[SciPy-User] scipy.optimize lmder_ and dlopen / dlsym

[Gregor Thalhammer]

Am 20.1.2011 um 17:05 schrieb Sebastian Haase:

> Hi Gregor,
> 
> sure. If you don't mind I would like to take a look at your code.
> My images are time series of 512x512 pixels. But my spots are only few
> pixels in diameter, so that I can - after some filtering and
> segmentation - crop the image into many small 7x7 piclets. Those are
> the ones where I have to do the 2d Gaussian fitting.
> Do you have any pointer on where I can learn about the part where you
> say "solving the normal equations instead of qr " ... ?
> 

I like very much this manuscript: 

METHODS FOR NON-LINEAR LEAST SQUARES PROBLEMS

In your case the Gaussian fit function depends linearly on the peak value, an nonlinear on the position. In this case I found this "variable projection" algorithm for very useful: http://www2.imm.dtu.dk/~hbn/publ/TR0001.ps
In my experience the main advantage of this algorithm is the better robustness of the fit against poor starting values. 

Gregor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20110121/dd7f8ccc/attachment.html>