// check the complex case described in the hp paper by playing through all
// sorts of arrangements of finite elements on one coarse and one refined cell
+//
+// this code in particular tests some compensating code in
+// dof_tools.cc, where we have to make sure that we select a suitable
+// set of master dofs. this is mostly trivial in 2d and for most fe
+// combinations in 3d as well. the exceptions are that it doesn't work
+// as easily in 3d for the combinations Q4/Q3, Q5/Q3, and
+// Q5/Q4. Higher order finite elements in 3d will probably only
+// exacerbate the problem, but the code there appears to be robust.
char logname[] = "hp-crash_12/output";
fe.push_back (FE_Q<dim>(2));
fe.push_back (FE_Q<dim>(3));
fe.push_back (FE_Q<dim>(4));
+ fe.push_back (FE_Q<dim>(5));
hp::DoFHandler<dim> dof_handler(triangulation);