Unfortunately, when my ConstraintMatrix tries to close, I trigger the
exception "Cycle in constraints detected!" (line 304 of
source/lac/constraint_matrix). After doing some digging, I think the
corners are getting being overconstrained, but I can't figure out why. I
don't have this problem in 2D, in 3D with Dirichlet boundary conditions
on the flat faces, or even in 3D with using the quarter_hyper_shell and
my own QuarterHyperShellBoundary written in the same manner as my chunk
boundary description. The first two I can understand working, but the
third case should be essentially the same as my case, which leads me to
believe the problem is in my grid or colorize function. I can view the
grid in paraview (switch to Dirichlet on the flat faces in code below),
and visually it looks as it should, and I don't think I have vertex/cell
numbering issues.
I've included a minimal code that will produce this error; a little bit
of the boundary functionality was left in for the visualization, though
it is unnecessary for producing the error. Any assistance in finding
where I'm going wrong would be greatly appreciated. Thanks in advance,
A good friend of mine, upon having been caught having written a
completely nonsensical sentence in a grant proposal, admitted that he
must have been smoking the carpet when he wrote it. The sentence must be
a literal translation of something you'd say in French, but it also
matches well what I must have been doing when I wrote the current code
that your testcase triggers: it can not possibly have worked except in
the situation that all faces of the cell are parallel to the axes.
In any case, I think I have fixed it in the current sources and it works
with your testcase (correctly, I may add). Can you please run it on your
real problem and see whether it works there too?
Best & thanks for the short testcase!
W.
------------------------------------------------------------------------
Wolfgang Bangerth email: [email protected]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
