6

Is there a way to take the interection of two NPDAs?

I can't seem to find anything that can make that happen, but it seems like the type of thing that is should be relatively trival.

Raphael
  • 72,336
  • 29
  • 179
  • 389
soandos
  • 1,143
  • 2
  • 10
  • 23
  • Possible interesting follow-up question: if we know that the intersection of the two languages is context-free, can we compute a PDA for it? – Raphael Mar 26 '14 at 08:50
  • @Raphael, while I agree that it's interesting, I have not the faintest clue of how to go about finding it. Also, I might be naive, but can't a PDA be found for all context-free languages? – soandos Mar 26 '14 at 10:06
  • 1
    Me neither! This question was not intended for you, but maybe for you to ask on the site? It's certainly true that there is a PDA for the intersection (if it's context-free) but whether you can compute if from PDA of the original languages is not clear. See for instance this answer: even if we know that the language of a given PDA is in DCFL, we can not compute a DPDA for it. – Raphael Mar 26 '14 at 10:28

2 Answers2

6

The intersection of two context-free languages can be non-context-free. The classical example is $$ \{ a^n b^n c^m : n,m \geq 0 \} \cap \{ a^m b^n c^n : n,m \geq 0 \} = \{ a^n b^n c^n : n \geq 0 \}. $$ So in general you cannot simulate the intersection of two NPDAs with an NPDA.

Yuval Filmus
  • 276,994
  • 27
  • 311
  • 503
2

Intersection of two NPDA means intersection of their languages. Of course, Context-free languages are not closed under intersection, so that you will not generally get a NPDA for the intersection. If interested in the topic, you may look at Range Concatenation Grammars, which are closed under intersection and parsable in cubic time. And, of course, they include Context-Free grammars.

babou
  • 19,445
  • 40
  • 76