Roger Hui wrote:
> How about solving the problem for large integers?
> If the argument is less than 2^31 a table-driven
> approach should be most efficient (but much less
> interesting).

I'm not sure how I would tackle this problem for
a domain which is not a part of the domain of
p:inv

I could probably use 1&p: to probe regions around
the number in question -- I can imagine this working
for many values up to around 2^62x 

That said, my npp does not work for numbers 
within the domain of p:inv, and it would also
be interesting (albeit, more complicated) to 
include these numbers in its domain.

   npp 2^30
|out of memory: npp
|       npp 2^30

One simple approach would be to have a "closest
match so far" (initially 1), and work through 
the powers one row at a time.

Another approach would involve partitioning
the primes such that implausible roots are
ignored in the iterations.

That said, I don't have the time right now to
pursue these implemtations.

-- 
Raul

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to