Permutations in expm

[John W. Eaton]

On 25-Jan-2008, Marco Caliari wrote:

| I investigated how the reverse permutation works in expm. The 
| enclosed patches (against 3.0.0) are much simpler than my previous ones. In 
| practice, there was an error just in the order of the computation of 
| iperm. I tested the patches on random permutations of balanced 
| matrices, both real and complex, up to order 100.
| The ChangeLog could be the same
| 
|   2008-01-25  Marco Caliari  <marco.caliari at univr.it>
| 
|         * dMatrix.cc (Matrix::expm): Correctly perform reverse permutation.
|         * CMatrix.cc (ComplexMatrix::expm): Likewise.

This patch does not apply cleanly for me.  I'm now confused about
preciesely what changes should be made.  So will you please get the
latest versions of dMatrix.cc and CMatrix.cc from CVS, apply your
changes to those files, and resubmit the patch?

Thanks,

jwe