+1 vote
148 views Ans. D

| 148 views
0
Please discuss the approach for this question
0
L(r1) = {0,01,011,0111......} , L(r2) = {1,01,001,0001......}

in question it is giving that L is any regular language obtained with quotient of L(r1) and L(r2)

then if you will consider right quotient ,

L = L(r1)/L(r2) = {$\epsilon$, 0,01,011,....} = $(01^{*})^{*}$

f you will consider left quotient ,

L = L(r1)\L(r2) = {$\epsilon$, 1,01,001,....} = $(0^{*}1)^{*}$

0
Can you explain how u did  L(r1)/L(r2) ??? Please
0
by default, Regular Language follow left/right quotient???
0
Are you asking about that regular languages are closed under left/right quotient?
0
yes, left/right or both??
0
yes regular languages are closed. but not closed under cfl.

+1 vote

L(r1) = {0,01,011,0111......} , L(r2) = {1,01,001,0001......}

in question it is giving that L is any regular language obtained with quotient of L(r1) and L(r2)

then if you will consider right quotient ,

L = L(r1)/L(r2) = {0,01,011,0111.....}/{1,01,001,0001......}

now in L only those string will come which follow below property, (from wikipedia)

means there is one x in L2 by which you can drive wx in L1.

-> Now in L2 first string is 1 and in L1 you can drive 0(from the string 01(w=0,x=1)), 01(from the string 011(w=01,x=1)),  011(from the string 0111(w=011,x=1))like wise go further...

-> Now in L2 second string is 01 and in L1 you can drive only $\epsilon$(from the string 01(w=$\epsilon$,x=01)) nothing else.

-> after second string in L2 third is 001. Now you can not drive any string from this. Same for others also

So finally L will contain strings like-

={$\epsilon$, 0,01,011,....} = $(\epsilon + 01^{*})$

-----------------------------------------------------------------------------

if you will consider left quotient ,

L = L(r1)\L(r2) = {0,01,011,0111.....}\{1,01,001,0001......}

= {$\epsilon$, 1,01,001,....} = $(\epsilon + 0^{*}1)$

now in L only those string will come which follow below property, For better understanding - https://math.stackexchange.com/questions/871662/finding-right-quotient-of-languages

by Loyal (9.2k points)
edited
+1

i think it will be $\left ( \epsilon + 01^{*} \right )$ instead of $\left ( 01^{*} \right )^{*}$