3
$\begingroup$

If a language like $L$ is regular, then any homomorphism of $L$ is regular too. So, if $h(L)$ is not regular, then we can conclude that $L$ is not regular.

Assume that the language $L=\{yy:y \in \{0,1\}^*\}$ is given. Can you provide a homomorphism for $L$ like $h$ that $h(L)$ is not regular?

Note : I don't want a simple homomorphism. I want a good homomorphism that $h(L)$ is obviously not regular. So there should be no need to use pumping lemma to prove that $h(L)$ is not regular. But you can use the pigeonhole principle.

$\endgroup$
7
  • $\begingroup$ Define "obvious". $L$ is well known to be not context-free, so from where I'm standing it's obvious that $L$ is not regular. $\endgroup$
    – Raphael
    Commented Apr 7, 2016 at 8:53
  • $\begingroup$ @Raphael target is to find a better way to prove that L is not regular. In the class, our teacher proved this using pumping lemma. But I thought how good it could be if we can find a homomorphism for this language that proving " $h(L)$ is not regular " is simpler than proving " $L$ is not regular " $\endgroup$ Commented Apr 7, 2016 at 8:55
  • $\begingroup$ This is probably not the best example for this, given that the proof using the Pumping lemma is as immediate as it is. How much better can it get? $\endgroup$
    – Raphael
    Commented Apr 7, 2016 at 8:59
  • $\begingroup$ @Raphael The problem of pumping lemma is that we are not sure that the string we chose always works . you chose a string. you partition the word to 3 substrings. and then you pray for this substring to be not regular. i think it's not a systematic way. $\endgroup$ Commented Apr 7, 2016 at 9:08
  • $\begingroup$ I am not sure what you're asking. I thought a homomorphism is a function. How can a homomorphism be a regular language? By $h(L)$ do you mean the homomorphic image of $L$? $\endgroup$
    – Nobody
    Commented Apr 7, 2016 at 9:49

2 Answers 2

3
$\begingroup$

As illustrated by other answer and comments there seems to be no easy way just using morphisms. Still it is a good observation that simple closure properties can be of help. In your case I would suggest intersection with regular languages as a tool: $L\cap 0^*10^*1 = \{0^n10^n1 \mid n\ge 0\}$ whch some recognize immediately as non-regular.

Alternatively one could include morphisms, inverse morphisms and intersection with regular languages (or, equivalently, finite state transductions). With these regularity preserving operations $L$ can be transformed in to $ \{a^nb^n \mid n\ge 0\}$, mother of all non-regular examples.

$\endgroup$
2
$\begingroup$

You can not prove this using homomorphisms only. There are exactly $2^2=4$ functions $\{0,1\} \rightarrow \{0,1\}$. All larger co-domains can be ignored by restricting the function to its image.

These functions are:

  1. $h_0 = \{0 \mapsto 1, 1 \mapsto 1\}$: $h(L) = \{(11)^{n}|n\geq 0 \}$ which is regular
  2. $h_1 = \{0 \mapsto 1, 1 \mapsto 0\}$: $h(L) = L$
  3. $h_2 = \{0 \mapsto 0, 1 \mapsto 1\}$: $h(L) = L$
  4. $h_3 = \{0 \mapsto 0, 1 \mapsto 0\}$: $h(L) = \{(00)^{n}|n\geq 0 \}$ which is regular

So, morphism do improve your prove in any way.

Can you provide a homomorphism for L like h that h(L) is not regular?

The answer to this question is "yes", because $L$ is not regular and $h_1$ and $h_2$ map $L$ to $L$.

$\endgroup$
3
  • 1
    $\begingroup$ A homomorphism is allowed to replace a character by a string. $\endgroup$ Commented Apr 7, 2016 at 8:46
  • $\begingroup$ The definition of homomorphism is not what you wrote sir :) a homomorphism is a function between two alphabets like $ \Sigma_1$ and $ \Sigma_2$. This function should be from $ \Sigma_1$ to $ \Sigma_2^*$. $\endgroup$ Commented Apr 7, 2016 at 8:48
  • $\begingroup$ Yes, but even morphism $\{0,1\} \rightarrow A^*$ suffer from the same problem: Cases 1 and 4 remain the regular, same for mappings to the empty word. In cases 1 and 2 you do not change the structural constraints. The general approach of using morphisms is to replace some structural constraint by some other (mostly using non-injective mappings). The intention of my answer was to clarify this problem. A morphism can not alter the structure of this specific language in a way, that could produce a language different enough to use reasoning that could not have been used on the initial one. $\endgroup$ Commented Apr 7, 2016 at 9:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.