No, I didn't parallelize. I should add that the program has two steps - first to create the polynomials, and then to multiply them. To create the polynomials, I could have used GiNaC, but I didn't have to do anything more complicated algebraically than multiply by x. So I used GMP directly for that part. According to the documentation, GiNaC uses GMP for its floating point arithmetic. And GMP seems to allow powers of ten way larger than plus or minus 10000. Cristobal Navarro wrote:
amazing!
im also in the same problematic but close to zero, 1^-10000, i have not tested yet but based on your experiments i guess i will have good results too just being curious, did you paralelize ?
On Tue, Aug 10, 2010 at 4:35 PM, Stephen Montgomery-Smith <stephen@missouri.edu <mailto:stephen@missouri.edu>> wrote:
Hi Guys,
I just wrote some programs that use GiNaC to multiply polynomials whose degrees can get into the 1000's, and whose coefficients are as large as 10^10000. GiNaC performed extraordinarily well, and even went a little faster than Mathematica! I am very impressed!
(If you are interested, I am trying to analyze results of Luria-Delbruck experiments using Bayesian statistics.)
Stephen _______________________________________________ GiNaC-list mailing list GiNaC-list@ginac.de <mailto:GiNaC-list@ginac.de> https://www.cebix.net/mailman/listinfo/ginac-list