text
stringlengths
30
4k
source
stringlengths
60
201
} • Path to accepting state: ε a Æ a Æ a Æ e Æ f Æ g 1 0 0 0 Viewing computations as a tree 0, 1 a ε ε b e 0 1 c f 1 0 d g Input w = 01 a 0 In general, accept if there is a path labeled by the entire input string, possibly interspersed with εs, leading to an accepting state. Here, lea...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
it isn’t accepted. • L(M), the language recognized by NFA M, = { w | w is accepted by M}. NFAs vs. FAs NFAs vs. DFAs • DFA = Deterministic Finite Automaton, new name for ordinary Finite Automata (FA). – To emphasize the difference from NFAs. • What languages are recognized by NFAs? • Since DFAs are special ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
δ1( r, a ) ) • Starting from states in S, δ2( S, a ) gives all states M1 could reach after a and possibly some ε-transitions. – M2 recognizes L(M1): At any point in processing the string, the state of M2 represents exactly the set of states that M1 could be in. Example: NFA Æ DFA • M1: 0,1 0 a b 1 c • S...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
ited Closure under operations • The last example suggests we retry proofs of closure of FA languages under concatenation and star, this time using NFAs. • OK since they have the same expressive power (recognize the same languages) as DFAs. • We already proved closure under common set- theoretic operations---u...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
2), a) = (δ1(q1, a), δ2(q2, a)) – q03 = (q01, q02) – F3 = { (q1,q2) | q1 ∈ F1 or q2 ∈ F2 } Closure under union • Theorem: FA-recognizable languages are closed under union. • New Proof: – Start with NFAs M1 and M2. – Get another NFA, M3, with L(M3) = L(M1) ∪ L(M2). ε Add new start state ε M1 M2 Use f...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
* = { x | x = y1 y2 … yk for some k ≥ 0, every y in L } = L0 ∪ L1 ∪ L2 ∪ … • Theorem: FA-recognizable languages are closed under • star. Proof: – Start with FA M1. – Get an NFA, M2, with L(M2) = L(M1)*. ε M1 Add new start state; it’s also a final state, since ε is in L(M1)*. ε Use final states from M1 an...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
– ε, – ∅, – ( R1 ∪ R2 ), where R1 and R2 are smaller regular expressions, – ( R1 ° R2 ), where R1 and R2 are smaller regular expressions, or – ( R1* ), where R1 is a smaller regular expression. • A recursive definition. Regular expressions • Definition: R is a regular expression over alphabet Σ exactly if...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
R1* ) = ( L(R1) )* • Example: Expression ( ( 0 ∪ 1 ) ε )* ∪ 0 denotes language { 0, 1 }* ∪ { 0 } = { 0, 1 }*, all strings. • Example: ( 0 ∪ 1 )* 111 ( 0 ∪ 1 )* denotes { 0, 1 }* { 111 } { 0, 1 }*, all strings with substring 111. More examples • Definition: – L(a) = { a }; one string, with one symbol a. ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
� 1 )* Abbreviate: Σ* 01 Σ* ∪ Σ* 10 Σ* • Example: L = strings with neither substring 01 or 10. – Can’t write complement. – But can write: 0* ∪ 1*. • Example: L = strings with no more than two consecutive 0s or two consecutive 1s – Would be easy if we could write complement. ( ε ∪ 1 ∪ 11 ) (( 0 ∪ 00 ) (1 ∪ ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
of R: • Show for the three base cases. • Show how to construct NFAs for more complex expressions from NFAs for their subexpressions. – Case 1: R = a • L(R) = { a } a – Case 2: R = ε • L(R) = { ε } Accepts only a. Accepts only ε. Theorem 1 • Theorem 1: If R is a regular expression, then L(R) is a regul...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
: b: a • ab: • a*: a ε • ab ∪ a*: ε a ε ε ε b a ε b ε a ε b Theorem 2 • Theorem 2: If L is a regular language, then there is a regular expression R with L = L(R). • Proof: – For each NFA M, define a regular expression R with L(R) = L(M). – Show with an example: b a a x a b b y z – Convert ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
. Theorem 2 a ∪ bb* a b*a q0 b a* y qf • Finally, remove y: b*a (a ∪ bb* a)* b a* q0 qf • New label describes all strings that can move the machine from q0 to qf, visiting (just) y any number of times. • This final label is the needed regular expression. Theorem 2 • Define a generalized NFA (gNFA). ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
Lemma • Algorithms that answer questions about FAs. • Reading: Sipser, Section 1.4; some pieces from 4.1 MIT OpenCourseWare http://ocw.mit.edu 6.045J / 18.400J Automata, Computability, and Complexity Spring 2011 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b6d125bc94344e53285a1cdbdb33371_MIT6_045JS11_lec04.pdf
6.045: Automata, Computability, and Complexity Or, Great Ideas in Theoretical Computer Science Spring, 2010 Class 7 Nancy Lynch Today • Basic computability theory • Topics: – Decidable and recognizable languages – Recursively enumerable languages – Turing Machines that solve problems involving FAs – Undecidability ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
