Hello everyone, Sorry to revive/hijack this thread, but it seems to have the exact information/people that might be able to help.
My current goal is to perform some shape optimization with some h(p)-refinement. I am currently using Prof. Heltai's method described in this issue <https://github.com/dealii/dealii/issues/5506#issuecomment-352429820> summarized below to extract a MappingFEField from a known Triangulation and Manifold. This initial Triangulation can(must) be linear, and I would use a TransfiniteInterpolation to curve the initial grid. - have a DoFHandler with spacedim FE_Q/FE_Bernstein describing the *absolute* geometry - attach manifolds (as expensive as you wish) - call VectorTools::get_position_vector - create MappingFEField with the computed position vector - discard manifolds and only use the MappingFEField afterwards However, I would now like to refine the mesh. This means that the grid refinement process needs a Manifold to query the new points. Therefore, it seems like I would need to derive some kind of PolynomialManifold (as mentionned in previous post) from the Manifold class, which would take a reference to the MappingFEField in its constructor. I would then use MappingFEField to compute the necessary values in the overridden virtual function of the Manifold class. As a result, I would have a PolynomialManifold that would allow my triangulation to be refined using the current polynomial mapping of the element. This would go back and forth between the refining process and the deformation process. - When a deformation occurs, simply displace the Triangulation->vertices and the MappingFEField->euler_vector. The PolynomialManifold, having a reference to the mapping would automatically update. - When refinement occurs, extract a new MappingFEField using Prof. Heltai's method, construct a new PolynomialManifold, and assign it to the Triangulation. Does that sound like a feasible plan? Is my implementation description of the PolynomialManifold as simple as Prof. Heltai hinted at in the previous post? Best regards, Doug On Thursday, September 10, 2015 at 2:02:48 AM UTC-4, Praveen C wrote: > > Dear all > > I would like to use high order meshes for compressible flow computations > around objects like airfoils. I would like to use Gmsh for this since it is > improving its high order meshing features. There is a series of workshops > on high order CFD, see the next one here > > http://how4.cenaero.be > > and they also provide/recommend using Gmsh. E.g., in the Joukowski airfoil > test case > > http://how4.cenaero.be/content/bl1-laminar-joukowski-airfoil-re1000 > > they provide some python code which generates meshes in Gmsh format. If > you run Laminar.py it generates mesh with degree 4 quadrilaterals. In each > quad, 5x5 = 25 points are generated. > > To use this in deal.II, I could create a preprocessor > > 1. Read in high order mesh. I cannot use GridIn for this since it probably > does not read high order elements. > 2. Create a triangulation using create_triangulation > 2. Create a MappingQEulerian field > 3. Save triangulation to file using GridOut > 4. Save the MappingQEulerian field into another file > > Then in my application program, I will read in the mesh and deformation > fields. > > Is this possible to do ? > > Even if this were possible, what happens for parallel computations, if I > want to use parallel::distributed::Triangulation ? Mesh is ok, but the > deformation field has to be partitioned. > > Thanks > praveen > -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/fe9356db-d85a-4d77-a299-2a0f8df5f28a%40googlegroups.com.
