]> https://gitweb.dealii.org/ - dealii-svn.git/commitdiff
use covariant transformation
authorguido <guido@0785d39b-7218-0410-832d-ea1e28bc413d>
Fri, 31 Mar 2006 09:30:55 +0000 (09:30 +0000)
committerguido <guido@0785d39b-7218-0410-832d-ea1e28bc413d>
Fri, 31 Mar 2006 09:30:55 +0000 (09:30 +0000)
git-svn-id: https://svn.dealii.org/trunk@12794 0785d39b-7218-0410-832d-ea1e28bc413d

deal.II/deal.II/source/fe/fe_poly_tensor.cc
deal.II/deal.II/source/fe/fe_raviart_thomas.cc

index ba3131dfd1b9c05434584e51f1f98efc160e1a7d..45b67b5cc507640e3af612c09b2b86bd8c572050 100644 (file)
@@ -287,22 +287,86 @@ FE_PolyTensor<POLY,dim>::fill_fe_values (
       const unsigned int first = data.shape_function_to_row_table[i];
       
       if (flags & update_values)
-       for (unsigned int k=0; k<n_quad; ++k)
-         for (unsigned int d=0;d<dim;++d)
-         data.shape_values(first+d,k) = fe_data.shape_values[i][k][d];
+       {
+         switch(mapping_type)
+           {
+             case independent:
+             case independent_on_cartesian:
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_values(first+d,k) = fe_data.shape_values[i][k][d];
+               break;
+             case covariant:
+             case contravariant:
+               if (true)
+                 {
+                                                    // Use auxiliary vector for transformation
+                   std::vector<Tensor<1,dim> > shape_values (n_quad);
+                   if (mapping_type == covariant)
+                     mapping.transform_covariant(fe_data.shape_values[i], 0,
+                                                 shape_values, mapping_data);
+                   else
+                     mapping.transform_contravariant(fe_data.shape_values[i], 0,
+                                                     shape_values, mapping_data);
+                   
+                                                    // then copy over to target:
+                   for (unsigned int k=0; k<n_quad; ++k)
+                     for (unsigned int d=0; d<dim; ++d)
+                       data.shape_values(first+d,k) = shape_values[k][d];
+                 }
+               break;
+             default:
+               Assert(false, ExcNotImplemented());
+           }
+       }
       
       if (flags & update_gradients)
        {
          std::vector<Tensor<2,dim> > shape_grads1 (n_quad);
-         mapping.transform_covariant(fe_data.shape_grads[i], 0,
-                                     shape_grads1,
-                                     mapping_data);
-         for (unsigned int k=0; k<n_quad; ++k)
-           for (unsigned int d=0;d<dim;++d)
-             data.shape_gradients[first+d][k] = shape_grads1[k][d];
+         std::vector<Tensor<2,dim> > shape_grads2 (n_quad);
+         switch(mapping_type)
+           {
+             case independent:
+             case independent_on_cartesian:
+               mapping.transform_covariant(fe_data.shape_grads[i], 0,
+                                           shape_grads1,
+                                           mapping_data);
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             case covariant:
+               mapping.transform_covariant(fe_data.shape_grads[i], 0,
+                                           shape_grads1,
+                                           mapping_data);
+               for (unsigned int q=0; q<n_quad; ++q)
+                 shape_grads2[q] = transpose(shape_grads1[q]);
+                                                // do second transformation
+               mapping.transform_covariant(shape_grads2, 0, shape_grads1,
+                                           mapping_data);
+                                                // transpose back
+               for (unsigned int q=0; q<n_quad; ++q)
+                 shape_grads2[q] = transpose(shape_grads1[q]);
+               
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             case contravariant:
+               Assert(false, ExcNotImplemented());             
+               mapping.transform_covariant(fe_data.shape_grads[i], 0,
+                                           shape_grads1,
+                                           mapping_data);
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             default:
+               Assert(false, ExcNotImplemented());
+           }
        }
     }
-
+  
   const typename QProjector<dim>::DataSetDescriptor dsd;
   if (flags & update_second_derivatives)
     this->compute_2nd (mapping, cell, dsd.cell(),
@@ -352,21 +416,76 @@ FE_PolyTensor<POLY,dim>::fill_fe_face_values (
       const unsigned int first = data.shape_function_to_row_table[i];
       
       if (flags & update_values)
-        for (unsigned int k=0; k<n_quad; ++k)
-         for (unsigned int d=0;d<dim;++d)
-           data.shape_values(first+d,k) = fe_data.shape_values[i][k+offset][d];
+       {
+         switch(mapping_type)
+           {
+             case independent:
+             case independent_on_cartesian:
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_values(first+d,k) = fe_data.shape_values[i][k+offset][d];
+               break;
+             case covariant:
+             case contravariant:
+               if (true)
+                 {
+                                                    // Use auxiliary vector for transformation
+                   std::vector<Tensor<1,dim> > shape_values (n_quad);
+                   if (mapping_type == covariant)
+                     mapping.transform_covariant(fe_data.shape_values[i], offset,
+                                                 shape_values, mapping_data);
+                   else
+                     mapping.transform_contravariant(fe_data.shape_values[i], offset,
+                                                     shape_values, mapping_data);
+                   
+                                                    // then copy over to target:
+                   for (unsigned int k=0; k<n_quad; ++k)
+                     for (unsigned int d=0; d<dim; ++d)
+                       data.shape_values(first+d,k) = shape_values[k][d];
+                 }
+               break;
+             default:
+               Assert(false, ExcNotImplemented());
+           }
+       }
       
