This question already has an answer here:

Is it possible, in general, to determine whether two regular expressions admit the same set of strings?

Also, in particular, is it possible to determine whether a given regular expression admits the same set of strings as the regular expression ".*" (i.e., accepts anything).


marked as duplicate by jmite, Kyle Jones, Luke Mathieson, Raphael Jun 25 '13 at 8:16

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

migrated from stackoverflow.com Jun 24 '13 at 20:58

This question came from our site for professional and enthusiast programmers.

  • 4
    $\begingroup$ It is an equivalent question to this stackoverflow.com/questions/6905043/… regular expression ~ regular automaton $\endgroup$ – Jiri Kremser Jun 24 '13 at 20:19
  • $\begingroup$ Is there a procedure for converting a regular expression to an equivalent regular automaton? Is the proof of the equivalence of regular expressions and regular automatons constructive? $\endgroup$ – Paul Reiners Jun 24 '13 at 20:30
  • $\begingroup$ ^ Yes, the procedure is given as part of Kleene's proof that regular expressions generate the same set of languages accepted by finite automata. It is constructive in that it provides an algorithm for constructing the NFA corresponding to a given DFA, yes. $\endgroup$ – Patrick87 Jun 24 '13 at 21:26

This seems like a dupe, but I couldn't find one in a cursory search. Given that, we can solve your problem as follows:

  1. Generate NFAs $N_1$ and $N_2$ for the regular expressions $r_1$ and $r_2$;
  2. Generate DFAs $M_1$ and $M_2$ for the NFAs $N_1$ and $N_2$;
  3. Generate DFAs $D_1$ and $D_2$ such that $L(D_1) = L(M_1) \setminus L(M_2)$ and $L(D_2) = L(M_2) \setminus L(M_1)$;
  4. Determine whether $L(D_1) = L(D_2) = \emptyset$; if so, $L(r_1) = L(r_2)$; else, $L(r_1) \neq L(r_2)$.

You can do (1) by using Kleene's theorem demonstrating that regular expressions and finite automata have the same expressive power. To each of the operations union, concatenation, and Kleene closure, there corresponds an automaton-based construction which accepts what the regular expression generates. By recursivly applying these constructions to subexpressions you can build an NFA to accept the language generated by a regular expression.

To do (2), you typically will want to use the powerset construction (sometimes called the subset construction). This involves constructing a DFA whose set of states equals the set of subsets of stats from the NFA. Transitions now connect subsets of states which are reachable from each other. The resulting DFA may have up to $2^{|Q|}$ states, where $Q$ is the set of states in your NFA.

To do (3), you can use the Cartesian product machine construction to generate machines with up to $|Q_1| \times |Q_2|$ states. Every state in your product machine will correspond to a pair of states from the operand machines. You must choose the set of accepting states to effect the correct operation; in our case, set difference.

To do (4), you can try all strings of length up to and including $|Q|$. If the machine accepts anything, it must accept some string of such a length. If you check all the strings and nothing's accepted, you know that $L(M) = \emptyset$. If that's the case, by construction, then the one language is a subset of the other. If both languages are subsets of each other, the languages are equal, by definition of set equality.

  • $\begingroup$ The duplicate has been found. Since your answer looks more detailed than the ones already there, maybe you want to repost? $\endgroup$ – Raphael Jun 25 '13 at 8:16
  • $\begingroup$ The question here also mentions an interesting subcase, equality test to ".*" (or $\Sigma^*$ for theoreticians). Probably for that the algorithm would be: (1) create NFA for the expression (2) create DFA (3) check whether all states are final. $\endgroup$ – Hendrik Jan Jun 25 '13 at 9:37

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