The last example suggests the conjecture that if a language L is

regular, so are L2, L3, .... We will see later that this is indeed correct.

EXERCISES

1. Which of the strings 0001, 01101, 00001101 are accepted by the dfa
in Figure 2.1?
2. Translate the graph in Figure 2.5 into δ- notation.
3. For Σ = {a, b}, construct dfa’s that accept the sets consisting of
(a) all strings of even length.
(b) all strings of length greater than 5.
(c) all strings with an even number of a’s.
(d) all strings with an even number of a’s and an odd number of b’s.
4. For Σ = {a, b}, construct dfa’s that accept the sets consisting of
(a) all strings with exactly one a.
(b) all strings with at least two a’s.
(c) all strings with no more than two a’s.
(d) all strings with at least one b and exactly two a’s.
(e) all the strings with exactly two a’s and more than three b’s.
5. Give dfa’s for the languages
(a) L = {ab4wb2 : w ∈ {a, b}*}.
(b) L = {abnam : n ≥ 3, m ≥ 2}.
(c) L = {w1abbw2 : w1 ∈ {a, b}*, w2 ∈ {a, b}*}.
(d) L = {ban : n ≥ 1, n ≠ 4}.
6. With Σ = {a, b}, give a dfa for L = {w1aw2 : |w1| ≥ 3, |w2| ≤ 4}.
7. Find dfa’s for the following languages on Σ = {a, b}.
(a) L = {w : |w| mod 3 ≠ 0}.
 (b) L = {w : |w| mod 5 = 0}.
(c) L = {w : na (w) mod 3 < 1}.
(d) L = {w : na (w) mod 3 < nb (w) mod 3}.
(e) L = {w : (na (w) − nb (w)) mod 3 = 0}.
(f) L = {w : (na (w)+ 2nb (w)) mod 3 < 1}.
(g) L = {w : |w| mod 3 = 0, |w| ≠ 5}.
8. A run in a string is a substring of length at least two, as long as
possible, and consisting entirely of the same symbol. For instance,
the string abbbaab contains a run of b’s of length three and a run of
a’s of length two. Find dfa’s for the following languages on {a, b}:
(a) L = {w : w contains no runs of length less than three}.
(b) L = {w : every run of a’s has length either two or three}.
(c) L = {w : there are at most two runs of a’s of length three}.
(d) L = {w : there are exactly two runs of a’s of length 3}.
9. Show that if we change Figure 2.6, making q3 a nonfinal state and
making q0, q1, q2 final states, the resulting dfa accepts .
10. Generalize the observation in the previous exercise. Specifically,
show that if M = (Q, Σ, δ, q0, F) and = (Q, Σ, δ, q0, Q − F) are two
dfa’s, then .
11. Consider the set of strings on {0, 1} defined by the requirements
below. For each, construct an accepting dfa.
(a) Every 00 is followed immediately by a 1. For example, the strings
101, 0010, 0010011001 are in the language, but 0001 and 00100
are not.
(b) All strings that contain the substring 000, but not 0000.
(c) The leftmost symbol differs from the rightmost one.
(d) Every substring of four symbols has, at most, two 0’s. For
example, 001110 and 011001 are in the language, but 10010 is
not because one of its substrings, 0010, contains three zeros.
(e) All strings of length five or more in which the third symbol from
 the right end is different from the leftmost symbol.
(f) All strings in which the leftmost two symbols and the rightmost
two symbols are identical.
(g) All strings of length four or greater in which the leftmost two
symbols are the same, but different from the rightmost symbol.
12. Construct a dfa that accepts strings on {0, 1} if and only if the value
of the string, interpreted as a binary representation of an integer, is
zero modulo five. For example, 0101 and 1111, representing the
integers 5 and 15, respectively, are to be accepted.
13. Show that the language L = {vwv : v, w ∈ {a, b}*, |v| = 3} is regular.
14. Show that L = {an : n ≥ 3} is regular.
15. Show that the language L = {an : n ≥ 0, n ≠ 3} is regular.
16. Show that the language L = {an : n is either a multiple of three or a
multiple of 5} is regular.
17. Show that the language L = {an : n is a multiple of three, but not a
multiple of 5} is regular.
18. Show that the set of all real numbers in C is a regular language.
19. Show that if L is regular, so is L −{λ}.
20. Show that if L is regular, so is L ∪{aa}, for all a ∈ Σ.
21. Use (2.1) and (2.2) to show that

δ* (q, wv) = δ* (δ* (q, w), v)

for all w, v ∈ Σ*.

