<div class="gmail_quote">On Fri, Mar 2, 2012 at 04:18, Manfred Gratt <span dir="ltr"><<a href="mailto:manfred.gratt@uibk.ac.at">manfred.gratt@uibk.ac.at</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
thank you for your quick response. The petsc-dev did not solve the
segfault. I checked what the difference from ex5 to my code was and
after I changed my code to ex5 the segfault was gone when I used
MatZeroEntries instead of destroying and creating a new Matrix for the
second solve. This seems to be logical but, what I do not understand
is why it works with destroying and creating a new Matrix on more than
one processor fine?</blockquote><div><br></div><div>I can't understand from this description exactly what works and what doesn't work for you. There shouldn't be a SEGV for any "reasonable" sequence of calls you make, so we should clarify the confusion and either fix the bug or make a better/earlier error message.</div>
</div><br><div>Is there a way you can modify ex5 (or another example, or send your own code) to show the problem?</div>