Formal Languages and

Automata Theory

Nonregular Languages

Modified and presented by Asst. Prof. Dr.

Mohamed Kurdi
1.3 Properties of Regular
Languages
 Pumping Lemma
 Closure properties
 Decision properties
 Minimization of DFAs
Limits on the Power of FA
 Is every language describable by a
sufficiently complex regular
expression?
 If someone asks you to design an FA
for a language that seems hard, how
do you know when to give up?
 An FA is limited on finite memory, i.e.
finite set of states
Non-Regular Languages
 Is this language regular?
▪ B = {0 1 | n  0}
n n
Non-Regular Languages

B = {0 n1n | n  0}
 Non-Regular Languages
 Doesn't the argument already given prove non-
regularity because the number of 0s is unlimited?
 It does not. Just because the language appears to
require unbounded memory doesn't mean that it is
necessarily so.
 It does happen to be true for the language B, but
other languages seem to require an unlimited
number of possibilities, yet actually they are regular.
 Non-Regular Languages
 Is this language regular?
◼ L = {w | w has equal numbers of occurrences of ‘01’ and ‘10’}
◼ Thus 101 ∈ L because 101 contains a single 01 and a single 10,
but 1010 ∈ L because1010 contains two 10s and only one 01.
◼ L=1 + 0 + ε + 0(0+1)*0 + 1(0+1)*1
◼ Claim:
◼ A string w has equal occurrences of 01 and 10 is a string that starts
and ends with the same symbol.
 Non-Regular Languages
 Is this language regular?
L = {w | w has equal numbers of ‘0’ and ‘1’}
 ?
 How to prove that there is no Finite
Automaton recognizing a given language?
 Every regular language satisfies the pumping
lemma
 Pumping Lemma
 If L is a regular language, then there exists a
constant n (pumping length) such that every string
w in L with |w| >= n, can be written as w = xyz,
where:
◼ |y| > 0 (y is not ε, i.e. y can not be empty. But yi can be ε )
◼ |xy| <= n (the first two pieces occur at the start of the string)
◼ For all i >= 0, xyiz is also in L (the middle piece y can be
replicated (“pumped”) zero or more times)

 Note: both of or one of x and z may be ε

 That means each such string contains a section that
can be repeated any number of times with the
resulting string remaining in the language.
Proof of Pumping Lemma
 L is regular, then
◼ Let M be a FA that recognizes L.
◼ Set n = number of states of M.
◼ Consider wL, say w = a1a2…am (m>=n).
◼ Let q0q1…qm be the sequence of states of M when
processing W.
◼ Since there are only n different states, two of q0q1…qm
(m>=n) must be the same (pigeon hole principle); say
qi=qj where 0<=i<j<=n (among the first n+1 states).
◼ Pigeonhole principle: if p pigeons are placed into fewer
than p holes, some hole has to have more than one
pigeon in it.
Proof of Pumping Lemma
(cont’d)
◼ Let x=a1…ai; y=ai+1…aj; z =aj+1…am.
◼ Then xyz=a1…am and |y| > 0.
◼ Then by repeating the loop from qi to qi
with label ai+1…aj zero or more times, we
can show that xyiz is accepted by A.
y
q0 qm

x qi
z M
Use Pumping Lemma
 Use pumping lemma to prove that L is not
regular.
◼ Assume L is regular;
◼ Then there exists a pumping length n.
 We may not know what n is, but we can work
the rest of the “game" with n as a parameter.
◼ Select a string wL such that |w|>=n.
◼ Applying the PL, we know w can be broken into
xyz, satisfying the PL properties.
◼ We derive a contradiction by picking i>=0 such
that xyiz is not in L (whatever x, y, z are)
 Example 1.38:
 B = {0n1n | n>=0}
 Assume B is regular, let p be the pumping length
 Choose w = 0p1p because it is in B and (|w|=2p > p)
 Applying PL, w = xyz, where |y|>0, such that xyiz in B for
all i >= 0
 Three possible cases for y:
◼ y = 0k, (k>0), then xyyz = 0p+k1p, not in B
◼ y = 1k, (k>0), then xyyz = 0p1k+p, not in B
◼ y = 0k1k, (k+l>0), then xyyz = 0p1k0k1p, not in B

 Contradiction, B is not regular

 Example 1.38:
 B = {0n1n | n>=0}
 Assume B is regular, let p be the pumping length
 Choose w = 0p1p because it is in B and (|w|=2p > p)
 Applying PL, w = xyz, where |y|>0, such that xyiz in B for
all i >= 0 Recall that
1.|y| > 0
 We consider only the first case: 2.|xy| <= p
3.i >= 0, xyiz is also in L
◼ y = 0k, (k>0), then xyyz = 0p+k1p, not in B
◼ y = 1k, (k>0), then xyyz = 0p1k+p, not in B
◼ y = 0k1k, (k+l>0), then xyyz = 0p1k0k1p, not in B

 Contradiction, B is not regular

Example 1.39:
 C = {w | w has equal number of 0s and 1s}
 Assume C is regular, let p be the pumping length
 Choose w = 0p1p which is in C and (|w|=2p > p)
◼ We did not choose w = (01)p, why?

 Applying PL, w = xyz, where |y|>0, |xy|<=p, such that

xyiz in C for all i >= 0
 Since |xy|<=p, then y = 0k, (k>0) Then xyyz = 0p+k1p,
not in C
 Contradiction, C is not regular
Example 1.39:
 C = {w | w has equal number of 0s and 1s}
 Assume C is regular, let p be the pumping length
 If we had chosen w= (01)p which is also in C and (|w|=2p>p)
◼ We have run into trouble.
◼ Because we need a string that can be pumped. But that sting
actually can be pumped.

 One way to do so sets x = ε, y = 01, and z = (01) p-1. Then xyiz ∈ C

for every value of i.
 If you fail on your first attempt to find a string that cannot be
pumped, don't despair. Try another one!
Example 1.40:
 F = {ww | w{0, 1}*}
 Assume F is regular, let p be the pumping length
 Choose w = 0p10p1 in F (|w| > p)
◼ We did not choose w = 0p0p

 Applying PL, w = xyz, where |y|>0, |xy|<=p, such that

xyiz in F for all i >= 0
 Since |xy|<=p, then y = 0k, (k>0)
 Then xyyz = 0p+k10p1, not in F
 Contradiction, F is not regular
Example 1.42:
 E = {0i1j | i > j}
 Assume E is regular, let p be the pumping length
 Choose w = 0p+11p because w in E and (|w| > p)
 Applying PL, w = xyz, where |y|>0, |xy|<=p, such that
xyiz in D for all i >= 0
 Since |xy|<=p, then y=0k, (k>0)
 Then xyyz = 0p+1+k1p, in E…
 But when k=0, xy0z = xz = 0p+1-k1p, not in E
 Contradiction, E is not regular