Re: Octave & C++/Octave and symbolic math

[Steve Hill]

Dirk,
I thoroughly understand your point. LINPACK has a long and honourable
past, and is maintained by specialists. But, at the moment, I can't do
some of the things I'd like to in octave (like finding the
symbolic eigenvalues/vectors of a matrix of symbolic terms) - or maybe it
is possible, but ugly. The suggestion of linking a symbolic math package
to octave is good. However, to make it work well we need to be able to do
matrix operations on symbolic terms. We all know that there are quick ways
of doing many of these things computationally. My basic point is that if
code is going to have to be written to implement quick computational
matrix operations on symbolic terms, then why not produce a decent
integrated representational format that will handle both numerical and
symbolic terms, and write the code for that. 

I think we'd all like to avoid re-inventing the wheel, and there is
certainly no point in writing a whole load of code in C++ when it already
exists in Fortran. I'm just not sure that the LINPACK libraries, or
whatever, can accomodate an appropriate range of symbolic math (correct
me if I'm wrong!). If symbolic math is going to be added to octave,
wouldn't it be good to know that whatever the calculation involves, it
will be handled in a uniform manner?

Cheers, steve

========================================================================
Like a war or a great fire, the effects of a storm go rippling outwards 
through webs of people for years, even generations. It breaches lives 
like coastlines and nothing is ever again the same.

Sebastian Junger