commons-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilles <gil...@harfang.homelinux.org>
Subject Re: [math] LevenbergMarquardtOptimizer unable to perform Q.R decomposition on the 107x2 jacobian matrix
Date Wed, 02 Oct 2013 11:26:56 GMT
Hi.

On Tue, 01 Oct 2013 15:47:11 +0200, Christian wrote:
> Hello
>
> I really need your help to get the LevenbergMarquardtOptimizer up and
> running. Please see the attached compileable code.

[You seem to use two Commons Math versions (cf. "math" and "math3" in 
the
"import" statements). It's better to use one and the latest version 
when
posting here.]

>
> No matter how I try to return the gradients an Exception will raise
> ... hmmm get stuck ...

Is it correct that coefficients "A", "B", "C1" and "C2" depends on
parameters "m" and "w"?
In the affirmative, shouldn't they contribute to the gradient?

Even if not, perhaps it is a wrong usage of the "DerivativeStructure".
It seems that the operations you apply on "f" won't produce the 
intended
function: e.g.
   f = f.add(B).multiply(tc-t).pow(m);
will produce
   ((A + B) * (tc - t))^m
not
   A + B * (tc - t)^m

Regards,
Gilles

> [...]


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@commons.apache.org
For additional commands, e-mail: user-help@commons.apache.org


Mime
View raw message