}.
\begin{figure}[bt]
\includegraphics[width=8.0cm]{fig11_theta8.eps} \vspace{-1.0cm}
\caption{(Color online)
Scattering angle $\theta_8$ of the excited $^8$B prior to
breakup, as reconstructed from the proton and $^7$Be vectors. The
full histogram has been calculated in first-order perturbation
theory assuming pure E1 multipolarity, the dashed one assuming
E1+E2 multipolarity.} \label{theta8}
\end{figure}
We present in Fig.~\ref{ptin} the distribution
of $p_t^{in}$ for three different upper limits in $\theta_8$,
0.62$^{\circ}$, 1.0$^{\circ}$, and 2.5$^{\circ}$.
In classical Rutherford scattering, this corresponds to impact parameters
of 30 fm, 18.5 fm, and 7 fm, respectively.
Relative energies between p and $^7$Be
up to 1.5 MeV were selected.
\begin{figure}[t]
\vspace*{-8mm}
\includegraphics[width=9.0cm]{fig12_ptin.eps}
\vspace*{-0.8cm}
\caption{(Color online) In-plane transverse momenta, $p_t^{in}$,
of the breakup protons for three different cuts in $\theta_8$. The
theoretical curves (full red lines: E1 multipolarity, dashed blue
lines: E1+E2 multipolarity) have been calculated in first-order
perturbation theory. They were normalized individually to the data
sets in each frame. } \label{ptin}
\end{figure}
The experimental data for all three $\theta_8$-cuts can be
reproduced well by a PT calculation that includes only E1
multipolarity (full histograms in Fig.~\ref{ptin}, the theoretical
curves were normalized individually to the data sets). If
E1-plus-E2 multipolarity is used in the PT calculation, the
different impact-parameter dependences of E1 and E2 multipolarity
lead to markedly different shapes for the different
$\theta_8$-cuts (dashed histograms in Fig.\ref{ptin}). In
particular for large values of $\theta_8$, the latter
distributions show a large asymmetry with respect to $p_{t}^{in} =
0$ that is in clear disagreement with our data points.
By comparing Fig.\ref{ptin} with similar plots in our earlier
Letter (Fig.2 of Ref.\cite{Sch03}) one can see the improvement in
the GEANT simulation which was achieved by the modified
prescription for the p-Be hit resolution (see subsect.~IV.3.1).
The dips near $p_t^{in} \approx 0$ in the theoretical
distributions are now much closer to the experimental ones (though
small residual discrepancies are still visible in the rightmost
panel).
Fig.\ref{theta} depicts the experimental
$\theta_{cm}$ distributions for three different $E_{rel}$ bins,
as indicated in the figure. A ``safe'' $\theta_8$ limit of 1$^{\circ}$
was chosen.
\begin{figure}[bt]
\vspace{-0.8cm}
\includegraphics[width=9.0cm]{fig13_theta.eps}
\vspace*{-1.0cm}
\caption{(Color online)
Top: Experimental proton polar angle ($\theta_{c.m.}$)
distributions for three different bins of the p-$^7$Be relative
energy, $E_{rel}$. The full red curves denote a first-order
perturbation-theory calculation for E1 multipolarity, the dashed
blue ones for E1+E2. All theoretical curves were individually
normalized to the data points in each frame. Bottom: the same data
compared to dynamical calculations, again for E1 (full curves) and
for E1+E2 (dashed curves) multipolarities (see text for details).
} \label{theta}
\end{figure}
As expected, these distributions are mostly isotropic at low
$E_{rel}$ (indicative of $s$-waves) and become increasingly
anisotropic for larger values (contributions from higher orbital
angular momenta). As in Fig.\ref{ptin}, also for the $\theta_{cm}$
distributions the calculations for pure E1 multipolarity fit all
spectra well; inclusion of an E2 component may lead to a slightly
better fit at low $E_{rel}$, but diverges clearly for the
large-$E_{rel}$ bin where E2 should play a major role. The
calculations with a dynamical model will be discussed below.
\subsubsection{Comparison to dynamical calculations}
As mentioned above, Esbensen \textit{et al.}~\cite{Esb04,Esb05}
suggested that dynamical calculations are required to properly
describe CD and to evaluate $S_{17}$ from the measured CD cross
sections. A sensitive test if such a theory describes the
experimental data better than first-order PT calculations is given
by comparing the dynamical predictions (using the model described
in subsect.~II.B.2) to the same angular distributions (bottom part
in Fig.\ref{theta}). In all three frames shown, our E1-only
dynamical calculations do not agree well with the data points.
Dynamical calculations with E1+E2 seem to introduce a slight
improvement as long as the effect of E2 multipolarity is small,
but a major discrepancy shows up when E2 should have a stronger
influence (rightmost lower panel in Fig.\ref{theta}).
In general, one would expect that the more complete description of
the Coulomb breakup within the semiclassical dynamical approach
leads to a better agreement with the experimental data than the
simpler perturbative treatment. However, in the dynamical
calculation more model parameters that are not really constrained
have to be specified than in the first-order approach. E.g., the
results for the angular distribution in the dynamical calculation
depend crucially on the assumed E2 strength, i.e.\ there is a
considerable model dependence. With sufficicently precise
experimental data it would be possible to determine this strength
in a fitting procedure but this requires extensive calculations.
Additionally, one has to keep in mind that a full quantal
treatment of the breakup process could lead to different results.
We conclude that within the limits of our experimental conditions
the simplest model (first-order PT with E1 multipolarity only)
still gives the best agreement with the measured center-of-mass
proton angular distributions. This is in line with conclusions
drawn by Kikuchi \textit{et al.}~\cite{Kik98} and by Iwasa
\textit{et al.}~\cite{Iwa99} from their respective $\theta_8$
distributions (which are, however, less sensitive to a small E2
component than the present angular correlations). Our findings
contradict the conclusions of Davids \textit{et al.}~\cite{Dav01}
that a substantial E2 cross section has to be subtracted from the
total measured CD cross section.
What remains to be done is to find a physical explanation for the
small E2 strength compared to the model calculations (both, the
potential model and the cluster model, predict almost equal
$S_{17}^{E2}$ values). At the same time, one has to find a
different way to explain the asymmetries found in inclusive
longitudinal-momentum distributions~\cite{Dav01} and attributed
either to a quenched~\cite{Dav01} or
enhanced~\cite{Dav03,Mor02,Sum05} E2 strength relative to the
respective model calculations.
\subsection{Energy-differential dissociation yields}
The measured momentum vectors of the outgoing p and $^7$Be
particles allowed to calculate $E_{rel}$ according to Eq.(1), from
which we have constructed the energy-differential dissociation
yields of the excited $^8$B$^*$ system prior to breakup
(Fig.\ref{dsde}).
\begin{figure}[bt]
\includegraphics[width=9.0cm]{fig15_dsde.eps}
\vspace*{-1.0cm}
\caption{(Color online)
Energy-differential Coulomb-dissociation yields for
equal-sized $E_{rel}$ bins of 100 keV each. The thick outermost
histogram results from our GEANT simulation including E1 and M1
multipolarity, scaled by a factor of 0.82. The thin(dash-dotted)
histograms show the separate contributions from E1(M1)
multipolarity. } \label{dsde}
\end{figure}
In line with our findings of a negligible E2 contribution
discussed above, we compare this spectrum to a simulated one that
contains contributions from E1 and M1 multipolarities only. The
latter contribution was calculated using the M1 resonance
parameters as determined by Filippone \textit{et
al.}~\cite{Fil83}. As expected, M1 contributes only in a narrow
energy range around the peak of the spectrum. In plotting
Fig.\ref{dsde}, we have restricted the Rutherford scattering
angles $\theta_8$ to values below 1.0$^{\circ}$ to ensure both
dominance of CD and reduction of the effect of any possible E2
contribution.
It should be noted that in CD, starting from the $^8$B ground
state, both the ground state and the first excited state at 429
keV in $^7$Be can be observed as a result from first-order E1
excitation. The relative amount of these contributions to the CD
are determined by the relative spectroscopic factors of the two
${}^{7}$Be states in the ${}^{8}$B ground state and the different
photon spectra due to the different excitation energies. This
component, which can be traced experimentally by observing the
coincident 429 keV $\gamma$-rays, needs to be subtracted before
calculating $S_{17}$ from differential CD cross sections.
Numerical values for this branching have been kindly provided by
Kikuchi \textit{et al.}~\cite{Kik98} and were scaled to the
present bombarding energy using Weizs\"acker-Williams theory.
Since the shape of the theoretical $d\sigma/dE$ distribution is
better defined than its absolute magnitude, we have normalized
both distributions to each other; the resulting scaling factor is
$f=0.82$. With this renormalization, the experimental and
simulated distributions agree rather well (Fig.~\ref{dsde}). Small
deviations between the data points and the histogram indicate
discrepancies between the assumed $S_{17}$ factor from our
potential model and the true one, as will be discussed in the next
section.
\section{The astrophysical $S_{17}$ factor}
The measured quantity in CD of $^8$B is the distribution of
energy-differential cross sections, Fig.\ref{dsde}. This
distribution is related to $S_{17}$ via a theoretical model. We
assume that at the high incident energy used in our experiment and
for the low $Q$-value of the reaction, first-order perturbation
theory is adequate to describe Coulomb dissociation. This has been
investigated in detail in Ref.~\cite{Typ02}. In analyzing our
results, we also assume that the GEANT simulations describe all
experimental effects quantitatively, in particular the feeding of
neighboring bins due to the relatively bad $E_{rel}$ resolution.
We have verified this assumption by comparing data and simulations
for several raw observables, e.g. the $\theta_{17}$ distribution
of Fig.\ref{theta17} or the $\theta_8$ distribution of
Fig.\ref{theta8}. Based on the good agreement, we conclude that
any remaining discrepancies between the two histograms in
Fig.\ref{dsde} can be attributed to a deviation of the true E1
$S_{17}$ factor from the one used in our simulation. Thus, the
true $S_{17}$ factor for each bin was obtained by multiplying the
theoretical one (averaged over this bin width) by the ratio of
observed and simulated counts. The bins were chosen in accordance
with the $E_{rel}$ resolution (Fig.\ref{minv_effic}) to be roughly
one FWHM wide, i.e. between 0.2 and 0.3 MeV. The resulting
$S_{17}$ factors as a function of $E_{rel}$ are visualized in
Fig.~\ref{s17_cd} and listed in Table \ref{s17_table}.
\begin{figure}[bt]
\includegraphics[width=9.0cm]{fig16_s17_cd.eps}
\caption{(Color online)
Comparison between $S_{17}$ values from
Coulomb-dissociation experiments. The full (open) circles indicate
the present (previous) GSI CD experiment labelled GSI-1 (GSI-2).
