Re: [eclipse-clp-users] Large integers and delayed goals

From: Joachim Schimpf <jschimpf_at_...311...>
Date: Sat, 17 Mar 2012 17:01:40 +0100
Sergey Dymchenko wrote:
> Hi Joachim,
> 
> Thank you for detailed response.
> 
> There is a solution for A^4 + B^4 + C^4 = D^4, but it's probably too
> large to find it in feasible amount of time:
> 
> a = 95800, b = 217519, c = 414560, d = 422481
> 
> As for your first suggestion to solve with ic in floating point
> arithmetics and then test with exact arithmetic. Of course we will
> weed out false positives with this approach, but isn't it possible
> that a solution in exact arithmetic will not be solution in floating
> point arithmetic?

For plain floating point arithmetic, that would indeed be the case.
But library(ic) and the breal data type use safe interval arithmetic,
where the lower bound is obtained by rounding down, and the upper
bound by rounding up.  Possible solutions are therefore guaranteed
to be within the bounds.

-- Joachim
Received on Sat Mar 17 2012 - 16:01:53 CET

This archive was generated by hypermail 2.2.0 : Mon Jul 09 2018 - 02:05:29 CEST