turing machines barath raghavan

35
Theory of Computation Turing machines Barath Raghavan CS 361 Fall 2009 Williams College

Upload: trinhdiep

Post on 02-Feb-2017

256 views

Category:

Documents


5 download

TRANSCRIPT

Page 1: Turing machines Barath Raghavan

Theory ofComputation

Turing machines

Barath RaghavanCS 361 Fall 2009Williams College

Page 2: Turing machines Barath Raghavan

MISC

Homework 2 back

Page 3: Turing machines Barath Raghavan

What isCOMPUTATION?

Page 4: Turing machines Barath Raghavan

1. Computation without storage.(Finite automata, regular expressions)

2. Computation with storage.(Pushdown automata, context-free grammars)

3. Limits of computation.(Turing machines, undecidability)

4. Complexity of computation.(Complexity classes, reductions)

Page 5: Turing machines Barath Raghavan

1. Computation without storage.(Finite automata, regular expressions)

2. Computation with storage.(Pushdown automata, context-free grammars)

3. Limits of computation.(Turing machines, undecidability)

4. Complexity of computation.(Complexity classes, reductions)

Page 6: Turing machines Barath Raghavan

What isCOMPUTATION?

Page 7: Turing machines Barath Raghavan

ALANTURING

Page 8: Turing machines Barath Raghavan

What were the limitations of automata?

Page 9: Turing machines Barath Raghavan

Suppose you have

infinite storage.

Page 10: Turing machines Barath Raghavan

Direct access to storage.

Page 11: Turing machines Barath Raghavan

Σ, ΓQ set of states

input, tape alphabet

transitionfunction

q0 ∈ Q start state

accept / rejectstates

TMs

δ : Q× Γ→ Q× Γ× {L, R}

Qaccept ∈ QQreject ∈ Q

Page 12: Turing machines Barath Raghavan

Properties of a Turing Machine1. Has an infinite tape.

2. Can read/write from/to the tape.3. Can move left/right on the tape.

4. Has states and transitions that govern its actions.

Page 13: Turing machines Barath Raghavan

L = {w#w | w ∈ {0, 1}∗}

How do we recognize this language?

Page 14: Turing machines Barath Raghavan

Turing Machine Operation

1. Input starts on the tape.2. Rest of tape is blank.3. No blanks in input.4. TM follows transition rules.5. TM cannot move past left end of tape.6. TM halts on accept/reject.

Page 15: Turing machines Barath Raghavan

L = {w#w | w ∈ {0, 1}∗}

How do we recognize this language?

Page 16: Turing machines Barath Raghavan

A language is

Turing RecognizableorRecursively Enumerable if some TM recognizes it.

RELanguage Class

Page 17: Turing machines Barath Raghavan

A language is

Decidableor

Recursiveif some TM decides it.

Language Class R

Page 18: Turing machines Barath Raghavan

Turing machines are more powerful than

pushdown automata.

Page 19: Turing machines Barath Raghavan

Let’s build something more powerful than a

Turing machine.

Page 20: Turing machines Barath Raghavan

Suppose a Turing machine had

multiple tapes.

Page 21: Turing machines Barath Raghavan

Suppose a Turing machine were

nondeterministic.

Page 22: Turing machines Barath Raghavan

What if we had a machine that

enumeratedthe strings in a language?

Page 23: Turing machines Barath Raghavan

A language is in

only if some enumerator

enumerates it.

RE

Page 24: Turing machines Barath Raghavan

What else is equivalent to Turing Machines?

Page 25: Turing machines Barath Raghavan

Universal Models of Computation

Lambda Calculus [Church 1936]

Turing Machines [Turing 1936]

Random Access Machines [von Neumann/Turing 1940s]

Programming Languages [Backus 1954]

Cellular Automata [Ulam/von Neumann 1966]

Page 26: Turing machines Barath Raghavan

Every universal model of computation allows for

the design of algorithms.

Page 27: Turing machines Barath Raghavan

Church-Turing thesis

For every algorithm that solves some problem there is a Turing Machine for that same problem.

Page 28: Turing machines Barath Raghavan

Implications of the Church-Turing thesis:

Page 29: Turing machines Barath Raghavan

1. No conventional language is more powerful than any other.

Page 30: Turing machines Barath Raghavan

2. A problem that can’t be solved on a Turing machine can’t be solved.

Page 31: Turing machines Barath Raghavan

Broad unresolved, potential implications of the Church-Turing thesis:

Page 32: Turing machines Barath Raghavan

1. The universe is a Turing machine(or if not a Turing machine it still cannot be used for non-recursive computation.)

Page 33: Turing machines Barath Raghavan

2. There is no problem solvable by a human brain that cannot be solved by a Turing machine.

Page 34: Turing machines Barath Raghavan

As with all science,

this theory is falsifiable:

1. Devise a task a computer cannot do.2. Show that a physical process cannot be replicated by a digital computer.

Page 35: Turing machines Barath Raghavan

Reading: Sipser 3.1-3.3