22. Let L be the language accepted by the automaton in Figure 2.2.
Find a dfa that accepts L3.
23. Let L be the language accepted by the automaton in Figure 2.2.
Find a dfa for the language L2 − L.
24. Let L be the language in Example 2.5. Show that L* is regular.
25. Let GM be the transition graph for some dfa M. Prove the following:
(a) If L (M) is infinite, then GM must have at least one cycle for
which there is a path from the initial vertex to some vertex in the
 cycle and a path from some vertex in the cycle to some final
vertex.
(b) If L (M) is finite, then no such cycle exists.
26. Let us define an operation truncate, which removes the rightmost
symbol from any string. For example, truncate (aaaba) is aaab. The
operation can be extended to languages by

truncate (L) = {truncate (w) : w ∈ L}.

Show how, given a dfa for any regular language L, one can construct
a dfa for truncate (L). From this, prove that if L is a regular language
not containing λ, then truncate (L) is also regular.
27. While the language accepted by a given dfa is unique, there are
normally many dfa’s that accept a language. Find a dfa with exactly
six states that accepts the same language as the dfa in Figure 2.4.
28. Can you find a dfa with three states that accepts the language of the
dfa in Figure 2.4? If not, can you give convincing arguments that no
such dfa can exist?
 In the same vein, nondeterminism is an effective mechanism for
describing some complicated languages concisely. Notice that the
definition of a grammar involves a nondeterministic element. In

S → aSb|λ

we can at any point choose either the first or the second production. This
lets us specify many different strings using only two rules.
Finally, there is a technical reason for introducing nondeterminism. As
we will see, certain theoretical results are more easily established for nfa’s
than for dfa’s. Our next major result indicates that there is no essential
difference between these two types of automata. Consequently, allowing
nondeterminism often simplifies formal arguments without affecting the
generality of the conclusion.

EXERCISES

1. Construct an nfa that accepts all integer numbers in C. Explain why

your construct is an nfa.
2. Prove in detail the claim made in the previous section that if in a
transition graph there is a walk labeled w, there must be some walk
labeled w of length no more than Λ + (1 + Λ) |w|.
3. Find a dfa that accepts the language defined by the nfa in Figure 2.8.
4. Find a dfa that accepts the complement of the language defined by
the nfa in Figure 2.8.
5. In Figure 2.9, find δ* (q0, 1011) and δ* (q1, 01).
6. In Figure 2.10, find δ* (q0, a) and δ* (q1, λ).
7. For the nfa in Figure 2.9, find δ* (q0, 1010) and δ* (q1, 00).
8. Design an nfa with no more than five states for the set {ababn : n ≥ 0}
∪ {aban : n ≥ 0}.
 9. Construct an nfa with three states that accepts the language {ab,
abc}*.
10. Do you think Exercise 9 can be solved with fewer than three states?
11. (a) Find an nfa with three states that accepts the language

L = {an : n ≥ 1} ∪ {bmak : m ≥ 0, k ≥ 0}.

(b) Do you think the language in part (a) can be accepted by an nfa
with fewer than three states?
12. Find an nfa with four states for L = {an : n ≥ 0}∪{bna : n ≥ 1}.
13. Which of the strings 00, 01001, 10010, 000, 0000 are accepted by
the following nfa?