Open stars depict Ref.\protect\cite{Kik98}, open squares
Ref.\protect\cite{Dav01} (E2 contribution subtracted). The
theoretical curves are described in the text. } \label{s17_cd}
\end{figure}
\begin{table}
\caption{\label{s17_table}Numerical values of $S_{17}$ (in eV b)
as a function of $E_{rel}$ (in MeV). The 1-sigma errors include
all $E_{rel}$-dependent terms. A common systematic error of 5.6\%
has to be added in quadrature to each data point (see text).}
\begin{ruledtabular}
\begin{tabular}{ccc}
$E_{rel}$(\text{MeV})& $S_{17}$\text{(eV b)} & $\sigma_{S17}$ \text{(eV b)}\\
\hline
0.160 & 17.5 & 2.1\\
0.316 & 19.3 & 1.2\\
0.507 & 20.6 & 1.3\\
0.695 & 22.9 & 1.7\\
0.942 & 23.6 & 1.6\\
1.244 & 25.2 & 1.9\\
1.540 & 25.6 & 2.2\\
1.841 & 27.1 & 2.5\\
2.187 & 27.9 & 2.8\\
2.582 & 29.8 & 3.4\\
2.988 & 56.8 & 7.1\\
\end{tabular}
\end{ruledtabular}
\end{table}
The error bars shown in Fig.~\ref{s17_cd} and listed in Table
\ref{s17_table} contain all $E_{rel}$-dependent terms, resulting
from counting statistics, from the error of the geometrical
efficiency as determined by the GEANT simulations, and from the
error of the feeding of the excited state in $^7$Be. In addition,
uncertainties in determining $E_{rel}$ and $\theta_8$ are
included. An $E_{rel}$-independent systematic error of 5.6\% has
to be added for all data points, reflecting an estimated error of
the dead-time correction (0.6\%), of the number of incident $^8$B
projectiles (1.4\%), and of the analysis (5.4\%). The analysis
error consists of the combined errors related with choosing the
appropriate gates to identify a $^7$Be fragment in the $\Delta E -
ToF$ spectra (1.8\%) and to identify a proton via the vertex
reconstruction (5.1\%). The latter contribution reflects the
uncertainty in choosing the low-energy cutoff in the
proton-$\Delta E$ spectra to remove the noise, which at the same
time leads to the loss of some real proton events. More details
can be found in Ref.~\cite{Sch02}.
\subsection{Comparison with other CD experiments}
Fig.\ref{s17_cd} shows the astrophysical $S_{17}$ factors as
deduced from the three other CD experiments published so far
\cite{Kik98,Iwa99,Dav01} (the data of Ref.\cite{Dav01} represent
their E1-$S_{17}$ factors after subtraction of the
E2-contribution). The CD $S_{17}$ factors are in reasonable
agreement with each other, though both the Kikuchi \textit{et
al.}~\cite{Kik98} and the Davids \textit{et al.}~\cite{Dav01} data
are systematically lower. We note that also our earlier CD
experiment~\cite{Iwa99} and the present one are in good agreement
up to $E_{rel} \approx 1.5$ MeV, marked discrepancies occur only
at higher $E_{rel}$ values. Compared to our previous results given
in Ref.~\cite{Sch03}, the lowest three data points have been
increased by 6.7\%, 10\%, and 5.8\%, respectively. The remaining
data points remain largely unaffected. As a consequence, the slope
of our $S_{17}$ factors as a function of $E_{rel}$ becomes smaller
and fits much better than previously to the energy dependence of
Descouvemont's cluster model; we will discuss this aspect in more
detail below.
In the above comparison with other published CD results, we have
plotted $S_{17}$ values as deduced by the authors in their
analyses. Other evaluations of the same data sets may lead to
different results. As an example, we quote the recent re-analysis
of the energy- and angle-differential cross sections, $d\sigma/dE$
and $d\sigma/d\theta_8$, of the RIKEN-2 experiment~\cite{Kik98} by
Ogata \textit{et al.}~\cite{Oga05}. The former authors deduced a
zero-energy factor of $S_{17}(0)=18.9 \pm 1.8$ eV b based on
first-order perturbation theory. Ogata \textit{et al.} obtain
$S_{17}(0)=21.4^{+2.0}_{-1.9}$ eV b from the same experimental
data, by taking into account interference of nuclear, E1 and E2
contributions and higher-order processes.
\subsection{Comparison with direct-capture experiments}
Fig.~\ref{s17_cd_pgamma} compares our data
to those of the recent $^7$Be(p,$\gamma$)$^8$B measurements where
the authors have subtracted the contribution from the M1 resonance
(Refs.~\cite{Ham01,Jun03,Bab03}). (Since we do not intend to dwell on
discrepancies among the results from direct-proton-capture
experiments, we restrict ourselves to the three data sets shown).
\begin{figure}[bt]
\includegraphics[width=9.0cm]{fig17_s17_cd_pgamma.eps}
\caption{(Color online)
$S_{17}$ from this work in comparison with the
(p,$\gamma$) experiments of Ref.~\protect\cite{Ham01} (squares),
Ref.~\protect\cite{Jun03} (stars), and Ref.~\protect\cite{Bab03}
(triangles). The latter data were corrected for the
contribution of the M1 resonance by the authors. The theoretical
curves are from Descouvemont~\protect\cite{Des04} and have been
fitted to the Seattle data (upper curve, Ref.~\cite{Jun03}) and
the present data (lower curve), respectively. See text for more
details.} \label{s17_cd_pgamma}
\end{figure}
With the modifications of the lowest-$E_{rel}$ data points
discussed above, our dataset follows now closely the (p,$\gamma$)
data of Junghans \textit{et al.}~\cite{Jun03} over their {\em
entire} energy range. Previously~\cite{Sch03} we noted good
agreement with Refs.\cite{Jun03,Bab03} only for the data points
above the $M1$ resonance. This solves partly a puzzle that
Junghans \textit{et al.} claim to have observed, a systematic
discrepancy between the slope of the CD $S_{17}$ factors and those
from direct p-capture experiments. It also removes the
experimental basis for recommendations by Esbensen \textit{et
al.}~\cite{Esb05} to modify the deduced slope of $S_{17}(E_{rel})$
on the basis of a fully dynamical calculation.
\subsection{Extrapolation to zero relative energy}
To extrapolate to zero energy, all recent (p,$\gamma$) experiments
have chosen the cluster model of Descouvemont and
Baye~\cite{Des94}. As mentioned above, Descouvemont~\cite{Des04}
has recently refined the cluster-model description of $^8$B (we
refer to this model below as D04). In this refined approach, the
curve resulting from the Minnesota force (MN) is closer to the
experimental data and has been used in Fig.\ref{s17_cd_pgamma} to
fit both the Seattle data~\cite{Jun03} and our present results
over the energy range up to $E_{rel} = 1.5$ MeV. The fits yield
practically identical results within their respective errors. The
D04 normalization factor for our data set is $0.837 \pm 0.013$
with a reduced $\chi^2$ of 0.40. Note that Descouvemont has
investigated the error introduced by scaling the $S$ factor and
found it negligible~\cite{Des04}.
Our previous data set~\cite{Sch03} was found to be best compatible
with the potential-model calculation of Typel as discussed in
subsect.~II.B. of the present paper or in Davids and
Typel~\cite{Dav03} (referred to below as DT03). It is obvious that
with the modified low-energy data points of the present paper, the
agreement with this model is less satisfactory. The black dashed
curve in Fig.\ref{s17_cd} visualizes a fit of this theory to our
data. Though from a purely statistical point of view the fit with
the DT03 curve is acceptable, we prefer to describe our data with
the D04 theory for the following reasons:
\begin{enumerate}
\item The two-cluster structure of $^7$Be is related to its
intrinsic deformation, so that a model of $^8$B based on this
feature should be more realistic; \item the $S_{17}$ energy
dependence from all modern direct p-capture experiments can be
described consistently and fitted with high confidence with the
cluster model, thus corroborating the above conjecture; \item D04
allows to fit our lower 4 to 9 data points with practically equal
results; i.e. the scaling factor does not depend on the fit range.
In contrast, the DT03 scaling factor changes continuously with
increasing fit range, reflecting the different shape of the curve;
\item using D04, we find $\chi^2_{red} < 1$ for a fit range up to
2 MeV; the DT03 fit yields $\chi^2_{red}
> 1$ already if the fit range is extended above 1.3 MeV.
\end{enumerate}
When we fit our lowest 8 data points, up to $E_{rel}$ = 2 MeV, to
the D04 model, we obtain $S_{17}(0) = 20.6 \pm 0.8 $ eV b. The
same result within error bars is obtained if we use any number of
data points between four and eight. As mentioned above, a
systematic error of 5.6\% has to be added, yielding $S_{17}(0) =
20.6 \pm 0.8 (stat) \pm 1.2 (syst)$ eV b. Not included in these
numbers is the theoretical uncertainty given by
Descouvemont~\cite{Des04} as about 5-10\% depending on the
relative energy.
This result overlaps perfectly with a fit of D04 to the full data
set of Junghans \textit{et al.}~\cite{Jun03} which gives
$S_{17}(0) = 21.2 \pm 0.5 $ eV b. A fit of the Baby \textit{et
al.} (p,$\gamma$) data to the D04 model yields a very similar
result, $S_{17}(0)=19.8 \pm 1.0$ eV b. When fitting D04 to the
Hammache \textit{et al.}~\cite{Ham01} data set, we obtain a
smaller central value of $S_{17}(0)=18.4\pm 1.7$ eV b, but the
error bar still overlaps with ours.
\section{Conclusions}
We conclude that at sufficiently high incident energy, a
high-resolution exclusive Coulomb-dissociation experiment can
provide a rather precise value for the low-energy
$^7$Be(p,$\gamma$)$^8$B cross section. Among other conditions to
be fulfilled, the efficiency of the method as a function of
proton-$^7$Be relative energy has to be modelled precisely. By
setting tight constraints to the scattering angle $\theta_8$ and
analyzing proton-$^7$Be angular correlations, a significant
contribution from E2 multipolarity could be excluded. Compared to
our first study of $^8$B Coulomb dissociation~\cite{Iwa99}, we
could base this conclusion on carefully measured angular
distributions. In contrast to our earlier
publication~\cite{Sch03}, our re-analyzed results for the
astrophysical $S_{17}$ factor follow closely the energy dependence
as predicted by the refined cluster-model description of
Descouvemont~\cite{Des04}. This finding is in line with the most
recent measurements of the $^7$Be(p,$\gamma$)$^8$B reaction. The
combined statistical and systematic errors of our fit value for
$S_{17}(0)$ amounts to 6.6\%; a similar error contribution of
about 5\% comes from the model uncertainty~\cite{Des04}.
\acknowledgments
The authors wish to thank K.-H.~Behr, K.~Burkard, and
A.~Br{\"u}nle for technical assistance. Vivid discussions with
B.~Davids, P.~Descouvemont, M.~Hass, and A.~Junghans are
gratefully acknowledged.
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 6,834
|
\section{Introduction}
\emph{Regular model
checking}~\cite{kesten2001,bouajjani2000b,voronkov-csl03} is a
popular paradigm for the symbolic verification of models with infinite
state space. It has been applied to varied families of systems
ranging from distributed algorithms and channel systems to hybrid
systems and programs handling dynamic data structures.
