[SciPy-User] Revisit Unexpected covariance matrix from scipy.optimize.curve_fit

Tom Aldcroft aldcroft@head.cfa.harvard....
Fri Feb 22 16:28:26 CST 2013


Thanks for the clarifications and discussion here.  Just to close this
out, I updated the notebook with the suggestion from Christoph (which
is formally the same as the one from Eric Moore).  This gives the
expected result.

http://nbviewer.ipython.org/5014170

I just made a documentation update PR
(https://github.com/scipy/scipy/pull/446).  Hopefully this is not
controversial and can be merged.  Then Christoph and/or I will make a
pull request to add a scale_pcov parameter.

Thanks,
Tom

On Fri, Feb 22, 2013 at 1:53 PM, Tom Aldcroft
<aldcroft@head.cfa.harvard.edu> wrote:
> On Fri, Feb 22, 2013 at 1:30 PM, Christoph Deil
> <deil.christoph@googlemail.com> wrote:
>> How about adding an option "scale_pcov" to curve_fit whether this scale
>> factor should be applied.
>> To keep backward compatibility the default would have to be scale_pcov=True.
>> The advantage of this option would be that the issue is explained in the
>> docstring and that people with "sigma=error" instead of "sigma=relative
>> weight" can easily get what they want.
>> Should I make a pull request?
>
> +1


More information about the SciPy-User mailing list