Reductions Prof. Busch - LSU 1 Problem X is reduced to problem Y If we can solve problem Y then we can solve problem X Prof. Busch - LSU 2 A Definition: Language A is reduced to language B w B f (w ) There is a computable function f (reduction) such that: w A f (w ) B Prof. Busch - LSU 3 Recall: Computable function f : There is a deterministic Turing machine M which for any string w computes f (w) Prof. Busch - LSU 4 Theorem: If: a: Language A is reduced to B b: Language B is decidable Then: A is decidable Proof: Basic idea: Build the decider for A using the decider for B Prof. Busch - LSU 5 Decider for A Input string w Reduction compute f (w ) Decider f (w ) for B YES accept (halt) NO reject (halt) YES accept (halt) NO reject (halt) w A f (w ) B END OF PROOF Prof. Busch - LSU 6 Example: EQUALDFA { M1 , M2 : M1 and M2 are DFAs that acceptthe same languages} is reduced to: EMPTYDFA { M : M is a DFA that accepts the empty language} Prof. Busch - LSU 7 We only need to construct: M1, M2 Turing Machine for reduction f M1, M2 EQUALDFA Prof. Busch - LSU f M1 , M2 M DFA M EMPTYDFA 8 Let Let M1, M2 L1 be the language of DFA M1 L2 be the language of DFA M2 Turing Machine for reduction f f M1 , M2 M DFA construct DFA M by combining M1 and M2 so that: L(M) (L1 L2 ) (L1 L2 ) Prof. Busch - LSU 9 L(M) (L1 L2 ) (L1 L2 ) L1 L2 M1, M2 EQUALDFA L(M ) M EMPTYDFA Prof. Busch - LSU 10 Decider for EQUALDFA Input string Reduction compute M1, M2 f M1, M2 YES M Decider EMPTYDFA Prof. Busch - LSU NO YES NO 11 Theorem (version 1): If: a: Language A is reduced to B b: Language A is undecidable Then: B is undecidable (this is the negation of the previous theorem) Proof: Suppose B is decidable Using the decider for B build the decider for A Contradiction! Prof. Busch - LSU 12 If B is decidable then we can build: Decider for A Input string w Reduction compute f (w ) Decider f (w ) for B YES accept (halt) NO reject (halt) YES accept (halt) NO reject (halt) w A f (w ) B CONTRADICTION! Prof. Busch - LSU END OF PROOF 13 Observation: In order to prove that some language B is undecidable we only need to reduce a known undecidable language A to B Prof. Busch - LSU 14 State-entry problem Input: •Turing Machine •State q •String w M Question: Does M enter state q while processing input string w? Corresponding language: STATETM { M ,w , q : M is a Turing machine that enters state q on inputstring w } Prof. Busch - LSU 15 Theorem: STATETM is undecidable (state-entry problem is unsolvable) Proof: Reduce HALTTM (halting problem) to STATETM (state-entry problem) Prof. Busch - LSU 16 Halting Problem Decider Decider for HALTTM Reduction M ,w state-entry problem decider Compute Mˆ, q ,w f M ,w Given the reduction, if STATETM is decidable, then HALTTM is decidable Prof. Busch - LSU Decider STATETM YES YES NO NO A contradiction! since HALTTM is undecidable 17 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ, q ,w f M,w So that: Mˆ,w , q STATETM M,w HALTTM Prof. Busch - LSU 18 Construct Mˆ from M : special halt state M halting states Mˆ q qi x x ,R A transition for every unused tape symbol x of qi Prof. Busch - LSU 19 Mˆ special halt state M halting states M halts q qi Mˆ halts on state q Prof. Busch - LSU 20 M halts on input w Therefore: Mˆ halts on state q on input w Equivalently: Mˆ,w , q STATETM M,w HALTTM END OF PROOF Prof. Busch - LSU 21 Blank-tape halting problem Input: Turing Machine Question: Does M M halt when started with a blank tape? Corresponding language: BLANKTM { M : M is aTuringmachine that halts when started on blank tape } Prof. Busch - LSU 22 Theorem: BLANKTM is undecidable (blank-tape halting problem is unsolvable) Proof: Reduce HALTTM (halting problem) to BLANKTM (blank-tape problem) Prof. Busch - LSU 23 Halting Problem Decider Decider for HALTTM blank-tape problem decider Reduction M ,w Compute f M ,w Mˆ Decider BLANKTM Given the reduction, If BLANKTM is decidable, then HALTTM is decidable Prof. Busch - LSU YES YES NO NO A contradiction! since HALTTM is undecidable 24 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ f M,w So that: Mˆ BLANKTM M,w HALTTM Prof. Busch - LSU 25 Construct Mˆ from Tape is blank? M ,w : no yes Write w Mˆ Accept and halt Run M on tape with input w If M halts then halt Prof. Busch - LSU 26 Mˆ Tape is blank? no yes Write w on tape Accept and halt Run M with input w M halts on input w Mˆ halts when started on blank tape Prof. Busch - LSU 27 M halts on input w Mˆ halts when started on blank tape Equivalently: Mˆ BLANKTM M,w HALTTM END OF PROOF Prof. Busch - LSU 28 Theorem (version 2): If: a: Language A is reduced to B b: Language A is undecidable