Scribd
Upload
28 views4 pages

Decidability - Lesson of 18th March: Decision Problems

This document discusses decidability and how formalizing problems as languages can help determine if they are decidable or not. It presents an example of formalizing an arithmetic expression problem as a language L, and shows how a membership problem can be solved to check if a string is in the language. While finite state machines can solve some problems, Turing machines may be needed to solve more complex problems that could loop indefinitely. Formalizing problems allows researching decidable versions that could lead to algorithms to solve the original problems in a finite time.

Uploaded by

Daven
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
28 views4 pages

Decidability - Lesson of 18th March: Decision Problems

This document discusses decidability and how formalizing problems as languages can help determine if they are decidable or not. It presents an example of formalizing an arithmetic expression problem as a language L, and shows how a membership problem can be solved to check if a string is in the language. While finite state machines can solve some problems, Turing machines may be needed to solve more complex problems that could loop indefinitely. Formalizing problems allows researching decidable versions that could lead to algorithms to solve the original problems in a finite time.

Uploaded by

Daven
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/ 4

Decidability - Lesson of 18th March

Decision problems
They answer to question like 'Is a graph cyclic?’, etc. So the answer of a decision problem could be yes or no.

Research problems
They answer to real problems, like ‘solve an equation’, ‘find a path for a graph’, etc.

Formalization
A necessary step to solve a decidability problem is formalize our problem.
For example, we could think about formalize an arithmetic expression as a problem. The expression has to be
defined in a formal way and encoded as a language.

An arithmetic expression (AE) could be defined as follow:


AE := NUM | AE + AE | AE * AE | etc

φ is an arithmetic expression
<φ> is a suitable encoding (one of the possible encodings)

For example, a good encoding could be a binary encoding over a ternary alphabet.

Symbol Encoding

0 000

1 001

+ 010

- 011

* 100

÷ 101

% 110

Example

This is our expression:


φ=3+2

So, in binary it should be encoded as:


<φ> = 11 + 10

And through our encoding we can write it as:


<φ> = 001 001 010 001 000

What is the language asssociated to this kind of equation?


L = {<φ> # <n> | φ is an A.E. over Q and {+, -, *, ÷, %} ∧
n∈Q∧
φ=n}
Notes:

1. we use Q because we use also the division (so we need rational numbers). Otherwise we can use Z.
2. φ = n means that the value of the equation has to be correct.

How to encode rational numbers?


We can encode rational numbers (Q) through a triple
(s, n, d), where s is the symbol
n is the numerator (binary encoding)
d is the denominator (binary encoding)

And using a rich alphabet

Γ = {+, 1, 0, −, ∗, ÷, %, ), (, #}

Example

3
The number 3 can be seen as + 1 , so it can be encoded as (+, 11, 1).
2
The number 2 can be seen as + 1 , so it can be encoded as (+, 10, 1).
5
The number 5 can be seen as + 1 , so it can be encoded as (+, 101, 1).

So..
(+, 11, 1) + (+, 10, 1) # (+, 101, 1) ∈ L
Now it belongs to L !

L captures all the strings that are good-encoded, and the expression represented is equal to the number after the
#.

Why this should help us?


This encoding help us classifying a problem and understanding if a certain problem can be solved through a
machine.
(?) • Check for every enumeration of the nunmbers if the strings belongs to the language (so if it’s satisfied).
The membership problem
∗
Given a string w ∈ ϵ , decide wether w ∈ L.
If the language L is simple we could use a final state machine. Otherwise, when we meet a difficult problem, a
FSM could not be enough.

Example

+
L = L(a b ) ∈ REG
We can define a procedure to check if we can reach a final state:

1 boolean solveL (string w) {


2 set Ml = <ε, K, S0, ς, F> //or any other machine for L
3 parse w w.r.t. Ml
4
5 if you reach a state in F
6 return true
7 else
8 return false
9 }

and we can use this to compute our result solving an arithmetic expression

1 rational computeResult (AE p) {


2 for each p ∈ ℕ {
3 n := int2frac(p)
4
5 if (solveL (<p> # <n>) == true)
6 return n;
7 }
8 }

But…we don’t have always a solution (for example in a disequation)!

For example x 2 + 2a + 1 < 0 will loop forever.

Sometime we could have a bound and our procedure will not loop infinitely, for example in the case of finding a
prime number (we iterate from 1 to the sqrt of n).

Every research problem can be solved if we have a language associated to the problem and a
solution of the form

L = {< in > # < out > | relation}, where 'relation' is a relation between the input and the output.

We don’t use this brute-force enumeration in real problems, but we know that if there’s a decisional version
of a problem, there’s an algorithm that in a finite time will solve the research version of the
problem.

Turing machine
There’s some differences between a Final state machine and a Touring machine:

FSM =< ϵ, K, S 0 , δ, F >


TM =< Γ, K, S 0 , δ >
δ : Γ × K → Γ × K × {←−, −→}

Notes and differences with the FSM:

we can change the content of the tape


we can move right or left on the tape
there’s no final state: we have two special state YES (Y) and NO (N)

+
L (ab )
Example of FSM

Example of TM

We can imagine a Turing machine as a tape

At the beginning the cursor is under the blank cell before the first input, then it moves.

You might also like