[SciPy-User] ODEINT/ODE solvers redesign--anyone for a sprint at SciPy 2010?

Dag Sverre Seljebotn dagss@student.matnat.uio...
Fri Jun 4 14:42:50 CDT 2010


Dag Sverre Seljebotn wrote:
> Warren Weckesser wrote:
>   
>> It's about time we tackled the issue of the ODE solvers in SciPy.   Some 
>> notes about the issue are on the wiki: 
>> http://projects.scipy.org/scipy/wiki/OdeintRedesign
>>
>> This would be a great topic for a sprint at the SciPy conference.  I 
>> just added it to the list of suggested sprint topics, so give it a vote 
>> if you are going to be there and are interested in working on this.
>>   
>>     
> I'm not going to be there, but I have an interest in this...
>
> Anyway here's an idea that may or may not be beyond the scope of this 
> redesign: It would be great with support for fast Cython callbacks, so 
> that one can implement a class in Cython and have the whole process run 
> in compiled code without any interpreter steps per loop.
>
> Basically, just do
>
> cdef class DoubleFunction: ...
>
> if not isinstance(callback, DoubleFunction):
>     callback = CallPythonFunctionDoubleFunction(callback)
>
> Perhaps one could accept ctypes function pointers too...
>   
Nah, sorry, this is better left as a later independent task. Perhaps 
worth keeping in mind while designing the API and writing any new code 
though?

Dag Sverre


More information about the SciPy-User mailing list