AEiE0803 Turing machine¶
Turing-machine fundamentals¶
- A Turing machine (TM) is the standard abstract model of general computation.
- It has a finite control, an infinite tape, and a read-write head.
- A TM may accept, reject, or run forever.
- A language \(L\) is recursively enumerable or Turing-recognizable if some TM accepts exactly the strings in \(L\); on nonmembers it may reject or loop forever.
- A language is decidable if some TM halts on every input and accepts exactly the strings in the language.
Typical tuple form:
\[
M=(Q, \Sigma, \Gamma, \delta, q_0, q_{acc}, q_{rej})
\]
Typical transition notation:
\[
\delta(q,a)=(p,b,R)
\]
meaning read \(a\), write \(b\), move right, and enter state \(p\).
Roles and variants¶
| Role or variant | Meaning | Recognition cue |
|---|---|---|
| Recognizer | Accepts language members | May loop on nonmembers |
| Decider | Always halts | Accept or reject for every input |
| Computing function | Produces output on tape | Input-output viewpoint |
| Enumerator | Prints language strings | Generates members one by one |
| Multi-track TM | Several tracks in one tape cell | Same power as standard TM |
| Multi-tape TM | Several tapes and heads | Same power, often faster |
| Nondeterministic TM | Branching computation | Same decidable and RE classes |
| Universal TM | Simulates encoded TMs | Stored-program idea |
- These variants improve convenience or efficiency, not computability class.
- The universal TM reads an encoding \(\langle M,w \rangle\) and simulates machine \(M\) on input \(w\).
Church-Turing thesis, acceptance, and complexity¶
- Church-Turing thesis is a thesis, not a proved theorem.
- It states that any effectively computable procedure can be performed by a TM.
- A string is accepted if the machine enters an accept state.
- A recognizer may loop forever on a nonmember.
- Time complexity counts transition steps as a function of input length \(n\).
- Space complexity counts the number of tape cells used.
Decidability limits¶
- Some problems are undecidable: no TM decides them for all inputs.
- The Halting Problem is the standard undecidable problem.
- Recursively enumerable is weaker than decidable.
- Reducibility is used to transfer undecidability from a known hard problem.
Recognition cues for Turing machines¶
- "Always halts" indicates decider, not mere recognizer.
- "Can enumerate all strings in the language" refers to an enumerator.
- "Simulates every TM" points to universal TM.
- "Undecidable" means no algorithm exists for every instance, not just that the problem is difficult.
Common traps in Turing machines¶
- Nondeterminism does not add computability power to TMs.
- Multi-track and multi-tape TMs are equivalent in computational power to the standard TM.
- Church-Turing thesis cannot be proved purely mathematically because it links a formal notion to an informal one.
- A recognizer is not automatically a decider.
One-step examples for Turing machines¶
- A TM for \(\{0^n1^n\}\) can repeatedly mark one 0 and one 1 until nothing unmatched remains.
- An enumerator for all binary strings may output \(\varepsilon,0,1,00,01,10,11,\ldots\) in length-lexicographic order.
- A universal TM is the abstraction behind an interpreter that runs many different programs given as data.
Turing-machine revision box¶
TM is the baseline model of general computation. Recognizer may loop; decider must halt. Multi-tape, multi-track, and nondeterministic TMs do not exceed the ordinary TM in computability power. Halting is the canonical undecidable problem.