On Wednesday 09 April 2008, parisse wrote: > On 8 avr, 21:25, "Mike Hansen" <[EMAIL PROTECTED]> wrote: > > > I have added a benchmark link with Fermat gcd tests, giac seems 5 to > > > 10 * faster than maxima. I don't have magma, it is most probably > > > > > > another factor of 10 * faster. > > > > I think that comparison with Magma is a little optimistic, especially > > in the modular case. > > Right, I thought the Opteron was faster than it really is. I have > compiled giac over an Opteron (244) to be able to make a more > meaningfull comparison and updated the timings. The improvements of > magma between version 2.09 (where giac has comparable timings in the > non modular case) and 2.12 suggests that either new algorithms were > found (but I never heard of them) or existing algorithms were improved > (tested with these benchmarks). And maybe, magma 2.12 has speed > improvements related to the 64 bit architecture. > Anyway, using giac would improve the current sage gcd speed and > working on efficiency would benefit to both. Now, if sage developers > are not interested, it's not a main concern for me, giac has it's own > interfaces, but I would find a little bit stupid that these two open- > source softwares which have a common target to provide an alternative > to maple/mathematica would not cooperate.
Hi, sorry for joining in so late. Generally speaking, it is no necessarily easy to determine what the "Sage developers" want, since we are all scattered across the globe and all have different agendas/preferences/ideas. Also, all one needs is one Sage developer (oneself can be that developer) to start working on something. Thus, the lack of response doesn't mean that "the Sage developers" don't want to collaborate. Each one of us might just have too much on her plate right now. To the topic: I don't really get the timings you posted, i.e. I cannot really relate them to the timings posted on the Sage Wiki, could you try to provide data that is easier to relate, or maybe I'm missing something. I would say the best way to cooporate (since both projects aim to be a frontend and IIRC you also ship Pari etc.) is to help each other out on the C(++) library level. Especially for multivariate factoring, it would make sense to colaborate on C/C++ library all projects could you. This way everybody wins. I don't really know the area but the "Factory" might be a good starting point. Factory is the stand-alone (!) library used by Singular to do multivariate factoring. Of course the defects of Factory started this thread in the first place, but it might still have some sort of framework such that one can start developing the missing algorithms. I would like to hear from people with more experience in that area what they think about Factory's data structures (Roman?). Another big plus for contributing to Factory is that it is somehow widely used: Singular, Macaulay2 and Sage. Btw. Michael, what is the status of multivariate factoring in CoCoALib? To put some code where my mouth is, I can provide the following: We had a direct interface to the factory once and I could revive it: I.e. I could provide a Python interface to Factory that looks as much as possible like the C++ interface in terms of class names, methods, etc. Then, people can start going crazy and implement their favorite factoring algorithms, heuristics, tricks in the Sage interpreter rather than on C level and make use of all the Sage library (I guess linear algebra could be relevant). As a crucial second step -- if the result doesn't totally suck -- this all gets ported down to C++. I can't do the "creative" steps, where one has to write new algorithms etc., but I can certainly help for the first (to Python) and the last (from Python) step. But maybe this whole idea is just bollocks, I don't know. Thoughts? Martin PS: In any case, If anyone wants to work on an optional GIAC spkg, please speak up! -- name: Martin Albrecht _pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99 _www: http://www.informatik.uni-bremen.de/~malb _jab: [EMAIL PROTECTED] --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