Definition: In fact, these two notions define different language classes: – L is Turing-recognizable – L is Turing-decidable if there is some TM that recognizes L. if there is some TM that decides L. • The classes of Turing-recognizable and Turing-decidable languages are different. • Theorem 2: If L is Turing-decid...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
Idea: Run both M1 and M2 on w. • One must accept. • If M1 accepts, then M accepts. • If M2 accepts, then M rejects. – But, we can’t run M1 and M2 one after the other because the first one might never halt. – Run them in parallel, until one accepts? – How? We don’t have a parallel Turing Machine model. Decidable and r...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
to renaming). Examples • Example: Every regular language L is decidable. – Let M be a DFA with L(M) = L. – Design a Turing machine M′ that simulates M. – If, after processing the input, the simulated M is in an accepting state, M′ accepts; else M′ rejects. Examples • Example: Let X = be the set of binary represen...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
sequence of finite strings (possibly infinitely many) • • • on output tape. It may print the same string more than once. If E is an enumerator, then define L(E) = { x | x is printed by E }. If L = L(E) for some enumerator E, then we say that L is recursively enumerable (r.e.). Recursively enumerable languages • Int...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
steps for 1st, 2nd, and 3rd inputs, ε, 0 and 1. • … • Run more and more steps for more and more inputs. – Eventually succeeds in reaching qacc for each accepting computation of M, so enumerates all elements of L. Recursively enumerable languages • Theorem 8: L is recursively enumerable if and only if L is Turing-rec...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
Turing Machines that solve graph problems • Consider the problem of whether a given graph has a cycle of length > 2. • Graph = (V,E), V = vertices, E = edges, undirected. • Write <G> for the encoding of G. • Using this representation for the input, we can write an algorithm to determine whether or not a given graph...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
inputs using standard encodings and formalize the algorithms that we’ve already discussed, using Turing machines. Turing Machines that solve DFA problems • Example: Equivalence for DFAs L2 = { < M1, M2 > | L(M1) = L(M2) } is Turing-decidable. • Elements of L2 are bit-string representations of pairs of DFAs that rec...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
DFA and L(M) = ∅ }, decidable by Turing machine that searches M’s digraph. { < M, w > | M is a DFA, w is a word in M’s alphabet, and w ∈ L(M) }, decidable by a Turing machine that emulates M on w. • Turing machines compute only on strings, but we can regard them as computing on DFAs by encoding the DFAs as strings (...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
accepts exactly < M, w> encodings for which M accepts w. • U is sometimes called a universal Turing machine because it runs all TMs. – Like an interpreter for a programming language. The Acceptance Problem • AccTM = { < M, w > | M is a TM and M accepts w }. • U: On input < M, w >: – Simulate M on input w. – If M acce...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
H(<M,w>) accepts if M accepts w, rejects if M rejects w or if M loops on w. – Use H to construct another TM H′ that decides a special case of the same language. – Instead of considering whether M halts on an arbitrary w, just consider M on its own representation: – H′(<M>): • accepts if M accepts <M>, • rejects if ...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
, with TMs labeling rows and strings that represent TMs labeling columns. • The major diagonal describes results for M(<M>), for all M. • D is a diagonal machine, constructed explicitly to differ • from the diagonal entries: D(<M>)’s result differs from M(<M>)’s. Implies that D itself can’t appear as a label for a r...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
: – Suppose for contradiction that HaltTM is Turing- decidable, say by Turing machine R: • R(<M,w>): – accepts if M halts on (accepts or rejects) w, – rejects if M loops (neither accepts nor rejects) on w. – Using R, define new TM S to decide AccTM: • S: On input <M,w>: – Run R on <M,w>; R must either accept or reject...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
Suppose HaltTM is Turing-decidable by TM R. – S: On input <M,w>: – Run R on <M,w>; R must either accept or reject; can’t loop, by definition of R. – If R accepts then M must halt (accept or reject) on w. Then simulate M on w, knowing this must terminate. If M accepts, accept. If rejects, reject. – If R rejects, then...
