| Re: L1 Norm? [message #45997 is a reply to message #45913] |
Thu, 20 October 2005 22:26  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
"Graham" <GrahamWilsonCA@yahoo.ca> writes:
> Hi all,
>
> This is probably really just a question for Craig, but I thought I'd
> ask here rather than in an e-mail. Is it possible to use mpfit and
> minimize for something other than the L2 norm? In particular, I'm
> looking to minimize the L1 norm as I have some outliers that are
> pulling my fit but are quite difficult to routinely remove.
The short answer is, not it's not possible. The L2-ness of MPFIT is
built deeply into the algorithm.
One option is to try the RESDAMP keyword of MPFIT. I don't really use
it, but it is designed to suppress residual outliers beyond a certain
size that you choose. It does assume that your initial conditions are
close enough to the correct ones that most good data points are not
flagged as outliers.
Another option is to use a general function minimizer (like perhaps
TNMIN) where you can minimize an arbitrary function.
Good luck,
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@REMOVEcow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
