Lecture 5: NFA Equivalence

CSC 320: Foundations of Computer Science

Quinton Yong
quintonyong@uvic.ca
NFAs and DFAs
• A DFA can be considered a special case of an NFA
• We just have to change the transition function definition slightly

1 𝑞2 0, 1 1 𝑞2 0, 1

𝑞1 𝑞3 0, 1 𝑞1 𝑞3 0, 1

• Identical state
0 0 diagram
DFA 𝑴 NFA 𝑴′
𝑞1 , 𝑞2 , 𝑞3 , 0, 1 , 𝛿, 𝑞1 , 𝑞3 with 𝑞1 , 𝑞2 , 𝑞3 , 0, 1 , 𝛿, 𝑞1 , 𝑞3 with
𝛿: 𝑄 × Σ → 𝑄 defined by 𝛿: 𝑄 × Σ ∪ 𝜀 → 𝒫(𝑄) defined by
𝜹 𝟎 𝟏 𝜹 𝟎 𝟏 𝜺
𝑞1 𝑞3 𝑞2 𝑞1 {𝑞3 } {𝑞2 } ∅ • Next states are
𝑞2 𝑞3 𝑞3 𝑞2 {𝑞3 } {𝑞3 } ∅ singleton sets
𝑞3 𝑞3 𝑞3 𝑞3 {𝑞3 } {𝑞3 } ∅ • No 𝜀 transitions
NFAs and DFAs

• Recall that this state diagram was not a valid DFA

• 𝒒𝟑 does not have an outgoing transition for symbol 2

𝑞4 0, 1, 2

𝑞3 0, 1

• However, this is a valid state diagram for an NFA

• In NFAs, we do not require outgoing transitions for each alphabet symbol from
every state
NFAs and DFAs

• Definition: Let 𝑴𝟏 and 𝑴𝟐 each be a DFA or an NFA. We call 𝑴𝟏 and 𝑴𝟐

equivalent if 𝑳 𝑴𝟏 = 𝑳(𝑴𝟐 ).

• Clearly, for every DFA there exists an equivalent NFA

• Recall, we claimed that the computational power of DFAs and NFAs are equal

• So, we will also prove that for every NFA there also exists an equivalent DFA
Equivalence of NFAs and DFAs
First: We show that for every DFA, there exists an equivalent NFA

Proof: (easy)
Let 𝑫 = (𝑸𝑫, 𝚺, 𝜹𝑫, 𝒒𝑫, 𝑭𝑫) be a DFA.
We can build NFA 𝑵 = 𝑸𝑵 , 𝚺, 𝜹𝑵 , 𝒒𝑵 , 𝑭𝑵 with 𝑳 𝑫 = 𝑳 𝑵 as follows:
• 𝑸𝑵 ≔ 𝑸𝑫

• 𝒒𝑵 ≔ 𝒒𝑫 States, start state, and accept states are the same

• 𝑭𝑵 ≔ 𝑭𝑫

• 𝜹𝑵 : 𝑸𝑵 × 𝚺 ∪ 𝜺 → 𝓟 𝑸𝑵 with • Exact same transitions, but written as set

• 𝜹𝑵 𝒒, 𝒂 ≔ 𝜹𝑫 𝒒, 𝒂 for all 𝒂 ∈ 𝚺 containing the outgoing state in DFA
• 𝜹𝑵 𝒒, 𝜺 ≔ ∅ • No 𝜺-transitions
Equivalence of NFAs and DFAs
Next: We show that for every NFA, there exists an equivalent DFA

Proof: Let 𝑵 = 𝑸𝑵 , 𝚺, 𝜹𝑵 , 𝒒𝑵 , 𝑭𝑵 be an NFA. Construct DFA 𝑫 = (𝑸𝑫 , 𝚺, 𝜹𝑫 , 𝒒𝑫 , 𝑭𝑫)

such that 𝑳 𝑫 = 𝑳 𝑵 .
0, 1 1
𝑵
0 𝜀
𝑞0 𝑞1 𝑞2

Idea: Build 𝑫 such that is simulates the computation of 𝑵.

