On Dec 6, 7:57 am, "David Joyner" <[EMAIL PROTECTED]> wrote:
> On Sat, Dec 6, 2008 at 8:54 AM, mabshoff <[EMAIL PROTECTED]> wrote:
<SNIP>
Hi David, rjf
> Sorry. I missed your post and emailed the Maxima list anyway.
> (I don't know what to do to test if the problem is really on the maxima side
> anyway, but I did not word my post as a bug report either.)
Yes, I saw the post. By the way: no need to CC for any email to the
Maxima list since I am subscribed.
> Meanwhile Richard Fateman's reply on this thread is very useful I think.
I agree, that will help narrow down the problem. I did not know as rjf
told us that Maxima "labels" results which would explain the more or
less linear increase in compute time since the garbage collection is
likely linear in the number of objects in the heap. So it looks like
the problem is caused by a mix of using a non-optimal common lisp
implementation and us not killing the results when we should.
> Let me know if you want to plot fourier series methods rewritten numerically,
> trying to avoid maxima calls. However, for me, it is not a problem to wait
> 10 or 20 seconds for a plot, since I typically do no more than one or two
> per day while teaching that material.
Well, this problem was exposed by the plotting code and I guess in
this particular case using _fast_float for plotting might be an
option. I don't consider resolving the problem particularly pressing
since no one has complained yet, but for most plot situations
_fast_float should be the better solution.
But the general issue here is that long running computations using
Maxima from Sage cause issues due to the labeling, so this is
something we need to deal with by
(a) taking advantage of the already computed results when it makes
sense
and
(b) killing labelled results to not use excessive amount of memory.
I.e. running the same integration over and over again from Sage in
Maxima should not result in use using up all available memory.
Finding the balance between (a) and (b) is likely tricky.
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
-~----------~----~----~----~------~----~------~--~---