https://ocw.mit.edu/courses/6-045j-automata-computability-and-complexity-spring-2011/5b9f68b5081985b135f6ad521a93d81d_MIT6_045JS11_lec07.pdf
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Lecture 3 Fall 2018 CONDITIONING AND INDEPENDENCE Most of the material in this lecture is covered in [Bertsekas & Tsitsiklis] Sec- tions 1.3-1.5 and Problem 48 (or problem 43, in the 1st edition), available at http://athenasc.com/Prob-2nd-Ch1.pdf. Solutions to ...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
we have ∪∞P i=1 � Ai | B =  ∞ X i=1 P(Ai | B). As a result, suppose PB : F → [0, 1] is defined by PB(A) = P(A | B). Then, PB is a probability measure on ( , F). (b) Let A be an event. If the events Bi, i ∈ N, form a partition of P(Bi) > 0 for every i, then , and P(A) = ∞ X i=1 P(A | Bi)P(Bi). In parti...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
ned. Proof. (a) We have P( | B) = P( ∩ B)/P(B) = P(B)/P(B) = 1. Also Ω Ω ∪∞P i=1 � P B ∩ (∪∞ Ai) �  P(B) Since the sets B ∩ Ai, i ∈ N are disjoint, countable additivity, applied to the P(∪∞ (B ∩ Ai)) P(B) Ai | B = i=1 i=1 =  . 2 right-hand side, yields ∪∞ i=1 P � Ai | B =  P ∞ i=1 P(B ∩ Ai) ...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
and the result from part (c). (d) Note that the sequence of events ∩n is decreasing and converges to ∩∞ Ai. By the continuity property of probability measures, we have P ∩∞ i=1 i=1 Ai = limn→∞ P ∩i n =1 Ai . Note that i=1Ai �  �  ∩n i=1 P � Ai = P(A1) ·  P(A1) · P(A1 ∩ A2) P(A1 ∩ A2) P(A1 ∩ A2 ∩ A3...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
condition is P(A) = P(A | B). ⊥ (b) Let S be an index set (possibly infinite, or even uncountable), and let {As | s ∈ S} be a family (set) of events. The events in this family are said to be independent if for every finite subset S0 of S, we have P ∩s∈S0 As = �  P(As). Y s∈S0 (c) Let F1 ⊂ F and F2 ⊂ F be two ˙-...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
the ith toss resulted in a “1”. If i 6= j, the events Ai and Aj are independent. (b) The events in the (infinite) family {Ai | i ∈ N} are independent. This statement captures the intuitive idea of “independent” coin tosses. (c) Let Fn be the collection of all events whose occurrence can be decided by looking at the...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
F2 turn out to be independent. This statement captures the intuitive idea that knowing the results of the tosses at odd times provides no information on the results of the tosses at even times. 4 2.1 How to check independence of ˙-algebras? p-systems. How can one establish that two complicated ˙-fields (e.g., as ...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
, A, B ∈  ⇒ A ∩ B ∈ ) is called a p-system. Examples of p-systems are intervals (a, b) on R, rectangles (a, b) × (c, d) on R2 , convex sets in Rd , etc. Theorem 2. Let 1 and 2 be p-systems and Fi = ˙(i), i = 1, 2. If P(A ∩ B) = P(A)P(B) (1) for every A ∈ 1, B ∈ 2, then F1 and F2 are independent. Proof. F...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
P(B)P(A) , n→∞ n→∞ implying A ∈ LB. Similar argument holds for An ց A. 5 It turns out that 1 and 2 imply that LB contains (1) (see Proposition to follow). Thus by the monotone class theorem LB = F1. Thus (1) holds for all A ∈ F1 and B ∈ 2. By symmetry it also holds for all A ∈ 1 and B ∈ F2. And applying th...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
it is sufficient to prove D0 is a p- system. Fix C ∈  and let LC = {A ∈ D0 : A ∩ C ∈ D0} On one hand, LC contains  and . On the other hand, LC is closed under punching holes: For A ⊂ B we have (B \ A) ∩ C = (B ∩ C) \ (A ∩ C). Thus LC = D0 by minimality of D0. Hence D0 is closed under intersections with elements ...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
form) (C) consists of Ø, and all sets that can be written in the Ω α (A1,1 ∩ . . . ∩ A1,n1 ) ∪ . . . ∪ (Am,1 ∩ . . . ∩ Am,nm ) , with Ai,j ∈ C or Ac ∈ C. i,j 6 3 THE BOREL-CANTELLI LEMMA The Borel-Cantelli lemma is a tool that is often used to establish that a certain event has probability zero or...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
(b) If P Remark: The result in part (b) is not true without the independence assumption. Indeed, consider an arbitrary event C such that 0 < P(C) < 1 and let An = C for all n. Then P({An i.o.}) = P(C) < 1, even though P(An) = ∞. n P The following lemma is useful here and in may other contexts. Lemma 1. Suppose ...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
This is true for every k. By taking the limit as k → ∞, we obtain log pi) = −∞, and (1 − pi) = 0. For the converse statement, note that under pi < 1 we have ∞ i=1 Q ∞ i=1 (1 − Q ∞ (1 − pi) = 0 ⇐⇒ ∀n Y i=1 ∞ (1 − pi) = 0 Y i=n If pi 6→ 0 the result is automatic. Hence, we may also assume pi → 0. Then takin...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
P ∞ n=1 ∞ i=n ∞ P(A) ≤ P � ∪∞ i=n Ai ≤  P(Ai). X i=n We take the limit of both sides as n → ∞. Since the right-hand side con- verges to zero, P(A) must be equal to zero. 8 (b) Let Bn = ∪∞ Ai, and note that A = ∩∞ Bn. We claim that P(Bc ) = 0. i=n n=1 n This will imply the desired result because P(Ac) ...