• Since states can go to a set of states in NFAs, create a state in 𝑫 for every
possible subset of states of 𝑸𝑵
𝑫 0 …
{𝑞0 } {𝑞0, 𝑞1, 𝑞2}
 NFA to DFA Construction
Proof continued: Given NFA 𝑵 = 𝑸𝑵 , 𝚺, 𝜹𝑵 , 𝒒𝑵 , 𝑭𝑵 , build DFA 𝑫 = (𝑸𝑫 , 𝚺, 𝜹𝑫 , 𝒒𝑫, 𝑭𝑫):
For now, let’s ignore the 𝜺-transitions of 𝑵.

{𝑞0, 𝑞1, 𝑞2} • 𝑸𝑫 ≔ 𝓟 𝑸𝑵 New states are all possible subsets of states in NFA

• 𝒒𝑫 ≔ {𝒒𝑵} Start state is the singleton set containing NFA start state

Next state is the state corresponding to set of all reachable

• 𝜹𝑫 is defined as follows: states in NFA from the states in current DFA state
• Let 𝑺 ∈ 𝑸𝑫 be a state of the DFA 𝑫 and let 𝒂 ∈ 𝚺. Recall, 𝑺 ⊆ 𝑸𝑵 .

• 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}

• 𝑭𝑫 ≔ {𝑺 ∈ 𝑸𝑫 ∣ there exists a 𝒒 ∈ 𝑺 with 𝒒 ∈ 𝑭𝑵 } Accept states are the states which

contain an accept state in NFA
NFA to DFA Construction (𝜺-transitions)
Proof continued: Next, modify the 𝑫 to simulate the 𝜺-transitions in NFA
All states in 𝑺 plus states reachable by 𝜺-transitions (0 or more) from states in 𝑺
• For any state 𝑺 ∈ 𝑸𝑫 , let 𝑬 𝑺 = {𝒒 ∣ 𝒒 can be reached from some state in 𝑺 by
traveling 𝟎 or more 𝜺-transitions}
𝑵 0, 1 1 State 𝑺 in 𝑫
0 𝜀 {𝑞0 , 𝑞1 } 𝑬 𝑺 = {𝑞0 , 𝑞1 , 𝒒𝟐 }
𝑞0 𝑞1 𝑞2

• Modify 𝑫 as follows:
• 𝒒𝑫 ≔ 𝑬 𝒒𝑵 Start state is 𝐸() of start state in NFA
• 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}

{𝑞0 , 𝑞1 }
1 {𝑞0, 𝑞1, 𝑞2}
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏
∅
𝑞1
𝑞2
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3
𝑞2 , 𝑞3 𝑸𝑫 ≔ 𝓟 𝑸𝑵
𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏
∅
𝑞1
𝑞2
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3
𝑞2 , 𝑞3 𝒒𝑫 ≔ 𝑬 𝒒𝑵
𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏
∅ ∅ ∅ {𝑞1 , 𝑞3 } 0,1
𝑞1
𝑞2
∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 DFA state ∅ represents when an NFA state has
𝑞2 , 𝑞3 no outgoing transitions for a symbol
𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏
∅ ∅ ∅ {𝑞1 , 𝑞3 } 0,1
𝑞1
𝑞2
∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏
∅ ∅ ∅ {𝑞1 , 𝑞3 } 0,1
𝑞1
𝑞2 0 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏 1
∅ ∅ ∅ {𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1
𝑞1
𝑞2 0 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏 1
∅ ∅ ∅ {𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1
0
𝑞1
𝑞2 0 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1

Transition table for DFA 𝑫

𝜹𝑫 𝟎 𝟏 1
∅ ∅ ∅ {𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1
0
𝑞1
𝑞2 0 1 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 }

