On Dec 4, 2008, at 10:51 PM, Jan Groenewald wrote: > Hi, > > On Thu, Dec 04, 2008 at 02:47:22AM -0800, Mike Hansen wrote: >> I don't have a whole lot of time, but I put some improvements up at >> http://sagenb.org:8000/home/pub/94/ and included some timings there. >> h6.eigenvalues() took 0.11 seconds to run on my machine. It looks >> like everything you're doing can be done with exact arithmetic so >> that >> you don't have to use CDF. Most of the matrices are over QQ, and the >> ones that are complex are over QQ[sqrt(-1)]. > > Thanks, this is great. This way the h6.eigenvalues took less than 1 > second, > and h6.eigenvectors_right took 18 seconds. > > I don't understand why complex floats are so much slower than > rationals: > 18 seconds versus 4+ hours (unfinished)! In the end the eigenvalues > are > real (floats).
One factor could be that the linear algebra over the default real and complex rings (RR and CC) have not been optimized at all (for speed or precision). Also, things cancel out exactly over QQ[i] which may help some (and your problem is sparse/small enough that it doesn't run into coefficient explosion). Definitely send us bug reports about this kind of thing so it can get fixed. - Robert --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
