[Numpy-discussion] Advanced indexing advice

[Cristi Constantin]

Thank you very much Robert !
For really big matrices, "where" is 4 times faster ! For small matrices, it's slower than my old solution, but it really doesn't matter, it's fast enough. :)
I will think about "put" too. I used it before, but i must adapt all functions to obtain the necessary indexes.

Have a nice day !

From: Robert <kxroberto at googlemail.com>
Subject: Re: [Numpy-discussion] Advanced indexing advice?
To: numpy-discussion at scipy.org
Date: Friday, June 19, 2009, 6:49 AM

well its not really slow.
yet with np.where it seems to be 2x faster for large arrays :

a[1:4,1:4] = np.where(mask,b,a[1:4,1:4])

otherwise consider Cython: 
http://docs.cython.org/docs/numpy_tutorial.html#tuning-indexing-further


Robert

>     2009/6/18 Cristi Constantin <darkgl0w at yahoo.com
>     </mc/compose?to=darkgl0w at yahoo.com>>:
>      > I have a question about advanced indexing.
>      >
>      > I have 2 matrices :
>      >
>      >>>>
>      > a=array([[ 0,  1,  2,  3,  4,  5],
>      >             [ 6,  7,  8,  9, 10, 11],
>      >          [12, 13, 14, 15, 16, 17],
>      >          [18, 19, 20, 21, 22, 23]])
>      >
>      > b=array([[1, 0, 1],
>      >          [0, 2, 0],
>      >          [0, 0, 3]])
>      >>>>
>      >
>      > I want to put all NON-zero elements from array B into array A,
>     but use offset!




      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20090619/ce6f4284/attachment.html>