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
m Y � i=n 1 − P(Ai) = 0,  where the second equality made use of the continuity property of probability measures. 9 MIT OpenCourseWare https://ocw.mit.edu 6.436J / 15.085J Fundamentals of Probability Fall 2018 For information about citing these materials or our Terms of Use, visit: https://ocw.mit.edu/terms
https://ocw.mit.edu/courses/6-436j-fundamentals-of-probability-fall-2018/5ba99e3803701c145f1d06c26dcf61b8_MIT6_436JF18_lec03.pdf
6.581 / 20.482 J Foundations of Algorithms and Computational Techniques in Systems Biology Adjoint Sensitivity Analysis for Optimization From: Cao Y, Li ST, Petzold L, Serban R, Adjoint sensitivity analysis or differential-algebraic equations: The adjoint DAE system and its numerical solution, SIAM Journal on Scie...
https://ocw.mit.edu/courses/20-482j-foundations-of-algorithms-and-computational-techniques-in-systems-biology-spring-2006/5bb25cdfb8333d9ef2d0631fcffc8e01_adjoint.pdf
& x dp { p dp { { Easy { { Easy + F x dx& dp Hard Hard Hard dt 2 6.581J / 20.482JFoundations of Algorithms and Computational Techniques in Systems BiologyProfessor Bruce TidorProfessor Jacob K. White Trick 2: Integration By Parts dG dp = T ∫0 g dx x dp { Hard dx + g p +λ*(F + F dt ) x& p dp ...
https://ocw.mit.edu/courses/20-482j-foundations-of-algorithms-and-computational-techniques-in-systems-biology-spring-2006/5bb25cdfb8333d9ef2d0631fcffc8e01_adjoint.pdf
⎝ dx & dp x t T= t =0 g − x T d (λ Fx& ) +λ F = 0 dt T x x x T T ⎧ d (λ Fx& ) = λ F + g ⎪ dt ⎪ dG⎨ ⎪ ⎪ dp ⎩ ( g p +λ F p ) ∫0 14243 Easy = T T dt + ⎛ ⎜λT F x& ⎝ dx dp t T= t =0 3 6.581J / 20.482JFoundations of Algorithms and Computational Techniques in Systems BiologyProfessor Bruce TidorProfessor Jac...
https://ocw.mit.edu/courses/20-482j-foundations-of-algorithms-and-computational-techniques-in-systems-biology-spring-2006/5bb25cdfb8333d9ef2d0631fcffc8e01_adjoint.pdf
TidorProfessor Jacob K. White What Do you Need To Know? For Your Dynamic System: ( , f x t , ( , f x t x f p x t ( , + p A x A x x B u B2 x ) = 1 + 2 ⊗ + 1 ) = ⊗ x p A A I , + ⊗ ) + + 2 ( x I ⊗ 1 ( p) ⊗ + 1 p A x A x x B u B u ( ) = 1 + 1 B I 2 + 1 u ⊗ ( p) , p ( ) u For Your Objective Function...
https://ocw.mit.edu/courses/20-482j-foundations-of-algorithms-and-computational-techniques-in-systems-biology-spring-2006/5bb25cdfb8333d9ef2d0631fcffc8e01_adjoint.pdf
Testing 6.170 Lecture 5 Fall 2005 Reading: Chapter 10 of Program Development in Java by Barbara Liskov 1 Program verification techniques and input space partitioning The goal of testing — like that of all other program verification techniques — is to ensure that a program functions correctly. Testing cannot prove ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
check each output for correctness, then we are guaranteed that the program is correct: it operates correctly no matter what input it is given. Checking the output is usually done by recording the expected answer and comparing the actual answer against that, though the checking may also be done procedurally. The desi...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
the equivalence group, and if it works for any member of the equivalence group, then it works for all members. Testing one member is equivalent, in terms of finding errors, to testing every member of the group. Thus, a test suite that includes one (arbitrary) input from each equivalence group provides complete testi...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
to partition the input space. The end result is the same: we have guidance about how to build a test suite. Three key heuristics for test suite selection are coverage, boundary cases, and duplicates. • Coverage This is the key heuristic used in testing. Coverage criteria aims to make some input partition (some test...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
domain requires having an element in each part of it. For a monolithic domain, this requires only one test case. Some domains have a natural structure, and in such a case 2 all parts should be covered. For instance, given an integer output, it would be wise to supply negative, positive, and zero inputs. • Boundary...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
� 22� 23� Actual partition� As another example, you might reason that another program behaves one way for inputs less than 10 and another way for inputs greater than or equal to 10; you presume that the test suite consisting of 5 and 15 is exhaustive. Suppose that your reasoning was faulty and the program has thre...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
) is sometimes called “stress testing.” It is particularly useful when the system under test does not check its representation invariants, preconditions, or postconditions. If something goes wrong in such a system, it may not be immediately apparent. A lengthy run of the program gives the problem time to manifest it...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
you should take their cross-product; for instance, if one heuristic suggests two partitions (say, over one input) and another suggests three partitions (say, over another input), then the test suite should include at least six test cases. 3 Black-box testing Black-box test cases are generated by examining the spec...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
