[SciPy-user] C/Fortran Integrate/minimization?

Pearu Peterson pearu at scipy.org
Wed Mar 30 00:37:40 CST 2005



On Tue, 29 Mar 2005, Greg Novak wrote:

> Some time ago I inquired about the ability to integrate a system of
> differential equations by specify a C function instead of a python
> function to compute derivatives.  This speeds up the calculation
> enormously.  There seemed to be interest in such a feature on the
> mailing list.  I posted code to do this here:
> http://www.ucolick.org/~novak/integrate.tgz.
>
> Are there any plans to include such a feature in scipy?

Yes, in fact, its partly already there. f2py has a _cpointer feature that 
allows one to pass C/Fortran function pointers directly to computational 
routines taking callback arguments giving C/Fortran speed also for
callback arguments. In scipy this feature is available for 
codes that are wrapped with f2py 2.45.241_1926 or up. Some codes in scipy 
are still handwrapped but in future they will be replaced with f2py 
generated wrappers.

Pearu



More information about the SciPy-user mailing list