[Dune] Changes in Quadraturerules?

[Christian Engwer]

Dear Patrick,

> Can anybody explain me what's the effect of these changes on my code's
> results (using quadrature order 1 on tetrahedra, ALUGrid)?
> I observe different finite element matrices with 3 times slower convergence.

The difference is that the quadrature rules for order 1, 2 and 3 on
simplex 3d were fixed. Due to bugs in the quadrature rules we were
using order 4 rules, when rules of order 4 or lower were requested.

If you have slower convergence rates, I assume that you request the
wrong order of integration. Did you take all transformations etc. into
account?

Christian