Hey, thanks for all the responses. I use Cython occasionally, but i tend to restrict myself to basic C types and simple functions which do not import anything -- it seems I can learn an awful lot by studying your examples!
Can i ask the following questions about Jason's code? ** why so you 'import' CDF while you 'cimport' ComplexDoubleElement? what is the difference? ** and why do these imports anyway, are they necessary for "cdef complex..." to work? ** I didn't know PY_NEW. What is the difference between res= PY_NEW(Matrix2) and simply res= Matrix2(0,0,0,0) (say) ? and what about using PY_NEW with a class that *requires* parameters to its __init__? ** more generally, is there a Cython tutorial that would be (just) sufficiently advanced to cover such an example? thanks! Pierre On 13 fév, 20:13, William Stein <wst...@gmail.com> wrote: > On Mon, Feb 13, 2012 at 11:07 AM, Jason Grout > > > > > > > > > > <jason-s...@creativetrax.com> wrote: > > On 2/13/12 12:41 PM, William Stein wrote: > > >> On Mon, Feb 13, 2012 at 10:29 AM, Robert Bradshaw > >> <rober...@math.washington.edu> wrote: > > >>> On Mon, Feb 13, 2012 at 10:06 AM, William Stein<wst...@gmail.com> wrote: > > >>>> On Mon, Feb 13, 2012 at 9:59 AM, Pierre<pierre.guil...@gmail.com> > >>>> wrote: > > >>>>> I see. Well I *do* have hundreds of 2x2 matrices to multiply out so > >>>>> i'm better off storing them as numpy matrices throughout... thanks for > >>>>> your explanations though. > > >>>>> Pierre > > >>>> You might consider using Cython and writing a custom 2x2 matrix class. > >>>> It wouldn't be difficult... so I'll write one right now and respond > >>>> with the benchmarks. > > >> Here it is: http://480.sagenb.org/home/pub/97/ > > >> I ended up using GSL's complex matrix data type and the BLAS level 3 > >> routine to do the multiplication. I did not add any other > >> convenience functions to the class, so some more will probably be > >> needed for your application. > > > I'm curious why you didn't just store the 4 complex numbers in C. > > I was concerned about "numerical stability" and figured BLAS would > deal with that. But maybe that doesn't matter. > > Most importantly, I forgot about Cython's complex type, which I now > remember Robert Bradshaw wrote for his thesis work. I wasn't looking > forward to writing your line: > > res.m00=self.m00*right.m00+self.m01*right.m10 > > but against the gsl library. But using the cython type in complex, > you get the above at C speed with easy notation, which is pretty > awesome. > > > I tried > > it and got a much bigger speedup: 17x faster than numpy and 150x faster than > > Sage. Seehttp://sagenb.org/home/pub/4303/ > > Nice. > > > > > Thanks, > > > Jason > > > -- > > To post to this group, send email to sage-support@googlegroups.com > > To unsubscribe from this group, send email to > > sage-support+unsubscr...@googlegroups.com > > For more options, visit this group at > >http://groups.google.com/group/sage-support > > URL:http://www.sagemath.org > > -- > William Stein > Professor of Mathematics > University of Washingtonhttp://wstein.org -- To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to sage-support+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-support URL: http://www.sagemath.org
