tsvector minimum match using index

Would it be possible without writing a very long tsquery to exploit
the index to retrieve the tsvectors that contain at least N lexemes?

If not exploiting the index... any suggestion to improve performance
of such a query?

computing rank still requires retrieving a lot of tsvectors and
compute the rank for each of them.

--
Ivan Sergio Borgonovo
http://www.webthatworks.it