CSE507. Course Introduction. Computer-Aided Reasoning for Software. Emina Torlak

Transcription

1 Computer-Aided Reasoning for Software CSE507 courses.cs.washington.edu/courses/cse507/14au/ Course Introduction Emina Torlak

2 Today What is this course about? Course logistics Review of basic concepts 2

3 about Tools for building better software, more easily

4 more reliable, faster, more energy efficient Tools for building better software, more easily

5 Tools for building better software, more easily automatic verification, debugging & synthesis

6 Tools for building better software, more easily class List { Node head;! void reverse() { Node near = head; Node mid = near.next; Node far = mid.next;! near.next = far; while (far!= null) { mid.next = near; near = mid; mid = far; far = far.next; }! mid.next = near; head = mid; } }! class Node { Node next; String data; } Is this list reversal procedure correct? 4

7 Tools for building better software, more easily class List { Node head;! void reverse() { Node near = head; Node mid = near.next; Node far = mid.next;! near.next = far; while (far!= null) { mid.next = near; near = mid; mid = far; far = far.next; }! mid.next = near; head = mid; } }! class Node { Node next; String data; } verification Is this list reversal procedure correct? 4

8 Tools for building better software, more easily class List { Node head;! void reverse() { Node near = head; Node mid = near.next; Node far = mid.next;! near.next = far; while (far!= null) { mid.next = near; near = mid; mid = far; far = far.next; }! mid.next = near; head = mid; } }! class Node { Node next; String data; } Which lines of code are responsible for the buggy behavior? 5

9 Tools for building better software, more easily class List { Node head;! void reverse() { Node near = head; Node mid = near.next; Node far = mid.next;! near.next = far; while (far!= null) { mid.next = near; near = mid; mid = far; far = far.next; }! mid.next = near; head = mid; } }! class Node { Node next; String data; } debugging Which lines of code are responsible for the buggy behavior? 5

10 Tools for building better software, more easily class List { Node head;! void reverse() { Node near = head; Node mid = near.next; Node far = mid.next;! near.next =??; while (far!= null) { mid.next = near; near = mid; mid = far; far = far.next; }! mid.next = near; head = mid; } }! class Node { Node next; String data; } Is there a way to complete this code so that it is correct? 6

11 Tools for building better software, more easily class List { Node head;! void reverse() { Node near = head; Node mid = near.next; Node far = mid.next;! near.next = null; while (far!= null) { mid.next = near; near = mid; mid = far; far = far.next; }! mid.next = near; head = mid; } }! class Node { Node next; String data; } synthesis Is there a way to complete this code so that it is correct? 6

12 goal By the end of this course, you ll be able to build computer-aided tools for any domain!

13 biology education goal By the end of this course, you ll be able to build computer-aided tools for any domain! hardware systems security networking databases low-power computing high-performance computing

14 logistics Topics, structure, people

15 Course overview program question tools logic automated reasoning engine 9

16 Course overview program question verifier, synthesizer, fault localizer logic SAT, SMT, model finders & checkers Drawing from Decision Procedures by Kroening & Strichman 9

17 Course overview program question study (part I) verifier, synthesizer, fault localizer logic SAT, SMT, model finders & checkers Drawing from Decision Procedures by Kroening & Strichman 9

18 Course overview program question study (part I) verifier, synthesizer, fault localizer logic build! (part II) SAT, SMT, model finders & checkers Drawing from Decision Procedures by Kroening & Strichman 9

19 Grading structure 3 individual homework assignments (50%) conceptual problems & proofs (Tex) study (part I) implementations in various programming languages may discuss problems with others but solutions must be your own Course project (50%) build a computer-aided reasoning tool for a domain of your choice teams of 2-3 people strongly encouraged see the course web page for timeline, deliverables and other details build! (part II) 10

20 Reading and references Required readings posted on the course web page Complete each reading before the lecture for which it is assigned Recommended text books Bradley & Manna, The Calculus of Computation Kroening & Strichman, Decision Procedures Related courses Isil Dillig: Automated Logical Reasoning (2013) Viktor Kuncak: Synthesis, Analysis, and Verification (2013) Sanjit Seshia: Computer-Aided Verification (2012) 11

21 Advice for doing well in 507 Come to class (prepared) Lecture notes are enough to teach from, but not enough to learn from Participate Ask and answer questions Meet deadlines Turn homework in on time Start homework and project sooner than you think you need to Follow instructions for submitting code (we have to be able to run it) 12

22 People instructor TA Emina Torlak PLSE CSE 596 Wednesdays 1-2 Mert Saglam Theory CSE 618 Thursdays

23 People instructor TA students! Your name Research area Survey Emina Torlak PLSE CSE 596 Wednesdays 1-2 Mert Saglam Theory CSE 618 Thursdays

24 review Propositional logic: syntax, semantics & proof methods

25 Syntax of propositional logic Atom truth symbols: ( true ), ( false ) propositional variables: p, q, r, Literal an atom α or its negation α Formula a literal or the application of a logical connective to formulas F not (negation) F1 F2 and (conjunction) F1 F2 or (disjunction) F1 F2 implies (implication) if and only if (iff) F1 F2 ( p ) (q ) 16

26 Syntax of propositional logic ( p ) (q ) Atom truth symbols: ( true ), ( false ) propositional variables: p, q, r, Literal an atom α or its negation α Formula a literal or the application of a logical connective to formulas F not (negation) F1 F2 and (conjunction) F1 F2 or (disjunction) F1 F2 implies (implication) if and only if (iff) F1 F2 16

27 Syntax of propositional logic ( p ) (q ) Atom truth symbols: ( true ), ( false ) propositional variables: p, q, r, Literal an atom α or its negation α Formula a literal or the application of a logical connective to formulas F not (negation) F1 F2 and (conjunction) F1 F2 or (disjunction) F1 F2 implies (implication) if and only if (iff) F1 F2 16

