Detailed Description

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>
struct ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >

Common parts of the implementation of Python boundary conditions and source terms.

The implementation of Python boundary conditions and source terms are almost the same, however, their interfaces differ slightly. This struct contains all common parts of the local assemblers.

Definition at line 34 of file BcAndStLocalAssemblerImpl.h.

#include <BcAndStLocalAssemblerImpl.h>

Collaboration diagram for ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >:

[legend]

Public Types
using	Traits

Public Member Functions
	BcAndStLocalAssemblerImpl (MeshLib::Element const &e, NumLib::GenericIntegrationMethod const &integration_method_, bool is_axially_symmetric, BcOrStData const &data)
void	assemble (std::size_t const boundary_element_id, NumLib::LocalToGlobalIndexMap const &dof_table_boundary, double const t, GlobalVector const &x, GlobalVector &b, GlobalMatrix *const Jac) const

Public Attributes
BcOrStData const &	bc_or_st_data
MeshLib::Element const &	element
NumLib::GenericIntegrationMethod const &	integration_method
std::vector< typename Traits::NsAndWeight > const	nss_and_weights

Private Types
using	ShapeMatrixPolicy = typename Traits::ShapeMatrixPolicy

Private Member Functions
std::array< double, 3 >	interpolateCoords (typename Traits::NsAndWeight const &ns_and_weight, NumLib::TemplateIsoparametric< ShapeFunction, ShapeMatrixPolicy > const &fe) const
void	assembleLocalRhs (double const flux, typename Traits::NsAndWeight const &ns_and_weight, Eigen::VectorXd &local_rhs) const
void	assembleLocalJacobian (std::vector< double > const &dFlux, typename Traits::NsAndWeight const &ns_and_weight, Eigen::MatrixXd &local_Jac) const

Member Typedef Documentation

◆ ShapeMatrixPolicy

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

using ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::ShapeMatrixPolicy = typename Traits::ShapeMatrixPolicy

private

Definition at line 41 of file BcAndStLocalAssemblerImpl.h.

◆ Traits

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

using ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::Traits

Initial value:

NsAndWeightsTraits<ShapeFunction, LowerOrderShapeFunction, GlobalDim>

ProcessLib::BoundaryConditionAndSourceTerm::Python::NsAndWeightsTraits

Collects common type aliases needed when working with NsAndWeight.

Definition NsAndWeight.h:125

Definition at line 37 of file BcAndStLocalAssemblerImpl.h.

Constructor & Destructor Documentation

◆ BcAndStLocalAssemblerImpl()

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::BcAndStLocalAssemblerImpl	(	MeshLib::Element const &	e,
		NumLib::GenericIntegrationMethod const &	integration_method_,
		bool	is_axially_symmetric,
		BcOrStData const &	data )

inline

Definition at line 44 of file BcAndStLocalAssemblerImpl.h.

        : bc_or_st_data(data),
          element(e),
          integration_method(integration_method_),
          nss_and_weights(
              computeNsAndWeights<ShapeFunction, LowerOrderShapeFunction,
                                  GlobalDim>(e, is_axially_symmetric,
                                             integration_method))
    {
    }

References bc_or_st_data, ProcessLib::BoundaryConditionAndSourceTerm::Python::computeNsAndWeights(), element, integration_method, and nss_and_weights.

Member Function Documentation

◆ assemble()

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

void ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::assemble	(	std::size_t const	boundary_element_id,
		NumLib::LocalToGlobalIndexMap const &	dof_table_boundary,
		double const	t,
		GlobalVector const &	x,
		GlobalVector &	b,
		GlobalMatrix *const	Jac ) const

inline

Assembles the local rhs and (possibly) Jacobian for this BC or ST and adds them to the global rhs vector and Jacobian matrix, respectively.

Definition at line 61 of file BcAndStLocalAssemblerImpl.h.

    {
        // Variables for input data
        Eigen::MatrixXd const primary_variables_mat = ProcessLib::
            BoundaryConditionAndSourceTerm::Python::collectDofsToMatrix(
                element, bc_or_st_data.bc_or_st_mesh.getID(),
                dof_table_boundary, x);
 
        // Variables for output data
        auto const& indices_this_component =
            dof_table_boundary(boundary_element_id,
                               bc_or_st_data.global_component_id)
                .rows;
        std::vector<GlobalIndexType> indices_all_components_for_Jac;
 
        auto const num_dof_this_component = indices_this_component.size();
 
        Eigen::VectorXd local_rhs =
            Eigen::VectorXd::Zero(num_dof_this_component);
        Eigen::MatrixXd local_Jac;
 
        if (Jac)
        {
            indices_all_components_for_Jac =
                NumLib::getIndices(boundary_element_id, dof_table_boundary);
 
            auto const num_dof_all_components =
                indices_all_components_for_Jac.size();
 
            local_Jac = Eigen::MatrixXd::Zero(num_dof_this_component,
                                              num_dof_all_components);
        }
 