In regular model checking, one works with regular sets of
states and handles them via
finite descriptions, e.g., finite-state automata or regular expressions. Models
amenable to regular model checking are such that, when $S\subseteq {\mathit{Conf}}$
is regular, then $\Post(S)$ (or $\Pre(S)$), the set of 1-step successors
(resp., predecessors), is again a regular set that
can be computed effectively from $S$. Since regular sets are
closed under Boolean operations, one can\footnote{
Actually, such symbolic computations are possible with any class of
representation closed under, and providing algorithms for, $\Pre$ or
$\Post$, Boolean operations, vacuity~\cite{kesten2001,henzinger2005}.
} try to compute the reachability set
$\Post^*(\mathit{Init})$, as the
limit of the sequence
\begin{gather}
\tag{*}
\label{eq-post*}
S_0 := \mathit{Init};
\quad
S_1 := S_0\cup\Post(S_0);
\quad
\ldots
\quad
S_{n+1} := S_n\cup\Post(S_n);
\quad
\ldots
\end{gather}
Since equality of regular sets is decidable, the computation of
\eqref{eq-post*} can contain a test that detects if the limit is reached in
finite time, i.e., if $S_{n+1}=S_n$ for some $n\in\Nat$,
With infinite-state models, the main difficulty is \emph{convergence}. It
is very rare that a fixpoint computation like \eqref{eq-post*} converges in
finite time, and innovative techniques that try to compute directly, or guess and
check, or approximate the limit set $\Post^*(\mathit{Init})$, are currently under
active
scrutiny~\cite{BLW03,boigelot2003,bouajjani2004,habermehl04,BFLS-atva2005}.
\emph{Well-structured transition systems} (WSTS) are a generic family of
models for which the co-reachability set $\Pre^*({{\mathit{Final}}})$ can be computed
symbolically with a backward-chaining version of
\eqref{eq-post*}~\cite{abdulla2000c,finkel98b}. For WSTS's, convergence of
the fixpoint computation is ensured by WQO theory: one handles
upward-closed sets, and increasing sequences of upward-closed sets always
converge in finite time when the underlying ordering is a
well-quasi-ordering (a WQO), as is the case with WSTS's.
Computing $\Pre^*({{\mathit{Final}}})$ for reachability analysis is just a special
case of fixpoint computation. When dealing with richer temporal
properties, one is interested in more complex fixpoints. E.g., the set
of states satisfying the CTL formula $\exists[{{\mathit{Cond}}}{\mathsf{U}}{{\mathit{Goal}}}]$ is
definable via a least-fixpoint expression: $\mu
X.{{\mathit{Goal}}}\cup({{\mathit{Cond}}}\cap\Pre(X))$. For game-theoretic properties, similar
fixpoints are involved. E.g., the states from which the first player
in a turn-based game can enforce reaching a goal is given by $\mu
X.{{\mathit{Goal}}}\cup\Pre(\overline{\Pre(\overline{X})})$.
\paragraph*{\bf Our contribution.}
In this paper, we define a notion of $\mu$-expressions where recursion is
guarded by upward-closure operators, and give a general finite convergence
theorem for all such expressions. The consequence is that these fixpoint
expressions can be evaluated symbolically by an iterative procedure. The
guarded fragment we isolate is very relevant for the verification of
well-structured transition systems as we demonstrate by providing several
new decidability results on channel systems.
\paragraph*{\bf Related work.}
Henzinger \textit{et al.} give general conditions for the convergence of
fixpoints computations for temporal~\cite{henzinger2005} or
game-theoretic~\cite{alfaro2001b} properties, but the underlying framework
(finite quotients) is different and has different applications (timed and
hybrid systems). Our applications to well-structured transition systems
generalize results from ~\cite{abdulla2003b,raskin03,raskin05,KucSch-TCS}
that rely on more ad-hoc finite convergence lemmas.
\section{A guarded mu-calculus}
\label{sec-guarded-mucalculus}
We assume basic understanding of $\mu$-calculi techniques (otherwise
see~\cite{arnold2001}) and of well-quasi-ordering (WQO) theory (otherwise
see~\cite{milner85,kruskal72}, or simply~\cite[sect.~2.1]{finkel98b}).
Let $(W,\sqsubseteq)$ be a well-quasi-ordered set. A subset $V$ of $W$ is
\emph{upward-closed} if $w\in V$ whenever $v\sqsubseteq w$ for some $v\in
V$. From WQO theory, we mostly need the following result:
\begin{fact}[Finite convergence]
\label{fact-fc}
If $V_0\subseteq V_1\subseteq V_2\subseteq \cdots$ is an infinite
increasing sequence of upward-closed subsets of $W$, then
for some index $k\in\Nat$, $\bigcup_{i\in\Nat}V_i =V_k$.
\end{fact}
The \emph{upward-closure} of $V\subseteq W$, denoted $C_{\uparrow}(V)$, is the
smallest upward-closed set that contains $V$. The \emph{upward-kernel} of
$V$, denoted $K_{\uparrow}(V)$, is the largest upward-closed set included in $V$.
There are symmetric notions of \emph{downward-closed} subset of $W$, of
\emph{downward-closure}, $C_{\downarrow}(V)$, and of \emph{downward-kernel}, $K_{\downarrow}(V)$,
of $V$. The complement of an upward-closed subset is downward-closed.
Observe that $C_{\uparrow}(V)=V=K_{\uparrow}(V)$ iff $V$ is upward-closed, and that $C_{\uparrow}$ and
$K_{\downarrow}$ (resp., $C_{\downarrow}$ and $K_{\uparrow}$) are dual:
\begin{xalignat}{2}
\label{eq-Cu-duality}
W\setminus K_{\uparrow}(V) &= C_{\downarrow}(W\setminus V),
&
W\setminus K_{\downarrow}(V) &= C_{\uparrow}(W\setminus V).
\end{xalignat}
\paragraph*{\bf Monotonic region algebra.}
In symbolic model-checking, a \emph{region algebra} is a family of
sets of states (subsets of $W$) that is closed under Boolean and other
operators like images or inverse images~\cite{henzinger2005}.
Here we consider regions generated by a family $O=\{o_1,o_2,\ldots\}$ of
(monotonic) operators. By a $k$-ary \emph{operator}, we mean a monotonic
mapping $o:(2^W)^k\to 2^W$ that associates a subset
$o(V_1,\ldots,V_k)\subseteq W$ with any $k$ subsets $V_1,\ldots,V_k$.
Monotonicity means that $o(V_1,\ldots,V_k)\subseteq o(V'_1,\ldots,V'_k)$
when $V_i\subseteq V'_i$ for $i=1,\ldots,k$. We allow nullary operators,
i.e., fixed subsets of $W$. Finally, we require that $O$ contains at least
four special unary operators: $C_{\uparrow}$, $C_{\downarrow}$, $K_{\uparrow}$, $K_{\downarrow}$, and two special
nullary operators: $\emptyset$ and $W$.
The \emph{region algebra generated by $O$}, denoted with $\R_O$, or simply
$\R$, is the set of all the subsets of $W$, called \emph{regions}, that can
be obtained by applying operators from $O$ on already constructed regions,
starting with nullary operators. Equivalently, $\R$ is the least subset of
$2^W$ that is closed under $O$.
We say the region algebra generated by $O$ is \emph{effective} if there are
algorithms implementing the operators in $O$ and an effective membership
algorithm saying whether $w\in R$ for some $w\in W$ and some region $R\in\R_O$.
Such effectiveness assumptions presuppose a finitary encoding of regions
and elements of $W$: if there are several possible encodings for a same
region, we assume an effective equality test.
\paragraph*{\bf Extending the region algebra with fixpoints.}
Let $\chi = \{X_1,X_2,\cdots\}$ be a countable set of variables.
$L_\mu(W,\sqsubseteq,O)$, or shortly $L_\mu$ when $(W,\sqsubseteq)$ and $O$
are clear from the context, is the set of \emph{$O$-terms with least and
greatest fixpoints} given by the following abstract syntax:
\begin{gather*}
L_\mu \ni \phi,\psi \ ::=
o(\phi_1,\ldots,\phi_k) \ \big| \
X \ \big| \
\mu X.\phi \ \big| \
\nu X. \phi \ \big| \
C_{\uparrow}(\phi) \ \big| \
C_{\downarrow}(\phi) \ \big| \
K_{\uparrow}(\phi) \ \big| \
K_{\downarrow}(\phi)
\end{gather*}
where $X$ runs over variables from $\chi$, and $o$ over operators from $O$.
$\mu X.\phi$ and $\nu X.\phi$ are fixpoint expressions. Free and bound
occurrences of variables are defined as usual. We assume that no variable
has both bound and free occurrences in some $\phi$, and that no two
fixpoint subterms bind the same variable: this can always be ensured by
renaming bound variables. (The abstract syntax for $L_\mu$ could be shorter
but we wanted to stress that $C_{\uparrow}$, $C_{\downarrow}$, $K_{\uparrow}$, and $K_{\downarrow}$ are required to
be present in $O$.)
The meaning of $L_\mu$ terms is as expected: an \emph{environment} is a
mapping ${\textit{env}} : \chi \to 2^W$ that interprets each variable $X \in \chi$ as
a subset of $W$. Given ${\textit{env}}$, a term $\phi\in L_\mu$ denotes a subset of
$W$, written $\rep{\phi}_{\textit{env}}$ and defined by induction on the structure of
$\phi$:
\begin{xalignat*}{2}
\rep{X}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} {\textit{env}}(X)
&
\rep{o(\phi_1,\ldots,\phi_k)}_{{\textit{env}}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}
o(\rep{\phi_1}_{\textit{env}},\ldots,\rep{\phi_k}_{\textit{env}})
\\
\rep{C_{\uparrow}( \phi)}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}
C_{\uparrow}\bigl( \rep{\phi}_{{\textit{env}}} \bigr)
&
\rep{C_{\downarrow}( \phi)}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}
C_{\downarrow} (\rep{\phi}_{{\textit{env}}})
\\
\rep{K_{\uparrow} (\phi)}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}
K_{\uparrow}( \rep{\phi}_{{\textit{env}}})
&
\rep{K_{\downarrow} (\phi)}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}
K_{\downarrow}\bigl( \rep{\phi}_{{\textit{env}}}\bigr)
\\
\rep{\mu X. \phi}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} \mathrm{lfp} \bigl(\Omega[\phi,X,{\textit{env}}]
\bigr)
\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
&
\rep{\nu X. \phi}_{\textit{env}} & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} \mathrm{gfp} \bigl(\Omega[\phi,X,{\textit{env}}]
\bigr)
\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
\end{xalignat*}
where $\Omega[\phi,X,{\textit{env}}] : 2^W \to 2^W$ is a unary operator defined by
$\Omega[\phi,X,{\textit{env}}](V) \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} \rep{\phi}_{{\textit{env}}[X:= V]}$, using the
standard variant notation ``${\textit{env}}[X:=V]$'' for the environment that agrees
with ${\textit{env}}$ everywhere except on $X$ where it returns $V$. As usual,
$\rep{\phi}_{\textit{env}}$ does not depend on ${\textit{env}}(X)$ if $X$ is not free in
$\phi$, so that we may shortly write $\rep{\phi}$ when $\phi$ is a closed
term, i.e., a term with no free variables.
