Re: math problem: matrix exponential not following the rules in ML or Octave?

[Mike Miller]

On Fri, Aug 07, 2015 at 18:37:40 -0400, Nicholas Jankowski wrote:
> Ok, crickets on that one. Could someone used to navigating the Octave
> source point me to the source on Savannah for either mpower or the ^
> operator (assuming one's just an alias to the other), want to see if
> they're set to do the eigenvector route or if they call something like A^x
> = expm(x*logm(A)), as calling those gives me the same output.

I can help with that one, take a look here:

http://hg.savannah.gnu.org/hgweb/octave/file/9c7dd58b57e4/libinterp/corefcn/xpow.cc#l188

(I assume you mean a matrix raised to a scalar, but other variants are
defined in the same file).

-- 
mike