#3 spmatrix.ll_mat(A)

open
nobody
5
2009-11-18
2009-11-18
Rodrigo
No

Dear Sirs,

I have tried to a convert a full matrix 11000x11000 (this can be sparse depending on processing) into the pysparse linked list format. All mehtods I use where very poor in performance. A summary of what I have tried:

a) reading from a file spmatrix.ll_mat_from_mtx
this was very very time consuming
b) extracting the indexes and the values of the matrix an then using put(V,index1,index2)
also time consuming
c) with a double loop
for ii in xrange(N):
for jj in xrange(N):
Al[ii,jj]=K[ii,jj]
also very time consuming

Is there any possibility to have a method like:
Al = spmatrix.ll_mat(A) where A can be a full matrix
or something like
Al=spmatrix.coo_toll(V,I,J)

For such big matrices (remember they are full) this could mean a dramatic improvement in many cases (I use of such matrices for eigenvalue calculations using jdsym).

Thank you very much

Rodrigo

Discussion