Skip to content

linearized scheme and D b.c. in outer loops like Uzawa

Claus-Justus Heine requested to merge bugfix/resurrect-linearized-scheme into master
  • make the linearized.hh scheme work again (including python bindings)
  • allow for setup() without argument to linearize around zero
  • replace rhs argument to solve with named argument rightHandSide which correctly handles the right hand side vector together with DirichletBC

Description

Given a scheme with dirichlet boundary conditions g then calling scheme.solve(target,rhs) always returns target=g an the boundary. Calling scheme(u,w)returnsw=u-gon the boundary so the correct behavior fir thesolvemethod is to returntarget=rhs+gon the boundary. This is now done when callingsolvewith the named argumentrightHandSide. The original version with rhs` argument is deprecated. The original behavior can be reproduced by calling

scheme.setConstraints(0,rhs)
scheme.solve(target=target,rightHandSide=rhs)
Edited by Andreas Dedner

Merge request reports