[SciPy-user] filling array without loop...

fred fredmfp@gmail....
Mon Apr 23 08:06:40 CDT 2007


Anne Archibald a écrit :
> On 22/04/07, fred <fredmfp@gmail.com> wrote:
>
>   
>> Forget that is a convolution (because in fact, it is really not),
>> simply a scalar product (inner product says scipy) in each cells of an
>> array, as I wrote in my example.
>>
>> Does it changes something ?
>>     
>
> Uh, maybe I'm confused - 
Ok, so let me explain a little more...

I have two 2D matrices, say A et B, with same dims = 571x876.
B is to be computed, A is known.
Each cell of the matrix B is computed from a "scalar product" with a few 
cells of the matrix A and
a "weights" vector W:

  B[i,j] = \sum_0^8 w_n a_n

where

 - W = [w_n] is a 1D vector (computed from solve(a,b)) with dim = 9

 - a_n = A[i,j]

a_n are selected from a submatrix of A (much smaller than A), say A':

        a_2 -- a_5 -- a_8
          |         |        |
A' = a_1 -- a_4 -- a_7
          |         |        |
        a_0 -- a_3 -- a_6  


My idea is to write B[i,j] as a scalar product between W and A' 
flattened to a 1D vector:

B[i,j] = dot (W,A')

(ok, inner() method was unadapted)

The problem, from my own opinion, is once again, the two (three in 
general case) loops to compute each cell
of the matrix B.

So my question is : is it possible to fill the B matrix without loops ?

I hope I more clear.


Thanks in advance.


Cheers,

-- 
http://scipy.org/FredericPetit



More information about the SciPy-user mailing list