We recall that the semantics of the fixpoint terms is well-defined since,
for every $\phi$, $X$ and ${\textit{env}}$, $\Omega[\phi,X,{\textit{env}}]$ is monotonic (and
since $(2^W,\subseteq)$ is a complete lattice). Moreover, if ${\textit{env}}$ and
${\textit{env}}'$ are such that ${\textit{env}}(X)\subseteq{\textit{env}}'(X)$ for all $X\in\chi$,
shortly written ${\textit{env}}\subseteq{\textit{env}}'$, then $\mathrm{lfp}(\Omega[\phi,X,{\textit{env}}])
\subseteq \mathrm{lfp}(\Omega[\phi,X,{\textit{env}}'])$ and $\mathrm{gfp}(\Omega[\phi,X,{\textit{env}}])
\subseteq \mathrm{gfp}(\Omega[\phi,X,{\textit{env}}'])$.
\begin{definition}[Upward- and downward-guardedness]
\begin{enumerate}
\item A variable $X$ is \emph{upward-guarded} in $\phi$ if all free
occurrences of $X$ in $\phi$ are in the scope of either a $C_{\uparrow}$ or a
$K_{\uparrow}$ operator, i.e., appear in a subterm of the form $C_{\uparrow}(\psi)$ or $K_{\uparrow}
(\psi)$.
\item Dually, $X$ is \emph{downward-guarded} in $\phi$ if all its
free occurrences are in the scope of a
$C_{\downarrow}$ or a $K_{\downarrow}$ operator.
\item A term $\phi$ is \emph{guarded} if all its least-fixpoint subterms
$\mu X.\psi$ have $X$ upward-guarded in $\psi$, and all its
greatest-fixpoint subterms $\nu X.\psi$ have $X$ downward-guarded in $\psi$.
\end{enumerate}
\end{definition}
Given some $\phi$, $X$ and ${\textit{env}}$, the approximants of
$\mathrm{lfp}(\Omega[\phi,X,{\textit{env}}])$ are given by the sequence $(M_i)_{i\in\Nat}$ of
subsets of $W$ defined inductively with $M_0= \emptyset$ and $M_{i+1} =
\rep{\phi}_{{\textit{env}}[X:=M_i]}$. Monotonicity yields
\begin{gather}
\label{eq-mon}
M_0 \subseteq M_1 \subseteq M_2 \subseteq \cdots \subseteq \mathrm{lfp}(\Omega[\phi,X,{\textit{env}}]).
\end{gather}
Similarly we define $(N_i)_{i\in\Nat}$ by $N_0 = W$ and $N_{i+1} =
\rep{\phi}_{{\textit{env}}[X:=N_i]}$, so that
\begin{gather}
N_0 \supseteq N_1 \supseteq N_2 \supseteq \cdots \supseteq \mathrm{gfp}(\Omega[\phi,X,{\textit{env}}]).
\end{gather}
\begin{lemma}[Finite convergence of approximants]
\label{lem-finite-approx}
If $X$ is upward-guarded in $\phi$, then there exists an index $k\in\Nat$ such that
\begin{gather}
\label{eq-pfa}
\rep{\mu X.\phi}_{\textit{env}} = M_k = M_{k+1} = M_{k+2} = \ldots
\end{gather}
Dually, if $X$ is downward-guarded in $\phi$, then there exists a $k'\in\Nat$
such that
\begin{gather}
\rep{\nu X.\phi}_{\textit{env}} = N_{k'} = N_{k'+1} = N_{k'+2} = \ldots
\end{gather}
\end{lemma}
\begin{proof}
We only prove the first half since the other half is dual. Let
$\psi_1,\ldots,\psi_m$ be the maximal subterms of $\phi$ that are
immediately under the scope of a $C_{\uparrow}$ or a $K_{\uparrow}$ operator. Then $\phi$ can
be decomposed under the form
\begin{gather*}
\phi \:\equiv\: \Phi(\Uparrow \psi_1,\ldots,\Uparrow \psi_m)
\end{gather*}
where the context $\Phi(Y_1,\ldots,Y_m)$ uses fresh variables
$Y_1,\ldots,Y_m$ to be substituted in, and where $\Uparrow\psi_i$ is either
$C_{\uparrow}(\psi_i)$ or $K_{\uparrow}(\psi_i)$, depending on how $\psi_i$ appears in
$\phi$. In either case, and for any environment ${\textit{env}}'$, the set
$\rep{\Uparrow \psi_i}_{{\textit{env}}'}$ is upward-closed.
For $V_1,\ldots,V_m \subseteq W$ we shortly write
$\rep{\Phi}(V_1,\ldots,V_m)$ for $\rep{\Phi}_{{\textit{env}}[Y_1 :=
V_1,\ldots,Y_m:=V_m]}$. Since $X$ is upward-guarded in $\phi$, it has no
occurrence in $\Phi$, only in the $\psi_i$'s, so that
\begin{align*}
M_{i+1} = \rep{\phi}_{{\textit{env}}[X:=M_i]} & =
\rep{\Phi}(\rep{\Uparrow \psi_1}_{{\textit{env}}[X:=M_i]},\ldots,
\rep{\Uparrow \psi_m}_{{\textit{env}}[X:=M_i]}) \\
& =
\rep{\Phi}(L_{i,1},\ldots, L_{i,m})
\end{align*}
writing $L_{i,j}$ for $\rep{\Uparrow\psi_j}_{{\textit{env}}[X:=M_i]}$. From
$M_0\subseteq M_1\subseteq M_2 \subseteq \cdots$, we deduce $L_{0,j}
\subseteq L_{1,j} \subseteq L_{2,j} \subseteq \cdots$ Since $K_{\uparrow}$ and $C_{\uparrow}$
return upward-closed sets, the $L_{i,j}$'s are upward-closed subsets of
$W$. For all $j=1,\ldots,m$, Fact~\ref{fact-fc} implies that there is an
index $k_j$ such that $L_{i,j} = L_{k_j,j}$ for all $i \geq k_j$. Picking
$K=\max(k_1,\ldots,k_j)$ gives for any $i\geq K$
\[
M_{i+1} = \rep{\Phi}(L_{i,1},\ldots, L_{i,m})
= \rep{\Phi}(L_{k_1,1},\ldots,L_{k_m,m})
= \rep{\Phi}(L_{K,1},\ldots,L_{K,m})
= M_{K+1}.
\]
Thus, $\bigcup_{i\in\Nat}M_i=M_{K+1}=M_{K+2}$ and $M_{K+1}$ is a fixpoint
of $\Omega[\phi,X,{\textit{env}}]$, hence the least one thanks to \eqref{eq-mon}.
Picking $k=K+1$ satisfies \eqref{eq-pfa}.
\qed
\end{proof}
\paragraph*{\bf Regions with guarded fixpoints.}
We can now prove our main result: subsets defined by $L_\mu$ terms are
regions (and can be computed effectively if the underlying region
algebra is effective).
By a \emph{region-environment} we mean an environment ${\textit{env}} : \chi \to \R$
that associates regions with variables. If ${\textit{env}}$ is a region-environment,
and $\phi$ has only free variables, i.e., has no fixpoints subterms, then
$\rep{\phi}_{\textit{env}}$ is a region.
\begin{theorem}
\label{th-guarded-effective}
If $\phi\in L_\mu$ is guarded and ${\textit{env}}$ is a region-environment then
$\rep{\phi}_{\textit{env}}$ is a region. Furthermore, if the region algebra is
effective, then $\rep{\phi}_{\textit{env}}$ can be computed effectively from $\phi$
and ${\textit{env}}$.
\end{theorem}
\begin{proof}
By structural induction on the structure of $\phi$. If $\phi=o()$ is a
nullary operator, the result holds by definition of the region algebra. If
$\phi = o(\phi_1,\cdots,\phi_k)$, the $\rep{\phi_i}_{\textit{env}}$'s are
(effectively) regions by induction hypothesis, so that $\rep{\phi}_{\textit{env}}$ is
an (effective) region too by definition. In particular, this argument
applies when $o$ is a nullary operator, or is one of the unary operators we
singled out: $C_{\uparrow}$, $C_{\downarrow}$, $K_{\uparrow}$, and $K_{\downarrow}$.
If $\phi = \mu X. \psi$, we can apply Lemma~\ref{lem-finite-approx} after
we have proved that each one of the approximants $M_0, M_1, M_2, \ldots$,
of $\rep{\phi}_{\textit{env}}$ are regions. In particular, $M_0=\emptyset$ is a
region, and if $M_i$ is a region, then $M_{i+1}=\rep{\psi}_{{\textit{env}}[X:=M_i]}$
is one too, since ${\textit{env}}'={\textit{env}}[X:=M_i]$ is a region-environment, and since
by induction hypothesis $\rep{\psi}_{{\textit{env}}'}$ is a region when ${\textit{env}}'$ is a
region-environment. When $\R_O$ is effective, the $M_i$ can be computed
effectively, and one can detect when $M_k=M_{k+1}$ since region equality is
decidable by definition. Then $\rep{\phi}_{\textit{env}} = M_k$ can be computed
effectively. Finally, the case where $\phi=\nu X.\psi$ is dual.
\qed
\end{proof}
\begin{corollary}[Decidability for guarded $\L_\mu$ properties]
The following problems are decidable for effective
monotonic region algebras:
\begin{description}
\item[Model-checking:]
``Does $w \in \rep{\phi}$?'' for a $w\in
W$ and a closed and guarded $\phi\in L_\mu$.
\item[Satisfiability:]
``Is $\rep{\phi}$ non-empty?'' for a closed and guarded $\phi\in L_\mu$.
\item[Universality:]
``Does $\rep{\phi}=W$?'' for a closed and guarded $\phi\in L_\mu$.
\end{description}
\end{corollary}
\paragraph*{\bf A region algebra of regular languages.}
Consider $W=\Sigma^*$, the set of finite words over some finite alphabet
$\Sigma$. The \emph{subword ordering}, defined by ``$u\sqsubseteq v$ iff $u$
can be obtained by erasing some letters from $v$'', is a WQO (Higman's
Lemma). Regular languages over $\Sigma$ are a natural choice for regions:
observe that the closure operators $C_{\uparrow}$ and $C_{\downarrow}$ preserve regularity and
have effective implementations.\footnote{
From a FSA for $R$, one obtains a FSA for $C_{\uparrow}(R)$ simply by adding loops
$q\step{a}q$ on all states $q$ of the FSA and for all letters
$a\in\Sigma$. A FSA for $C_{\downarrow}(R)$ is obtained by adding
$\epsilon$-transitions $q\step{\epsilon}q'$ whenever there is a
$q\step{a}q'$. From this, $K_{\uparrow}$ and $K_{\downarrow}$ can be implemented using
\eqref{eq-Cu-duality}.
