Theory of Computation

ND HUNG

Sirindhorn International Institute of Technology

[email protected]

October 10, 2021

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 1 / 26

1 Turing Machines (TM)

2 Combining TMs

3 Recursive languages

4 Recursive Functions

5 Recursively Enumerable Languages

6 Extensions of Turing Machines

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 2 / 26

 Turing Machines (TM)

Turing Machine (TM)

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 3 / 26

 Turing Machines (TM)

Turing Machine (TM)

#
I of states (finite
. .

the set
. .

the alphabet out of

Definition ↑/ It gets
infinite
A Turing machine is a 5-tuple M = (K , ⌃, , s, H) where t an

loop.
atsee
- -

K is a finite set of states, Lithentofhating

⌃ is an alphabet containing the blank symbol t and the left end
symbol ., but not containing the symbols ! and ,
s 2 K is the initial state,
H ✓ K is the set of halting states, inputstateinput
Occurrent syma
new symbol
, the A
transition operation
function, is a function from (K H) ⇥ ⌃ to
memory
K ⇥ (⌃ [ { , !}) such that
- - -start ofThe tape*
D
next
for all q 2 K H there exists p such that (q, .) = (p, !), and -

state for all q 2 K H, and a 2 ⌃ {.}, if (q, a) = (p, b) then b 6= .

It

D
never

onto
writes down
the memory
.
OnlyaHart anysnot A
ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 4 / 26
 Turing Machines (TM)

Turing Machine (TM)

h : halting state

transition &1 J
↓

"
D/
-
-
,
aa a a

O air . (80 , Danaa) I

↓ (g ,
Daaca) [D/)
↓ (gI , Dana) [a/W]
↓
(00 DWaaa) [r/-)
,

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 5 / 26

 Turing Machines (TM)

Turing Machine (TM)

a)

w/r
Do-

68 h
0
2
DI-

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 6 / 26

 Turing Machines (TM)

Turing Machine (TM)

Example: A Turing machine for accepting start state
L = {! 2 {a, b}⇤ &
| |!| is even} is M = (K , ⌃, , s, H) where
K = {s, q0 , q1 , y , n}, H = {y, n}
D
↑ initial config
becap reject (S
⌃ = {a, b}
: (K H) ⇥ ⌃ ! K ⇥ (⌃ [ { , !})
,
D) A

cursor
input
q (q, ) M accepts aa: string
s t (q0 , !) ↓ &
.
Z

&
- (s, .taa) `M (q0 , taa) `M
q0 t (y, t) 24 - -

>
-
(q1 , taa) `M (q0 , taat) `M
q0 a (q1 , !)
M
1

(y , taat) sop/LT
q0 b
z
(q1 , !)
q1 t (n, t) M rejects a:
q1 a (q0 , !)
Z
(s, .ta) `M (q0 , ta) `M
q1 b (q0 , !) (q1 , tat) `M (n, tat)
2
STOP
/HALT
ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 7 / 26
 Turing Machines (TM)

Turing Machine (TM) syntactic >

-
(G2 DWabac,
D
sugar =
(G ,
Dra , back

A configuration of a Turing machine M = (K , ⌃, , s, H) is a member

Acurrent
state
of (82 Draca WWL) , -

For eDra,
K ⇥ .⌃⇤ ⇥ (⌃⇤ ⇥ (⌃ {t}) [ {e})

Definition
-
empty strings
DLet (q1 , !1 a1 u1 ) and (q2 , !2 a2 u2 ) be configurations of M. Then
it. we can
ignore
(q1 , !1 a1 u1 ) `M (q2 , !2 a2 u2 )

if and only if for some b 2 ⌃ [ { , !}, (q1 , a1 ) = (q2 , b) and one of

the following cases holds 5 : symbol -
> replace the cell's etr.

D : more ptr to the LHS

1. b 2 ⌃, !1 = !2 , u1 = u2 and a2 = b
,

D - i m RHS

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 8 / 26

 Turing Machines (TM)

Turing Machine (TM)

Definition
2. b = , !1 = !2 a2 , and either
a. u2 = a1 u1 , if a1 6= t or u1 6= e, or
b. u2 = e, if a1 = t and u1 = e
3. b =!, !2 = !1 a1 , and either
a. u1 = a2 u2
b. u1 = u2 = e and a2 = t

(p Dabc)
,
p ,
Dab

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 9 / 26

 Turing Machines (TM)

Turing Machine (TM)

the machine returns a result .

result

A halted configuration is of the form (h, !1 a!2 ) with h 2 H.

