[m-dev.] Re: lp_solve distribution

[Thomas Charles CONWAY]

Tyson Richard DOWD, you write:
> For completeness, the issues are:
> 	1. Numerical stability of the solver (probably not a problem
> 	   with our datasets).

Chris and I discussed this some time ago, and I think he said that
it would be unstable where the values of the coefficients differ by
many orders of magnitude. Aparently the datasets from the termination
analyser are not like this - is it the case that they will *never*
be like this? (ie is there some charactaristic of programs that prevents
this from being a problem?)

> 	Did I forget anything?

Not that I can think of.

Thomas
-- 
ZZ:wq!
^X^C
Thomas Conway               				      conway at cs.mu.oz.au
AD DEUM ET VINUM	  			      Every sword has two edges.