14 REGULAR LANGUAGES
= {w 1 (3u)[uw E Ll]} u {w 1 (3u)[uw E L21) = L1’ u L2’.
(b) (L1L2)’ = {w 1 (~U)[UW E ~51~521)
= {w ] either w E L2’ or w = zy for some z E Ll’, y E L2)
= L2’ u LI’L:!.
(c) (Ll*)’ = ( E Ll”)’ = E(Ll”)’
i=o i=o
bY Part (4
= {E} u (G(L1’ u LI’LI u * l l u L1’Lf’))
i=l
bY Part (b)
ccl
II Ll’ l
(u ({E} u L1 u ’ l l u Lli-l ,>
= L&*.
i=l
Therefore, (L1 U Lz)‘, (L1 Lz)‘, and (Ll*)’ are regular sets. This completes
the induction step. cl
Remark. What is wrong with the following proof for part (c) of Example 1.21?
Proof for part (c). From part (a), we know that
(Ll*)’ = ( 6 Ll”)’ = E(L1”)‘.
i=o i=o
From part (b), we know that each (Li)’ is regular for i > 0. There-
fore, (L;)’ is the union of regular languages (Li)‘, for z7= 0, 1,... ,
and so it is regular.
In the last step of the above “proof,” the rule of
[A, B are regular 3 A U B is regular]
has been illegally extended to
Ao, A1, A2, l. l are regular 3 6 A; ’ is regular.
i=o^1