E

A computation is a sequence of configurations C1 , C2 , . . . , Cn

such that C1 `M C2 `M · · · `M Cn . We say that the computation
has length n and C1 yields Cn .
M is said to halt on input ! iff (s, .t!) yields some halted
D configuration.
*

Dah Du
·

IS ,
,

be the
same

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 10 / 26

 Combining TMs

Symbol writing machine

Ma tape
write's to the
RHS
. Me Me
For each symbol a 2 ⌃ [ { , !} {.}, Ma is a Turing machine
which writes the symbol a into the scanned tape square, i.e.
Ma = ({s, h}, ⌃, , s, {h}) where (s, b) = (h, a) for each
b 2 ⌃ {.}.
Notations: for short, we often write
a for Ma : write 'a' to the current cell
=

L for M : more pr to LHS .

R for M! : more ptr to RHS ,

-
Lt : Finds the first blank square to the left of the currently scanned
symbol (p Dw abcdWefg) +in (p X Wabed Lefg)
, ,

Rt : Finds the first blank square to the right of the currently scanned
symbol (p DWabedWefgWhij) thr
,

Exercises: Study Examples 4.1.5,>

4.1.6, 4.1.7, 4.1.8, 4.1.9, 4.1.10 in
-

textbook (p D WabedLefgWhij)
,

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 11 / 26

 Combining TMs

The rules for Combining Machines

Suppose M1 = (K1 , ⌃, 1 , s1 , H1 ), M2 = (K2 , ⌃, 2 , s2 , H2 ) and
M3 = (K3 , ⌃, 3 , s3 , H3 )
① Let M work
,
D until it halfs,
↑
- if clauses & Peek the curr · all .

③ If the all atm is

let Me work,
·
a
,

① If the all oth is

ob Let M, work
M = (K , ⌃, , s, H) where K = K1 [ K2 [ K3 , s = s1 , H = H2 [ H 3 . For
,

each 2 ⌃, q 2 K H, (q, ) is defined as follows

If qnon-halting
state
2 K1 H1 , then (q, ) = 1 (q, )

YAAD
-

If q 2 K2 H2 , then (q, ) = 2 (q, )

If q 2 K3 H3 , then (q, ) = 3 (q, )
Ifhalting
state of
q 2 H1 , then (q,Ma
-
) if = a; (q, ) = (s3 , )bif = b
) = (s2 , nd
a
ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 12 / 26
 Recursive languages

Recursive languages
re

defined 81
.
Lang rec can be as a recursive
-
An initial configuration is of the form (s, .t!) recursive
algo ,

Definition
Given is a Turing machine M = (K , ⌃, , s, H) with H = {y, n}. Any
halting configuration whose state is y is said to be an accepting (y ) . -

configuration while a halting configuration whose state is n is said to

be a rejecting configuration. (n - ,

For an input ! 2 (⌃ {t, .})⇤ , we say that

M accepts ! if (s, .t!) yields an accepting configuration. t(y ,+
M rejects ! if (s, .t!) yields an rejecting configuration. (n - ]

2) M gets stuck in an O-loop ,

we won't get answer .

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 13 / 26

 Recursive languages

Recursive languages do
ALT
-M
has .

M tell if it
*
Deciding : can

Definition
accepts/rejects an input string.
Let ⌃0 ✓ ⌃ {t, .} be an alphabet, called input alphabet. We say
that M decides a language L ✓ ⌃⇤0 if for any string ! 2 ⌃⇤0 ,
If ! 2 L then M accepts !
If ! 62 L (i.e. ! 2 ⌃⇤0 L) then M rejects !
A language L is called recursive if it is decided by some Turing
machine.
Note: if TM M decides a language L ✓ ⌃⇤0 , then M halts on any ! 2 ⌃⇤0 ,
but M is not required to halt on ! 2 ⌃⇤ ⌃⇤0 (see supplementary note
for further discussion)

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 14 / 26

 Recursive languages

Recursive languages
Example: Show that L0 = {an bn | n 0} is recursive.
(s Dabb) in it config aabb
,. t dabb `d
.