14. What is the complement of the language accepted by the nfa in

Figure 2.10?
15. Let L be the language accepted by the nfa in Figure 2.8. Find an nfa
that accepts L ∪ {a5}.
16. Find an nfa for L*, where L is the language in Exercise 15.
17. Find an nfa that accepts {a}* and is such that if in its transition graph
a single edge is removed (without any other changes), the resulting
automaton accepts {a}.
18. Can Exercise 17 be solved using a dfa? If so, give the solution; if not,
give convincing arguments for your conclusion.
19. Consider the following modification of Definition 2.6. An nfa with
multiple initial states is defined by the quintuple
 M = (Q, Σ, δ, Q0, F),

where Q0 ⊆ Q is a set of possible initial states. The language

accepted by such an automaton is defined as

L (M) = {w : δ* (q0, w) contains qf, for any q0 ∈ Q0, qf ∈ F}.

Show that for every nfa with multiple initial states there exists an nfa
with a single initial state that accepts the same language.
20. Suppose that in Exercise 19 we made the restriction Q0 ∩ F = ∅.
Would this affect the conclusion?
21. Use Definition 2.5 to show that for any nfa

for all q ∈ Q and all w, v ∈ Σ*.

22. An nfa in which (a) there are no λ-transitions, and (b) for all q ∈ Q
and all a ∈ Σ, δ (q, a), it contains at most one element, is sometimes
called an incomplete dfa. This is reasonable because the conditions
make it such that there is never any choice of moves.
For Σ = {a, b}, convert the incomplete dfa below into a standard dfa.

23. Let L be a regular language on some alphabet Σ, and let Σ1 ⊂ Σ be a

smaller alphabet. Consider L1, the subset of L whose elements are
made up only of symbols from Σ1, that is,
Show that L1 is also regular.
 we have the partially constructed automaton shown in Figure 2.15.
Since there are still some missing edges, we continue until we obtain
the complete solution in Figure 2.16.

One important conclusion we can draw from Theorem 2.2 is that

every language accepted by an nfa is regular.

EXERCISES

1. Use the construction of Theorem 2.2 to convert the nfa in Figure

2.10 to a dfa. Can you see a simpler answer more directly?
2. Convert the nfa in Exercise 13, Section 2.2, into an equivalent dfa.
3. Convert the nfa defined by

with initial state q0 and final state q2 into an equivalent dfa.

4. Convert the nfa defined by

with initial state q0 and final state q2 into an equivalent dfa.

5. Convert the nfa defined by
 with initial state q0 and final state q2 into an equivalent dfa.
6. Carefully complete the arguments in the proof of Theorem 2.2.
Show in detail that if the label of (q0, w) contains qf, then (q0,
w) also contains qf.
7. Is it true that for any nfa M = (Q, Σ, δ, q0, F), the complement of L
(M) is equal to the set {w ∈ Σ* : δ* (q0, w) ∩ F = ∅}? If so, prove it. If
not, give a counterexample.
8. Is it true that for every nfa M = (Q, Σ, δ, q0, F), the complement of L
(M) is equal to the set {w ∈ Σ* : δ* (q0, w) ∩ (Q − F) ≠ ∅}? If so,
prove it; if not, give a counterexample.
9. Prove that for every nfa with an arbitrary number of final states
there is an equivalent nfa with only one final state. Can we make a
similar claim for dfa’s?
10. Find an nfa without λ-transitions and with a single final state that
accepts the set {a}∪{bn : n ≥ 2}.
11. Let L be a regular language that does not contain λ. Show that an
nfa exists without λ-transitions and with a single final state that
accepts L.
12. Define a dfa with multiple initial states in an analogous way to the
corresponding nfa in Exercise 18, Section 2.2. Does an equivalent
dfa with a single initial state always exist?
13. Prove that all finite languages are regular.
14. Show that if L is regular, so is LR.
15. Give a simple verbal description of the language accepted by the dfa
 in Figure 2.16. Use this to find another dfa, equivalent to the given
one, but with fewer states.
16. Let L be any language. Define even (w) as the string obtained by
extracting from w the letters in even-numbered positions; that is, if

w = a1a2a3a4...,

then

even (w) = a2a4....