Transition table for DFA 𝑫 0

𝜹𝑫 𝟎 𝟏 1
∅ ∅ ∅ {𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1
0
𝑞1 𝑞1 , 𝑞3
𝑞2 0 1 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 }
1
Transition table for DFA 𝑫 0
𝜹𝑫 𝟎 𝟏 1
∅ ∅ ∅ {𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1
0
𝑞1 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 0 1 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 }
1
Transition table for DFA 𝑫 0
𝜹𝑫 𝟎 𝟏 1
{𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1 {𝑞2 }
0 {𝑞3 }
∅ ∅ ∅
0
𝑞1 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 𝑞3 0 1 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 }
1
Transition table for DFA 𝑫 0
𝜹𝑫 𝟎 𝟏 1
{𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1 {𝑞2 }
0 {𝑞3 }
∅ ∅ ∅
0
𝑞1 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
1
𝑞2 𝑞3 ∅ 0 1 ∅
𝑞3
𝑞1 , 𝑞2
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝜹𝑫 𝑺, 𝒂 ≔ {𝒒 ∈ 𝑸𝑵 ∣ 𝒒 ∈ 𝑬 𝜹𝑵 𝒔, 𝒂 for some 𝒔 ∈ 𝑺}
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 } {𝑞2 , 𝑞3 }
1
Transition table for DFA 𝑫 0 0,1
𝜹𝑫 𝟎 𝟏 1
{𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1 {𝑞2 }
0 {𝑞3 }
∅ ∅ ∅
0
𝑞1 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
0 1 1
𝑞2 𝑞3 ∅ 0 {𝑞1 , 𝑞2 } 1 ∅ 0,1
𝑞3 𝑞3 𝑞3
𝑞1 , 𝑞2 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞1 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝑞2 , 𝑞3 𝑞3 𝑞3
𝑞1 , 𝑞2 , 𝑞3 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 } {𝑞2 , 𝑞3 }
1
Transition table for DFA 𝑫 0 0,1
𝜹𝑫 𝟎 𝟏 1
{𝑞1 , 𝑞3 } {𝑞1, 𝑞2, 𝑞3} 0,1 {𝑞2 }
0 {𝑞3 }
∅ ∅ ∅
0
𝒒𝟏 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
0 1 1
𝑞2 𝑞3 ∅ 0 {𝑞1 , 𝑞2 } 1 ∅ 0,1
𝒒𝟑 𝑞3 𝑞3
𝒒𝟏 , 𝒒𝟐 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝒒𝟏 , 𝒒𝟑 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝒒𝟐 , 𝒒𝟑 𝑞3 𝑞3 𝑭𝑫 ≔ {𝑺 ∈ 𝑸𝑫 ∣ there exists a 𝒒 ∈ 𝑺 with 𝒒 ∈ 𝑭𝑵 }
𝒒𝟏 , 𝒒𝟐 , 𝒒𝟑 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
NFA to DFA Example
𝑵 1 𝑞2 0

𝑞1 𝑞3
𝑫
𝜀
0,1 0,1
{𝑞1 } {𝑞2 , 𝑞3 }
1
Transition table for DFA 𝑫 0 0,1
𝜹𝑫 𝟎 𝟏 1
{𝑞1 , 𝑞3 } {𝑞1 , 𝑞2 , 𝑞3 } 0,1 {𝑞2 }
0 {𝑞3 }
∅ ∅ ∅
0
𝒒𝟏 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
0 1 1
𝑞2 𝑞3 ∅ 0 {𝑞1 , 𝑞2 } 1 ∅ 0,1
𝒒𝟑 𝑞3 𝑞3
𝒒𝟏 , 𝒒𝟐 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝒒𝟏 , 𝒒𝟑 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
𝒒𝟐 , 𝒒𝟑 𝑞3 𝑞3
𝒒𝟏 , 𝒒𝟐 , 𝒒𝟑 𝑞1 , 𝑞3 𝑞1 , 𝑞2 , 𝑞3
Equivalence of NFAs and DFAs

• We have shown that:

• For every DFA there exists an equivalent NFA
• For every NFA there exists an equivalent DFA

• So, we know that DFAs and NFAs produce the same set of languages

• Therefore, the languages recognized by NFAs is exactly the set of

regular languages
Regular Languages are Closed under Concatenation

Recall: Given languages 𝑳𝟏 and 𝑳𝟐 over alphabet 𝚺, their concatenation denoted

𝑳𝟏 𝑳𝟐 is defined as 𝑳𝟏 𝑳𝟐 = 𝒘 ∈ 𝚺 ∗ | 𝒘 = 𝒙𝒚 for some 𝒙 ∈ 𝑳𝟏 and 𝒚 ∈ 𝑳𝟐
• E.g. Let 𝑳𝟏 = {0, 10} and 𝑳𝟐 = 0, 11 . Then 𝑳𝟏 𝑳𝟐 = 00, 011, 100, 1011

Theorem: If 𝑳𝟏 and 𝑳𝟐 are regular languages over alphabet Σ, then the language
𝑳𝟏 𝑳𝟐 is a regular language

