Scribd
Upload
47 views8 pages

Church's Thesis

The document discusses Church's thesis, which states that the intuitive notion of computation is equivalent to computation by Turing machines. It introduces universal Turing machines that can simulate any Turing machine. It defines decision problems and states that a problem is decidable if the set of strings having the desired property is recursive/decidable by a terminating computer program. The halting problem, which asks whether a Turing machine halts on a given input, is proven to be undecidable. Many other problems about formal languages generated by grammars and Turing machines are also shown to be undecidable.

Uploaded by

theresa.painter
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
47 views8 pages

Church's Thesis

The document discusses Church's thesis, which states that the intuitive notion of computation is equivalent to computation by Turing machines. It introduces universal Turing machines that can simulate any Turing machine. It defines decision problems and states that a problem is decidable if the set of strings having the desired property is recursive/decidable by a terminating computer program. The halting problem, which asks whether a Turing machine halts on a given input, is proven to be undecidable. Many other problems about formal languages generated by grammars and Turing machines are also shown to be undecidable.

Uploaded by

theresa.painter
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Churchs Thesis

(Church-Turing thesis)

Intuitive notion of computation

equals

computation by Turing Machines

We outlined how to encode TMs as strings. Notation: M {0, 1} M encoded as a string.

(also X {0, 1} symbol X encoded as a string, etc)

Mx the TM encoded by string x. M such that M = x Mx = M0 if x is not the encoding of any TM

Universal Turing Machine (UTM)


a TM U such that L(U ) = { M, x | x L(M ) } U simulates any given TM. Construction of U . 3 tapes: M state of M tape of M Put M , s and x on the tapes. Simulate M step-by-step.

Decision problems
Problems like Is a given CFG ambiguous? Is a given graph acyclic? Does a given TM accept a given string? Does x + y = z ? (CFGs, graphs, etc can be encoded as strings) A problem is decidable if the set of strings having the property in question is recursive. (This set sometimes called the language of the problem.) Decidable problems can be solved by a terminating computer program. Undecidable cannot.
4

The halting problem is undecidable


Halting problem (HP): Given M - a TM, x - a string, does M halt on x ? Assume HP decidable. This means: The question can be answered by a total TM. In other words, the language of the problem: HP = { M, x | M halts on x } is recursive. (HP = L(K ) for some total TM K ). We can construct following TMs: K: K : N: Does M halt on x ? Does Mx halt on x ? N loops on x i Mx halts on x.

Hence N loops on N i N halts on N . Contradiction!


5

Language HP is r.e. but not recursive.

Th. If both L and L are r.e. then L (and L) is recursive. Why? Let L = L(M1 ), L = L(M2 ). One can construct a TM M simulating M1 and M2 in parallel (a step of M1 , a step of M2 , etc), M accepts when M1 accepts and rejects when M2 accepts. M is total.

Thus HP is not r.e.

Reduction of HP to MP (membership problem)


MP = L(U ) = { M, x | x L(M ) } For any TM M we can construct a TM M such that: x L(M ) i M halts on x. Then we can construct a total TM K which out of any M, x produces M , x . If there exists a total TM N accepting MP then by composing K and N we can construct a total TM for HP . The latter means the halting problem is decidable, we know it is not. Hence such N does not exists. Thus MP is undecidable.

Examples of undecidable problems


CFGs: G ambiguous? L(G) deterministic CFL? L(G) regular? L(G) = ? (Decidable: L(G) = ?, x L(G) ? ) TMs: M total? L(M ) = ? x L(M ) ? L(M ) nite? L(M ) ? L(G1 ) = L(G2 ) ? L(G1 ) L(G2 ) = ? L(G1 ) L(G2 ) context-free? L(G) context-free?

L(M ) regular (CF, recursive)? Any nontrivial property of r.e. languages is undecidable! [Chapter 34] Programming: Termination of programs Equivalence of programs Dead code (is given program fragment ever executed?)
8

You might also like