On Nov 26, 2008, at 2:13 AM, root wrote: > Tim, > > Were you just interested in integration or do you intend to work on a > full test suite? If you're interested in building a larger test suite > I'd be happy to work with you (or others) on areas that interest both > Axiom and Sage. Indeed, this might be a good way for students and > mathematicians who don't program to contribute to Sage. > > As you know, the CATS idea is that we establish a common test suite > among the various systems so that we can leverage the test effort > from any one system for the benefit of all. Testing is important and > in very short supply. > > Ideally we would have independently created and verified mathematics > (like the standard integration tables, ODEs, polynomial factorization) > organized in some rational way. Then we show that every system using > the suite gets equivalent answers. For areas where Axiom implements > the mathematics I'm willing to do the heavy lifting of getting the > original mathematics into an initial, machine readable form. And I'm > willing to run the test suite through Axiom to create an initial set > of results.
Tim, My main interests are in terms of CAS capabilities are integration, differentiation, and ODEs/PDEs. So, I'm quite happy to help out in all of these areas. Right now, the approach I'm taking for the integrals is pretty simplistic and a more general framework would work not only for integration, but for the others as well. For example, in integration, the specific test needs to specify a) the integrand, b) variables in integrand, c) desired result, d) comparison procedure. For a general framework we would need a) the function to be tested, b) function inputs, c) variables in function, d) desired result, e) comparison procedure. Even in your CATS there tend to be specific strategies utilized for comparison. So, one can package those up in a function and the test suite is told to call that function for comparison purposes. I've been a little lucky in that Maxima's simplify_full() has done most of what I've needed. However, for testing the Sage-SymPy interface, I'll have to do something else. Once a basic framework is done then it should be fairly simple to implement a test suite (just time consuming) for different areas of mathematics. However, I don't plan on doing all of it myself. After I've finished your tests, I'm going to see if I can find others to have a broader coverage. At the moment, I've only finished* 10 of the 34. (*) I don't support multiple branches in the solution yet. > > When Sage decides to develop various facilities (like integration) > in native form these test suites would be very valuable. The test > suites would also be useful as user documentation for Sage. > My tests aren't that useful as user documentation, because my approach is somewhat different than the one you took. > Axiom already has over 600 input files used to test various parts > of the system. Comparing these results with other subsystems of > Sage would provide a valuable cross-check. That's certainly a good idea. It would also be good in providing upstream bug reports to the subsystems that Sage interfaces to. Cheers, Tim. --- Tim Lahey PhD Candidate, Systems Design Engineering University of Waterloo --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