        // Variables for temporary data
        auto const num_comp_total =
            dof_table_boundary.getNumberOfGlobalComponents();
        std::vector<double> prim_vars_data(num_comp_total);
        auto prim_vars = MathLib::toVector(prim_vars_data);
 
        // Variables for the integration loop
        unsigned const num_integration_points =
            integration_method.getNumberOfPoints();
        auto const fe = NumLib::createIsoparametricFiniteElement<
            ShapeFunction, ShapeMatrixPolicy>(element);
 
        for (unsigned ip = 0; ip < num_integration_points; ip++)
        {
            auto const& ns_and_weight = nss_and_weights[ip];
            auto const coords = interpolateCoords(ns_and_weight, fe);
 
            ProcessLib::BoundaryConditionAndSourceTerm::Python::interpolate(
                primary_variables_mat,
                bc_or_st_data.all_process_variables_for_this_process,
                nss_and_weights[ip], prim_vars);
 
            auto const [flag, flux, dFlux] =
                bc_or_st_data.getFlagAndFluxAndDFlux(t, coords, prim_vars_data);
 
            if (!flag)
            {
                // No flux value for this integration point. Skip assembly of
                // the entire element.
                return;
            }
 
            assembleLocalRhs(flux, ns_and_weight, local_rhs);
 
            if (static_cast<int>(dFlux.size()) != num_comp_total)
            {
                // This strict check is technically mandatory only if a Jacobian
                // is assembled. However, it is done as a consistency check also
                // for cases without Jacobian assembly.
                OGS_FATAL(
                    "The Python BC or ST must return the derivative of the "
                    "flux w.r.t. each component of each primary variable. "
                    "{:d} components expected. {:d} components returned from "
                    "Python."
                    "The expected number of components depends on multiple "
                    "factors. In the case of vectorial primary variables (such "
                    "as displacement), it will depend on the space dimension. "
                    "In the case of staggered coupling, the number of "
                    "components can be different for each staggered process "
                    "and different from monolithic coupling.",
                    num_comp_total, dFlux.size());
            }
 
            if (Jac)
            {
                assembleLocalJacobian(dFlux, ns_and_weight, local_Jac);
            }
        }
 
        b.add(indices_this_component, local_rhs);
 
        if (Jac)
        {
            MathLib::RowColumnIndices<GlobalIndexType> rci{
                indices_this_component, indices_all_components_for_Jac};
            Jac->add(rci, local_Jac);
        }
    }

References MathLib::EigenMatrix::add(), MathLib::EigenVector::add(), assembleLocalJacobian(), assembleLocalRhs(), bc_or_st_data, ProcessLib::BoundaryConditionAndSourceTerm::Python::collectDofsToMatrix(), NumLib::createIsoparametricFiniteElement(), element, NumLib::getIndices(), NumLib::LocalToGlobalIndexMap::getNumberOfGlobalComponents(), integration_method, ProcessLib::BoundaryConditionAndSourceTerm::Python::interpolate(), interpolateCoords(), nss_and_weights, OGS_FATAL, NumLib::LocalToGlobalIndexMap::size(), and MathLib::toVector().

◆ assembleLocalJacobian()

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

void ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::assembleLocalJacobian	(	std::vector< double > const &	dFlux,
		typename Traits::NsAndWeight const &	ns_and_weight,
		Eigen::MatrixXd &	local_Jac ) const

inlineprivate

Assembles the term \(N_\mathrm{this}^T \cdot \mathrm{dFlux}_c \cdot \mathrm{weight} \cdot N_c \) for each component \(c\) of each primary variable and adds it to the passed local Jacobian.

The shape functions \(N_\alpha\) of different primary variables \(\alpha\) might differ, e.g., for Taylor-Hood elements.

The index this refers to the primary variable to which this boundary condition or source term belongs. The index \(c\) (called other in the source code in the method body) runs over all components of all primary variables of the ProcessLib::Process to which this boundary condition or source term belongs.

Definition at line 205 of file BcAndStLocalAssemblerImpl.h.

    {
        auto const& pv_refs =
            bc_or_st_data.all_process_variables_for_this_process;
        auto const weight = ns_and_weight.weight();
 
        auto const this_shp_fct_order = bc_or_st_data.shape_function_order;
        auto const N_this = ns_and_weight.N(this_shp_fct_order);
 
        Eigen::Index left = 0;
 
        for (auto pv_ref : pv_refs)
        {
            auto const& pv = pv_ref.get();
            auto const other_num_comp = pv.getNumberOfGlobalComponents();
            auto const other_shp_fct_order = pv.getShapeFunctionOrder();
            auto const N_other = ns_and_weight.N(other_shp_fct_order);
            auto const width = N_other.size();
 
            for (decltype(+other_num_comp) other_comp = 0;
                 other_comp < other_num_comp;
                 ++other_comp)
            {
                // The assignment -= takes into account the sign convention
                // of 1st-order in time in ODE systems in OpenGeoSys.
                local_Jac.middleCols(left, width).noalias() -=
                    N_this.transpose() * (dFlux[other_comp] * weight) * N_other;
 
                left += width;
            }
        }
 
        assert(left == local_Jac.cols());
    }

References bc_or_st_data, ProcessLib::BoundaryConditionAndSourceTerm::Python::NsAndWeight< ShapeFunction, LowerOrderShapeFunction, ShapeMatrix, LowerOrderShapeMatrix >::N(), and ProcessLib::BoundaryConditionAndSourceTerm::Python::NsAndWeight< ShapeFunction, LowerOrderShapeFunction, ShapeMatrix, LowerOrderShapeMatrix >::weight().