Corresponding to this, we can define a language

even (L) = {even (w) : w ∈ L}.

Prove that if L is regular, so is even (L).

17. From a language L, we create a new language, chopleft (L), by
removing the leftmost symbol of every string in L. Specifically,

chopleft (L) = {w : vw ∈ L, with |v| = 1}.

Show that if L is regular, then chopleft (L) is also regular.

18. From a language L, we create a new language chopright (L), by
removing the rightmost symbol of every string in L. Specifically,

chopright (L) = {w : wv ∈ L, with |v| = 1}.

Show that if L is regular, then chopright (L) is also regular.

 EXERCISES

1. Consider the dfa with initial state q0, final state q2 and

Find a minimal equivalent dfa.

2. Minimize the number of states in the dfa in Figure 2.16.
3. Find minimal dfa’s for the following languages. In each case, prove
that the result is minimal.
(a) L = {anbm : n ≥ 1, m ≥ 2}.
(b) L = {anb : n ≥ 1}∪{bna : n ≥ 1}.
(c) L = {an : n ≥ 0, n ≠ 2}.
(d) L = {an : n ≠ 3 and n ≠ 4}.
(e) L = {an : n mod 3 = 1}∪{an : n mod 5 = 1}.
4. Show that the automaton generated by procedure reduce is
deterministic.
5. Show that if L is a nonempty language such that any w in L has
length at least n, then any dfa accepting L must have at least n + 1
states.
6. Prove or disprove the following conjecture. If M = (Q, Σ, δ, q0, F) is
a minimal dfa for a regular language L, then = (Q, Σ, δ, q0, Q − F)
is a minimal dfa for .
7. Show that indistinguishability is an equivalence relation but that
distinguishability is not.
8. Show the explicit steps of the suggested proof of the first part of
Theorem 2.4, namely, that is equivalent to the original dfa.
9. Prove the following: If the states qa and qb are indistinguishable, and
if qa and qc are distinguishable, then qb and qc must be
distinguishable.
expressions. We say the two regular expressions are equivalent if they
denote the same language. One can derive a variety of rules for
simplifying regular expressions (see Exercise 20 in the following exercise
section), but since we have little need for such manipulations we will not
pursue this.

EXERCISES

1. Find all strings in L ((a + bb)∗) of length five.

2. Find all strings in L ((ab + b)∗ b (a + ab)∗) of length less than four.
3. Find an nfa that accepts the language L (aa∗ (ab + b)).
4. Find an nfa that accepts the language L (aa∗ (a + b)).
5. Does the expression ((0 + 1) (0 + 1)∗)∗ 00 (0 + 1)∗ denote the
language in Example 3.5?
6. Show that r = (1 + 01)∗ (0+1∗) also denotes the language in Example
3.6. Find two other equivalent expressions.
7. Find a regular expression for the set {anbm : n ≥ 3, m is odd}.
8. Find a regular expression for the set {anbm : (n + m) is odd}.
9. Give regular expressions for the following languages.

(a) L1 = {anbm, n ≥ 3, m ≤ 4}.

(b) L2 = {anbm : n < 4, m ≤ 4}.
(c) The complement of L1.
(d) The complement of L2.

10. What languages do the expressions (⊘∗)∗ and a ⊘ denote?

11. Give a simple verbal description of the language L ((aa)∗ b (aa)∗ + a
(aa)∗ ba (aa)∗).
12. Give a regular expression for LR, where L is the language in Exercise
 2.
13. Give a regular expression for L = {anbm : n ≥ 2, m ≥ 1, nm ≥ 3}.
14. Find a regular expression for L = {abnw : n ≥ 4, w ∈ {a, b}+}.
15. Find a regular expression for the complement of the language in
Example 3.4.
16. Find a regular expression for L = {vwv : v, w ∈ {a, b}∗, |v| = 2}.
17. Find a regular expression for L = {vwv : v, w ∈ {a, b}∗, |v| ≤ 4}.
18. Find a regular expression for

L = {w ∈ {0, 1}∗ : w has exactly one pair of consecutive zeros}.

