Article View: sci.math.num-analysis
Article #988Re: non-linear least-squares fitting
From: Walter Polkosnik
Date: Thu, 21 Jan 1993 08:26
Date: Thu, 21 Jan 1993 08:26
25 lines
1446 bytes
1446 bytes
In <1jkb5nINN8qn@iskut.ucs.ubc.ca> holland@geop.ubc.ca (Stephen Holland) writes:
>Does anyone out there know where I might find a good public domain routine
>for doing non-linear least-squares fitting WITHOUT derivative information.
>I'm looking for something along the lines of POWELL in Numerical Recipes or
>CHIFIT in Bevington. ForTran source would be nice since my software is
>writen in ForTran but i can translate from C if need be.
>thanks
>steve holland
I have used subroutine LMDIF from MINPACK (which you can get from netlib) and
NL2SOL (which you can also get from netlib, as part of TOMS) extensively and
for all my applications they work great. You can always do some kind of
difference approximation to the derivatives if you don't want to calculate them
explicitly (which is what those two subroutines do).
Those two routines are also very simple to program around also.
--
Walter Polkosnik
Physics Department banana@panix.com
Queens College, CUNY {cmcl2,apple}!panix!banana
Technical Details
Message-ID:
<C173FK.5DI@panix.com>Path:
arc1-ca.pugleaf.net!sparky!uunet!usc!howland.reston.ans.net!news.ans.net!cmcl2!panix!bananaReferences:
<1jkb5nINN8qn@iskut.ucs.ubc.ca>Show Full Headers
Complete Article Headers
Newsgroups: sci.math.num-analysis Path: arc1-ca.pugleaf.net!sparky!uunet!usc!howland.reston.ans.net!news.ans.net!cmcl2!panix!banana From: banana@panix.com (Walter Polkosnik) Subject: Re: non-linear least-squares fitting Message-ID: <C173FK.5DI@panix.com> Organization: PANIX Public Access Unix, NYC References: <1jkb5nINN8qn@iskut.ucs.ubc.ca> Date: Thu, 21 Jan 1993 08:26:07 GMT Lines: 25