Line data    Source code

       1           0 : // Distributed under the MIT License.
       2             : // See LICENSE.txt for details.
       3             : 
       4             : #pragma once
       5             : 
       6             : #include <array>
       7             : #include <boost/functional/hash.hpp>
       8             : #include <iterator>
       9             : #include <limits>
      10             : #include <memory>
      11             : #include <pup.h>
      12             : #include <type_traits>
      13             : #include <unordered_map>
      14             : #include <utility>
      15             : 
      16             : #include "DataStructures/DataVector.hpp"
      17             : #include "DataStructures/Tags.hpp"
      18             : #include "DataStructures/Tensor/Tensor.hpp"
      19             : #include "Domain/Structure/Direction.hpp"
      20             : #include "Domain/Structure/DirectionMap.hpp"
      21             : #include "Domain/Structure/Element.hpp"
      22             : #include "Domain/Structure/ElementId.hpp"
      23             : #include "Domain/Structure/OrientationMap.hpp"
      24             : #include "Evolution/DiscontinuousGalerkin/Limiters/MinmodHelpers.hpp"
      25             : #include "Evolution/DiscontinuousGalerkin/Limiters/MinmodType.hpp"
      26             : #include "NumericalAlgorithms/LinearOperators/MeanValue.hpp"
      27             : #include "NumericalAlgorithms/Spectral/Mesh.hpp"
      28             : #include "Utilities/ErrorHandling/Assert.hpp"
      29             : #include "Utilities/Gsl.hpp"
      30             : #include "Utilities/Literals.hpp"
      31             : #include "Utilities/Numeric.hpp"
      32             : #include "Utilities/TMPL.hpp"
      33             : 
      34             : namespace Limiters::Minmod_detail {
      35             : 
      36             : // This function combines the evaluation of the troubled-cell indicator with the
      37             : // computation of the post-limiter reduced slopes. The returned bool indicates
      38             : // whether the slopes are to be reduced. The slopes themselves are returned by
      39             : // pointer.
      40             : //
      41             : // Note: This function is only made available in this header file to facilitate
      42             : // testing.
      43             : template <size_t VolumeDim>
      44             : bool minmod_limited_slopes(
      45             :     gsl::not_null<DataVector*> u_lin_buffer,
      46             :     gsl::not_null<BufferWrapper<VolumeDim>*> buffer,
      47             :     gsl::not_null<double*> u_mean,
      48             :     gsl::not_null<std::array<double, VolumeDim>*> u_limited_slopes,
      49             :     Limiters::MinmodType minmod_type, double tvb_constant, const DataVector& u,
      50             :     const Mesh<VolumeDim>& mesh, const Element<VolumeDim>& element,
      51             :     const std::array<double, VolumeDim>& element_size,
      52             :     const DirectionMap<VolumeDim, double>& effective_neighbor_means,
      53             :     const DirectionMap<VolumeDim, double>& effective_neighbor_sizes);
      54             : 
      55             : // Implements the minmod limiter for one Tensor<DataVector> at a time.
      56             : template <size_t VolumeDim, typename Tag, typename PackagedData>
      57             : bool minmod_impl(
      58             :     const gsl::not_null<DataVector*> u_lin_buffer,
      59             :     const gsl::not_null<BufferWrapper<VolumeDim>*> buffer,
      60             :     const gsl::not_null<typename Tag::type*> tensor,
      61             :     const Limiters::MinmodType minmod_type, const double tvb_constant,
      62             :     const Mesh<VolumeDim>& mesh, const Element<VolumeDim>& element,
      63             :     const tnsr::I<DataVector, VolumeDim, Frame::ElementLogical>& logical_coords,
      64             :     const std::array<double, VolumeDim>& element_size,
      65             :     const std::unordered_map<DirectionalId<VolumeDim>, PackagedData,
      66             :                              boost::hash<DirectionalId<VolumeDim>>>&
      67             :         neighbor_data) {
      68             :   // True if the mesh is linear-order in every direction
      69             :   const bool mesh_is_linear = (mesh.extents() == Index<VolumeDim>(2));
      70             :   const bool minmod_type_is_linear =
      71             :       (minmod_type != Limiters::MinmodType::LambdaPiN);
      72             :   const bool using_linear_limiter_on_non_linear_mesh =
      73             :       minmod_type_is_linear and not mesh_is_linear;
      74             : 
      75             :   // In each direction, average the size of all different neighbors in that
      76             :   // direction. Note that only the component of neighor_size that is normal
      77             :   // to the face is needed (and, therefore, computed). Note that this average
      78             :   // does not depend on the solution on the neighboring elements, so could be
      79             :   // precomputed outside of `limit_one_tensor`. Changing the code to
      80             :   // precompute the average may or may not be a measurable optimization.
      81             :   const auto effective_neighbor_sizes =
      82             :       compute_effective_neighbor_sizes(element, neighbor_data);
      83             : 
      84             :   bool some_component_was_limited = false;
      85             :   for (size_t i = 0; i < tensor->size(); ++i) {
      86             :     // In each direction, average the mean of the i'th tensor component over
      87             :     // all different neighbors in that direction. This produces one effective
      88             :     // neighbor per direction.
      89             :     const auto effective_neighbor_means =
      90             :         compute_effective_neighbor_means<Tag>(i, element, neighbor_data);
      91             : 
      92             :     DataVector& u = (*tensor)[i];
      93             :     double u_mean = std::numeric_limits<double>::signaling_NaN();
      94             :     std::array<double, VolumeDim> u_limited_slopes{};
      95             :     const bool reduce_slopes = minmod_limited_slopes(
      96             :         u_lin_buffer, buffer, make_not_null(&u_mean),
      97             :         make_not_null(&u_limited_slopes), minmod_type, tvb_constant, u, mesh,
      98             :         element, element_size, effective_neighbor_means,
      99             :         effective_neighbor_sizes);
     100             : 
     101             :     if (reduce_slopes or using_linear_limiter_on_non_linear_mesh) {
     102             :       u = u_mean;
     103             :       for (size_t d = 0; d < VolumeDim; ++d) {
     104             :         u += logical_coords.get(d) * gsl::at(u_limited_slopes, d);
     105             :       }
     106             :       some_component_was_limited = true;
     107             :     }
     108             :   }  // end for loop over tensor components
     109             : 
     110             :   return some_component_was_limited;
     111             : }
     112             : 
     113             : }  // namespace Limiters::Minmod_detail