19. Give regular expressions for the following languages on Σ = {a, b, c}:
(a) All strings containing exactly two a’s.
(b) All strings containing no more than three a’s.
(c) All strings that contain at least one occurrence of each symbol in
Σ.
20. Write regular expressions for the following languages on {0, 1}:
(a) All strings ending in 10.
(b) All strings not ending in 10.
(c) All strings containing an odd number of 0’s.
21. Find regular expressions for the following languages on {a, b}:
(a) L = {w : |w| mod 3 ≠ 0}.
(b) L = {w : na (w) mod 3 = 0}.
(c) L = {w : na (w) mod 5 > 0}.

22. Determine whether or not the following claims are true for all
regular expressions r1 and r2. The symbol ≡ stands for equivalence
of regular expressions in the sense that both expressions denote the
same language.
 (a)
(b)
(c)
(d)

23. Give a general method by which any regular expression r can be

changed into such that .
24. Prove rigorously that the expressions in Example 3.6 do indeed
denote the specified language.
25. For the case of a regular expression r that does not involve λ or ⊘,
give a set of necessary and sufficient conditions that r must satisfy if
L (r) is to be infinite.
26. Formal languages can be used to describe a variety of two-
dimensional figures. Chain-code languages are defined on the
alphabet Σ = {u, d, r, l}, where these symbols stand for unit-length
straight lines in the directions up, down, right, and left, respectively.
An example of this notation is urdl, which stands for the square with
sides of unit length. Draw pictures of the figures denoted by the
expressions (rd)∗, (urddru)∗, and (ruldr)∗.
27. In Exercise 27, what are sufficient conditions on the expression so
that the picture is a closed contour in the sense that the beginning
and ending points are the same? Are these conditions also necessary?
28. Find a regular expression that denotes all bit strings whose value,
when interpreted as a binary integer, is greater than or equal to 40.
29. Find a regular expression for all bit strings, with leading bit 1,
interpreted as a binary integer, with values not between 10 and 30.
 A challenging task in such an application is to write an efficient
program for recognizing string patterns. Searching a file for
occurrences of a given string is a very simple programming exercise,
but here the situation is more complicated. We have to deal with an
unlimited number of arbitrarily complicated patterns; furthermore,
the patterns are not fixed beforehand, but created at run time. The
pattern description is part of the input, so the recognition process
must be flexible. To solve this problem, ideas from automata theory
are often used.
If the pattern is specified by a regular expression, the pattern-
recognition program can take this description and convert it into an
equivalent nfa using the construction in Theorem 3.1. Theorem 2.2
may then be used to reduce this to a dfa. This dfa, in the form of a
transition table, is effectively the pattern-matching algorithm. All the
programmer has to do is to provide a driver that gives the general
framework for using the table. In this way we can automatically handle
a large number of patterns that are defined at run time.
The efficiency of the program must also be considered. The
construction of finite automata from regular expressions using
Theorems 2.1 and 3.1 tends to yield automata with many states. If
memory space is a problem, the state reduction method described in
Section 2.4 is helpful.

EXERCISES

1. Use the construction in Theorem 3.1 to find an nfa that accepts the
language L (a∗a + ab).
2. Use the construction in Theorem 3.1 to find an nfa that accepts the
language L ((aab)∗ab).
3. Use the construction in Theorem 3.1 to find an nfa that accepts the
language L (ab∗aa + bba∗ab).
 4. Find an nfa that accepts the complement of the language in Exercise
3.
5. Give an nfa that accepts the language L ((a + b)∗ b (a + bb)∗).
6. Find dfa’s that accept the following languages:
(a) L (aa∗ + aba∗b∗).
(b) L (ab (a + ab)∗ (a + aa)).
(c) L ((abab)∗ +(aaa∗ + b)∗).
(d) L (((aa∗)∗ b)∗).
(e) L((aa∗)∗ + abb).
7. Find dfa’s that accept the following languages:
(a) L = L (ab∗a∗) ∪ L ((ab)∗ ba).
(b) L = L (ab∗a∗) ∩ L ((b)∗ ab).

