Hello, Asimulo seems great: it does exactly what I wanted to do: integrate Radau5 in Sage. Radau5 is from far the best method for the integration of (very) stiff ODEs...
Le 13/05/2011 10:30, Guilherme a écrit : > Dear all, > > I did two simple spkg (Sage package) one for SUNDIALS and another for > Assimulo. > These are very thin wrappers on the original source code, no patches. > > Here are the links, (Dropbox in the lack of a better place :) ) > http://dl.dropbox.com/u/29028094/sundials-2.4.0.spkg > http://dl.dropbox.com/u/29028094/assimulo-1.4b2.spkg > http://dl.dropbox.com/u/29028094/tutorialIDA.p0.py > > Install them in the above order. > I will keep this files on the above public links till we decide what > to do with this files. > > I tested them on OSX 10.6.7 and RHEL 5.4. > There is plenty of room for improvement on the scripts and compiler > flags related to architectures and optimization... > > The tutorial is the one from the Assimulo website, except that I > patched the example with a few lines to save the figure. > I think the Sage's Matplotlib is default to Agg and it won't show the > figure interactively... > > I run the example with: >> sage -python tutorialIDA.p0.py > > A few commands I used to quickly edit/compile/break/fix the spkg. >> tar xvfz <package>.spkg > Will unpack and show you the files. You can edit the spkg-install if > needed. > >> sage -pkg <package> > With create the above tarball with the edited files. > >> sage -f <package>.spkg > Force compile and install the package. > > More on > http://www.sagemath.org/doc/developer/producing_spkgs.html#creating-a-new-spkg > I was rather easy actually, given that I started with Sage one week > ago... > > Further discussion: > I received positive replies from Assimulo developers and the python- > sundials developer. > > Here is a snap (with his ack.) of the email I got from Runar Tenfjord > (python-sundials): > >> Please go ahead to include this in Sage if it fits your needs. >> >> The background for this wrapper was that the current support >> for ODE/DAE is very poor in Scipy. The solver does not support >> proper root finding and is not reentrant. This is worked on and >> the future result could be of interest to you as the Scipy code >> has a large audience and the code is supported. >> >> Due to the state of the Scipy solver i wrote a limited wrapper for Sundials. >> Several areas of Sundials is not wrapped: sparsity, sensivity analysis etc. >> I have no needs currently for these functions and have no plans >> to include it. > > Here is a snap (without their ack.) of the email from Christian > Andersson (Assimulo): > >> We (Me, Claus Führer, Johan Åkesson) do not have any objection to such >> an development. We would instead like to encourage you to choose >> Assimulo and try to support the work as best as we can. >> >> The only thing we require is that we are acknowledge for the work on >> Assimulo. > > > Final remarks. > I will release a spkg also for python-sundials. I have already > something, but somehow I am mixing x86_64 with i368 libs on the OSX... > Assimulo besides of having more developers they also have a trac and > more documentation. > > That is it. What should I do next with these packages?? > Looking forward for further discussion and corrections. > > Regards, > Guilherme > -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
<<attachment: tdumont.vcf>>
