Torbjorn Granlund tg at
Wed Mar 16 14:45:49 CET 2011

James Wanless <james at> writes:

  On 16 Mar 2011, at 12:17, Torbjorn Granlund wrote:
  > James Wanless <james at> writes:
  >  Code should go _something_ like this (I don't pretend this is a
  > genuine patch)
  >  File mpz/perfpow.c should replace its only function implementation,
  > mpz_perfect_power_p, thusly:
  > I don't follow you at all.  We have quite good code for detection of
  > perfect powers already.
  This returns the actual root?
  [though, like I said, maybe your version is just as fast :)]
We are not communicating very well, are we?  :-(

I thought the operation you asked for was computation of the nth root of
some (large) integer a given that it is known by the caller that the
exact result will be an integer.

Was this what you mean?

Then you quote mpz_perfect_power_p which does not seem to have much to
do with this.  We have great code for mpz_perfect_power_p (by means of a
quite new mpn_perfect_power_p).

(It is quite possible that some code from mpn_perfect_power_p can be
reused from this new function, but that's a different story.)


More information about the gmp-discuss mailing list