. t dadb `d
↑

a .tdadb `Lt
. t dadb `R
+ a b b
. t dadb `d
=>

-
. t dddb `R
b
. t dddb `R +
. t dddb `d
a
Sample run: *
~.taabb `
R
. t dddd `Lt + o of $
-. t aabb `d .tdddd `R *
. t dabb `R . t dddd `⇤R -
Y. t dabb `R D . t ddddt `y
D . t ddddt

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 15 / 26

 Recursive languages

Recursive languages

Example: Show that L1 = {an bn c n | n 0} is recursive.

- -

oo
Dr aabbxx

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 16 / 26

 Recursive languages

Recursive languages

Example: Any regular language is recursive.

Let M = (K , ⌃, , s, F ) be a DFA. We can build a TM
M 0 = (K [ {y, n}, ⌃0 , 0 , s0 , {y , n}) that decides L(M) as follows
⌃0 = ⌃ [ {., t} a

0 :K ⇥ ⌃0 ! (K [ {y , n}) ⇥ (⌃0 [{ , !}) where -

>
&
(s0 , t) = (s, !)
P q
0
(p, a) = ( (p, a), !) for p 2 K , a 2 ⌃
(p, t) = (y, t) for p 2 F G(p a)
0
=
(p, t) = (n, t) for p 62 F
0 ,

(g -)

w
=
,

-
>
scan through
- make state trans .

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 17 / 26

 Recursive Functions

Recursive Functions f(w,, W, Wa

I ...
.,
. .

Modern comp
:
We treat math &1 as

Definition tape manipulation .

M
Let M = (K , ⌃, , s, {h}), ⌃0 ✓ (⌃ {t, .} and ! 2 ⌃⇤0 . If

(s, .t!) `⇤M (h, .tu)

then u is called the output of M on input ! and denoted by M(!). We

say that
1. M computes function f : ⌃⇤0 ! ⌃⇤0 if for any ! 2 ⌃⇤0 , M(!) = f (!).
2. M computes function f : Nk ! N (k 1) if for all binary numbers
!1 , . . . , !k , M(!1 ; . . . ; !k ) = f (!1 , . . . , !k ) where f (!1 , . . . , !k ) is
also in binary representation.
A function f is recursive if there is a Turing machine computing f .
X / v U
+,
-

, , , , , ...

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 18 / 26

 Recursive Functions

Recursive Functions
Example: Show that succ(n) = n + 1 and d(n) = 2n are recursive
functions. succ(101) 110 =

0
= 1
D 0
R⊔ L 1 L⊔
L
D ↑
⊔
Me Sa
1 S! L ⊔
Revised

tru
I
DLI01
the shift-right machine
-

*
+ DW/10
ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 19 / 26
 Recursively Enumerable Languages

Recursively Enumerable Languages

definite of y/n yet
.
~
answer

Definition to beek the current state

have

Let M = (K , ⌃, , s, H), ⌃0 ✓ (⌃ {t, .} and L ✓ ⌃⇤0 . We say that M

semidecides L if for any string ! 2 ⌃⇤0 :

! 2 L iff M halts on input !

A language L is called recursively enumerable if it is semidecided by

some Turing machine.
T o
Lang rec . can be explained/tenbor
Theorem: algo mud
gives yu by a recursive

↑
answer ,

If a language is recursive then it is also recursively enumerable.

If a language is recursive then its complement is also recursive.
L= Carba/n > on / I
ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 20 / 26
 Extensions of Turing Machines

There are several important extensions of Turing machines. However,

all these extensions of TM can be simulated by the standard model.
1 Several tapes instead of one
2 Random access memory
-addr >
- data
(no more wait)
-

3 Nondeterministism : transition f -
> tranation rel1
We focus on the multi-tape extension.

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 21 / 26

 Extensions of Turing Machines

Multi-tape Turing Machines

A k-tape Turing machine consists of a finite control together with k

infinite tapes. Each tape is scanned by a read/write head. The
machine can in one step sense the symbols scanned by all its heads
and then depending on those symbols and it s current state, rewrite
some of those scanned squares or move some of the heads to the left
or right.

