[deal.II] Re: Periodic Boundary Condition

2017-01-06 Thread Daniel Arndt
Hamed, [...] > Considering for example the term [*M*+k*theta**A*]**U* in my right-hand > side, since the mass matrix(M) and laplace matrix(A) in the local level > have the size of *dofs_per_cell * dofs_per_cell* I need to have a local > solution vector to multiply with them. However, > FEVal

[deal.II] Tensor Output

2017-01-06 Thread Ehsan
Hello, I have two questions related to DataPostprocessor class, compute_derived_quantities_vector method and add_data_vector. I would appreciate it if someone help me and explain them to me. 1- As far as I tested and understood, when we write data_out.add_data_vector (solver.solution, Ur); we p

Re: [deal.II] Boundary ID in a parallel distributed triangulation & Neumann boundary conditions

2017-01-06 Thread Seyed Ali Mohseni
Thanks. Works like a charm. ;) *Hidden question: Why can you respond to my recent post, although I deleted it a while ago since I solved my problem already shortly afterwards?* Additional question: if ( std::abs(cell->face(i)->center()[0] - 1.0) < tol_boundary ) cell->face(i)->set_boundary

Re: [deal.II] Boundary ID in a parallel distributed triangulation & Neumann boundary conditions

2017-01-06 Thread Daniel Arndt
Seyed, *Hidden question: Why can you respond to my recent post, although I deleted > it a while ago since I solved my problem already shortly afterwards?* > If you are subscribed to this group, you get an email for each post and you can just answer that mail. Since you deleted the topic a new to

[deal.II] Re: Tensor Output

2017-01-06 Thread Daniel Arndt
Ehsan, 1- As far as I tested and understood, when we write data_out.add_data_vector > (solver.solution, Ur); we pass a solution vector which has > dof_handler.n_dofs() members to add_data_vector and then inside > compute_derived_quantities_vector method when I print uh the values > relate to t

Re: [deal.II] Boundary ID in a parallel distributed triangulation & Neumann boundary conditions

2017-01-06 Thread Seyed Ali Mohseni
For a cube discretized with 4 elements, I receive the following output: CELL NUMBER: 0_1:0 CELL CENTER X: 0 CELL CENTER Y: 0.25 CELL CENTER X: 0.25 CELL CENTER Y: 0 CELL NUMBER: 0_1:1 CELL CENTER X: 1 CELL CENTER Y: 0.25 CELL CENTER X: 0.75 CELL CENTER Y: 0 CELL NUMBER: 0_1:2 CELL CENTER X: 0 CE

[deal.II] Adding new finite element to the library

2017-01-06 Thread Eldar Khattatov
Hello, I was working on implementing new finite element space, which can roughly be described as RT + curl bubbles. I wrote the new FE class and the class for the underlying polynomials space. I wanted to incorporate these to my local installation of Deal.II in order to test it, so my question

[deal.II] Re: Adding new finite element to the library

2017-01-06 Thread Bruno Turcksin
Hi, you also need to add your .cc and .inst.in files to the CMakeLists.txt https://github.com/dealii/dealii/blob/master/source/fe/CMakeLists.txt Then you need to reconfigure deal, so that the .inst files can be created, and finally you can recompile. Best, Bruno On Friday, January 6, 2017 a

[deal.II] Re: Adding new finite element to the library

2017-01-06 Thread Denis Davydov
Hi Eldar, The main thing to add your new FE class is source/fe/CMakeLists.txt That should be all you need to do (plus, of course, don't forget to instantiate it for all dim/spacedim as done for other FE spaces). You should also consider adding tests and checking that things like shape values, g