. Supplying Integer.MIN VALUE would uncover a bug in our implementation: no Java function can satisfy the specification, because Integer.MIN VALUE = -2147483648 and Integer.MAX VALUE = 2147483647. – output: if the output is viewed as the ints, we would attempt to produce outputs of Integer.MIN VALUE and Integer.MAX...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
// returns the length of the longest stutter in s int stutterCount(String s) We can apply the three heuristics: • coverage – input: include some test; the string input domain is monolithic. (Inputting the empty string is a wise choice and implicitly suggested; it will also be covered by other heuristics, below.) ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
they don’t hurt. • duplicates A string (of whatever length) containing only a single character achieves the goal of making the stutter be the same as the whole string. Another kind of duplication is having multiple stutters of the same length, either with the same character or with different char­ acters. Such stutte...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
places. Consider the following routine: 6 // returns the maximum of its arguments int max3(int x, int y, int z) { if (x>y) { if (x>z) return x; else return z; } else { if (y>z) return y; else return z; } } The declarative specification is much better than an operational (procedural) one, but it only suggests three ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
achieve branch coverage (and those two cases achieve every variety of glass-box coverage). Likewise, branch coverage does not imply path coverage. Two executions can ensure branch (and statement) coverage of if (p1) a; else b; if (p2) c; else d; if the first sets p1 and p2 to true and the second sets them to false. ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
may read about various types or levels of testing, such as unit, module, system, and acceptance testing. These refer to testing larger and larger portions of an entire system and are described in many software engineering books. Testing can be performed either bottom-up, in which subpieces are verified before they ar...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
of this approach is that design errors can be identified early rather than only when most of the system is built (and the pieces still don’t work together). In general, the right approach to testing is to do the variety which suits your development style. Testing as you code is more effective because you remember the ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
are violated, then trouble is very likely later on. The more quickly such an error is detected (in terms of how many lines of code have executed between the occurrence of the error and its detection), the easier it is to locate. Locating an error is frequently much, much harder than fixing it. 6 Static verification ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
programmers to realize that their implementation does not handle a particular case. • ask a friend. Code inspections, in which a programmer who did not write the code reads and reasons about it, are remarkably effective at finding errors. You aren’t able to do this in the first half of 6.170 — that’s what the LAs and ...
https://ocw.mit.edu/courses/6-170-laboratory-in-software-engineering-fall-2005/5bb92f87c17d1cb32a055db3af9cb4cc_lec5.pdf
2.094 — Finite Element Analysis of Solids and Fluids Fall ‘08 Lecture 18 - Solution of F.E. equations Prof. K.J. Bathe In structures, F (u, p) = R. In heat transfer, F (θ) = Q In fluid flow, F (v, p, θ) = R In structures/solids � F (m) � � F = = m m 0V (m) Elastic materials Example p. 590 textbook MIT O...
https://ocw.mit.edu/courses/2-094-finite-element-analysis-of-solids-and-fluids-ii-spring-2011/5bbd789c4c008f43a78ffc97d314ed5d_MIT2_094S11_lec18.pdf
0ρ 0L = tρ tL ⇒ 0 tL tS11 = 0L � �2 0L tL 0L tτ11 = tL tτ11 ∴ 0 L tτ11 = E˜ · tL 1 2 �� t u 1 + 0 L �2 � − 1 ⇒ tτ11A = tP = �� 1 + E˜A 2 tu �2 0L � � − 1 1 + tu � 0L This is because of the material-law assumption (18.5) (okay for small strains . . . ) Hyperelasticity tW = f (Green-Lagrange...
https://ocw.mit.edu/courses/2-094-finite-element-analysis-of-solids-and-fluids-ii-spring-2011/5bbd789c4c008f43a78ffc97d314ed5d_MIT2_094S11_lec18.pdf
of (18.1) (similarly (18.2) and (18.3)) Newton-Raphson Find U ∗ as the zero of f (U ∗) f (U ∗) = − R t+Δt � t+Δt t+Δt F � (i−1) + U = f ∂f ∂U � � � � t+Δt (i−1) U � U ∗ − t+ΔtU (i−1) + H.O.T. · � where t+ΔtU (i−1) is the value we just calculated and an approximation to U ∗. Assume t+ΔtR is independent ...
https://ocw.mit.edu/courses/2-094-finite-element-analysis-of-solids-and-fluids-ii-spring-2011/5bbd789c4c008f43a78ffc97d314ed5d_MIT2_094S11_lec18.pdf
MIT 2.094 18. Solution of F.E. equations Pictorially for a single degree of freedom system i = 1; i = 2; tK Δu(1) = t+ΔtR − tF t+ΔtK (1)Δu(2) = t+ΔtR − t+ΔtF (1) Convergence Use �ΔU (i)�2 < � �a�2 = �� (ai)2 i But, if incremental displacements are small in every iteration, need to also use �t+ΔtR − t+ΔtF...
