When trying to calculate the errornorm on larger meshes I get the known error “UMFPACK V5.7.1 (Oct 10, 2014): ERROR: out of memory” and there seems to be no way to change the solver that errornorm uses. Is the only workaround to implement errornorm myself or is there a way to change the default solver in 2019.1.0?
Errornorm should not use a solver, as it uses interpolation and assembly:
https://bitbucket.org/fenics-project/dolfin/src/99f847d29589a4258ee3385c26f9bc0f4c47dc42/python/dolfin/fem/norms.py#lines-154
Do you have a minimal example?
Oh, I was a bit too quick to assume the error comes from a call to a solver! I get the error on line 263 in norms.py, where the higher order function space is created. I’ll make a minimal example.
You can change degree_rise Kwarg to something lower than 3.
Perfect, that fixed it!