[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: C++ wrapper for NPSOL

From: Al Goldstein
Subject: Re: C++ wrapper for NPSOL
Date: Fri, 3 Oct 1997 15:18:39 -0700 (PDT)

On Thu, 2 Oct 1997, John W. Eaton wrote:

> I am hoping to find an optimization expert to help provide freely
> availble optimization tools.  Would anyone like to help provide
> something in this area?

NPSOL was, the last time I perused it, based on ancient optimization
techniques which work well for most problems. Code based on modern
algorithms should be freely available. For the most part these are based
on optimizing with the aid of a differentiable logarithmic penalty function. 
These methods work well and there is more interest in using them.
They are called interior penalty methods.

I have a code using octave for the Chebyshev approximation of inconsistent
systems of linear equations. It has been run in octave for systems as
large as 1000X100. Beyond 1000 results in octave failure due to memory
problems. Perhaps this could be improved if I knew the cause of the failure.
The code can also be used to solve systems of linear inequalities.  I
regard the code to be in alpha status, but I'm willing to share it.
The algorithm works on a new principle; the paper describing it has been
submitted but not yet accepted.

If you would like to get programs and someone is willing to write wrappers
for them I'll be glad to advertise for them.


> Thanks,
> jwe

reply via email to

[Prev in Thread] Current Thread [Next in Thread]