28 Syntax of propositional logic ( p ) (q ) Atom truth symbols: ( true ), ( false ) propositional variables: p, q, r, Literal an atom α or its negation α Formula a literal or the application of a logical connective to formulas F not (negation) F1 F2 and (conjunction) F1 F2 or (disjunction) F1 F2 implies (implication) if and only if (iff) F1 F2 16

29 Interpretations of propositional formulas An interpretation I for a propositional formula F maps every variable in F to a truth value: I : { p, q, } 17

30 Interpretations of propositional formulas An interpretation I for a propositional formula F maps every variable in F to a truth value: I : { p, q, } I is a satisfying interpretation of F, written as I F, if F evaluates to under I. I is a falsifying interpretation of F, written as I F, if F evaluates to under I. 17

31 Semantics of propositional logic Base cases: I I I p iff I[p] = I p iff I[p] = 18

32 Semantics of propositional logic Base cases: I Inductive cases: I I p iff I[p] = I p iff I[p] = 18

33 Semantics of propositional logic Base cases: I Inductive cases: I F iff I F I I p iff I[p] = I p iff I[p] = 18

34 Semantics of propositional logic Base cases: I I Inductive cases: I F iff I F I F1 F2 iff I F1 and I F2 I p iff I[p] = I p iff I[p] = 18

35 Semantics of propositional logic Base cases: I I I p iff I[p] = I p iff I[p] = Inductive cases: I F iff I F I F1 F2 iff I F1 and I F2 I F1 F2 iff I F1 or I F2 I F1 F2 iff I F1 or I F2 I F1 F2 iff I F1 and I F2, or I F1 and I F2 18

36 Semantics of propositional logic: example F: (p q) (p q) I: {p, q }? 19

37 Semantics of propositional logic: example F: (p q) (p q) I: {p, q } I F 19

38 Satisfiability & validity of propositional formulas F is satisfiable iff I F for some I. F is valid iff I F for all I. 20

39 Satisfiability & validity of propositional formulas F is satisfiable iff I F for some I. F is valid iff I F for all I. Duality of satisfiability and validity: F is valid iff F is unsatisfiable. 20

40 Satisfiability & validity of propositional formulas F is satisfiable iff I F for some I. F is valid iff I F for all I. Duality of satisfiability and validity: If we have a procedure for checking satisfiability, then we can also check validity of propositional formulas, and vice versa. F is valid iff F is unsatisfiable. 20

41 Techniques for deciding satisfiability & validity Search Deduction SAT solver 21

42 Techniques for deciding satisfiability & validity! Search Enumerate all interpretations (i.e., build a truth table), and check that they satisfy the formula. Deduction SAT solver 21

43 Techniques for deciding satisfiability & validity! Search! Deduction Enumerate all interpretations (i.e., build a truth table), and check that they satisfy the formula. Assume the formula is invalid, apply proof rules, and check for contradiction in every branch of the proof tree. SAT solver 21

44 Proof by search (truth tables) F: (p q) (p q) p q p q q p q F

45 Proof by search (truth tables) F: (p q) (p q) p q p q q p q F Valid

46 Proof by deduction (semantic arguments) Example proof rules: I F I F I F1 F2 I F1 I F2 I F I F1 F2 I F I F1 I F2 23

47 Proof by deduction (semantic arguments) Example proof rules: F: p q I F I F I F1 F2 I F1 I F2 I F I F1 F2 I F I F1 I F2 23

48 Proof by deduction (semantic arguments) Example proof rules: F: p q I F I F I F1 F2 I F1 I F2 1. I p q (assumption) I F I F1 F2 I F I F1 I F2 23

49 Proof by deduction (semantic arguments) Example proof rules: F: p q I F I F I F1 F2 I F1 I F2 1. I p q (assumption) a. I p (1, ) I F I F1 F2 I F I F1 I F2 23

50 Proof by deduction (semantic arguments) Example proof rules: F: p q I F I F I F1 F2 I F1 I F2 1. I p q (assumption) a. I p (1, ) b. I q (1, ) I F I F1 F2 I F I F1 I F2 23

51 Proof by deduction (semantic arguments) Example proof rules: F: p q I F I F I F I F1 F2 I F1 I F2 I F1 F2 1. I p q (assumption) a. I p (1, ) b. I q (1, ) i. I q (1b, ) I F I F1 I F2 23

52 Proof by deduction (semantic arguments) Example proof rules: F: p q I F I F I F I F1 F2 I F1 I F2 I F1 F2 1. I p q (assumption) a. I p (1, ) b. I q (1, ) i. I q (1b, ) I F I F1 I F2 Invalid; I is a falsifying interpretation. 23

53 Semantic judgements Formulas F1 and F2 are equivalent, written F1 F2, iff F1 F2 is valid. Formula F1 implies F2, written F1 F2, iff F1 F2 is valid. 24

54 Semantic judgements Formulas F1 and F2 are equivalent, written F1 F2, iff F1 F2 is valid. Formula F1 implies F2, written F1 F2, iff F1 F2 is valid. F1 F2 and F1 F2 are not propositional formulas (not part of syntax). They are properties of formulas, just like validity or satisfiability. 24

55 Semantic judgements Formulas F1 and F2 are equivalent, written F1 F2, iff F1 F2 is valid. Formula F1 implies F2, written F1 F2, iff F1 F2 is valid. If we have a procedure for checking satisfiability, then we can also check for equivalence and implication of propositional formulas. 24

56 Summary Today Course overview & logistics Review of propositional logic Next Lecture (by Zach Tatlock) Normal forms A basic SAT solver Take the course survey Read Chapter 1 of Bradley & Manna 25