[Dune] ISTL solver causes segfault

Patrick Leidenberger patrick.leidenberger at psi.ch
Wed Dec 13 15:44:53 CET 2006


Hi Markus,

> Why do you use a 2x2 Matrix if you do not use the second row?
Our code uses a 4x4 matrix and should be able to work with different
polynomial degree. For p=1 I get only degrees of freedom in the edges,
for p=2 for edges and faces, ... So every row in the matrix represent
the base function to a topological entity and every column the test
function (We use Galerkins methode). If there is no base function for a
topological entity at the current polynomial order the respective matrix
entries are empty.

> This behaviour may not be optimal (at least in debugging mode there
> should be an exception), but how are we supposed to
> solve an indefinite matrix?
I don't know much about the processes working behind the dune interface,
but if I use the printmatrix function, the empty parts at the edges of
the matrix are 'ignored'. Isn't it possible that the solver behaves similar?

A question to your fix: What exact have you fixed? I get the same
segfault. What is the way to get dune in the debug mode?

Best regards Patrick




More information about the Dune mailing list