} Natural operators to be considered in $O$ are $\cup$ (union) and $\cap$
(intersection). However, any operation on languages that is monotonic,
preserve regularity, and has an effective implementation on regular
languages can be added. This includes concatenation (denoted $R.R'$),
star-closure (denote $R^*$), left- and right-residuals
($R^{-1}R'\stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}\{v~|~\exists u\in R,uv\in R'\}$), shuffle product (denoted
$R \parallel R'$), reverse (denoted $\overleftarrow{R}$), conjugacy
($\widetilde{R}\stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}\{vu~|~uv\in R\}$), homomorphic and
inverse-homomorphic images, and many more~\cite{perrin90}. Complementation
is not allowed in $O$ (it is not monotonic) but the duals of all
above-mentioned operators can be included in $O$ (without compromising
effectiveness) so that, for all practical purposes, complement can be used
with the restriction that bound variables in $L_\mu$ terms are under an
even number of complementations.
An application of Theorem~\ref{th-guarded-effective} is that, if $R_1$ and
$R_2$ are regular languages, then the language defined as $\mu X.\nu
Y.\bigl(K_{\uparrow}\bigl[R_1 \parallel (X^*\cap C_{\downarrow}(Y^{-1}\overleftarrow{X}\cap
X^{-1}R_2))\bigr]\bigr)$ is regular and a finite representation for it
(e.g., a regular expression or a minimal DFA) can be constructed from $R_1$
and $R_2$.
\section{Verification of lossy channel systems}
\label{sec-verif-LCS}
Theorem~\ref{th-guarded-effective} has several applications for regular
model checking of lossy channel systems~\cite{abdulla96b} (LCS) and other
families of well-structured systems~\cite{abdulla2000c,finkel98b}. In the
rest of this paper we concentrate on LCS's.
\subsection{Channel systems, perfect and lossy}
A channel system is a tuple $\L= (Q,\C,\Mess,\Delta)$ consisting of a
finite set $Q=\{p,q,\ldots\}$ of \emph{locations}, a finite set
$\C=\{c,\ldots\}$ of \emph{channels}, a finite \emph{message alphabet}
$\Mess=\{m,\ldots\}$ and a finite set $\Delta=\{\delta,\ldots\}$ of
\emph{transition rules}. Each transition rule has the form $q \step{{\mathit{op}}} p$
where ${\mathit{op}}$ is an \emph{operation}: $\c!m$ (sending message $m \in \Mess$
along channel $\c \in \C$), $\c?m$ (receiving message $m$ from channel
$\c$), or $\surd$ (an internal action to some process, no I/O-operation).
\paragraph*{\bf Operational semantics.}
Let $\L= (Q,\C,\Mess,\Delta)$ be a channel system. A \emph{configuration}
(also, a \emph{state}) is a pair $\sigma = (q,w)$ where $q \in Q$ is a
location and $w : \C \to \Mess^*$ is a channel valuation that associates
with any channel its content (a sequence of messages). The set $Q\times
{\Mess^*}^\C$ of all configurations is denoted by
${\mathit{Conf}}=\{\sigma,\rho,\ldots\}$. For a subset $V$ of ${\mathit{Conf}}$, we let
$\overline{V} \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} {\mathit{Conf}} \setminus V$.
Steps between configurations are as expected. Formally, $\sigma = (q,w)$
leads to $\sigma' = (q',w')$ by firing $\delta = p \step{{\mathit{op}}} r$, denoted
$\sigma \step{\delta}_{\mathrm{perf}} \sigma'$, if and only if $q=p$, $q'=r$ and
$w'$ is obtained from $w$ by the effect of ${\mathit{op}}$ (the ``$\mathrm{perf}$'' subscripts
emphasizes that the step is perfect: without losses). Precisely,
$w'(c)=w(c)$ for all channels $c$ that are not touched upon by ${\mathit{op}}$, and
\[
w'(c) = \begin{cases}
w(c) m & \textrm{ if } {\mathit{op}} = c!m,\\
m^{-1}w(c) & \textrm{ if } {\mathit{op}} = c?m.
\end{cases}
\]
Thus, when ${\mathit{op}}=c?m$, $w'$ is only defined if $w(c)$ starts with $m$ and
indeed this is the intended condition for firing $\delta$. Whenever $\sigma
\step{\delta} \rho$ for some $\rho$, we say that $\delta$ is \emph{enabled}
in $\sigma$, written $\delta \in \Delta(\sigma)$.
Below we restrict our attention to LCS's where from each $q\in Q$ there is
at least one rule $q\step{{\mathit{op}}}p$ in $\Delta$ where ${\mathit{op}}$ is not a receiving
action: this ensures that the LCS has no deadlock states and simplifies
many technical details without losing any generality.
\paragraph*{\bf Lossy systems.}
In \emph{lossy} channel systems, losing messages is formalized via the
subword ordering, extended from $\Mess^*$ to ${\mathit{Conf}}$: $(q,w) \sqsubseteq
(q',w')$ if $q=q'$ and $w(c) \sqsubseteq w'(c)$ for all channels $c \in
\C$.
A (possibly lossy) step in the LCS is made of a perfect step followed by
arbitrary losses:\footnote{
\label{STEP-fn}
Note that, with this definition, message losses only occur \emph{after}
steps (thus, not in the initial configuration). The usual definition
allows arbitrary losses before and after a step. There is no essential
semantical difference between these two ways of grouping atomic events
into single ``steps''. The usual definition is technically smoother when
LCS's are viewed as nondeterministic systems, but becomes unnatural in
situations where several adversarial processes compete, e.g., in
probabilistic LCS's~\cite{BBS-acmtocl06} or other game-theoretical
settings we explore in sections~\ref{sec-turn-based}
and~\ref{sec-prob-lcs}.
} formally, we write $\sigma \step{\delta}\rho$ whenever there is a perfect
step $\sigma \step{\delta}_{\mathrm{perf}} \sigma'$ such that $\rho \sqsubseteq
\sigma'$. This gives rise to a labeled transition system $\LTS_\L \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}
({\mathit{Conf}},\Delta,\to)$, that can be given a WSTS structure by the following
relation: $\sigma\preceq \rho\;\stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{\Leftrightarrow}\;
\sigma\sqsubseteq\rho\:\cap\:\Delta(\sigma)=\Delta(\rho)$. Both
$\sqsubseteq$ and $\preceq$ turns ${\mathit{Conf}}$ into a WQO.
\begin{remark}
From now on we assume for the sake of simplicity that
$({\mathit{Conf}},\sqsubseteq)$ is the WQO on which $L_\mu$ is defined. All results
could be strengthened using $({\mathit{Conf}},\preceq)$.
\qed
\end{remark}
Following standard notations for transition systems $({\mathit{Conf}},\Delta,\to)$
labeled over some $\Delta$, we write
$\Pre[\delta](\sigma)\stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=}\{\rho\in{\mathit{Conf}}~|~\rho\step{\delta}\sigma\}$ for
the set of predecessors via $\delta$ of $\sigma$ in $\L$. Then
$\Pre(\sigma) \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} \bigcup_{\delta \in \Delta} \Pre[\delta](\sigma)$ has
all 1-step predecessors of $\sigma$, and $\Pre(V) = \bigcup_{\sigma \in V}
\Pre(\sigma)$ has all 1-step predecessors of states in $V$. The dual
$\wPre$ of $\Pre$ is defined by $\wPre(V)
=\overline{\Pre(\overline{V})}$. Thus $\sigma \in \wPre(V)$ iff
all 1-step successors of $\sigma$ are in $V$ (this includes the case where
$\sigma$ is a deadlock state).
Seen as unary operators on $2^{\mathit{Conf}}$, both $\Pre$ and $\wPre$ are
monotonic and even continuous for all transition systems~\cite{sifakis82}.
For LCS's, the following lemma states that $\Pre$ is compatible with the
WQO on states, which will play a crucial role later when we want to show
that some $L_\mu$ term is guarded.
\begin{lemma}
\label{lem-pre-guarded}
Let $V\subseteq{\mathit{Conf}}$ in the transition system $\LTS_\L$ associated with a
LCS $\L$. Then $\Pre(V) = \Pre(C_{\uparrow}(V))$ and $\wPre(V) =
\wPre(K_{\downarrow}(V))$.
\end{lemma}
\begin{proof}
$V\subseteqC_{\uparrow}(V)$ implies $\Pre(V) \subseteq \Pre(C_{\uparrow}(V))$. Now
$\sigma\in\Pre(C_{\uparrow}(V))$ implies that $\sigma\step{}\rho\sqsupseteq\rho'$
for some $\rho'\in V$. But then $\sigma\step{}\rho'$ by definition of lossy
steps and $\sigma\in\Pre(V)$. The second equality is dual.
\qed
\end{proof}
\paragraph*{\bf An effective region algebra for LCS's.}
We are now ready to apply the framework of
section~\ref{sec-guarded-mucalculus} to regular model checking of lossy
channel systems. Assume $\L= (Q,\C,\Mess,\Delta)$ is a given LCS. A region
$R\in\R$ is any ``regular'' subset of ${\mathit{Conf}}$. More formally, it is any set
$R\subseteq{\mathit{Conf}}$ that can be written under the form
\[
R = \sum_{i \in I} (q_i,R_{i}^1,\ldots,R_{i}^{|\C|})
\]
where $I$ is a \emph{finite} index set, the $q_i$'s are locations from $Q$,
and each $R_{i}^j$ is a regular language on alphabet $\Mess$. The notation
has obvious interpretation, with summation denoting set union (the empty
sum is denoted $\emptyset$). We are not more precise on how such regions
could be effectively represented (see~\cite{abdulla2001b}), but they could be handled as, e.g.,
regular expressions or FSAs over the extended alphabet $\Mess\cup
Q\cup\{\mathtt{'('},\mathtt{')'},\mathtt{','}\}$.
The set $O$ of operators includes union, intersection, $C_{\uparrow}$, $C_{\downarrow}$, $K_{\uparrow}$,
$K_{\downarrow}$: these are monotonic, regularity-preserving, and effective operators
as explained in our example at the end of
section~\ref{sec-guarded-mucalculus}. Operators specific to regular
model-checking are $\Pre$ and $\wPre$. That they are
regularity-preserving and effective is better seen by first looking at the
special case of perfect steps:
\begin{align*}
\Pre_{\mathrm{perf}}[p \step{c_i?m} r](q,R_p^1, \cdots, R_p^{|\C|}) &=
\begin{cases}
(p,R_p^1, \ldots, R_p^{i-1}, m R_p^i, R_p^{i+1}, \ldots, R_p^{|\C|}) & \textrm{ if } q=r,\\
\emptyset & \textrm{ otherwise.}
\end{cases}
\\
\Pre_{\mathrm{perf}}[q \step{c_i!m} q'](q,R_p^1, \cdots, R_p^{|\C|}) &=
\begin{cases}
(p,R_p^1, \ldots, R_p^{i-1}, R_p^i m^{-1}, R_p^{i+1}, \ldots, R_p^{|\C|}) & \textrm{ if } q=r,\\
\emptyset & \textrm{ otherwise.}
\end{cases}
\\
\Pre_{\mathrm{perf}}\Bigl(\sum_{i\in I} (q_i, R_i^1,\ldots,R_i^{|\C|})\Bigr)
& = \sum_{i\in I}\sum_{\delta\in\Delta} \Pre_{\mathrm{perf}}[\delta] (q_i,R_i^1,\ldots,R_i^{|\C|}).