https://ocw.mit.edu/courses/2-094-finite-element-analysis-of-solids-and-fluids-ii-spring-2011/5bbd789c4c008f43a78ffc97d314ed5d_MIT2_094S11_lec18.pdf
Massachusetts Institute of Technology Department of Electrical Engineering and Computer Science 6.245: MULTIVARIABLE CONTROL SYSTEMS by A. Megretski � Interpretations for Standard Optimization Setup1 This is the second lecture on standard feedback optimization setup. It describes a variety of ways to come up with...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
(transfer function 1/s) maps a signal f ≤ L1 to signals of the form g(t) = c0 + t 0 f (δ )dδ, where c0 is an arbitrary constant. In general, a system’s output is not necessarily unique because it may also depend on a set of auxiliary parameters (e.g. the initial states of the system), as in Figure 2.1. In the c...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
rate 1/M for every positive integer M (though, indeed, it will not be the result of sampling f (t) = cos(�t) at rate T = 1/M for M ≥= 1). An alternative way of representing a discrete time signal f = f (t) with sampling rate T > 0 is by specifying T and the sequence of its values f [k] at time instances tk = kT , ...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
f [k] into g[k] = f [k − 1], with the value of g[0] being arbitrary. ∞� lm [T ] + 2.1.3 Finite Order CT LTI Models A state-space model for a finite order CT LTI system H with input f = f (t), output g = g(t), and state x = x(t) has the form x˙ (t) = Ax(t) + Bf (t), g(t) = Cx(t) + Df (t), (2.1) (2.2) where A, ...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
(2.3) (2.4) where A, B, C, D are constant matrices with real entries. Here f, g, x are DT signals with same sampling rate T > 0. In terms of sampled value sequences f [k] = f (kT ), x[k] = x(kT ), and g[k] = g(kT ), equations have the form x[k + 1] = Ax[k] + Bf [k], g[k] = Cx[k] + Df [k], (2.5) (2.6) The trans...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
A)−1B, and, for an k-by-m complex matrix M , �max(M ) = max u→Cm , |u|=1 5 |M u|, and |v| denotes the standard Hermitean norm (length) of vector v. If H is a discrete time system, maximization over the imaginary axis is replaced with maximization over the unit circle: ∈H∈� = sup �max(H(ej�)). �→[−�,�] ...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
. The definition for discrete time systems is similar, with integrals replaced by sums: N inf N �0 {π2|f [k]| − |g[k]|2} > −→, 2 � k=0 6 where g = S(f ). The informal rationale behind the definition is as follows: for “zero initial conditions” (whatever this means), we expect the “energy” of the output to ...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
≤ (0, →) (for gT this is true since A is a Hurwitz matrix), and hence both have Fourier transforms fT and ˜T respectively. In addition, by causality, g(t) = gT (t) for t < T . Hence ˜ g |g(t)|2dt = T |g T (t)|2dt � T 0 � 0 |g T (t)|2dt 0 � = 1 2� −� � 1 2� −� = |˜T g (jσ)|2dσ = |H(jσ)f˜T (jσ)|2d...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
τ � 0 |g1(t)|2dt. To show that H-Infinity norm cannot be larger than L2 gain, consider zero initial conditions and sinusoidal inputs f (t) = f0 cos(σt), where frequency σ and vector f0 are appropriately chosen. 7 2.2.3 H2 norm and L2-to-L-Infinity gain L2-to-L-Infinity gain of a stable state space model is defined...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
[k] with zero mean, unit variance, and zero cross-corellation: Ef [k] = 0, Ef [k1]f [k2]≥ = �(k1, k2)I, where �(k1, k2) = 0 for k1 ≥= k2 and �(k, k) = 1 for all k. The continuous time white noise is a slightly more complicated concept: it is a general­ ized random process f = f (t) (something akin the Dirac delta in...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
through an input, and will in turn be driven by an output of an LTI system, as shown on Figure 2.2, left, where � is the nonlinear/time varying block, and G is the LTI system. � � z1 G � z2 w1 � w2 � �w1 w2 � G � z1 � z2 Figure 2.2: Small gain theorem The following theorem, a version of the famous small...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
for the standard setup shown on Figure 2.3, where G is an unstable LTI plant with delay, P (s) = −� s e , s − 1 and F is the controller to be designed to guarantee good tracking of the reference input r at lower frequencies. r � � � −� F � P � q Figure 2.3: Feedback design with delay We begin by approxi...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
P0=exp(-tau)/(s-1)+G0; g=5; d=150; [A,B,C,D]=linmod2(’lec2_ex1a’); p=pck(A,B,C,D); k=hinfsyn(p,1,1,0,20,0.01); 2.3.3 A small gain theorem for H2 H2 norm, too, can serve as a measure of robustness to white noise perturbations of the feedback loop gain. Remember that the usual small gain bound on a CT system � with i...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
every square integrable row vector function p = p(t). In the discrete time case, simply require that t2 t1 � � � � 2 = � � � � � k2 p[k]g[k] � k=k1 k2 � k=k1 |p[k]|2E|g[k]| 2 for every sequence p = p[k] of row vectors. Theorem 2.4 Let G be a stable finite order LTI system with two inputs w1, w2 and two ou...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
, v = [v1; v2] is a strict sense white noise signal, i.e. Evi[k]2 = 1, Evi[k] = 0, and it is known that vi[k] is independent of y[t], vi[t − 1] for all t � k. What is the asymptotic value of Ey[k]2 as k � →? This question can be answered easily without relying on the H2 small gain theorem, because Ey[k + 1]2 = (a...
