[Scipy-tickets] [SciPy] #1680: error with misleading msg: matrix must have float64 values

SciPy Trac scipy-tickets@scipy....
Fri Jun 22 12:22:45 CDT 2012


#1680: error with misleading msg: matrix must have float64 values
------------------------+---------------------------------------------------
 Reporter:  yarikoptic  |       Owner:  somebody   
     Type:  defect      |      Status:  new        
 Priority:  normal      |   Milestone:  Unscheduled
Component:  Other       |     Version:  0.10.0     
 Keywords:              |  
------------------------+---------------------------------------------------
 on a sparc (and s390x), ie big endians we ran into


 {{{
   File "/usr/lib/python2.7/dist-
 packages/scipy/sparse/linalg/dsolve/linsolve.py", line 85, in spsolve
     autoTranspose = True )
   File "/usr/lib/python2.7/dist-
 packages/scipy/sparse/linalg/dsolve/umfpack/umfpack.py", line 590, in
 linsolve
     self.numeric( mtx )
   File "/usr/lib/python2.7/dist-
 packages/scipy/sparse/linalg/dsolve/umfpack/umfpack.py", line 403, in
 numeric
     self.symbolic( mtx )
   File "/usr/lib/python2.7/dist-
 packages/scipy/sparse/linalg/dsolve/umfpack/umfpack.py", line 364, in
 symbolic
     indx = self._getIndx( mtx )
   File "/usr/lib/python2.7/dist-
 packages/scipy/sparse/linalg/dsolve/umfpack/umfpack.py", line 349, in
 _getIndx
     raise ValueError('matrix must have float64 values')
 ValueError: matrix must have float64 values
 }}}

 with scipy 0.10.1 (as in Debian sid). and that is due:

 {{{
 (Pdb) l
 344                        or mtx.indptr.dtype != np.dtype('l'):
 345                     raise ValueError('matrix must have long indices')
 346
 347             if self.isReal:
 348                 if mtx.data.dtype != np.dtype('<f8'):
 349  ->                 raise ValueError('matrix must have float64
 values')
 350             else:
 351                 if mtx.data.dtype != np.dtype('<c16'):
 352                     raise ValueError('matrix must have complex128
 values')
 353
 354             return indx
 (Pdb) print mtx.data.dtype
 float64
 (Pdb) print mtx.data.dtype == np.dtype('<f8')
 False
 (Pdb) print mtx.data.dtype == np.dtype('>f8')
 True
 }}}

 I haven't checked the code to see if that is just a matter of correcting
 the type check...

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/1680>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list