Theory of Computation

CSE 4040
Nondeterministic Finite Automata(NFA)

21-12-2023 1
 Recall, Deterministic Finite Automata

The processing begins in𝑀’s start state.

 The automaton receives the symbols from the input string one by one from left to
right.
After reading each symbol,𝑀 moves from one state to another along the transition that
has that symbol as its label.
When it reads the last symbol, 𝑀 produces its output.
 The output is accept if 𝑀 is now in a final state.
21-12-2023
 It is reject if it is not. 2
Recall, Deterministic Finite Automata
a a a a

a a
q0 q1 a a
𝒒𝟐
q0 q1 𝒒𝟐

𝛿 𝒒𝟎 , 𝝀 = 𝒒𝟎 𝛿 𝒒𝟏 , 𝑎 = 𝒒𝟐

a a a a

a a
q0
a
q1
a
𝒒𝟐
q0 q1 𝒒𝟐

𝛿 𝒒𝟐 , 𝝀 = 𝒒𝟐

21-12-2023𝛿 𝒒𝟎 , 𝑎 = 𝒒 𝟏 3
Recall, Deterministic Finite Automata
𝛿 ∗ 𝒒𝟎 , 𝑎𝑎 = 𝛿(𝛿 ∗ 𝒒𝟎 , 𝑎 , 𝑎)
= 𝛿(𝛿 ∗ 𝒒𝟏 , 𝑎 , 𝝀)
a a
= 𝛿 𝒒𝟐 , 𝝀 q0 q1 𝒒𝟐
= 𝒒𝟐
𝑎𝑎 ∈ 𝐿(𝑀)

21-12-2023 4
 Nondeterministic Finite Accepters(NFA)
Σ = {𝑎}
there are two transitions labelled a out of 𝑞0
and 𝑞3 is lacking any transition for input.
No Transition

𝛿 𝑞0 , 𝑎 = 𝑞1 , 𝛿 𝑞0 , 𝑎 = 𝑞4 , 𝛿 𝑞3 , 𝑎 = nowhere (as transition to the empty set)

𝛿 𝑞3 , 𝑎 =𝜙 (transition undefined)

21-12-2023 dead configuration 5

2.2 Nondeterministic Finite Accepters (𝜆 Transition)

a a a a

q0 a q1  q2 a q3 q0 a q1  q2 a q3

a a
a a

q0 a q1  q2 a q3 q0 a q1  q2 a q3

21-12-2023
The 𝜆 symbol is never seen on the input tape. 6
 2.2 Nondeterministic Finite Accepters (𝜆 − NFA)
Example Σ = {0,1}

 Accepting Strings 𝜆,10,1010,101010

 Reject : 0,1,11

 for 10 there are two alternative walks , one leading to

𝛿 𝑞0 , 𝜆 = 𝑞0 , 𝛿 𝑞0 , 𝜆 = 𝑞2 , 𝛿 𝑞0 , 1 = 𝑞1 , 𝑞0 , the other to 𝑞2 .
 Even though 𝑞2 is not a final state, the string is accepted
𝛿 𝑞1 , 0 = 𝑞0 , 𝛿 𝑞1 , 0 = 𝑞2 , 𝛿 𝑞1 , 1 = 𝑞2 because one walk leads to a final state.

𝛿 𝑞2 , 0 = 𝜙 , 𝛿 𝑞2 , 1 = 𝜙

Myanmar Institute of Information Technology 7

2.2 Nondeterministic Finite Accepters

multiple edges leaving one vertex labelled with the same symbol

no edge leaving vertex labelled with symbol

edges leaving vertex labelled with 𝜆

 can take 𝜆 − edge without reading any input symbol

Myanmar Institute of Information Technology 8

 2.2 Nondeterministic Finite Accepters
The three difference between dfas and nfas are
1. dfa : a single state of Q can be reached by the transition
nfa : the set of all possible states of Q can be reached by the transition
2. dfa : consumes input with each movement
nfa : can move without consuming an input symbol(𝜆)
3. dfa : moves for every input and state
nfa : there is no transition defined for some situation.

An nfa accepts a string if some possible sequence of moves ends in a final state.
An nfa rejects a string if no possible sequence of moves ends a final state.