\end{align*}
where the notation ``$m R$'' (for concatenation) and ``$R m^{-1}$'' (for
right-residuals) are as in section~\ref{sec-guarded-mucalculus}. For lossy
steps we use
\[
\Pre(R) = \Pre_\mathrm{perf}(C_{\uparrow}(R)).
\]
Clearly, both $\Pre_\mathrm{perf}$ and $\Pre$ are effective operators on regions.
\subsection{Regular model-checking for lossy channel systems}
Surprising decidability results for lossy channel systems is what launched
the study of this model~\cite{finkel94,abdulla96b,cece95}. We reformulate
several of these results as a direct consequence of
Theorem~\ref{th-guarded-effective}, before moving to new problems and new
decidability results in the next sections. Note that our technique is
applied here to a slightly different operational semantics (cf.\
footnote~\ref{STEP-fn}) but it would clearly apply as directly to the
simpler semantics.
\paragraph*{\bf Reachability analysis.}
Thanks to Lemma~\ref{lem-pre-guarded}, the co-reachability set can be
expressed as a guarded $L_\mu$ term:
\begin{equation}
\label{eq-pre-star}
\Pre^*(V) = \mu X. V \cup \Pre(X) = \mu X. V \cup \Pre(C_{\uparrow}(X)).
\end{equation}
\begin{corollary}
For regular $V\subseteq{\mathit{Conf}}$, $\Pre^*(V)$ is regular and effectively computable.
\end{corollary}
\paragraph*{\bf Safety properties.}
More generally, safety properties can be handled. In CTL, they can be
written $\forall (V_1 {\mathsf{R}} V_2)$. Recall that ${\mathsf{R}}$, the Release
modality, is dual to Until: a state $\sigma$ satisfies $\forall (V_1 {\mathsf{R}}
V_2)$ if and only if along all paths issuing from $\sigma$, $V_2$ always holds
until maybe $V_1$ is visited. Using Lemma~\ref{lem-pre-guarded},
$\rep{\forall (V_1 {\mathsf{R}} V_2)}$, the set of states where the safety property
holds, can be defined as a guarded $L_\mu$ term:
\begin{equation
\label{eq-forall-release}
\rep{\forall (V_1 {\mathsf{R}} V_2)}
= \nu X. \bigl(V_2 \cap (\wPre(X) \cup V_1)\bigr)
= \nu X. \bigl(V_2 \cap (\wPre(K_{\downarrow}(X)) \cup V_1)\bigr).
\end{equation}
\begin{corollary}
\label{coro-AR}
For regular $V_1,V_2\subseteq{\mathit{Conf}}$, $\rep{\forall (V_1 {\mathsf{R}} V_2)}$ is regular
and effectively computable.
\end{corollary}
Another formulation is based on the duality
between the ``$\forall{\mathsf{R}}$'' and the ``$\exists{\mathsf{U}}$'' modalities.
\begin{theorem}\cite[sect.~5]{KucSch-TCS}
\label{KS-theorem}
If $f$ is a temporal formula in the
${\mathsf{TL}}(\exists{\mathsf{U}},\exists{\mathsf{X}},\wedge,\neg)$ fragment of CTL (using
regions for atomic propositions), then $\rep{f}$ is regular and effectively
computable.
\end{theorem}
\begin{proof}
By induction on the structure of $f$, using
$\rep{\exists{\mathsf{X}} f}\stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} \Pre(\rep{f})$, and the fact that
regions are (effectively) closed under complementation.
\qed
\end{proof}
\paragraph*{\bf Beyond safety.}
Inevitability properties, and recurrent reachability can be stated in $L_\mu$.
With temporal logic notation, this yields
\begin{align*}
\rep{\forall \Diamond V} &= \mu X. \bigl(V \cup (\Pre({\mathit{Conf}}) \cap \wPre(X))\bigr),
\\
\rep{\exists \Box\Diamond V} &= \nu X. \bigl(\mu Y. ((V \cup \Pre(Y)) \cap \Pre(X))\bigr).
\end{align*}
These two terms are not guarded and Lemma~\ref{lem-pre-guarded} is of no
help here. However this is not surprising: firstly, whether
$\sigma\sat\exists\Box\Diamond V$ is undecidable~\cite{abdulla96c}; secondly,
and while $\sigma\sat\rep{\forall \Diamond V}$ is decidable, the set
$\rep{\forall \Diamond V}$ cannot be computed
effectively~\cite{mayr-unreliable}.
\subsection{Generalized lossy channel systems}
\label{ssec-extended-lcs}
Transition rules in LCS's do not carry guards, aka preconditions, beyond
the implicit condition that a reading action $c?m$ is only enabled when
$w(c)$ starts with $m$. This barebone definition is for simplification
purpose, but actual protocols sometimes use guards that probe the contents
of the channel before taking this or that transition. The simplest such
guards are emptiness tests, like ``$p\step{c=\epsilon?}q$'' that only
allows a transition from $p$ to $q$ if $w(c)$ is empty.
We now introduce \emph{LCS's with regular guards} (GLCS's), an extension of
the barebone model where any regular set of channel contents can be used to
guard a transition rule. This generalizes emptiness tests, occurrence tests
(as in~\cite{OuaknineWorrell06}), etc., and allows expressing priority
between rules since whether given rules are enabled is a regular condition.
Formally, we assume rules in $\Delta$ now have the form $p
\step{G:{\mathit{op}}} q$ with $p,q,{\mathit{op}}$ as before, and where $G$, the guard,
can be any regular region. The operational semantics is a expected:
when $\delta=p\step{G:{\mathit{op}}}q$, there is a perfect step
$\sigma\step{\delta}_\mathrm{perf}\theta$ iff $\sigma \in G$ and $\theta$ is
obtained from $\sigma$ by the rule $p \step{G:{\mathit{op}}} q$ (without any
guard). Then, general steps $\sigma\step{\delta}\rho$ are obtained
from perfect steps $\sigma\step{\delta}_\mathrm{perf}\sigma'$ by message
losses $\rho\sqsubseteq \sigma'$.
\paragraph*{\bf Verification of GLCS's.}
For GLCS's, $\Pre$ and $\Post$ are effective monotonic
regularity-preserving operators as in the LCS case since
\begin{align*}
\Pre[p\step{G:{\mathit{op}}}q](R)&=G\:\cap\:\Pre[p\step{{\mathit{op}}}q](R), \\
\Post[p\step{G:{\mathit{op}}}q](R)&=\Post[p\step{{\mathit{op}}}q](G\cap R).
\end{align*}
Observe that Lemma~\ref{lem-pre-guarded} holds for GLCS's as well, so that
Equations \eqref{eq-pre-star} and \eqref{eq-forall-release} entail a
generalized version of Theorem~\ref{KS-theorem}:
\begin{theorem}
For all GLCS's $\L$ and formulae $f$ in the
${\mathsf{TL}}(\exists{\mathsf{U}},\exists{\mathsf{X}},\wedge,\neg)$ fragment, $\rep{f}$ is
regular and effectively computable.
\end{theorem}
\section{Solving games on lossy channel systems}
\label{sec-turn-based}
In this section, we consider turn-based games on GLCS's where two players,
$A$ and $B$, alternate their moves. Games play a growing role in
verification where they address situations in which different agents have
different, competing goals. We assume a basic understanding of the
associated concepts: arena, play, strategy, etc. (otherwise
see~\cite{Graedel-Thomas-Wilke-LNCS}).
Games on well-structured systems have already been investigated
in~\cite{abdulla2003b,raskin03,raskin05}. The positive results in these
three papers rely on ad-hoc finite convergence lemmas that are special
cases of our Theorem~\ref{th-guarded-effective}.
\subsection{Symmetric LCS-games with controllable message losses}
\label{ssec-symmetric-games}
We start with the simplest kind of games on a GLCS: $A$ and $B$ play in
turn, choosing the next configuration, i.e., picking what rule
$\delta\in\Delta$ is fired, and what messages are lost.
Formally, a \emph{symmetric LCS-game} is a GLCS $\L=
(Q_A,Q_B,\C,\Mess,\Delta)$ where the set of locations $Q=Q_A \cup Q_B$ is
partitioned into two sets, one for each player, and where the rules ensure
strict alternation: for all $p \step{G:{\mathit{op}}} q \in \Delta$, $p\in Q_A$ iff
$q\in Q_B$. Below, we shortly write ${\mathit{Conf}}_A$ for
$Q_A\times{\Mess^*}^{|\C|}$, the regular region where it is $A$'s turn to
play. ${\mathit{Conf}}_B$ is defined similarly. Strict alternation means that the
arena, $\LTS_\L$, is a bipartite graph partitioned in ${\mathit{Conf}}_A$ and
${\mathit{Conf}}_B$.
\paragraph*{\bf \em Reachability games.}
Reachability and invariant are among the simplest objectives for
games. In a reachability game, $A$ tries to reach a state in some set
$V$, no matter how $B$ behaves. This goal is denoted $\Diamond V$. It
is known that such games are determined and that memoryless strategies
are sufficient~\cite{Graedel-Thomas-Wilke-LNCS}. The set of winning
configurations for $A$ is denoted with $\ex{A} \Diamond V$, and can be
defined in $L_\mu$:
\begin{equation}
\label{eq-reachgame}
\ex{A} \Diamond V =
\mu X.\Bigl[ V \cup \bigl[{\mathit{Conf}}_A \cap \Pre(X)\bigr]
\cup \bigl[{\mathit{Conf}}_B \cap \wPre(X)\bigr]\Bigr].
\end{equation}
The first occurrence of $X$ can be made upward-guarded by replacing
$\Pre(X)$ with $\Pre(C_{\uparrow}(X))$ (Lemma~\ref{lem-pre-guarded}). For the second
occurrence, we can unfold the term, relying on the fixpoint equation
$\rep{\mu X.\phi(X)}=\rep{\mu X.\phi(\phi(X))}$. This will replace
${\mathit{Conf}}_B\cap\wPre(X)$ in \eqref{eq-reachgame} with
\begin{equation}
\tag{+}
\label{eq+}
{\mathit{Conf}}_B\cap \wPre\Bigl(V \cup \bigl[{\mathit{Conf}}_A \cap
\Pre(X)\bigr] \cup \bigl[{\mathit{Conf}}_B \cap \wPre(X)\bigr]\Bigr).
\end{equation}
Now, the strict alternation between ${\mathit{Conf}}_A$ and ${\mathit{Conf}}_B$ lets us simplify
\eqref{eq+} into
\begin{equation}
{\mathit{Conf}}_B \cap \wPre\Bigl(V \cup \Pre(X) \Bigr).
\end{equation}
Hence \eqref{eq-reachgame} can be rewritten into
\begin{gather}
\label{eq-reachgame-var}
\tag{\ref{eq-reachgame}'}
\ex{A} \Diamond V = \mu X.\Bigl[ V \cup \bigl[{\mathit{Conf}}_A \cap
\Pre(C_{\uparrow}(X))\bigr] \cup \bigl[{\mathit{Conf}}_B \cap
\wPre(V\cup \Pre(C_{\uparrow}(X)))\bigr]\Bigr].
