On May 18, 2009, at 12:13 AM, Ondrej Certik wrote:
> > > * at the very end of the presentation there was a discussion about > numeric stuff. There are tons of computational programs in lots of > fields (atomic physics, quantum field theory, electromagnetics, > electronic structure calculations, fluid dynamics, atmospheric > sciences, nuclear engineering ....) and there is just no way how this > can all be in Sage. Nevertheless, people would like to teach with it, > let's say some electrodynamic course, or finite element course, or > (partial) differential equations course. Sage currently cannot do any > of that. As Ondrej well knows, I'm quite interested in FEM and PDEs, so I have an interest in using these with Sage. > One way out is just to install the particular package for it, but it's > not easy. For example for the finite element solvers that our group[0] > is doing we need (besides python, numpy, scipy, cython) also lots of > solvers (petsc, umfpack, ...), then 3d plotting (mayavi) and other > dependencies. And also I would like to have more than our FEM library, > I would like to include libmesh and other things, unified with one > Python interface, so that one can compare, etc. Adding all of this > into Sage, it would get twice as big. > Having a simple way to interface these packages to Sage and create the .spkg would be a good thing. That way, once someone figures out the process, they can bundle it and make it available for others. Also, if there is a wiki page where Sage can get URLs for 3rd-party packages (with the understanding that these are unsupported by Sage developers) it would facilitate sharing and make it easy to install. > So the approach that I took with SPD [1] that William was also > discussing in the presentation, was to throw out most of Sage, except > the build system and the scipy+notebook combo. That way our group can > add the finite element + related stuff and have our package that we > can try to get to the end users (e.g. the combination of the web > notebook and the actual program that works on all platforms), and > other groups can do the same for atomic physics, fluid dynamics, etc. > And it is still fully compatible with Sage, e.g. the base packages are > the same, and I think all the other custom packages would also just > work when installed into Sage. I'm interested in the SPD stuff because I'd like to see some of the FEM stuff available from Sage, although the difficult part of FEM for me is the derivation, not the solution. Whereas often the equations are already well known and it's the meshing/solution that presents the difficulty. > > I however still want to be able to represent and manipulate symbolics > (e.g. represent some ordinary or partial differential equation and > then solve it with our C++ solver), so we'll just use sympy for that, > because it's small and it can convert expression from and to Sage, if > needed. I hope we could figure out more ways to collaborate with > pynac, Fredrik (sympy and mpmath dev) is at SD15, so I hope they'll > discuss it. :) > Based upon the videos, it looks like sympy integration will be better fit into Sage 4.0 so I'll probably make some changes to my test framework to utilize the new features. This will make testing sympy easier (at least from Sage) so hopefully, that will help you as well. I'm very interested in this. Some of the big advantages of MATLAB and Mathematica are the toolboxes that are available. MATLAB at least makes it quite easy to package a new toolbox, if we can have a simple way to make and distribute new Sage toolboxes, that would go a long way. Then, if it's clear that particular toolboxes are seeing a lot of use, Sage developers can look at bringing them into the standard distribution. Cheers, Tim. --- Tim Lahey PhD Candidate, Systems Design Engineering University of Waterloo http://www.linkedin.com/in/timlahey --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