Myanmar Institute of Information Technology 9

 Nondeterministic Finite Accepters(NFA)
Definition of a Nondeterministic Accepters
A nondeterministic finite acceptor or nfa is defined by the quintuple
M=(Q,Σ, 𝛿, 𝑞0 , 𝐹) , where Q, Σ, 𝑞0 , F are defined as for deterministic finite
accepters ,but
𝛿: 𝑄 × Σ ∪ 𝜆 ⟶ 2𝑄 .

21-12-2023 10
 ∗
Extended transition function 𝜹

If 𝛿 ∗ 𝑞𝑖 , 𝑤 = 𝑄𝑗 then 𝑄𝑗 is the set of all possible states the

automaton may be in, having started in state 𝑞𝑖 and having read w.

21-12-2023 11
Extended transition function 𝜹 ∗
𝛿 ∗ 𝑞0 , 𝑎 = {𝑞1 }
𝛿 ∗ 𝑞0 , 𝑎𝑏 = {𝑞2 , 𝑞3 , 𝑞0 }

q4 q5
a a q4 q5
q0 a q1 b q2  q3 a a
q0 a q1 b q2  q3

𝛿 ∗ 𝑞0 , 𝑎𝑎 = {𝑞4 , 𝑞5 }

q4 q5
a a
q0 a q1 b q2  q3

21-12-2023  12
 Extended transition function 𝜹∗
 Use 𝛿 ∗ to describe how the string 01010 is processed by the NFA.
𝛿 ∗ 𝑞0 , 𝟎1010 = δ 𝛿 ∗ 𝑞0 , 0101 , 0

1 0,1 = δ 𝛿 ∗ 𝑞0 , 0101 , 0
q0 q1 𝒒𝟐
= δ 𝛿 ∗ 𝑞0 , 101 , 0

= δ 𝛿 ∗ {𝑞0 , 𝑞1 }, 01 , 0
M
= δ 𝛿 ∗ ( 𝑞0 , 0) ∪ (𝑞1 , 0), 1 , 0

M=({𝑞0 , 𝑞1 , 𝑞2 },{0,1}, 𝑞0 , δ, {𝑞2 }) = δ(𝛿 ∗ ({𝑞0 } ∪ {𝑞2 }, 1), 0)

where the transition function δ is
= δ(𝛿 ∗ ({𝑞0 , 𝑞2 }, 1), 0)
δ 0 1
= δ 𝛿 ∗ (𝑞0 , 1) ∪ (𝑞2 , 1), 0
𝑞0 {𝑞0 } {𝑞0 , 𝑞1 }
𝑞1 {𝑞2 } {𝑞2 } = δ ({𝑞0 , 𝑞1 } ∪ ∅), 0) = δ( 𝑞0 , 𝑞1 , 0)

𝑞2 ∅ ∅ = δ( 𝑞0 , 0 ∪ 𝑞1 , 0 , 𝜆) = δ({𝑞0 } ∪ {𝑞2 }, 𝜆)
= δ( 𝑞0 , 𝜆 ∪ 𝑞2 , 𝜆 )
L(M)= all the strings ending in 10 (or) 11. = {𝑞0 } ∪ {𝑞2 }
= {𝑞0 , 𝑞2 }
21-12-2023 13
Nondeterministic Automata NFA
Definition 2.6
The language L accepted by nfa 𝑀 = (𝑄, Σ, 𝛿, 𝑞0 , 𝐹) is defined as the set of all
strings accepted in the above sense. Formally,
𝐿 𝑀 = {𝑤 ∈ Σ ∗ : 𝛿 ∗ (𝑞0 , 𝑤) ∩ 𝐹 ≠ 𝜙}.
In words, the language consists of all strings w for which there is a walk labelled
w from the initial vertex of the transition graph to some final vertex.

NFAs accept the Regular Languages.

21-12-2023 14
 Example
Σ = {𝑎, 𝑏}
(1) 𝐿1 = {𝑠𝑡𝑎𝑟𝑡𝑠 𝑤𝑖𝑡ℎ ′𝑎′}

(2) 𝐿2 = {𝑐𝑜𝑛𝑡𝑎𝑖𝑛 ′𝑎′}

(3) 𝐿3 = {𝑒𝑛𝑑𝑠 𝑤𝑖𝑡ℎ ′𝑎′}

(4) 𝐿4 = {𝑠𝑡𝑎𝑟𝑡𝑠 𝑤𝑖𝑡ℎ ′𝑎𝑏′}

(4) 𝐿5 = {𝑐𝑜𝑛𝑡𝑎𝑖𝑛 ′𝑎𝑏′}

(5) 𝐿6 = {𝑒𝑛𝑑𝑠 𝑤𝑖𝑡ℎ ′𝑎𝑏′}

21-12-2023 15
Example (1): What is the language accepted by the automaton in fig?

