Recent changes to feature-requests

spmatrix.ll_mat(A)

Rodrigo — Wed, 18 Nov 2009 21:46:24 -0000

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

add import pysparse; pysparse.test() mechanism

Daniel Wheeler — Thu, 11 Dec 2008 18:26:14 -0000

add "import pysparse; pysparse.test()" mechanism that launches all the tests. Always useful, since most people, especially windows users and source package managers just install stuff on the fly and never actually unpack and run the tests. It makes support easier too.

matrix sum

Anonymous — Mon, 19 Dec 2005 16:35:02 -0000

Please add the possibility to sum the matrix!!!!