CS21 Decidability and Tractability

Transcription

1 CS21 Decidability and Tractability Lecture 14 February 7, 2018 February 7, 2018 CS21 Lecture 14 1

2 Outline Gödel Incompleteness Theorem February 7, 2018 CS21 Lecture 14 2

3 Background Hilbert s program (1920 s): formalize mathematics in axiomatic form derive all true statements mechanically from initial axioms would put mathematicians out of business! very influential proposal to start: try for all true statements about the natural numbers ( number theory ) February 7, 2018 CS21 Lecture 14 3

4 Background: Kurt Gödel (1931): it is not possible! no formalization of number theory can prove all true statements stunning result considered one of greatest 20 th century achievements in mathematics February 7, 2018 CS21 Lecture 14 4

5 Background We will prove using: RE languages and non-re languages reductions Idea: set of all theorems is RE set of all true statements is not RE This kind of proof of Gödel s result attributed to Turing (1937). February 7, 2018 CS21 Lecture 14 5

6 Number Theory formal language to express properties of N = {0, 1, 2, 3, } allowable symbols: parentheses, and variables x,y,z, ranging over N operators + (addition) and * (multiplication) constants 0 (additive id) and 1 (mult. identity) relation = (equality) quantifiers " (for all) and $ (exists) propositional operators Ù (and) Ú (or) (not) Þ (implies) Û (iff) February 7, 2018 CS21 Lecture 14 6

7 Number Theory can formalize syntax of allowable formulas (skip) defining comparison relations: x y º $z x + z = y x < y º $z x + z = y Ù (z = 0) February 7, 2018 CS21 Lecture 14 7

8 Number Theory Other natural concepts we will need: quotient q and remainder r when divide x by y INTDIV(x, y, q, r) º x = qy + r Ù r < y y divides x DIV(y, x) º $ q INTDIV(x,y,q,0) x is even EVEN(x) º DIV(1+1, x) x is odd ODD(x) º EVEN(x) February 7, 2018 CS21 Lecture 14 8

9 Number Theory Other natural concepts we will need: x is prime PRIME(x) º x (1+1) Ù "y (DIV(y, x) Þ (y = 1 Ú y = x)) x is a power of 2 POWER 2 (x) º "y (DIV(y, x) Ù PRIME(y)) Þ y = (1+1) y = 2 k and k th bit of x is 1 BIT(x, y) º POWER 2 (y) Ù "q "r (INTDIV(x, y, q, r) Þ ODD(q)) February 7, 2018 CS21 Lecture 14 9

10 Number Theory y = 2 k and k th bit of x is 1 BIT(x, y) º POWER 2 (y) Ù "q "r (INTDIV(x, y, q, r) Þ ODD(q)) y = x = q r February 7, 2018 CS21 Lecture 14 10

11 Number Theory A sentence is a formula with no unquantified variables every number has a successor: "x $y y = x + 1 every number has a predecessor: "x $y x = y + 1 not a sentence: x + y = 1 true false number theory = set of true sentences denoted Th(N) February 7, 2018 CS21 Lecture 14 11

12 Proof systems Proof system components: axioms (asserted to be true) rules of inference (mechanical way to derive theorems from axioms) axioms for manipulating symbols (e.g.): (j Ù y) Þ j ("x j(x)) Þ j(1+1+1) "x "y "z (x = y Ù y = z Þ x = z) others February 7, 2018 CS21 Lecture 14 12

13 Peano Arithmetic Peano Arithmetic: proof system for number theory. Axioms: 0 is not a successor "x (0 = x + 1) the successor function is one-to-one 0 is an identity for + "x "y (x+1 = y+1 Þ x = y) "x x + 0 = x February 7, 2018 CS21 Lecture 14 13

14 Peano Arithmetic + is associative "x "y x + (y + 1) = (x + y) + 1 multiplying by zero gives 0 "x x*0 = 0 * distributes over + "x "y x * (y + 1) = (x * y) + x induction axiom (j(0) Ù "x (j(x) Þ j(x+1))) Þ "x j(x) February 7, 2018 CS21 Lecture 14 14