Referenced by assemble().

◆ assembleLocalRhs()

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

void ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::assembleLocalRhs	(	double const	flux,
		typename Traits::NsAndWeight const &	ns_and_weight,
		Eigen::VectorXd &	local_rhs ) const

inlineprivate

Assembles the term \(N^T \cdot \mathrm{flux} \cdot \mathrm{weight} \) and adds it to the passed local rhs vector.

Definition at line 183 of file BcAndStLocalAssemblerImpl.h.

    {
        auto const w = ns_and_weight.weight();
        auto const N = ns_and_weight.N(bc_or_st_data.shape_function_order);
 
        local_rhs.noalias() += N.transpose() * (flux * w);
    }

References bc_or_st_data, ProcessLib::BoundaryConditionAndSourceTerm::Python::NsAndWeight< ShapeFunction, LowerOrderShapeFunction, ShapeMatrix, LowerOrderShapeMatrix >::N(), and ProcessLib::BoundaryConditionAndSourceTerm::Python::NsAndWeight< ShapeFunction, LowerOrderShapeFunction, ShapeMatrix, LowerOrderShapeMatrix >::weight().

Referenced by assemble().

◆ interpolateCoords()

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

std::array< double, 3 > ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::interpolateCoords	(	typename Traits::NsAndWeight const &	ns_and_weight,
		NumLib::TemplateIsoparametric< ShapeFunction, ShapeMatrixPolicy > const &	fe ) const

inlineprivate

Determines the coordinates that the point associated with the passed shape matrix by interpolating the element's node coordinates

Note: We use the shape function of higher order to interpolate coordinates.

Definition at line 172 of file BcAndStLocalAssemblerImpl.h.

    {
        auto const& N_higher = ns_and_weight.NHigherOrder();
        return fe.interpolateCoordinates(N_higher);
    }

References NumLib::TemplateIsoparametric< ShapeFunctionType_, ShapeMatrixTypes_ >::interpolateCoordinates(), and ProcessLib::BoundaryConditionAndSourceTerm::Python::NsAndWeight< ShapeFunction, LowerOrderShapeFunction, ShapeMatrix, LowerOrderShapeMatrix >::NHigherOrder().

Referenced by assemble().

Member Data Documentation

◆ bc_or_st_data

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

BcOrStData const& ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::bc_or_st_data

Definition at line 244 of file BcAndStLocalAssemblerImpl.h.

Referenced by BcAndStLocalAssemblerImpl(), assemble(), assembleLocalJacobian(), and assembleLocalRhs().

◆ element

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

MeshLib::Element const& ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::element

Definition at line 245 of file BcAndStLocalAssemblerImpl.h.

Referenced by BcAndStLocalAssemblerImpl(), and assemble().

◆ integration_method

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

NumLib::GenericIntegrationMethod const& ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::integration_method

Definition at line 247 of file BcAndStLocalAssemblerImpl.h.

Referenced by BcAndStLocalAssemblerImpl(), and assemble().

◆ nss_and_weights

template<typename BcOrStData, typename ShapeFunction, typename LowerOrderShapeFunction, int GlobalDim>

std::vector<typename Traits::NsAndWeight> const ProcessLib::BoundaryConditionAndSourceTerm::Python::BcAndStLocalAssemblerImpl< BcOrStData, ShapeFunction, LowerOrderShapeFunction, GlobalDim >::nss_and_weights

Definition at line 248 of file BcAndStLocalAssemblerImpl.h.

Referenced by BcAndStLocalAssemblerImpl(), and assemble().

The documentation for this struct was generated from the following file:

ProcessLib/BoundaryConditionAndSourceTerm/Python/Utils/BcAndStLocalAssemblerImpl.h

Detailed Description

Public Types

Public Member Functions

Public Attributes

Private Types

Private Member Functions

Member Typedef Documentation

◆ ShapeMatrixPolicy

◆ Traits

Constructor & Destructor Documentation

◆ BcAndStLocalAssemblerImpl()

Member Function Documentation

◆ assemble()

◆ assembleLocalJacobian()

◆ assembleLocalRhs()

◆ interpolateCoords()

Member Data Documentation

◆ bc_or_st_data

◆ element

◆ integration_method

◆ nss_and_weights