Proof: Since 𝑳𝟏 and 𝑳𝟐 are regular languages, then there exist DFAs 𝑴𝟏 and 𝑴𝟐
where 𝑳𝟏 = 𝑳(𝑴𝟏 ) and 𝑳𝟐 = 𝑳(𝑴𝟐 )

We can create an NFA 𝑴 that accepts the strings where the first part is accepted
by 𝑴𝟏 and the second part is accepted by 𝑴𝟐 .
Regular Languages are Closed under Concatenation
𝑴𝟏 = 𝑸𝟏 , 𝚺, 𝜹𝟏 , 𝒒𝟏 , 𝑭𝟏 𝑴𝟐 = 𝑸𝟐 , 𝚺, 𝜹𝟐 , 𝒒𝟐 , 𝑭𝟐

𝑓1 𝑓4
𝑴 𝑞1 𝑓2 𝑞2 𝑓5
𝑓3 𝑓6

Create NFA 𝑴 as follows:

• 𝑴 inherits all states from DFAs 𝑴𝟏 and 𝑴𝟐
Regular Languages are Closed under Concatenation
𝑴𝟏 = 𝑸𝟏 , 𝚺, 𝜹𝟏 , 𝒒𝟏 , 𝑭𝟏 𝑴𝟐 = 𝑸𝟐 , 𝚺, 𝜹𝟐 , 𝒒𝟐 , 𝑭𝟐

𝑓1 𝑓4
𝑴 𝑞1 𝑓2 𝑞2 𝑓5
𝑓3 𝑓6

Create NFA 𝑴 as follows:

• 𝑴 inherits all states from DFAs 𝑴𝟏 and 𝑴𝟐
• The start state of 𝑴 is the start state of 𝑴𝟏
Regular Languages are Closed under Concatenation
𝑴𝟏 = 𝑸𝟏 , 𝚺, 𝜹𝟏 , 𝒒𝟏 , 𝑭𝟏 𝑴𝟐 = 𝑸𝟐 , 𝚺, 𝜹𝟐 , 𝒒𝟐 , 𝑭𝟐

𝑓1 𝑓4
𝑴 𝑞1 𝑓2 𝑞2 𝑓5
𝑓3 𝑓6

Create NFA 𝑴 as follows:

• 𝑴 inherits all states from DFAs 𝑴𝟏 and 𝑴𝟐
• The start state of 𝑴 is the start state of 𝑴𝟏
• The accept states of 𝑴 are the accept states of 𝑴𝟐 (accept states of 𝑴𝟏 are
no longer accept states)
Regular Languages are Closed under Concatenation
𝑴𝟏 = 𝑸𝟏 , 𝚺, 𝜹𝟏 , 𝒒𝟏 , 𝑭𝟏 𝑴𝟐 = 𝑸𝟐 , 𝚺, 𝜹𝟐 , 𝒒𝟐 , 𝑭𝟐

𝑓1 𝜺 𝑓4
𝑓2 𝜺 𝑓5
𝑴 𝑞1
𝜺
𝑞2
𝑓3 𝑓6

Create NFA 𝑴 as follows: 𝑴 = 𝑸𝟏 ∪ 𝑸𝟐 , 𝚺, 𝜹, 𝒒𝟏 , 𝑭𝟐

• 𝑴 inherits all states from DFAs 𝑴𝟏 and 𝑴𝟐
• The start state of 𝑴 is the start state of 𝑴𝟏
• The accept states of 𝑴 are the accept states of 𝑴𝟐 (accept states of 𝑴𝟏 are
no longer accept states)
• The transitions of 𝑴 consist of:
• All transitions of 𝑴𝟏 and all transitions of 𝑴𝟐 (remain the same)
• Add 𝜺-transitions between the previous accept states in 𝑴𝟏 to the start
state of 𝑴𝟐
Regular Language Closure

We have now shown that given regular languages 𝑳𝟏 and 𝑳𝟐 :

• 𝑳𝟏 ∪ 𝑳𝟐 is a regular language

• 𝑳𝟏 ∩ 𝑳𝟐 is a regular language
• 𝑳𝟏 𝑳𝟐 is a regular language

Regular languages are also closed under complement and Kleene star:

• 𝑳𝟏 is a regular language

• 𝑳𝟏∗ is a regular language