15 Peano Arithmetic rules of inference: modus ponens j y j Þ y generalization j "x j February 7, 2018 CS21 Lecture 14 15

16 Proof systems a proof is a sequence of formulas j 1, j 2, j 3,, j n such that each j i is either an axiom, or follows from formulas earlier in list from rules of inference A sentence is a theorem of the proof system if it has a proof February 7, 2018 CS21 Lecture 14 16

17 Proof systems A proof system is sound if all theorems in that proof system are true (better have this) Peano Arithmetic (PA) is sound. true sentences = Th(N) false sentences = co-th(n) theorems of PA February 7, 2018 CS21 Lecture 14 17

18 Proof systems A proof system is complete if all true sentences are theorems in that proof system hope to have this (recall Hilbert s program) true sentences = Th(N) false sentences = co-th(n) theorems of a complete proof system February 7, 2018 CS21 Lecture 14 18

19 Incompleteness Theorem Theorem: Peano Arithmetic is not complete. (same holds for any reasonable proof system for number theory) Proof outline: the set of theorems of PA is RE the set of true sentences (= Th(N)) is not RE February 7, 2018 CS21 Lecture 14 19

20 Incompleteness Theorem Lemma: the set of theorems of PA is RE. Proof: TM that recognizes the set of theorems of PA: systematically try all possible ways of writing down sequences of formulas accept if encounter a proof of input sentence (note: true for any reasonable proof system) February 7, 2018 CS21 Lecture 14 20

21 Incompleteness Theorem Lemma: Th(N) is not RE Proof: reduce from co-halt (show co-halt m Th(N)) recall co-halt is not RE what should f(<m, w>) produce? construct g such that M loops on w Û g is true February 7, 2018 CS21 Lecture 14 21

22 Incompleteness Theorem we will define VALCOMP M,w (v) º (details to come) so that it is true iff v is a (halting) computation history of M on input w then define f(<m, w>) to be: g º $v VALCOMP M,w (v) YES maps YES? <M, w> Î co-halt Þ g is true Þ g ÎTh(N) NO maps to NO? <M, w> Ï co-halt Þ g is false Þ g ÏTh(N) February 7, 2018 CS21 Lecture 14 22

23 Expressing computation in the language of number theory Last time: basic building blocks x < y º $z x + z = y Ù (z = 0) INTDIV(x, y, q, r) º x = qy + r Ù r < y DIV(y, x) º $ q INTDIV(x,y,q,0) PRIME(x) º x (1+1) Ù "y (DIV(y, x) Þ (y = 1 Ú y = x)) February 7, 2018 CS21 Lecture 14 23

24 Expressing computation in the language of number theory we ll write configurations over an alphabet of size p, where p is a prime that depends on M d is a power of p: POWER p (d) º "z (DIV(z, d) Ù PRIME(z)) Þ z = p d = p k and length of v as a p-ary string is k LENGTH(v, d) º POWER p (d) Ù v < d February 7, 2018 CS21 Lecture 14 24

25 Expressing computation in the language of number theory the p-ary digit of v at position y is b (assuming y is a power of p): DIGIT(v, y, b) º $u $a (v = a + by + upy Ù a < y Ù b < p) the three p-ary digits of v at position y are b,c, and d (assuming y is a power of p): 3DIGIT(v, y, b, c, d) º $u $a (v = a + by + cpy + dppy + upppy Ù a < y Ù b < p Ù c < p Ù d < p) February 7, 2018 CS21 Lecture 14 25

26 Expressing computation in the language of number theory the three p-ary digits of v at position y match the three p-ary digits of v at position z according to M s transition function (assuming y and z are powers of p): MATCH(v, y, z) º Ú (a,b,c,d,e,f) Î C 3DIGIT(v, y, a, b, c) Ù 3DIGIT(v, z, d, e, f) where C = {(a,b,c,d,e,f) : abc in config. C i can legally change to def in config. C i+1 } February 7, 2018 CS21 Lecture 14 26

27 Expressing computation in the language of number theory all pairs of 3-digit sequences in v up to d that are exactly c apart match according to M s transition function (assuming c, d powers of p) MOVE(v, c, d) º "y (POWER p (y) Ù yppc < d) Þ MATCH(v, y, yc) February 7, 2018 CS21 Lecture 14 27

