I thank you both for your replies to my question. The idea to initially use a coarse triangulation and DoFHandler is very good an will definitely save time.
Best regards Benjamin Am Montag, 26. September 2016 07:59:17 UTC+2 schrieb Wolfgang Bangerth: > > On 09/25/2016 11:31 PM, Jean-Paul Pelteret wrote: > > Hi Benjamin, > > > > I'm afraid that I don't have any knowledge of the save/load > functionality in > > deal.II, but maybe someone who has experience with this part of the code > (one > > of the ASPECT users, perhaps?) may be willing to comment on this. Just > to > > clarify though - this is not a bug per se, but more a matter of > efficiency. > > Before loading the DoFHandler (thereby, presumably, defining a DoF > numbering), > > there's no way to register an FE with it except by the initialize or > > distribute_dof calls, both of which actually do the job of distributing > DoFs. > > Is my interpretation correct? > > Yes, that is correct. It is true that that incurs unnecessary work. I > don't > think anyone found that to be too much work in practice to bother. I agree > that it is conceptually unpleasant. > > (The reason why it will be hard to change this behavior internally is that > in > the DoFHandler class, we always assume that "having a FE set" and "DoFs > have > been distributed" are equivalent. This invariant would break if one were > to > add a function that only attaches an FE object without distributing DoFs. > > If you think the unnecessary work is too much, you can always create a > triangulation with a single cell, attach the DoF handler, distributed > DoFs, > and then clear/recreate the triangulation with more useful content.) > > Best > W. > > > > > On Tuesday, September 20, 2016 at 10:24:06 AM UTC+2, Benjamin Brands > wrote: > > > > Dear deal.ii Users, > > > > I have a question regarding the dof_handler.load () function. > > Before calling the load function for a dof_handler with assigned > > triangulation I have to call the function distribute_dofs with the > FE > > object as argument. > > By doing this DoF distribution data is created just to be overridden > by > > the subsequent call of dof_handler.load (archive). > > I see that one has to call either distribute_dofs (with FE object as > > argument) or alternatively initialize (with triangulation and FE > object as > > arguments) to implicitly store a pointer to the FE object, but it > also > > causes (in my opinion) superfluous work. > > I wonder whether there is an alternative approach to the described > one? > > > > Best regards > > Benjamin > > > > > > > > > > -- > > 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 [email protected] <javascript:> > > <mailto:[email protected] <javascript:>>. > > For more options, visit https://groups.google.com/d/optout. > > > -- > ------------------------------------------------------------------------ > Wolfgang Bangerth email: [email protected] > <javascript:> > 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 [email protected]. For more options, visit https://groups.google.com/d/optout.