8. Find the minimal dfa that accepts L(abb)∗ ∪ L(a∗bb∗).

9. Find the minimal dfa that accepts L(a∗bb) ∪ L(ab∗ba).
10. Consider the following generalized transition graph.

(a) Find an equivalent generalized transition graph with only two

states.
(b) What is the language accepted by this graph?
11. What language is accepted by the following generalized transition
graph?
12. Find regular expressions for the languages accepted by the following
automata.
13. Rework Example 3.11, this time eliminating the state OO first.
14. Show how all the labels in Figure 3.14 were obtained.
15. Find a regular expression for the following languages on {a, b}.
(a) L = {w : na (w) and nb (w) are both odd}.
(b) L = {w : (na (w) − nb (w)) mod 3 = 2}.
(c) L = {w : (na (w) − nb (w)) mod 3 = 0}.
 (d) L = {w : 2na (w) + 3nb (w) is even}.

16. Prove that the construction suggested by Figures 3.11 and 3.12
generate equivalent generalized transition graphs.
17. Write a regular expression for the set of all C real numbers.
18. Use the construction in Theorem 3.1 to find nfa’s for L (a ⊘) and L
(⊘∗). Is the result consistent with the definition of these languages?
FIGURE 3.19

Each gives a complete and unambiguous definition of a regular language.

The connection between all these concepts is established by the four
theorems in this chapter, as shown in Figure 3.19.

EXERCISES

1. Construct a dfa that accepts the language generated by the grammar

S → abA,

A → baB,

B → aA|bb.
 2. Construct a dfa that accepts the language generated by the grammar

S → abS|A,

A → baB,

B → aA|bb.

3. Find a regular grammar that generates the language L (aa∗ (ab + a)∗).
4. Construct a left-linear grammar for the language in Exercise 1.
5. Construct right- and left-linear grammars for the language

L = {anbm : n ≥ 3, m ≥ 2}.

6. Construct a right-linear grammar for the language L ((aaab∗ab)∗).

7. Find a regular grammar that generates the language on Σ = {a, b}
consisting of all strings with no more than two a’s.
8. In Theorem 3.5, prove that .

9. Suggest a construction by which a left-linear grammar can be

obtained from an nfa directly.
10. Use the construction suggested by the above exercises to construct a
left-linear grammar for the nfa below.
11. Find a left-linear grammar for the language L ((aaab∗ba)∗).
12. Find a regular grammar for the language L = {anbm : n + m is odd}.
13. Find a regular grammar that generates the language

L = {w ∈ {a, b}∗ : na (w) + 3nb (w) is odd}.

14. Find regular grammars for the following languages on {a, b}:
(a) L = {w : na(w) is even, nb(w) ≥ 4}.
(b) L = {w : na (w) and nb (w) are both even}.)
(c) L = {w : (na (w) − nb (w)) mod 3 = 1}.
(d) L = {w : (na (w) − nb (w)) mod 3 ≠ 1}.
(e) L = {w : (na (w) − nb (w)) mod 3 ≠ 0}.
(f) L = {w : |na (w) − nb (w)| is odd}.

15. Show that for every regular language not containing λ there exists a
right-linear grammar whose productions are restricted to the forms

A → aB,

or

A → a,

where A, B ∈ V, and a ∈ T.
16. Show that any regular grammar G for which L (G) ⊘= ⊘ must have at
least one production of the form

A→x

where A ∈ V and x ∈ T ∗.
17. Find a regular grammar that generates the set of all real numbers in
C.
18. Let G1 = (V1, Σ, S1, P1) be right-linear and G2 = (V2, Σ, S2, P2) be a
left-linear grammar, and assume that V1 and V2 are disjoint.
Consider the linear grammar G = ({S} ∪ V1 ∪ V2, Σ, S, P), where S
is not in V1 ∪ V2 and P = {S → S1|S2} ∪ P1 ∪ P2. Show that L (G)
is regular.