https://ocw.mit.edu/courses/6-245-multivariable-control-systems-spring-2004/5bca558808b8b652b14b2be6128d8e7d_lec2_6245_2004.pdf
8.871 Lecture 3 M. Padi November 14, 2004 Let’s use dimensional analysis to make some general remarks about the tensions of various branes. We’ll start with the M2- and M5-branes. Note that the tension of an object that fills 2 space dimensions must have units of [L−3] = [L2] . The only length scale we have is lp, ...
https://ocw.mit.edu/courses/8-871-selected-topics-in-theoretical-particle-physics-branes-and-gauge-theory-dynamics-fall-2004/5bcdff3a3071c28befd43ad3ef4fc350_lec3.pdf
anes and D-branes using S-duality between M-theory compactified on a circle and Type IIA string theory. We use the fact that the M2-brane compactified over one of its dimensions becomes an F1-string in Type IIA. Thus we can write TF 1 = = 1 R l3 l2 p s . (3) An M2 which does not wrap the M theory circle becomes ...
https://ocw.mit.edu/courses/8-871-selected-topics-in-theoretical-particle-physics-branes-and-gauge-theory-dynamics-fall-2004/5bcdff3a3071c28befd43ad3ef4fc350_lec3.pdf
is S-dual to itself at coupling g� = 1/gs. Under this duality, the D1-brane becomes the F1-string, the D3-brane is self-dual and the D5-brane becomes the NS5-brane. So now we have the following equalities: s TD1 = TF 1 = 1 gsl2 s 1 l2 s = T � = F 1 = T � = D1 1 l�2 s 1 g� l�2 s s From these equations we ...
https://ocw.mit.edu/courses/8-871-selected-topics-in-theoretical-particle-physics-branes-and-gauge-theory-dynamics-fall-2004/5bcdff3a3071c28befd43ad3ef4fc350_lec3.pdf
Frequency response: Resonance, Bandwidth, Q factor Resonance. Let’s continue the exploration of the frequency response of RLC circuits by investigating the series RLC circuit shown on Figure 1. I R C Vs R + VR - The magnitude of the transfer function when the output is taken across the resistor is Figure 1 H ( ) ...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
the current through it ( cos( cos( ) tω are RV RI ) ) p t ( ) = = V R V I R R cos( ω t I ) R t ω 2 cos ( 6.071/22.071 Spring 2006, Chaniotakis and Cory cos( t ω ) ) (1.4) 2 And the average power becomes P ( ) ω = = 1 2 1 2 V I R R 2 I R R (1.5) Notice that this power is a function o...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
⎠ + 1 2 ω 0 ω 2 = R L 2 + 2 ⎛ ⎜ ⎝ R L 2 ⎞ ⎟ ⎠ + 1 2 ω 0 6.071/22.071 Spring 2006, Chaniotakis and Cory (1.8) (1.9) (1.10) 4 The bandwidth is the difference between the half power frequencies Bandwidth B = = − 1ω ω 2 (1.11) By multiplying Equation (1.9) with Equation (1.10) ...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
Vc dt = ω CA cos( t ω ) . The SE = 1 2 2 L C A ω 2 2 2 cos ( t ω ) + 2 CA 2 sin ( ω t ) 1 2 (1.15) At the resonance frequency where ω ω= 0 = 1 LC the energy stored in the circuit becomes SE = 2 CA 1 2 6.071/22.071 Spring 2006, Chaniotakis and Cory (1.16) 5 ...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
the resonance characteristics of the parallel RLC circuit. I s(t) IR(t) R L C Here the impedance seen by the current source is Figure 4 Z // = j L ω (1 − 2 ω LC ) + j L ω R (1.20) At the resonance frequency resistive. The parallel combination of the capacitor and the inductor act as an open circuit. Therefore at ...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
⎜ ⎝ 1 RC ⎞ ⎟ ⎠ 2 + 1 2 ω 0 ω 2 = 1 RC 2 + 2 ⎛ ⎜ ⎝ 1 RC ⎞ ⎟ ⎠ 2 + 1 2 ω 0 And the bandwidth for the parallel RLC circuit is Figure 5 The Q factor is PB − ω ω= 2 1 = 1 RC Q = ω 0 B P = ω 0 RC = R L ω 0 (1.23) (1.24) (1.25) (1.26) (1.27) 6.071/22.071 Spring 2006, Chaniotakis and Cory 8 ...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
2 + 2 ⎛ ⎜ ⎝ 1 RC ⎞ ⎟ ⎠ 2 + 1 2 ω 0 ω 2 = 1 RC 2 + 2 ⎛ ⎜ ⎝ 1 RC ⎞ ⎟ ⎠ 2 + 1 2 ω 0 Bandwidth B S − ω ω= 2 1 = R L PB − ω ω= 2 1 = 1 RC Q factor Q = 0 L ω ω 0 = R B S = 1 RC ω 0 Q = ω 0 B P = ω 0 RC = R L ω 0 6.071/22.071 Spring 2006, Chaniotakis and Cory 9 ...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
“notch” filter. If we are interested in suppressing a 60 Hz noise signal then 2 60 π = 1 LC 6.071/22.071 Spring 2006, Chaniotakis and Cory (1.29) (1.30) (1.31) 10 For L=47mH, the corresponding value of the capacitor is C=150µF. The plot of the transfer function wit...