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 22 / 26

 Extensions of Turing Machines

Multi-tape Turing Machines ouis

input
Gaddn . tapes

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 23 / 26

 Extensions of Turing Machines

Multi-tape Turing Machines

Definition
A k-tape Turing machine is a 5-tuple M = (K , ⌃, , s, H) where
statiodnetaaa
=
states

K , ⌃, s, H are as in the definition of ordinary Turing machine, and

, the transition function, is a function
Ult(K H) ⇥ ⌃k ! K ⇥ (⌃ [ { , !})k

Take a
-
>
u
memory
next operation
cuw ,

state meak peek state

on each tape
of le tapes
at the sam time

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 24 / 26

 Extensions of Turing Machines

Multi-tape Turing Machines

y/o answer

Exercise: Design a two-tapes machine to decide

L = {!c! | ! 2 {a, b}⇤ } where c 62 {a, b}
~

WCW
abaa
-
c
ama

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 25 / 26

 Extensions of Turing Machines

Multi-tape Turing Machines

Extending Turing machines with more than one tape does not increase

F
the power of the machines.

Theorem: Let k > 1 and M = (K , ⌃, , s, H) be a k-tape Turing

machine. Then there is a standard Turing machine
M 0 = (K 0 , ⌃0 , 0 , s0 , H) where ⌃0 ✓ ⌃, and such that for any
! 2 (⌃ {t, .})⇤

M halts on input x with output y on its first tape after t steps

if and only if

M 0 halts on input x with the same output after a polynomial number of

steps of t and the size of x.

ND HUNG (ICT, SIIT) Chapter 4: Turing Machines October 10, 2021 26 / 26

Useful Subroutines for Denigning TMs

L go left
3 of questions
type

R go right
① Compute arithmatic
functions :

Ly ecan to LHS until W is found f(n) = 2n + 1

Ry scan to RHS until W is found ② Check conditions :

E misd
f(n)-dif
s

Sh stiff entire tape to LHS

S stiff entire tape to RHS ③ Lang recog :

.

R
qwEZ /num)
*
L =

M& return O Dent@

M1 return 1 DWW + DL1

Arithmicinitialth
# 2
# f(n)
contain
- In
-n

&2 f(m) = 2n + 1 +Q1 -

Q

&3 f(m) = 2n +
3 @f(n) = n + 1 : mis var
len
Dr0100 + DW0101
.

& f(u)
-

= So *
D201011 D 2011
A
D41111 + D41004
25 f(a ,
b) = a + b

DLGD
L
Dr10000
Dr1d90
Espe
2

Sh misod
t
& flul =

Sh mise
s

& ful =

&3 f(m) = i mod ?

Tum
3

Q an yu *
(g Daabb)
.
+ (y ,
D ...
)
*
* (go Draab) + (n Dr )
an yo ,
Q2
...

n , ,

&3 an bncamym

anywanyn
I
Multi-tape Turing Machine

- Attach each instruction with the tape index .

(M go left

Ru go right
Idea
-

Ly ecan to LHS until W is found D Main tape shores

the input arguments

R scan to RHS until W is found land the output)

S stiff entire tape to LHS ② Extra tapes =

temporary variables

Su stiff entire tape to RHS

"
③ Config :

M& return O Pent@ (DUw ,

DX
,
. . . )
-
main extra

M1" return 1 DLN + DL1

tape tapes

>
Drop
ri
the take
simp :

- Trt
1
. Split the input arguments in the main tape.

Iman -

1
. Scand to tefts to find next W 4 Read t all
. .

writed y
.
2 Moved to teHS for I cell .
If not W
,
write it
?
KHS
More
move
.
3 to teHS for I cell
.
moved tells
1 2
go to 4
.
Otherwise scan
,
to LHS for L
,
2 .
Copy the
string in the main tape to the extra one .

aa
1
. Moved 200 RHS
,

.
2 Read" all
?
If not W write it
,
L
to 1
go
Otherwise scanti to LHS to find W
,
 Create of the the
.
3 a reversed
capy string in

store the result in the extra

main tape and one .

a
1
. Scan" to RHS to find W DRY-- > REir
.
2

3
.
More' to LHS

Move
?
to RHS
&
4 .
Read" all
If not W write? it
,

to 2
go
2
Otherwise scan to LHS
,
4 Check if the input is the WWWR .
.
etring in
language
Use to create wh before
the extra tape comparing
the second part of the string
.

F * second part

+
-
MOD
* RE'Ro rlf- Rhiz

& bur &

wer

42 MID ?

xt =
y -
-
create a
where return
reversed
copy x = W

-
compare

rymbd
each