28 Expressing computation in the language of number theory the string v starts with the start configuration of M on input w = w 1 w n padded with blanks out to length c (assuming c is a power of p): START(v, c) º Ù i = 0,1,2,, n DIGIT(v, p i, k i ) Ù p n < c Ù "y (POWER p (y) Ù p n < y < c Þ DIGIT(v, y, k)) where k 0 k 1 k 2 k 3 k n is the p-ary encoding of the start configuration, and k is the p-ary encoding of a blank symbol. February 7, 2018 CS21 Lecture 14 28

29 Expressing computation in the language of number theory string v has a halt state in it somewhere before position d (assuming d is power of p): HALT(v, d) º $y (POWER p (y) Ù y < d Ù Ú aîh DIGIT(v,y,a)) where H is the set of p-ary digits containing states q accept or q reject. February 7, 2018 CS21 Lecture 14 29

30 Expressing computation in the language of number theory string v is a valid (halting) computation history of machine M on string w: VALCOMP M,w (v) º $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) M does not halt on input w: $v VALCOMP M,w (v) February 7, 2018 CS21 Lecture 14 30

31 Gödel Incompleteness Theorem (an example) February 7, 2018 CS21 Lecture 14 31

32 Incompleteness Theorem v = VALCOMP M,w (v) º $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) February 7, 2018 CS21 Lecture 14 32

33 Incompleteness Theorem v = d = VALCOMP M,w (v) º $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) d = p k and length of v as a p-ary string is k LENGTH(v, d) º POWER p (d) Ù v < d February 7, 2018 CS21 Lecture 14 33

34 Incompleteness Theorem v = VALCOMP M,w (v) º kk n k 2 k 1 k 0 c = p n =1000 $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) v starts with the start configuration of M on input w padded with blanks out to length c: START(v, c) º Ù i = 0,, n DIGIT(v, p i, k i )Ù p n < c Ù "y (POWER p (y) Ù p n < y < c Þ DIGIT(v, y, k)) February 7, 2018 CS21 Lecture 14 34

35 Incompleteness Theorem v = VALCOMP M,w (v) º yc = y =1 $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) all pairs of 3-digit sequences in v up to d exactly c apart match according to M s transition function MOVE(v, c, d) º "y (POWER p (y) Ù yppc < d) Þ MATCH(v, y, yc) February 7, 2018 CS21 Lecture 14 35

36 Incompleteness Theorem v = VALCOMP M,w (v) º yc = y =10 $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) all pairs of 3-digit sequences in v up to d exactly c apart match according to M s transition function MOVE(v, c, d) º "y (POWER p (y) Ù yppc < d) Þ MATCH(v, y, yc) February 7, 2018 CS21 Lecture 14 36

37 Incompleteness Theorem v = VALCOMP M,w (v) º yc = y =100 $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) all pairs of 3-digit sequences in v up to d exactly c apart match according to M s transition function MOVE(v, c, d) º "y (POWER p (y) Ù yppc < d) Þ MATCH(v, y, yc) February 7, 2018 CS21 Lecture 14 37

38 Incompleteness Theorem halt state v = y = VALCOMP M,w (v) º $c $d (POWER p (c) Ù c < d Ù LENGTH(v, d) Ù START(v, c) Ù MOVE(v, c, d) Ù HALT(v, d)) string v has a halt state in it before pos n d: HALT(v, d) º $y (POWER p (y) Ù y < d Ù Ú aîh DIGIT(v,y,a)) February 7, 2018 CS21 Lecture 14 38

39 Incompleteness Theorem Lemma: Th(N) is not RE Proof: reduce from co-halt (show co-halt m Th(N)) recall co-halt is not RE constructed g such that M loops on w Û g is true February 7, 2018 CS21 Lecture 14 39

40 Summary full-fledged model of computation: TM many equivalent models Church-Turing Thesis encoding of inputs Universal TM February 7, 2018 CS21 Lecture 14 40

41 Summary classes of problems: decidable ( solvable by algorithms ) recursively enumerable (RE) co-re counting: not all problems are decidable not all problems are RE February 7, 2018 CS21 Lecture 14 41

42 Summary diagonalization: HALT is undecidable reductions: other problems undecidable many examples Rice s Theorem natural problems that are not RE Recursion Theorem: non-obvious capability of TMs: printing out own description Incompleteness Theorem February 7, 2018 CS21 Lecture 14 42