Scribd
Upload
249 views3 pages

Rice Theorem

Rice's Theorem states that for any non-trivial property P of formal languages, determining whether the language accepted by a Turing machine has property P is undecidable. A property is non-trivial if it is satisfied by some but not all recursively enumerable languages. The proof reduces the acceptance problem for Turing machines (which is undecidable) to the problem of determining whether a language has property P. Therefore, determining property P must also be undecidable.

Uploaded by

Uday Bhan Singh
Copyright
© © All Rights Reserved
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
249 views3 pages

Rice Theorem

Rice's Theorem states that for any non-trivial property P of formal languages, determining whether the language accepted by a Turing machine has property P is undecidable. A property is non-trivial if it is satisfied by some but not all recursively enumerable languages. The proof reduces the acceptance problem for Turing machines (which is undecidable) to the problem of determining whether a language has property P. Therefore, determining property P must also be undecidable.

Uploaded by

Uday Bhan Singh
Copyright
© © All Rights Reserved
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/ 3

1

1.1

Rices Theorem
Properties

Checking Properties Given M Does L(M ) contain M ? Is L(M ) non-empty? Undecidable Is L(M ) empty? Is L(M ) innite? Is L(M ) nite? Undecidable Is L(M ) co-nite (i.e., is L(M ) nite)? Is L(M ) = ?

None of these properties can be decided. This is the content of Rices Theorem. Properties Denition 1. A property of languages is simply a set of languages. We say L satises the property P if L P. Denition 2. For any property P, dene language LP to consist of Turing Machines which accept a language in P: LP = { M | L(M ) P} Deciding LP : deciding if a language represented as a TM satises the property P. Example: { M | L(M ) is innite}; Etm = { M | L(M ) = } Non-example: { M | M has 15 states} This is a property of TMs, and not languages!

Trivial Properties Denition 3. A property is trivial if either it is not satised by any r.e. language, or if it is satised by all r.e. languages. Otherwise it is non-trivial. Example 4. Some trivial properties: Pall = set of all languages Pr.e. = set of all r.e. languages P where P is trivial P = {L | L is recognized by a TM with an even number of states} = Pr.e. Observation. For any trivial property P, LP is decidable. (Why?) Then LP = or LP = .

1.2

Main Theorem

Rices Theorem Proposition 5. If P is a non-trivial property, then LP is undecidable. Thus { M | L(M ) P} is not decidable (unless P is trivial) We cannot algorithmically determine any interesting property of languages represented as Turing Machines! Properties of TMs Note. Properties of TMs, as opposed to those of languages they accept, may or may not be decidable. Example 6. { { { { M M M M | M has 193 states} Decidable | M uses at most 32 tape cells on blank input } | M halts on blank input} | on input 0011 M at some point writes the Undecidable symbol $ on its tape}

Proof of Rices Theorem Rices Theorem If P is a non-trivial property, then LP is undecidable. Proof. Suppose P non-trivial and P. If P, then in the following we will be showing LP is undecidable. Then LP = LP is also undecidable. Recall LP = { M | L(M ) satises P}. Well reduce Atm to LP . Then, since Atm is undecidable, LP is also undecidable. Broadly the idea behind the reduction is as follows. Since P is non-trivial, at least one r.e. language satises P. i.e., L(M0 ) P for some TM M0 . We will show a reduction f that maps an instance M, w for Atm , to N such that If M accepts w then N accepts the same language as M0 . Then L(M ) = L(M0 ) P If M does not accept w then N accepts . Then L(N ) = P Thus, M, w Atm i N LP . We now describe the reduction precisely. The reduction f maps M, w to N , where N is a TM that behaves as follows: On input x Ignore the input and run M on w If M does not accept (or doesnt halt) then do not accept x (or do not halt) If M does accept w then run M0 on x and accept x iff M0 does. 2

Notice that indeed if M accepts w then L(N ) = L(M0 ). Otherwise L(N ) = . Rices Theorem Recap Every non-trivial property of r.e. languages is undecidable Rices theorem says nothing about properties of Turing machines Rices theorem says nothing about whether a property of languages is recurisvely enumerable or not.

Big Picture . . . again

almost all properties! Languages Ld , Atm , Etm

Recursively Enumerable Atm , Etm , HALT Decidable CFL Regular Lanbncn L0n1n

You might also like