I assume so. My approach to ensure that is: I create a ConstraintMatrix called "Newton_constraints", initialize it as following:
newton_constraints.clear(); newton_constraints.reinit(solution_relevant_partitioning); DoFTools::make_hanging_node_constraints(dof_handler, newton_constraints); VectorTools::interpolate_boundary_values(dof_handler, 0, ZeroFunction<dim>( dim), newton_constraints); newton_constraints.close(); and then distribute the values like newton_constraints.distribute_local_to_global(cell_residual, local_dof_indices, local_residual); Does that do what I think, or does it do something else? Am Donnerstag, 5. Oktober 2017 17:38:31 UTC+2 schrieb mac: > > Are you ensuring that the entries in F(U) associated with the constrained > dofs are set to zero? As shown in > http://www.dealii.org/developer/doxygen/deal.II/step_15.html#MinimalSurfaceProblemcompute_residual > > > A > > On 5 Oct 2017, at 16:13, 'Maxi Miller' via deal.II User Group < > dea...@googlegroups.com <javascript:>> wrote: > > Yes, I forgot to clarify. I am calculating the l2-norm of the right hand > side, i.e. which contains F(U) (when staying in the terminology of > step-15). According to my knowledge that should go to zero, after that is > the whole goal of the Newton iteration (if my understanding of the method > is correct). Can it be related to (as mentioned) large value differences? > For small value differences the l2-norm goes below the threshold, for large > it does not. > > Am Donnerstag, 5. Oktober 2017 16:59:30 UTC+2 schrieb Wolfgang Bangerth: >> >> On 10/05/2017 08:52 AM, 'Maxi Miller' via deal.II User Group wrote: >> > Having a nonlinear, time-dependent heat equation, I tried to implement >> > it as in step-15 and in step-33. Everytime my l2-value goes below a >> > certain threshold, >> >> The l2 norm of what? The solution? The residual? >> >> >> > I make one step in time (or I refine the mesh). At >> > the beginning the L2-value goes down rather fast, but later it slows >> > down, and even stops going lower. >> >> Do you have theoretical evidence that it *should* decrease to zero? If, >> for example, you just look at the L2 norm of the solution, then there is >> no reason to believe that it should converge to zero -- it should just >> decrease to the L2 norm of the exact solution. >> >> Best >> W. >> >> -- >> ------------------------------------------------------------------------ >> Wolfgang Bangerth email: bang...@colostate.edu >> www: http://www.math.colostate.edu/~bangerth/ >> >> > > -- > The deal.II project is located at http://www.dealii.org/ > For mailing list/forum options, see > https://groups.google.com/d/forum/dealii?hl=en > --- > You received this message because you are subscribed to the Google Groups > "deal.II User Group" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to dealii+un...@googlegroups.com <javascript:>. > For more options, visit https://groups.google.com/d/optout. > > > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
