# Numerical Computation

Participants: Kirsten Fagnan, Josh Kantor, didier deshommes

Possible Projects depending on interests of participants. Unless more people join our group we will probably have to focus on one or two projects. I believe we or someone should definitely do the first one listed below. This would be mostly pure python coding. Projects 2,3,4 and 5 would require pyrex and maybe some C programming. The last one would be great to have someone do, but I'm not very good at writing interfaces.

- Extend numerical sparse matrix capabilities. In particular add additional constructors for matrices with special forms, such as tridiagonal matrices.
- Implement numerical complex sparse matrices based on Josh's real sparse matrices.
- Incorporate clapack into SAGE (Not much to do here, just to make an spkg and make sure it works). Actually there are a couple specialized routines in lapack that the numpy/gsl don't include which we could directly call. For example specialized solvers for banded matrices.
- Write an interface to ARPACK for computing eigenvalues of large spare matrices. This depends on clapack.
- Wrap the optimization/minimization/root finding/least squares routines in GSL. (These would all be very similar to the code in sage/gsl/ode.pyx).
- Improve SAGE's support for octave and matlab.

* look at chebfun.