> The problem is that deal.II is built as a collection of tools that know > nothing about how they (and the other tools) are used. In other words,
That makes sense. I have been playing with several fem libraries over the past few years and each one has its own flavor of doing the same thing. For example in libMesh, they perform the projection internally after refinement but as you rightly pointed out, unlike Deal, the Mesh in libMesh is aware of the dofs and the system is aware of the matrices/vectors in the discrete system. It's a little more tightly coupled in software and personally, I do like both and each method has its own pros and cons. Vijay On Thu, May 10, 2012 at 1:55 PM, Wolfgang Bangerth <[email protected]> wrote: > >> Just a thought: why is this not the default behavior when doing AMR ? >> Perhaps a user flag to control whether the transfer should be done >> would be a nicer API. > > > The problem is that deal.II is built as a collection of tools that know > nothing about how they (and the other tools) are used. In other words, > whenever a triangulation is refined, it doesn't know if and what DoFHandlers > are attached to it, and the DoFHandler have no knowledge which vectors exist > that store solution values (and not, for example, random data or right hand > side vectors, both of which one would have to treat differently). > > Best > > W. > > ------------------------------------------------------------------------ > Wolfgang Bangerth email: [email protected] > www: http://www.math.tamu.edu/~bangerth/ > > _______________________________________________ > dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
