Page MenuHomec4science
Files F121847033

Element.cpp
No OneTemporary
Actions

File Metadata

Created
Mon, Jul 14, 09:23

Element.cpp
View Options

/* =================================================================================================
(c - GPLv3) T.W.J. de Geus (Tom) | tom@geus.me | www.geus.me | github.com/tdegeus/GooseFEM
================================================================================================= */
#ifndef GOOSEFEM_ELEMENT_CPP
#define GOOSEFEM_ELEMENT_CPP
// -------------------------------------------------------------------------------------------------
#include "Element.h"
// ======================================= GooseFEM::Element =======================================
namespace GooseFEM {
namespace Element {
// -------------------------------------------------------------------------------------------------
ArrD elementVector(const MatS &conn, const MatD &global)
{
// extract dimensions
size_t nelem = static_cast<size_t>(conn .rows());
size_t nne = static_cast<size_t>(conn .cols());
size_t ndim = static_cast<size_t>(global.cols());
// allocate output element vector
ArrD local({nelem, nne, ndim});
// extract element vector for global vector
#pragma omp parallel for
for ( auto e = 0 ; e < nelem ; ++e )
for ( auto m = 0 ; m < nne ; ++m )
for ( auto i = 0 ; i < ndim ; ++i )
local(e,m,i) = global(conn(e,m),i);
return local;
}
// -------------------------------------------------------------------------------------------------
MatD assembleElementVector(const MatS &conn, const ArrD &local)
{
// check input
assert( local.ndim() == 3 ); // shape of the matrix [nelem, nne, ndim]
// extract dimensions
size_t nelem = static_cast<size_t>(conn.rows()); // number of elements
size_t nne = static_cast<size_t>(conn.cols()); // number of nodes per element
size_t ndim = local.shape(2); // number of dimensions
size_t nnode = conn.maxCoeff()+1; // number of nodes
// check input
assert( local.shape(0) == nelem );
assert( local.shape(1) == nne );
// allocate/initialize global vector
MatD out(nnode, ndim);
out.setZero();
// temporarily disable parallelization by Eigen
Eigen::setNbThreads(1);
// assemble
#pragma omp parallel
{
// - allocate/initialize global vector, per thread
MatD global(nnode, ndim);
global.setZero();
// - assemble, per thread
#pragma omp for
for ( auto e = 0 ; e < nelem ; ++e )
for ( auto m = 0 ; m < nne ; ++m )
for ( auto i = 0 ; i < ndim ; ++i )
global(conn(e,m), i) += local(e,m,i);
// - reduce "global" per thread to total "out"
#pragma omp critical
out += global;
}
// automatic parallelization by Eigen
Eigen::setNbThreads(0);
return out;
}
// -------------------------------------------------------------------------------------------------
}} // namespace ...
// =================================================================================================
#endif

Event Timeline

c4science · Help