L={(10)𝑛 : 𝑛 ≥ 0}

what happens w=110 ?

𝛿 ∗ 𝑞0 , 110 = ∅ 𝛿 𝑞2 , 0 = 𝜙 (transition undefined)

21-12-2023 16
2)Design an nfa with no more than five states for the set 𝑎𝑏𝑎𝑏 𝑛 : 𝑛 ≥ 0 ∪
{𝑎𝑏𝑎𝑛 : 𝑛 ≥ 0}.

21-12-2023 17
Equivalence of Deterministic and Nondeterministic Finite Automata

21-12-2023 18
Equivalence of DFA’s and NFA’s
 Every DFA is also an NFA by definition.

 There is no simply non-determinism.

 Every NFA can be converted into an equivalent DFA.

 Two equivalent machines recognize the same language.

 Non-intuitive, as we’d expect NFA’s to be more powerful.

 Useful, as describing an NFA is much simpler.

Proof is the subset construction.

The number of states of the DFA can be exponential in the number of states of the NFA.

21-12-2023 19
 Equivalence of Machines
• Definition 2.7
Two finite accepters 𝑀1 and 𝑀1 are said to be equivalent if
L(𝑀1 ) = L(𝑀2 ).
That is, if they both accept the same language.

21-12-2023 20
Equivalence of DFAs and NFAs

L(𝑀1 )={(10)𝑛 , 𝑛 ≥ 0}

𝑀1 (NFA)

L(𝑀2 )={(10)𝑛 , 𝑛 ≥ 0}

𝑀2 (DFA)
21-12-2023
Thus, 𝑀1 is equivalent to 𝑀2 . 21
Subset Construction

Given an NFA(𝑄, Σ, 𝛿𝑁 , 𝑞0 , 𝐹) , construct equivalent DFA with:

 States P(Q) (set of subsets of Q).
 Inputs Σ.
 Start state {𝑞0 } .
 Final states = all those with a member of F.

Note:

 The DFA states have names that are sets of NFA states.

 But as a DFA state, an expression like {𝑝, 𝑞} must read as a single symbol.

21-12-2023 22
Subset construction
• Example: We ‘ll construct the DFA for the following NFA which recognizes strings in {0,1}∗
containing a 1 in the second position from the end.

Solution:
Construction DFA 𝑀1 as 𝑀1 = (𝑄1 , 0,1 , 𝜹𝑫 , 𝑨 , 𝑭)
where 𝑄1 = 𝑃 𝑄 = 2𝑄 =
{∅, 𝐴 , 𝐵 , 𝐶 , 𝐴, 𝐵 , 𝐴, 𝐶 , 𝐵, 𝐶 , {𝐴, 𝐵, 𝐶}}

M=({A,B,C},{0,1}, 𝜹𝑵 ,A,F) 𝛿𝐷 𝐴 , 0 = {𝐴}

𝛿𝐷 𝐴 , 1 = {𝐴, 𝐵} (New single state)

𝜹𝑵 0 1 𝛿𝐷 𝐴, 𝐵 , 0 = 𝛿𝑁 𝐴 , 0 ∪ 𝛿𝑁 𝐵, 0 = 𝐴 ∪ 𝐶 = {𝐴, 𝐶} (New single state)

𝛿𝐷 𝐴, 𝐵 , 1 = 𝛿𝑁 𝐴 , 1 ∪ 𝛿𝑁 𝐵, 1 = 𝐴, 𝐵 ∪ 𝐶 = {𝐴, 𝐵, 𝐶} (New single state)
A {A} {A,B} 𝛿𝐷 𝐴, 𝐶 , 0 = 𝛿𝑁 𝐴 , 0 ∪ 𝛿𝑁 𝐶, 0 = 𝐴 ∪ Φ = {𝐴}
𝛿𝐷 𝐴, 𝐶 , 1 = 𝛿𝑁 𝐴 , 1 ∪ 𝛿𝑁 𝐶, 1 = 𝐴, 𝐵 ∪ Φ = {𝐴, 𝐵}
B {C} {C}
𝛿𝐷 𝐴, 𝐵, 𝐶 , 0 = 𝛿𝑁 𝐴 , 0 ∪ 𝛿𝑁 𝐵 , 0 ∪ 𝛿𝑁 𝐶 , 0 = 𝐴 ∪ {𝐶} ∪ Φ = {𝐴, 𝐶}
*C 𝜙 𝜙 𝛿𝐷 𝐴, 𝐵, 𝐶 , 1 = 𝛿𝑁 𝐴 , 1 ∪ 𝛿𝑁 𝐵 , 1 ∪ 𝛿𝑁 𝐶 , 1 = 𝐴, 𝐵 ∪ {𝐶} ∪ Φ = {𝐴, 𝐵, 𝐶}