https://ocw.mit.edu/courses/6-071j-introduction-to-electronics-signals-and-measurement-spring-2006/5bcec4bfba5f2e99754b77509e9e7ab4_resonance_qfactr.pdf
15.083J/6.859J Integer Optimization Lecture 3: Methods to enhance formulations 1 Outline • Polyhedral review • Methods to generate valid inequalities • Methods to generate facet defining inequalities Slide 1 2 Polyhedral review 2.1 Dimension of polyhedra • Definition: The vectors x1 , . . . , xk ∈ (cid:3)n uni...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf
P . • Let f (cid:2) x ≥ g be a valid inequality for a polyhedron P , and let F = {x ∈ P | f (cid:2) x = g}. Then, F is called a face of P and we say that f (cid:2) x ≥ g represents F . A face is called proper if F = Ø(cid:6) , P . • A face F of P represented by the inequality f (cid:2) x ≥ g, is called a facet of P if...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf
of dimension one: −x1 + 2x2 ≤ 4, and 2x1 + x2 ≤ 8. 3 Methods to generate valid inequalities 3.1 Rounding • Choose u = (u1, . . . , um)(cid:2) ≥ 0; Multiply ith constraint with ui and sum: Slide 5 Slide 6 Slide 7 (cid:2) n (u (cid:2)Aj )xj ≤ u (cid:2)b. j=1 • Since (cid:5)u (cid:2)Aj (cid:6) ≤ u (cid:2)Aj a...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf
e∈E(U ) • Round to (|U | is odd) 1 2 (cid:2) e∈δ(U ) 1 xe ≤ |U |. 2 (cid:2) xe + e∈E(U ) xe ≤ 1 |U |. 2 (cid:2) xe ≤ e∈E(U ) (cid:9) (cid:10) 1 |U | = 2 |U | − 1 , 2 3.2 Superadditivity 3.2.1 Definition A function F : D ⊂ (cid:3) (cid:8)n → (cid:3) is superadditive if for a1, a2 ∈ D,: a1 + a2 ∈ D : ...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf
F (Aj ) + F (Aj )(k − 1) ≤ F (Aj ) + F (Aj (k − 1)) ≤ F (Aj + Aj (k − 1)), by superadditivity, and the induction is complete. Therefore, n (cid:2) n (cid:2) F (Aj )xj ≤ F (Aj xj ). Slide 12 j=1 j=1 3 By superadditivity, ⎛ n (cid:2) F (Aj xj ) ≤ F ⎝ n (cid:2) Aj xj ⎞ ⎠ = F (Ax). j=1 j=1 Since...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf
4 x1 + 4x2 + x3 + x4 ≥ 8 is valid for S. (cid:5) • For d = 1, and aj are not integers. In this case, since x ≥ 0, we obtain n (cid:5)aj (cid:6)xj ≤ (cid:5)a0(cid:6), and thus the following j=1 (cid:5)aj (cid:6)xj ≤ a0. Since x ∈ Z, (cid:5) n j=1 inequality is valid for S (cid:2) n (aj − (cid:5)aj (cid:6))xj ...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf
is valid for S for some d ≥ 0, and the aj xj + c(xk − α − 1) ≤ b is valid for S for some c ≥ 0, then the aj xj ≤ b is valid for S. n j=1 n j=1 (cid:5) (cid:5) (−x1 + x2) + (x2 − 3) ≤ 1, (−x1 + x2) − (x2 − 2) ≤ 1. α = 2, −x1 + x2 ≤ 1 is valid. 3.5 Mixed integer rounding 3.5.1 Proposition • For v ∈ (cid:3), f ...
https://ocw.mit.edu/courses/15-083j-integer-programming-and-combinatorial-optimization-fall-2009/5be17dfa9a150e8db5b9fb76905d2408_MIT15_083JF09_lec03.pdf