Then: B is undecidable Proof: Suppose B is decidable Then B is decidable Using the decider for B build the decider for A Contradiction! Prof. Busch - LSU 29 Suppose B is decidable reject s Decider for B (halt) accept (halt) Prof. Busch - LSU 30 Suppose B is decidable Then B is decidable (we have proven this in previous class) Decider for B NO reject s Decider for B (halt) YES accept (halt) Prof. Busch - LSU YES accept (halt) NO reject (halt) 31 If B is decidable then we can build: Decider for A Input string w Reduction compute f (w ) Decider f (w ) for B YES accept (halt) NO reject (halt) YES accept (halt) NO reject (halt) w A f (w ) B CONTRADICTION! Prof. Busch - LSU 32 Alternatively: Decider for A Input string w Reduction compute f (w ) Decider f (w ) for B NO reject (halt) YES accept (halt) YES accept (halt) NO reject (halt) w A f (w ) B CONTRADICTION! Prof. Busch - LSU END OF PROOF 33 Observation: In order to prove that some language B is undecidable we only need to reduce some known undecidable language A (theorem version 1) to B or to B (theorem version 2) Prof. Busch - LSU 34 Undecidable Problems for Turing Recognizable languages Let L be a Turing-acceptable language • L is empty? • L is regular? • L has size 2? All these are undecidable problems Prof. Busch - LSU 35 Let L be a Turing-acceptable language • L is empty? • L is regular? • L has size 2? Prof. Busch - LSU 36 Empty language problem Input: Turing Machine M Question: Is L(M ) empty? L(M ) ? Corresponding language: EMPTYTM { M : M is aTuringmachine that accepts the empty language} Prof. Busch - LSU 37 Theorem: EMPTYTM is undecidable (empty-language problem is unsolvable) Proof: Reduce ATM (membership problem) to EMPTYTM (empty language problem) Prof. Busch - LSU 38 membership problem decider Decider for ATM empty problem decider Reduction M ,w Compute f M ,w Mˆ Decider EMPTYTM Given the reduction, if EMPTYTM is decidable, then ATM is decidable Prof. Busch - LSU YES YES NO NO A contradiction! since ATM is undecidable 39 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ f M,w So that: Mˆ EMPTYTM M,w ATTM Prof. Busch - LSU 40 Construct Mˆ from M ,w : s input string Tape of Turing Machine Mˆ Mˆ Accept s s Loui si ana? yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 41 The only possible accepted string Louisiana s Turing Machine Mˆ Accept s s Loui si ana? yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 42 M accepts w L(Mˆ) {Louisiana} does not M w accept L(Mˆ) Turing Machine Mˆ Accept s s Loui si ana? yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 43 Therefore: L(Mˆ) M accepts w Equivalently: Mˆ EMPTYTM M,w ATTM END OF PROOF Prof. Busch - LSU 44 Let L be a Turing-acceptable language • L is empty? • L is regular? • L has size 2? Prof. Busch - LSU 45 Regular language problem Input: Turing Machine M Question: Is L(M ) a regular language? Corresponding language: REGULARTM { M : M is aTuringmachinethat accepts a regularlanguage} Prof. Busch - LSU 46 Theorem: REGULARTM is undecidable (regular language problem is unsolvable) Proof: Reduce ATM (membership problem) to REGULARTM (regular language problem) Prof. Busch - LSU 47 membership problem decider Decider for ATM regular problem decider Reduction M ,w Compute f M ,w YES YES REGULARTM NO NO Decider Mˆ Given the reduction, If REGULARTM is decidable, then ATM is decidable Prof. Busch - LSU A contradiction! since ATM is undecidable 48 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ f M,w So that: Mˆ REGULARTM M,w ATTM Prof. Busch - LSU 49 Mˆ from Construct M ,w : s input string Tape of Turing Machine (for some k 0) s a b ? k k Mˆ Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 50 not regular M accepts w L(Mˆ) {a n b n : n 0} does not M w accept L(Mˆ) regular Turing Machine (for some k 0) s a b ? k k Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 51 Therefore: M accepts w L(Mˆ) is not regular Equivalently: Mˆ REGULARTM M,w ATTM END OF PROOF Prof. Busch - LSU 52 Let L be a Turing-acceptable language • L is empty? • L is regular? • L has size 2? Prof. Busch - LSU 53 Size2 language problem Input: Turing Machine M Question: Does L(M ) have size 2 (two strings)? | L(M ) | 2 ? Corresponding language: SIZE 2TM { M : M is aTuringmachine that accepts exactly two strings} Prof. Busch - LSU 54 Theorem: SIZE 2TM is undecidable (size2 language problem is unsolvable) Proof: Reduce ATM (membership problem) to SIZE 2TM (size 2 language problem) Prof. Busch - LSU 55 membership problem decider Decider for ATM size2 problem decider Reduction M ,w Compute f M ,w Mˆ Decider SIZE 2TM Given the reduction, If SIZE 2TM is