On Dec 6, 5:02 am, "David Joyner" <[EMAIL PROTECTED]> wrote:
Hi David, > I think the first computation of 10 seconds makes sense. You are > basically doing > around 30 definite integrals symbolically, multiplying each by a sine or > cosine, > adding them and plotting the result. If Maxima is doing something > "dumb", I guess > it depends on how you define the term. I'm *guessing* that it is the > Maxima evaluation > of the trig sum to create the plot which is the bottleneck here. This > is a finite > trig sum (and will always be a finite trig sum for any use of this method) > which is known to Python I, so Maxima is not needed for the evaluation. The concern here isn't relative performance. The bad thing is that Maxima seems to take consistently longer and longer to do the same work. I have seen discussions about MMA where the first computation would be faster than a repeat of the same computation. The time to repeat again and so would also increase somewhat and then eventually plateau. An explanation for that was MMA's garbage collection, but the person who stated that did not have access to MMA's internals, but it seems like a good educated guess. Now with Maxima it might be something similar, i.e. there are various explanations: (a) We do something dumb on Sage's end, but since the CPU time on our end is more or less consistent I doubt that there is the problem. (b) We tell Maxima do do something dumb, i.e. more work than the previous run since we might not clear some global variables or alike (c) Maxima has a bug in it that makes the computation we run slower and slower. That could be related to the global state as alluded to in (b) with the difference that it is Maxima internal, i.e. vanilla Maxima would show the problem (d) the garbage collection of clisp sucks and/or takes much longer to plateau So someone who knows what we do in Maxima when we call the above functions should see if running the computation on vanilla Maxima also shows the same problem. I case vanilla Maxima has this is trying a Maxima with is not build with clisp could potentially narrow down if (d) is the culprit. I would do it myself, but since (a) performance of calculus is a lost cause anyway when using Maxima and (b) I have more pressing things to do I was hoping that somebody else would poke around. Cheers, Michael --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