\end{gather}
\paragraph*{\bf \em Invariant games.}
In invariant games, $A$'s goal is to never leave some set
$V\subseteq{\mathit{Conf}}$, no matter how $B$ behaves. Invariant games are
dual to reachability games, and the set of winning configurations
$\ex{A}\Box V$ is exactly $\overline{\ex{B}\Diamond \overline{V}}$.
\paragraph*{\bf \em Repeated reachability games.}
Here $A$'s goal is to visit $V$ infinitely many times, no matter how $B$
behaves. The set of winning configurations is given by the following
$L_\mu$ term:
\begin{equation}
\ex{A} \Box \Diamond V = \nu Y. \ex{A}\Diamond \Bigl[ V \cap
(\varphi_A(Y) \cup \varphi_B(Y))\Bigr],
\end{equation}
where
\begin{align*}
\varphi_A(Y) & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} {\mathit{Conf}}_A \cap \Pre\bigl( C_{\uparrow}(\wPre(K_{\downarrow}(Y)))\bigr),
\\
\varphi_B(Y) & \stackrel{\mbox{\begin{scriptsize}def\end{scriptsize}}}{=} {\mathit{Conf}}_B \cap \wPre(K_{\downarrow}(Y)).
\end{align*}
and where we reuse \eqref{eq-reachgame-var} for $\ex{A} \Diamond [\ldots]$.
\paragraph*{\bf \em Persistence games.}
In a persistence game, $A$ aims at remaining inside $V$ from some moment
on, no matter how $B$ behaves. Dually, this can be seen as a repeated
reachability game for $B$. Note that $\ex{A} \Diamond \Box V \neq \ex{A}
\Diamond (\ex{A} \Box V)$.
\begin{theorem}[Decidability of symmetric LCS-games]
For symmetric LCS-games $\L$ and regular regions $V$, the four sets
$\ex{A}\Diamond V$, $\ex{A}\Box V$, $\ex{A}\Diamond\Box V$, and
$\ex{A}\Box\Diamond V$, are (effective) regions. Hence reachability,
invariant, repeated reachability, and persistence symmetric games are
decidable on GLCS's.
\end{theorem}
\begin{proof}[Sketch] The winning sets can be defined by guarded $L_\mu$
terms.
\end{proof}
\begin{remark}
There is no contradiction between the undecidability of
$\exists\Box\Diamond V$ and the decidability of $\ex{A}\Box\Diamond V$. In
the latter case, $B$ does not cooperate with $A$, making the goal harder to
reach for $A$ (and the property easier to decide for us).
\qed
\end{remark}
\subsection{Asymmetric LCS-games with 1-sided controlled loss of messages}
\label{ssec-1sided-games}
Here we adopt the setting considered in~\cite{abdulla2003b}. It varies from
the symmetric setting of section~\ref{ssec-symmetric-games} in that only
player $B$ can lose messages (and can control what is lost), while player
$A$ can only make perfect steps. Note that this generalizes games where $A$
plays moves in the channel system, and $B$ is an adversarial environment
responsible for message losses. We use the same syntax as for symmetric
LCS-games.
\paragraph*{\bf \em Reachability and invariant games.}
Let us first consider games where one player tries to reach a regular
region $V$ (goal $\Diamond V$), no matter how the other player behaves.
The configurations where $B$ can win a reachability game are given by:
\begin{align*}
\begin{split}
\ex{B} \Diamond V
& = \mu X. V \cup \Bigl({\mathit{Conf}}_B \cap \Pre(X)\Bigr) \cup \Bigl({\mathit{Conf}}_A \cap \wPre_\mathrm{perf}(X) \Bigr)\\
& = \mu X. V \cup \Bigl( {\mathit{Conf}}_B \cap \Pre(C_{\uparrow} (X))\Bigr) \cup
\Bigl({\mathit{Conf}}_A \cap \wPre_\mathrm{perf}(V \cup \Pre(C_{\uparrow}(X))) \Bigr)
\end{split}
\end{align*}
where guardedness is obtained via Lemma~\ref{lem-pre-guarded} and
unfolding.
When we consider a reachability game for $A$, the situation is not so clear:
\begin{equation*}
\ex{A} \Diamond V = \mu X. V \cup \Bigl({\mathit{Conf}}_A \cap
\Pre_\mathrm{perf}(X)\Bigr) \cup \Bigl({\mathit{Conf}}_B \cap \wPre(X) \Bigr).
\end{equation*}
Neither Lemma~\ref{lem-pre-guarded} nor unfolding techniques can turn
this into a guarded term. This should be expected since the set
$\ex{A} \Diamond V$ cannot be computed
effectively~\cite{abdulla2003b}.
\begin{theorem}[Decidability of asymmetric LCS-games~\cite{abdulla2003b}]
For asymmetric LCS-games $\L$ and regular regions $V$, the sets
$\ex{B}\Diamond V$ and $\ex{A}\Box V$ are (effective) regions. Hence
reachability games for $B$, and invariant games for $A$ are decidable on
GLCS's.
\end{theorem}
\begin{proof}[Sketch] Invariant games are dual to reachability games, and
the winning set $\ex{B}\Diamond V$ is defined by a guarded $L_\mu$
term.
\end{proof}
\section{Channel systems with probabilistic losses}
\label{sec-prob-lcs}
LCS's where messages losses follow probabilistic rules have been
investigated as a less pessimistic model of protocols with unreliable
channels (see~\cite{Sch-voss,ABRS-icomp,BBS-acmtocl06} and the references
therein).
In~\cite{BBS-acmtocl06}, we present decidability results for LCS's seen as
combining \emph{nondeterministic} choice of transition rules with
\emph{probabilistic} message losses. The semantics is in term of Markovian
decision processes, or $1\frac{1}{2}$-player games, whose solutions can be
defined in $L_\mu$. Indeed, we found the inspiration for $L_\mu$ and our
Theorem~\ref{th-guarded-effective} while extending our results in the MDP
approach to richer sets of regions.
In this section, rather than rephrasing our results on
$1\frac{1}{2}$-player games on LCS's, we show how to deal with
$2\frac{1}{2}$-player games~\cite{chatterje05} on LCS's, i.e., games
opposing players $A$ and $B$ (as in section~\ref{sec-turn-based}) but where
message losses are probabilistic.
\\
Formally, a \emph{symmetric probabilistic LCS-game}
$\L=(Q_A,Q_B,\C,\Mess,\Delta)$ is exactly like a symmetric LCS-game but
with an altered semantics: in state $\sigma\in{\mathit{Conf}}_A$, player $A$ selects
a fireable rule $\delta\in\Delta$ ($B$ picks the rule if
$\sigma\in{\mathit{Conf}}_B$) and the system moves to a successor state $\rho$ where
$\sigma\step{\delta}_\mathrm{perf}\sigma'\sqsupseteq\rho$ and $\rho$ is chosen
probabilistically in $C_{\downarrow}(\{\sigma'\})$. The definition of the probability
distribution $\bfP(\sigma,\delta,\rho)$ can be found
in~\cite{Sch-voss,BBS-acmtocl06} where it is called \emph{the local-fault
model}. It satisfies $\bfP(\sigma,\delta,\rho)>0$ iff
$\rho\sqsubseteq\sigma'$ (assuming $\sigma\step{\delta}_\mathrm{perf}\sigma'$).
Additionally it guarantees a \emph{finite-attractor property}: the set of
states where all channels are empty will be visited infinitely many times
almost surely~\cite{ABRS-icomp,BBS-ipl}.
\paragraph*{\bf \em Reachability games.}
Assume $A$ tries to reach region $V$ (goal $\Diamond V$) \emph{with
probability 1} no matter how $B$ behaves. The set $\ex{A} [\Diamond
V]_{=1}$ of states in which $A$ has an almost-sure winning strategy is
given by
\begin{equation}
\label{eq-reach-P1}
\ex{A} [\Diamond V]_{=1} =
\nu Y. \mu X. \left(
\begin{array}{rl}
V \cup & \Bigl[{\mathit{Conf}}_A \cap \Pre_\mathrm{perf} (C_{\uparrow}(X) \cap K_{\downarrow}(Y))\Bigr]
\\[.5em]
\cup & \Bigl[{\mathit{Conf}}_B \cap \wPre_\mathrm{perf} (C_{\uparrow}(X) \cap K_{\downarrow}(Y))\Bigr]
\end{array}
\right).
\end{equation}
\begin{remark}
Justifying \eqref{eq-reach-P1} is outside the scope of this paper, but we
can try to give an intuition of why it works: the inner fixpoint ``$\mu X.
V\cup \cdots$'' define the largest set from which $A$ has a strategy to
reach $V$ no matter what $B$ does \emph{if the message losses are
favorable}. However, whatever messages are lost, $A$'s strategy also
guarantees that the system will remain in $Y$, from which it will be
possible to retry the strategy for $\Diamond V$ as many times as necessary.
This will eventually succeed almost surely thanks to the finite-attractor
property.
\qed
\end{remark}
\paragraph*{\bf \em Invariant games.}
Assume now $A$ tries to stay in $V$ almost surely (goal $[\Box V]_{=1}$), no
matter how $B$ behaves. Then $A$ must ensure $\Box V$ surely and
we are considering a 2-player game where message losses are adversarial
and could as well be controlled by $B$. This leads to
\begin{equation}
\label{eq-Box=1}
\begin{split}
\ex{A}[\Box V]_{=1}
& = \nu X. V \cap
\Bigl(\bigl[{\mathit{Conf}}_A \cap
\Pre_\mathrm{perf}(K_{\downarrow}(X))\bigr] \cup \bigl[{\mathit{Conf}}_B \cap \wPre(X)\bigr]\Bigr)
\\
& = \nu X. V \cap
\Bigl(\bigl[{\mathit{Conf}}_A \cap
\Pre_\mathrm{perf}(K_{\downarrow}(X))\bigr] \cup \bigl[{\mathit{Conf}}_B \cap \wPre(K_{\downarrow}(X))\bigr]\Bigr).
\end{split}
\end{equation}
In \eqref{eq-Box=1}, the subterm $\Pre_\mathrm{perf}(K_{\downarrow}(X))$
accounts for states in which $A$ can choose a perfect move that will end
in $K_{\downarrow}(X)$, i.e., that can be followed by any adversarial message losses
and still remain in $X$. The subterm $\wPre(X)$ accounts for states
in which $B$ cannot avoid going to $X$, even with message losses under his
control. $\wPre(X)$ can be rewritten into
$\wPre(K_{\downarrow}(X))$ thanks to Lemma~\ref{lem-pre-guarded}, so that we end up
with a guarded term.
\paragraph*{\bf \em Goals to be satisfied with positive probability.}
In $2\frac{1}{2}$-player games, it may happen that a given goal can only be
attained with some non-zero probability~\cite{chatterje05}. Observe that,
since the games we consider are determined~\cite{martin98}, the goals
$[\Diamond V]_{>0}$ or $[\Box V]_{>0}$ are the opposite of goals asking for
probability 1:
\begin{xalignat*}{2}
\ex{A}[\Diamond V]_{>0} &= \overline{\ex{B}[\Box \overline{V}]_{=1}},
& \quad
\ex{A}[\Box V]_{>0} &= \overline{\ex{B}[\Diamond \overline{V}]_{=1}}.