decidable, then ATM is decidable Prof. Busch - LSU YES YES NO NO A contradiction! since ATM is undecidable 56 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ f M,w So that: Mˆ SIZE 2TM M,w ATTM Prof. Busch - LSU 57 Construct Mˆ from M ,w : s input string s {Baton, Rouge} ? yes Tape of Turing Machine Mˆ Mˆ Accept s yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 58 2 strings L(Mˆ) {Baton , Rouge } M accepts w does not M w accept L(Mˆ) 0 strings Turing Machine s {Baton, Rouge} ? yes Mˆ Accept s yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 59 Therefore: M accepts w L(Mˆ) has size 2 Equivalently: Mˆ SIZE 2TM M,w ATTM END OF PROOF Prof. Busch - LSU 60 RICE’s Theorem Undecidable problems: • L is empty? • L is regular? • L has size 2? This can be generalized to all non-trivial properties of Turing-acceptable languages Prof. Busch - LSU 61 Non-trivial property: A property P possessed by some Turing-acceptable languages but not all Example: P1 : L is empty? YES L NO L {Louisiana} NO L {Baton, Rouge} Prof. Busch - LSU 62 More examples of non-trivial properties: P2 : L is regular? YES L YES L {a n : n 0} n n NO L {a b : n 0} P3 : L has size 2? NO L NO L {Louisiana} YES L {Baton, Rouge} Prof. Busch - LSU 63 Trivial property: A property P possessed by ALL Turing-acceptable languages Examples: P4 : L has size at least 0? True for all languages P5 : L is accepted by some Turing machine? True for all Turing-acceptable languages Prof. Busch - LSU 64 We can describe a property P as the set of languages that possess the property If language Example: P {L1 } L has property P then L P P : L is empty? YES L1 NO L2 {Louisiana} NO L3 {Baton, Rouge} Prof. Busch - LSU 65 Example: Suppose alphabet is {a } P : L has size 1? NO YES { } {a } {aa } {aaa } NO {, a } {, aa } {a , aa } NO {, a , aa } {aa, aaa, aaaa} P {{ }, {a }, {aa }, {aaa }, {aaaa},} Prof. Busch - LSU 66 Non-trivial property problem Input: Turing Machine M Question: Does L(M ) have the non-trivial L(M ) P ? property P ? Corresponding language: PROPERTYTM { M : M is aTuringmachine such that L(M ) has the non - trivial property P , that is, L(M ) P } Prof. Busch - LSU 67 Rice’s Theorem: PROPERTYTM is undecidable (the non-trivial property problem is unsolvable) Proof: Reduce ATM (membership problem) to PROPERTYTM or PROPERTYTM Prof. Busch - LSU 68 We examine two cases: Case 1: P Examples: P : L(M ) is empty? P : L(M ) is regular? Case 2: P Example: P : L(M ) has size 2? Prof. Busch - LSU 69 Case 1: P Since P is non-trivial, there is a Turing-acceptable language X such that: X P Let MX be the Turing machine that accepts X Prof. Busch - LSU 70 Reduce ATM (membership problem) to PROPERTYTM Prof. Busch - LSU 71 membership problem decider Decider for ATM Non-trivial property problem decider Reduction M ,w Compute f M ,w Decider Mˆ PROPERTYTM Given the reduction, if PROPERTYTM is decidable, then ATM is decidable Prof. Busch - LSU YES YES NO NO A contradiction! since ATM is undecidable 72 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ f M,w So that: Mˆ PROPERTYTM M,w ATTM Prof. Busch - LSU 73 Construct Mˆ from M ,w : s input string s X ? Tape of Turing Machine Mˆ Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 74 For this we can run machine MX , that accepts language X , with input string s Turing Machine s X ? Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 75 M accepts w L(Mˆ) X P does not M w accept L(Mˆ) P Turing Machine s X ? Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 76 Therefore: L(Mˆ) P M accepts w Equivalently: M,w ATTM Mˆ PROPERTYTM Prof. Busch - LSU 77 Case 2: P Since P is non-trivial, there is a Turing-acceptable language X such that: X P Let MX be the Turing machine that accepts X Prof. Busch - LSU 78 Reduce ATM (membership problem) to PROPERTYTM Prof. Busch - LSU 79 membership problem decider Decider for ATM Non-trivial property problem decider Reduction M ,w Compute f M ,w YES YES PROPERTYTM NO NO Decider Mˆ Given the reduction, if PROPERTYTM is decidable, then ATM is decidable Prof. Busch - LSU A contradiction! since ATM is undecidable 80 We only need to build the reduction: Reduction M ,w Compute f M ,w Mˆ f M,w So that: Mˆ PROPERTYTM M,w ATTM Prof. Busch - LSU 81 Construct Mˆ from M ,w : s input string s X ? Tape of Turing Machine Mˆ Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 82 M accepts w L(Mˆ) X P does not M w accept L(Mˆ) P Turing Machine s X ? Mˆ Accept s yes yes w on tape, and •Simulate M on input w •Write Prof. Busch - LSU M accepts w ? 83 Therefore: L(Mˆ) P M accepts w Equivalently: M,w ATTM Mˆ PROPERTYTM END OF PROOF Prof. Busch - LSU 84