Page MenuHomec4science

MatrixDiagonal.h
No OneTemporary

File Metadata

Created
Sun, Nov 10, 23:14

MatrixDiagonal.h

/**
Diagonal matrix.
\file MatrixDiagonal.h
\copyright Copyright 2017. Tom de Geus. All rights reserved.
\license This project is released under the GNU Public License (GPLv3).
*/
#ifndef GOOSEFEM_MATRIXDIAGONAL_H
#define GOOSEFEM_MATRIXDIAGONAL_H
#include "Element.h"
#include "Matrix.h"
#include "config.h"
namespace GooseFEM {
/**
CRTP base class for a partitioned matrix with tying.
*/
template <class D>
class MatrixDiagonalBase {
public:
/**
Underlying type.
*/
using derived_type = D;
private:
auto derived_cast() -> derived_type&
{
return *static_cast<derived_type*>(this);
}
auto derived_cast() const -> const derived_type&
{
return *static_cast<const derived_type*>(this);
}
public:
/**
Solve \f$ x = A^{-1} b \f$.
\param b nodevec [nelem, ndim].
\return x nodevec [nelem, ndim].
*/
array_type::tensor<double, 2> Solve(const array_type::tensor<double, 2>& b)
{
array_type::tensor<double, 2> x = xt::empty_like(b);
derived_cast().solve_nodevec_impl(b, x);
return x;
}
/**
Solve \f$ x = A^{-1} b \f$.
\param b dofval [ndof].
\return x dofval [ndof].
*/
array_type::tensor<double, 1> Solve(const array_type::tensor<double, 1>& b)
{
array_type::tensor<double, 1> x = xt::empty_like(b);
derived_cast().solve_dofval_impl(b, x);
return x;
}
/**
Same as Solve(const array_type::tensor<double, 2>&)
but writing to preallocated data.
\param b nodevec [nelem, ndim].
\param x nodevec overwritten [nelem, ndim].
*/
void solve(const array_type::tensor<double, 2>& b, array_type::tensor<double, 2>& x)
{
derived_cast().solve_nodevec_impl(b, x);
}
/**
Same as Solve(const array_type::tensor<double, 1>&)
but writing to preallocated data.
\param b nodevec [nelem, ndim].
\param x nodevec overwritten [nelem, ndim].
*/
void solve(const array_type::tensor<double, 1>& b, array_type::tensor<double, 1>& x)
{
derived_cast().solve_dofval_impl(b, x);
}
};
/**
Diagonal matrix.
Warning: assemble() ignores all off-diagonal terms.
See Vector() for bookkeeping definitions.
*/
class MatrixDiagonal : public MatrixBase<MatrixDiagonal>,
public MatrixDiagonalBase<MatrixDiagonal> {
private:
friend MatrixBase<MatrixDiagonal>;
friend MatrixDiagonalBase<MatrixDiagonal>;
public:
MatrixDiagonal() = default;
/**
Constructor.
\tparam C e.g. `array_type::tensor<size_t, 2>`
\tparam D e.g. `array_type::tensor<size_t, 2>`
\param conn connectivity [#nelem, #nne].
\param dofs DOFs per node [#nnode, #ndim].
*/
template <class C, class D>
MatrixDiagonal(const C& conn, const D& dofs)
{
m_conn = conn;
m_dofs = dofs;
m_nelem = m_conn.shape(0);
m_nne = m_conn.shape(1);
m_nnode = m_dofs.shape(0);
m_ndim = m_dofs.shape(1);
m_ndof = xt::amax(m_dofs)() + 1;
m_A = xt::empty<double>({m_ndof});
m_inv = xt::empty<double>({m_ndof});
GOOSEFEM_ASSERT(xt::amax(m_conn)() + 1 <= m_nnode);
GOOSEFEM_ASSERT(m_ndof <= m_nnode * m_ndim);
}
private:
template <class T>
void assemble_impl(const T& elemmat)
{
GOOSEFEM_ASSERT(xt::has_shape(elemmat, {m_nelem, m_nne * m_ndim, m_nne * m_ndim}));
GOOSEFEM_ASSERT(Element::isDiagonal(elemmat));
m_A.fill(0.0);
for (size_t e = 0; e < m_nelem; ++e) {
for (size_t m = 0; m < m_nne; ++m) {
for (size_t i = 0; i < m_ndim; ++i) {
m_A(m_dofs(m_conn(e, m), i)) += elemmat(e, m * m_ndim + i, m * m_ndim + i);
}
}
}
m_changed = true;
}
template <class T>
void todense_impl(T& ret) const
{
ret.fill(0.0);
#pragma omp parallel for
for (size_t d = 0; d < m_ndof; ++d) {
ret(d, d) = m_A(d);
}
}
public:
/**
Set all (diagonal) matrix components.
\param A The matrix [#ndof].
*/
void set(const array_type::tensor<double, 1>& A)
{
GOOSEFEM_ASSERT(A.size() == m_ndof);
std::copy(A.begin(), A.end(), m_A.begin());
m_changed = true;
}
/**
Return matrix as diagonal matrix.
\param [#ndof].
*/
const array_type::tensor<double, 1>& Todiagonal() const
{
return m_A;
}
private:
template <class T>
void dot_nodevec_impl(const T& x, T& b) const
{
GOOSEFEM_ASSERT(xt::has_shape(x, {m_nnode, m_ndim}));
GOOSEFEM_ASSERT(xt::has_shape(b, {m_nnode, m_ndim}));
#pragma omp parallel for
for (size_t m = 0; m < m_nnode; ++m) {
for (size_t i = 0; i < m_ndim; ++i) {
b(m, i) = m_A(m_dofs(m, i)) * x(m, i);
}
}
}
template <class T>
void dot_dofval_impl(const T& x, T& b) const
{
GOOSEFEM_ASSERT(x.size() == m_ndof);
GOOSEFEM_ASSERT(b.size() == m_ndof);
xt::noalias(b) = m_A * x;
}
private:
template <class T>
void solve_nodevec_impl(const T& b, T& x)
{
GOOSEFEM_ASSERT(xt::has_shape(b, {m_nnode, m_ndim}));
GOOSEFEM_ASSERT(xt::has_shape(x, {m_nnode, m_ndim}));
this->factorize();
#pragma omp parallel for
for (size_t m = 0; m < m_nnode; ++m) {
for (size_t i = 0; i < m_ndim; ++i) {
x(m, i) = m_inv(m_dofs(m, i)) * b(m, i);
}
}
}
template <class T>
void solve_dofval_impl(const T& b, T& x)
{
GOOSEFEM_ASSERT(b.size() == m_ndof);
GOOSEFEM_ASSERT(x.size() == m_ndof);
this->factorize();
xt::noalias(x) = m_inv * b;
}
private:
array_type::tensor<double, 1> m_A; ///< The matrix.
array_type::tensor<double, 1> m_inv; /// Inverse of the matrix.
/**
Compute inverse (automatically evaluated by "solve").
*/
void factorize()
{
if (!m_changed) {
return;
}
#pragma omp parallel for
for (size_t d = 0; d < m_ndof; ++d) {
m_inv(d) = 1.0 / m_A(d);
}
m_changed = false;
}
};
} // namespace GooseFEM
#endif

Event Timeline