[SciPy-user] SciPy-user Digest, Vol 43, Issue 9

Travis Oliphant oliphant@ee.byu....
Tue Mar 6 11:24:43 CST 2007


Anne Archibald wrote:
> On 06/03/07, Paul Ray <paul.ray@nrl.navy.mil> wrote:
>   
>> On Mar 6, 2007, at 11:51 AM, scipy-user-request@scipy.org wrote:
>>
>>     
>>> It is easy enough to just pull interp out.  Some of the functions
>>> in arrayfns are in numpy.lib already (in the _compiled_base.c module.
>>> Perhaps we could put interp in there as well (it doesn't look too
>>> big).  A simple 1-d interpolation would probably be a useful thing
>>> to have in NumPy.   What do others think?
>>>       
>> It sounds good to me.  I often have to regrid irregularly spaced data
>> onto a uniform grid, which I use interpolate.splrep and
>> interpolate.splev from scipy to do.  This is often the only
>> dependency on scipy in my codes that otherwise use only numpy.  I'd
>> love having a 1-d interp in numpy.
>>     
>
> Why not just use scipy? splev works fine, and numpy doesn't need to
> grow any more... it seems to me that numpy should include just the
> basics of working with arrays (indexing functions, ufuncs, matrix
> multiplication, things like searchsorted) and all the mathematics
> (ffts, linear algebra, interpolation, integration) should go in scipy.
>   
That is generally what we believe as well.  The problem is that Numeric 
already included several additional features.   Trying to maintain some 
semblance of backward compatibility is why NumPy has not shrunk even more.

Because the interp function was already in Numeric and is not that big, 
perhaps it should be added to NumPy.

-Travis



More information about the SciPy-user mailing list