[SciPy-dev] Removing globals from minpack wrapper

Yosef Meller mellerf@netvision.net...
Thu Jun 26 12:07:47 CDT 2008


Travis E. Oliphant wrote:
> Yosef Meller wrote:
[snip]
>> The attached patch implements the trick in minpack_hybrj, and the 
>> Python code tests that it works.
>>
>> Do you think it's worth doing to the rest of the minpack functions?
> 
> That is pretty clever.   Amy I correct that the only time cost is an 
> extra array allocation (for ap_fvec) per complete calla?

That's about right. Maybe there's extra O(evals) pointer retrievals, but 
that's much smaller than O(n) if I understand this correctly.

> I think it is worth it.    Using globals requires a bit of rigging to 
> support re-entrance of the optimizers (so you could have a python 
> function that itself calls optimize).  

I'll try to keep the patches coming.

-- 
http://yosefm.imagekind.com/Eclectic


More information about the Scipy-dev mailing list