21-12-2023 𝛿𝑁∗ 𝐴, 10011 = 𝐶 ∈ 𝐹 23

Subset construction 𝜹𝑫 0 1
𝛿𝐷 𝐴 , 0 = {𝐴}
𝛿𝐷 𝐴 , 1 = {𝐴, 𝐵} {A} {A} {A,B}

𝛿𝐷 𝐴, 𝐵 , 0 = 𝛿𝑁 𝐴 , 0 ∪ 𝛿𝑁 𝐵, 0 = 𝐴 ∪ 𝐶 = {𝐴, 𝐶} {A,B} {A,C} {A,B,C}

𝛿𝐷 𝐴, 𝐵 , 1 = 𝛿𝑁 𝐴 , 1 ∪ 𝛿𝑁 𝐵, 1 = 𝐴, 𝐵 ∪ 𝐶 = {𝐴, 𝐵, 𝐶}
*{A,C} {𝐴} {𝐴, 𝐵}
𝛿𝐷 𝐴, 𝐶 , 0 = 𝛿𝑁 𝐴 , 0 ∪ 𝛿𝑁 𝐶, 0 = 𝐴 ∪ Φ = {𝐴} *{A,B,C} {A,C} {A,B,C}
𝛿𝐷 𝐴, 𝐶 , 1 = 𝛿𝑁 𝐴 , 1 ∪ 𝛿𝑁 𝐶, 1 = 𝐴, 𝐵 ∪ Φ = {𝐴, 𝐵}

𝛿𝐷 𝐴, 𝐵, 𝐶 , 0 = 𝛿𝑁 𝐴 , 0 ∪ 𝛿𝑁 𝐵 , 0 ∪ 𝛿𝑁 𝐶 , 0 = 𝐴 ∪ {𝐶} ∪ Φ = {𝐴, 𝐶}
𝛿𝐷 𝐴, 𝐵, 𝐶 , 1 = 𝛿𝑁 𝐴 , 1 ∪ 𝛿𝑁 𝐵 , 1 ∪ 𝛿𝑁 𝐶 , 1 = 𝐴, 𝐵 ∪ {𝐶} ∪ Φ = {𝐴, 𝐵, 𝐶}

𝛿𝐷∗ {𝐴}, 10011 = {𝐴, 𝐵, 𝐶} ∈ 𝐹

NFA
21-12-2023 24
DFA
 Equivalence of DFAs and NFAs
Every DFA is obviously an NFA.
 A language accepted by a DFA is also accepted by an NFA.

Any NFA can be converted to an equivalent DFA.

 A language accepted by a NFA is also accepted by an DFA.

Thus, DFAs and NFAs are equally powerful.

21-12-2023 25
Epsilon-Closures
 It is a set of reachable states from a given state with 𝜀 − 𝑡𝑟𝑎𝑛𝑠𝑖𝑡𝑖𝑜𝑛.
0 1 0 1

q0 𝜀 𝒒𝟏 𝜀 𝒒𝟐 q0 𝜀 𝒒𝟏 𝜀 𝑞2

0
1
 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞0 = {𝑞0 , 𝑞1 , 𝑞2 }
 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞1 = {𝑞1 , 𝑞2 } 𝒒𝟑
 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞2 = {𝑞2 }

 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞0 = {𝑞0 , 𝑞1 , 𝑞2 }
 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞1 = {𝑞1 , 𝑞2 }
 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞2 = {𝑞2 }
 𝜀 − 𝑐𝑙𝑜𝑠𝑢𝑟𝑒 𝑞3 = {𝑞3 }
21-12-2023 26
 Summary
DFA’s , NFA’s and 𝜆 − 𝑁𝐹𝐴′ 𝑠 all accept exactly the same set of languages:
the regular languages.

NFA types are easier to design and may have exponentially fewer states than a
DFA.

21-12-2023 27