we had a bug in the functions transforming a point from real
coordinates to reference coordinates. What happened in there was that
a Newton method would be run for 10 steps or until the residual would
reach a tiny value. Convergence of the Newton method was never tested.
In the bug fix, the code throws an exception if the residual after 10
steps is above the threshold, which I somewhat arbitrarily set to
10^-8 times cell diameter. While I expect this to be sufficient for
all applications, please let me know if an adjustment is needed.
Is this the bug Andrea and Seba found?
There seem to be a number of tests for which the loop doesn't converge
in the maximal number of iterations. Do you know what's happening there?
Best
W.
------------------------------------------------------------------------
Wolfgang Bangerth email: [email protected]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
