PySparse: Segmentation fault when using superlu.factorize(csr_mat)

[William Hunter]

Hi there;

I used to use my own compiled PySparse, which worked fine, but I saw the
other day that it's now in the repo's, which is very convenient, thanks.
However, I get a segmentation fault when I use superlu.factorize(csr_mat)
--- please see attached Python script for a simple, detailed example of
what's happening. Here's the output:

###
Initialise sparse matrix 'A':
A = ll_mat(general, [5,5]):
 --------  --------  --------  --------  --------
 --------  --------  --------  --------  --------
 --------  --------  --------  --------  --------
 --------  --------  --------  --------  --------
 --------  --------  --------  --------  --------


Values to be added to 'A' ...
vals = [ 0.14911707  0.13379686  0.28159129  0.5655244   0.66737709]

... at these indices:
indices = [4, 1, 2, 0, 3]

Now add the values in 'vals' to 'A':
Updated A = ll_mat(general, [5,5]):
 0.565524  --------  --------  --------  --------
 --------  0.133797  --------  --------  --------
 --------  --------  0.281591  --------  --------
 --------  --------  --------  0.667377  --------
 --------  --------  --------  --------  0.149117


Convert to CSR, otherwise we can't factorize:
A in CSR format = csr_mat([5,5], [(0,0): 0.565524, (1,1): 0.133797, (2,2):
0.281591, (3,3): 0.667377, (4,4): 0.149117])
Type of A = <type 'csr_mat'>

Now factorize 'A_CSR' so that we can solve:
Segmentation fault
###

Regards,
William Hunter
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-users/attachments/20080530/3c759b6a/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pysparse_superlu_factorize_bug.py
Type: text/x-python
Size: 931 bytes
Desc: not available
URL: <https://lists.ubuntu.com/archives/ubuntu-users/attachments/20080530/3c759b6a/attachment.py>