[Scipy-tickets] [SciPy] #311: The test results in a segmentation fault

SciPy scipy-tickets at scipy.net
Tue Jan 30 14:38:26 CST 2007


#311: The test results in a segmentation fault
----------------------------+-----------------------------------------------
 Reporter:  nils            |        Owner:  timl    
     Type:  defect          |       Status:  assigned
 Priority:  normal          |    Milestone:          
Component:  scipy.linsolve  |      Version:  devel   
 Severity:  normal          |   Resolution:          
 Keywords:                  |  
----------------------------+-----------------------------------------------
Comment (by mauger):

 Ok...I also get the segfault. The code still has two bugs in it (one of
 which I unsuccessfully tried to fix with the previous patch). The bug
 causing the segfault seen above could be "fixed" (really circumvented) by
 setting the "delta" parameter in get_perm_c() to 0. Otherwise, mmd.c could
 be patched to correctly deal with non-zero delta parameters.

 I've attached a new patch that takes the first approach and sets delta to
 zero. The patch also fixes the second bug. I've tested the patch on the
 following code with no failures.

 {{{
 from scipy import *
 from pylab import spy, show


 M = sparse.csr_matrix(arange(36).reshape((6,6)))
 lu = linsolve.splu(M)

 n = 20
 A = sparse.lil_matrix((n,n))
 for i in arange(n):
   A[i,:] = random.rand(n)
 A = A.tocsc()
 B = sparse.speye(n,n)
 linsolve.splu(A - B)

 spy(A.todense())
 show()
 }}}

 I don't have ARPACK installed so I cannot test the second half of the
 eigs1.py example, but I don't get any errors running the first half.

-- 
Ticket URL: <http://projects.scipy.org/scipy/scipy/ticket/311#comment:12>
SciPy <http://www.scipy.org/>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list