\end{xalignat*}
\begin{theorem}[Decidability of qualitative symmetric probabilistic LCS-games]
For symmetric probabilistic LCS-games $\L$ and regular regions $V$, the
sets $\ex{A}[\Diamond V]_{=1}$, $\ex{A}[\Diamond V]_{>0}$, $\ex{A}[\Box
V]_{=1}$, and $\ex{A}[\Box V]_{>0}$ are (effective) regions. Hence
qualitative reachability and invariant games are decidable on GLCS's.
\end{theorem}
\begin{proof}[Sketch]
These sets can be defined by guarded $L_\mu$ terms.
\qed
\end{proof}
\iffalse
\paragraph*{\bf \em Persistence games.}
$A$ attempts to ensure $\Diamond \Box V$ with probability 1,
no matter how $B$ behaves.
$$\ex{A} [\Diamond \Box V]_{=1}
\stackrel{\text{def}}{=}
\ex{A}\bigl[ \Diamond \ex{A} [\Box V]_{=1} \bigr]_{=1}
$$
$A$ attempts to ensure $\Diamond \Box V$ with probability $>0$,
no matter how $B$ behaves.
$$\ex{A} [\Diamond \Box V]_{>0}
\stackrel{\text{def}}{=}
\ex{A}\bigl[ \Diamond \ex{A} [\Box V]_{=1} \bigr]_{>0}
$$
\paragraph*{\bf \em Repeated reachability games.}
$A$'s goal is to visit $V$ infinitely often with probability 1,
no matter how $B$ behaves.
$$\ex{A} [\Box \Diamond V]_{=1}
\stackrel{\text{def}}{=}
\nu Y. \ex{A} [\Diamond (V \cap (\varphi_A(Y) \cup \varphi_B(Y)))]_{=1}
$$
where
$$\begin{array}{lcl}
\varphi_A(Y) & = & Q_A \cap \Pre_{\text{perfect}}
(K_{\downarrow}(Y)) \\
\varphi_B(Y) & = & Q_B \cap
\wPre_{\text{perfect}}(K_{\downarrow}(Y))
\end{array}
$$
\paragraph*{\bf \em Parity games.}
.....
\fi
\section{Conclusion}
\label{sec-concl}
We defined a notion of upward/downward-guarded fixpoint expressions that
define subsets of a well-quasi-ordered set. For these guarded fixpoint
expressions, a finite convergence theorem is proved, that shows how the
fixpoints can be evaluated with a finite number of operations. This has a
number of applications, in particular in the symbolic verification of
well-structured systems, our original motivation. We illustrate this in the
second part of the paper, with lossy channel systems as a target. For these
systems, we derive in an easy and uniform way, a number of decidability
theorems that extend or generalize the main existing results in the
verification of temporal properties or game-theoretical properties.
These techniques can be applied to other well-structured systems, with a
region algebra built on, e.g., upward-closed sets. Admittedly, many
examples of well-structured systems do not enjoy closure properties as nice
as our Lemma~\ref{lem-pre-guarded} for LCS's, which will make it more
difficult to express interesting properties in the guarded fragment of
$L_\mu$. But this can still be done, as witnessed
by~\cite{raskin03,raskin05} where the authors introduced a concept of
$B$-games and $BB$-games that captures some essential closure assumptions
allowing the kind of rewritings and unfoldings we have justified with
Lemma~\ref{lem-pre-guarded}.
|
{
"redpajama_set_name": "RedPajamaArXiv"
}
| 1,655
|
{"url":"http:\/\/www.ipam.ucla.edu\/abstract\/?tid=15318&pcode=HIL2019","text":"Arithmetic subspaces of moduli spaces of rank one local systems\n\nH\u00e9l\u00e8ne EsnaultFreie Universit\u00e4t Berlin\n\nWe show that Arithmetic subspaces of moduli spaces of rank one local systems are motivic (joint with Moritz Kerz).\n\nBack to Braids, Resolvent Degree and Hilbert's 13th Problem","date":"2022-05-19 04:55:52","metadata":"{\"extraction_info\": {\"found_math\": false, \"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\": 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\": 0.8519283533096313, \"perplexity\": 2360.7989524350396}, \"config\": {\"markdown_headings\": false, \"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-2022-21\/segments\/1652662525507.54\/warc\/CC-MAIN-20220519042059-20220519072059-00176.warc.gz\"}"}
| null | null |
North Austin Branch Library will host an African dance fusion session.
Austin-based BUILD Chicago will host an art show benefit fundraiser.
On Wednesday, Feb. 20, 5:30 p.m., at the North Austin Branch Library, 5724 W. North Ave., the library will host an African dance fusion session. It's a unique blend of African dance, South American movement and Egyptian dance, spiced up with a hint of Jazz, R&B and Line dance. African Dance Fusion is certain to get your body moving and your heart pumping. All levels are welcome. For more info, call (312) 746-4233.
On Saturday, Feb. 23, 6 p.m. to 9 p.m., at Chicago's West Loop, 1133 W. Fulton Market, Austin-based BUILD Chicago will host an art show benefit fundraiser to support BUILD's arts programs. All art is on silent auction and merchandise is for sale.
|
{
"redpajama_set_name": "RedPajamaC4"
}
| 8,275
|
{"url":"https:\/\/math.stackexchange.com\/questions\/2197404\/is-a-a-proper-set-of-its-power-set","text":"# Is A a proper set of its power set?\n\nI originally believed that the answer is yes, as all elements of A are subsets of A that would be included in its power set. And, the empty set would be in the power set but not in A. However, the textbook states that this is sometimes but more often not the case. Would someone mind explaining?\n\n\u2022 What is a proper set? \u2013\u00a0Jacob Wakem Mar 22 '17 at 0:03\n\nNo. There is a world of difference between a thing, and a set containing that thing. It is the same as the difference between a cat and the word cat.\n\nTake a set of children: A = {Sam, Claire}. Its power set is {{Sam, Claire}, {Sam}, {Claire}, {}}. Note that every element of the power set of A is a set. Every element of A is a child. Therefore A is not a subset of its power set, and not being a subset is not a proper set either.\n\n\u2022 Unfortunately for your explanation, set theory deals with pure sets that don't involve cats or children: if $\\mathtt{Sam} = \\emptyset$ and $\\mathtt{Claire} = \\{\\emptyset\\}$ and $A = \\{\\mathtt{Sam}, \\mathtt{Claire}\\} = \\{\\emptyset, \\{\\emptyset\\}\\}$ then $A \\subseteq \\Bbb{P}(A)$ \u2013\u00a0Rob Arthan Mar 21 '17 at 23:50\n\u2022 @RobArthan The standard axioms of set theory do not construct or deny the possibility of constructing sets out of things that have nothing to do with sets. (The existence or not of such sets is independent of ZFC.) But the normal usage of the word set in mathematics supports it, and it leads to clear intuition about the result. \u2013\u00a0btilly Mar 22 '17 at 3:26\n\u2022 well actually, the axiom of foundation effectively does deny that possibility. However, I did not write that set theory deals exclusively with pure sets. \u2013\u00a0Rob Arthan Mar 22 '17 at 6:24\n\u2022 @RobArthan Upon a close read you are right and I was wrong. I still think that my answer was more likely to be helpful, but yours is clearly more accurate in every key respect. \u2013\u00a0btilly Mar 23 '17 at 5:25\n\nSets that are subsets of their power sets are quite special. They are called transitive sets. If you work through the definitions you will find that $A \\subseteq \\Bbb{P}(A)$ means that for any $x \\in A$ and for any $y \\in x$, $y \\in A$, i.e., it says that members of members of $A$ are members of $A$ (hence the name transitive sets, because the membership relation on the members of $A$ and their members is transitive). If (as is usual in set theory) you use the von Neumann representation of the natural numbers: \\begin{align*} 0 &= \\emptyset \\\\ 1 &= \\{0\\} \\\\ 2 &= \\{0, 1\\}\\\\ &\\ldots \\end{align*} then the natural numbers are all represented by transitive sets while $\\{0, 2\\}$ gives you the simplest example of a set that is not transitive (because $2 \\in \\{0, 2\\}$ but $2 = \\{0, 1\\} \\not\\subseteq \\{0, 2\\}$).\n\n\u2022 This is the correct answer. \u2013\u00a0ChocolateAndCheese Mar 22 '17 at 0:03\n\u2022 @ChocolateAndCheese This is the answer that can be used to formally prove the existence of non-transitive sets from your favorite axiom system. For example ZF. However the formalism won't help someone who is lost and confused by the terminology. I was aiming to help intuition. \u2013\u00a0btilly Mar 22 '17 at 3:32\n\u2022 @btilly I agree that your accepted answer is most likely what the OP was looking for, however, that does not change that fact that it is not entirely correct. Set theory can be surprisingly subtle and often non-intuitive, and mistakes get made when we oversimplify. \u2013\u00a0ChocolateAndCheese Mar 22 '17 at 16:53\n\nSupppose $\\;\\;\\mathcal{P}(A)\\;\\;\\;$is the power set of $\\;\\;A\\;.\\;\\;$ Generally, the elements of $\\;\\;A\\;\\;$ and that of $\\;\\;\\mathcal{P}(A)\\;\\;\\;$ are not comparable and they are of different types. For example, if $\\;\\;A=\\{\\;1\\;,\\;2\\;\\}\\;\\;\\;$ then $\\;\\;\\mathcal{P}(A)\\;=\\;\\{\\;\\Phi\\;,\\;\\{1\\}\\;,\\;\\{2\\}\\;,\\;\\{1,2\\}\\;\\}\\;\\;\\;$and in this case $\\;A\\;\\not \\subset\\;\\mathcal{P}(A)\\;.\\;\\;$\n\nBut, we can easily construct an example of class in which an element can also be considered as a subset of the power set. Example, $\\;A\\;=\\{\\Phi\\;,\\;\\{\\Phi\\}\\;,\\;\\{\\Phi\\;,\\;\\{\\Phi\\}\\}\\;,...\\}\\;$\n\nThe indistinguishability of elements and subsets of classes leads to logical inconsistency, in general discussions, such classes are avoided by (restricting the class ) certain axioms like no set is an element of itself, no element of a set is subset of that set etc. The use of null set is also confusing for the beginners,as it is of different nature in different contexts, regarding the nature of its elements (though it has not a single element!)\n\n\u2022 How does this lead to logical inconsistencies? Also, such sets are not avoided by the usual axioms of ZFC, see Rob's answer. \u2013\u00a0Christian Matt Mar 22 '17 at 0:49\n\u2022 Maybe you want to use \\empyset instead of \\Phi to denote the empty set. \u2013\u00a0martin.koeberl Mar 22 '17 at 1:21","date":"2019-10-21 22:52:46","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\": 0.9598293304443359, \"perplexity\": 273.8097520064987}, \"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-2019-43\/segments\/1570987795253.70\/warc\/CC-MAIN-20191021221245-20191022004745-00441.warc.gz\"}"}
| null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.