Hi Daniel,

The pysparse in our machine is updated to 1.1, and the errors of "A.find()"
go away. Thank you very much!

I ran "python setup.py test --examples --Trilinos" again. It ends with the
following:

"Ran 166 tests in 384.578s

 OK"

So I guess we are good. However, there are a lot of "Aztec status AZ_loss:
loss of precision" during the test. Is this normal?

Also, I sometimes get the following messages after the calculation is done:

0 - MPI_ERRHANDLER_FREE : MPI_Errhandler argument is not a valid
errorhandler
Special bit pattern %x in errhandler is incorrect. May indicate an
out-of-order argument or a deleted error handler
[0]  Aborting program !
[0] Aborting program!

Thank you very much for your help! I really appreciate it.

Regards,
Zhiwen

On Thu, May 13, 2010 at 12:23 AM, Daniel Wheeler
<[email protected]>wrote:

>
> Update to pysparse 1.1
> <http://sourceforge.net/projects/pysparse/files/>. find() was added in
> Oct 2008, pysparse 1.0.1 dates from May 2008. Remember that fipy now
> needs pysparse even when using trilinos as the solver.
>
> On Wed, May 12, 2010 at 11:23 PM, Zhiwen Liang <[email protected]> wrote:
> > Hi Jon,
> > Thanks a lot for your reply. The equations I am solving are very very
> > long... I am not sure if you really want to look at it. Basically, they
> are
> > coupled equations of second order and contain many cross terms like
> > $\frac{\partial^2 u}{\partial x \partial y}$. My colleagues are now
> trying
> > to reproduce the error with simpler equations. At the mean time, I tried
> > running the test for trilinos using the command you provided. And I also
> got
> > the same "A.find()" errors, many of them actually They all look similar
> like
> > this:
> > File
> > "/home/liangz/Installs/FiPy-2.1/examples/phase/impingement/mesh40x1.py",
> > line 189, in examples.phase.impingement.mesh40x1
> > Failed example:
> >     for i in range(steps):
> >         theta.updateOld()
> >         phase.updateOld()
> >         thetaEq.solve(theta, dt = timeStepDuration)
> >         phaseEq.solve(phase, dt = timeStepDuration)
> >         if __name__ == '__main__':
> >             phaseViewer.plot()
> >             thetaProductViewer.plot()
> > Exception raised:
> >     Traceback (most recent call last):
> >       File "/usr/local/lib/python2.6/doctest.py", line 1241, in __run
> >         compileflags, 1) in test.globs
> >       File "<doctest examples.phase.impingement.mesh40x1[41]>", line 4,
> in
> > <module>
> >         thetaEq.solve(theta, dt = timeStepDuration)
> >       File "/home/liangz/Installs/FiPy-2.1/fipy/terms/term.py", line 131,
> in
> > solve
> >         solver._solve()
> >       File
> > "/home/liangz/Installs/FiPy-2.1/fipy/solvers/trilinos/trilinosSolver.py",
> > line 120, in _solve
> >         self._buildGlobalMatrix()
> >       File
> > "/home/liangz/Installs/FiPy-2.1/fipy/solvers/trilinos/trilinosSolver.py",
> > line 93, in _buildGlobalMatrix
> >         values, irow, jcol = A.find()
> >     AttributeError: find
> > I hope it helps to debug.
> > As to Daniel's question, the version of pysparse is "1.0.1". The system
> is
> > Ubuntu, with kernel version of "2.6.24-27-generic". And we are using
> python
> > 2.6.4.
> > Let me know if you need more information.
> > Thanks a lot! I really appreciate your help!
> > Regards,
> > Zhiwen
> >
> > On Wed, May 12, 2010 at 4:20 PM, Jonathan Guyer <[email protected]> wrote:
> >>
> >>
> >> On May 12, 2010, at 3:33 PM, I wrote:
> >>
> >> > On May 12, 2010, at 2:43 PM, Zhiwen Liang wrote:
> >> >
> >> >> I am using Fipy 2.1 to solve some partial differential equations. I
> got
> >> >> the following errors when I tried to "sweep", which ends right at
> A.find()
> >> >> when it's trying to assemble the matrix. I wonder if there is a way
> to fix
> >> >> it. Thank you very much for any advice!
> >> >
> >> > I think we'll need to see more to know why this is happening. Can you
> >> > send me a minimal script that produces this?
> >>
> >> ..also, if you run
> >>
> >>   python setup.py test --examples --Trilinos
> >>
> >> do you get this error on any of our examples?
> >>
> >
> >
>
>
>
> --
> Daniel Wheeler
>
>
>

Reply via email to