       if (flags & update_gradients)
        {
          std::vector<Tensor<2,dim> > shape_grads1 (n_quad);
-         mapping.transform_covariant(fe_data.shape_grads[i], offset,
-                                     shape_grads1, mapping_data);
-         for (unsigned int k=0; k<n_quad; ++k)
-           for (unsigned int d=0;d<dim;++d)
-             data.shape_gradients[first+d][k] = shape_grads1[k][d];
+         std::vector<Tensor<2,dim> > shape_grads2 (n_quad);
+         switch(mapping_type)
+           {
+             case independent:
+             case independent_on_cartesian:
+               mapping.transform_covariant(fe_data.shape_grads[i], offset,
+                                           shape_grads1, mapping_data);
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             case covariant:
+               mapping.transform_covariant(fe_data.shape_grads[i], offset,
+                                           shape_grads1,
+                                           mapping_data);
+               for (unsigned int q=0; q<n_quad; ++q)
+                 shape_grads2[q] = transpose(shape_grads1[q]);
+                                                // do second transformation
+               mapping.transform_covariant(shape_grads2, 0, shape_grads1,
+                                           mapping_data);
+                                                // transpose back
+               for (unsigned int q=0; q<n_quad; ++q)
+                 shape_grads2[q] = transpose(shape_grads1[q]);
+               
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             default:
+               Assert(false, ExcNotImplemented());
+           }
        }
     }
-
+  
   if (flags & update_second_derivatives)
     this->compute_2nd (mapping, cell, offset, mapping_data, fe_data, data);
 }
@@ -415,18 +534,73 @@ FE_PolyTensor<POLY,dim>::fill_fe_subface_values (
       const unsigned int first = data.shape_function_to_row_table[i];
       
       if (flags & update_values)
-        for (unsigned int k=0; k<n_quad; ++k)
-         for (unsigned int d=0;d<dim;++d)
-           data.shape_values(first+d,k) = fe_data.shape_values[i][k+offset][d];
+       {
+         switch(mapping_type)
+           {
+             case independent:
+             case independent_on_cartesian:
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_values(first+d,k) = fe_data.shape_values[i][k+offset][d];
+               break;
+             case covariant:
+             case contravariant:
+               if (true)
+                 {
+                                                    // Use auxiliary vector for transformation
+                   std::vector<Tensor<1,dim> > shape_values (n_quad);
+                   if (mapping_type == covariant)
+                     mapping.transform_covariant(fe_data.shape_values[i], offset,
+                                                 shape_values, mapping_data);
+                   else
+                     mapping.transform_contravariant(fe_data.shape_values[i], offset,
+                                                     shape_values, mapping_data);
+                   
+                                                    // then copy over to target:
+                   for (unsigned int k=0; k<n_quad; ++k)
+                     for (unsigned int d=0; d<dim; ++d)
+                       data.shape_values(first+d,k) = shape_values[k][d];
+                 }
+               break;
+             default:
+               Assert(false, ExcNotImplemented());
+           }
+       }
       
       if (flags & update_gradients)
        {
          std::vector<Tensor<2,dim> > shape_grads1 (n_quad);
-         mapping.transform_covariant(fe_data.shape_grads[i], offset,
-                                     shape_grads1, mapping_data);
-         for (unsigned int k=0; k<n_quad; ++k)
-           for (unsigned int d=0;d<dim;++d)
-             data.shape_gradients[first+d][k] = shape_grads1[k][d];
+         std::vector<Tensor<2,dim> > shape_grads2 (n_quad);
+         switch(mapping_type)
+           {
+             case independent:
+             case independent_on_cartesian:
+               mapping.transform_covariant(fe_data.shape_grads[i], offset,
+                                           shape_grads1, mapping_data);
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             case covariant:
+               mapping.transform_covariant(fe_data.shape_grads[i], offset,
+                                           shape_grads1,
+                                           mapping_data);
+               for (unsigned int q=0; q<n_quad; ++q)
+                 shape_grads2[q] = transpose(shape_grads1[q]);
+                                                // do second transformation
+               mapping.transform_covariant(shape_grads2, 0, shape_grads1,
+                                           mapping_data);
+                                                // transpose back
+               for (unsigned int q=0; q<n_quad; ++q)
+                 shape_grads2[q] = transpose(shape_grads1[q]);
+               
+               for (unsigned int k=0; k<n_quad; ++k)
+                 for (unsigned int d=0;d<dim;++d)
+                   data.shape_gradients[first+d][k] = shape_grads1[k][d];
+               break;
+             default:
+               Assert(false, ExcNotImplemented());
+           }
        }
     }
   
index 6182a5ea07f2b45cdcc9d69dbc3dd24e04852a21..bc6aaa047dbe6f529e1aafdd796de4fbee67fe37 100644 (file)
@@ -43,6 +43,7 @@ FE_RaviartThomas<dim>::FE_RaviartThomas (const unsigned int deg)
   Assert (dim >= 2, ExcImpossibleInDim(dim));
   const unsigned int n_dofs = this->dofs_per_cell;
   
+  this->mapping_type = this->covariant;
                                   // First, initialize the
                                   // generalized support points and
                                   // quadrature weights, since they

In the beginning the Universe was created. This has made a lot of people very angry and has been widely regarded as a bad move.

Douglas Adams


Typeset in Trocchi and Trocchi Bold Sans Serif.