belfem::fem::TimestepMatrices Class Reference

Container for element-level DENSE matrices in transient nonlinear FEM. More...

#include <cl_TimestepMatrices.hpp>

Public Member Functions
	TimestepMatrices ()=default
	~TimestepMatrices ()=default
void	initialize (const index_t aNumDofs)
	Initialize all matrices to given size.
void	reset ()
void	reset_flags ()
	Clear all flags ( unused; the IWG sets its flags once in the constructor ).
void	set_flag (const MatrixFlag aFlag)
	Set a flag after populating corresponding matrix.
bool	has_flag (const MatrixFlag aFlag) const
	Check if a matrix has been populated.
const Bitset< 8 > &	flags () const
	Direct access to flags bitset.
const Matrix< real > &	M () const
const Matrix< real > &	D () const
const Matrix< real > &	K () const
const Vector< real > &	f () const
const Matrix< real > &	dMdx_times_x () const
const Matrix< real > &	dMdx_times_h () const
const Matrix< real > &	dKdx_times_x () const
const Matrix< real > &	dfdx () const
const Matrix< real > &	J () const
const Matrix< real > &	dJdx () const
Matrix< real > &	M ()
Matrix< real > &	D ()
Matrix< real > &	K ()
Vector< real > &	f ()
Matrix< real > &	dMdx_times_x ()
Matrix< real > &	dMdx_times_h ()
Matrix< real > &	dKdx_times_x ()
Matrix< real > &	dfdx ()
void	assemble_J (const real adt)
void	assemble_dJdx (const real adt, const real aAlpha=1.0)
	Assemble the Newton correction dJ/dx.

Detailed Description

Container for element-level DENSE matrices in transient nonlinear FEM.

This class serves as interface between the physics-aware IWG and the time-stepping scheme (BDF). All matrices are DENSE and sized n_e × n_e where n_e is the number of element DOFs.

The global SPARSE Jacobian is assembled from these element contributions in a separate assembler class.

Workflow:

1. IWG constructor sets the flags for the matrices it produces
2. link_to_group() calls initialize( n_e ), which sizes the flagged matrices
3. per element: reset(), then the IWG fills M, K, f and the derivative blocks
4. assemble_dJdx() combines the flagged blocks with the BDF coefficients
5. Global assembler accumulates element J^e into sparse global Jacobian

Key design decisions:

• All matrices are DENSE (element-level, typically small: 4-27 DOFs)
• Derivatives are stored as contracted n_e × n_e matrices (dX/dx · x), not as n_e × n_e × n_e tensors. The IWG computes this contraction directly.
• Uses Bitset<8> to track which matrices are populated.

See also

IWG Module Usage Guide

Constructor & Destructor Documentation

TimestepMatrices()

belfem::fem::TimestepMatrices::TimestepMatrices ( )

default

~TimestepMatrices()

belfem::fem::TimestepMatrices::~TimestepMatrices ( )

default

Member Function Documentation

assemble_dJdx()

void belfem::fem::TimestepMatrices::assemble_dJdx	(	const real	adt,
		const real	aAlpha = 1.0 )

Assemble the Newton correction dJ/dx.

Parameters

adt	length of the current timestep
aAlpha	BDF coefficient of the current time level; scales the (dM/dx)·x term (Hairer & Wanner 1996, II.4). One for all schemes that scale M by one (BDF1, Crank-Nicolson, ...)

assemble_J()

void belfem::fem::TimestepMatrices::assemble_J

(

const real

adt

)

D() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::D ( )

inline

D() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::D ( ) const

inline

dfdx() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::dfdx ( )

inline

dfdx() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::dfdx ( ) const

inline

dJdx()

const Matrix< real > & belfem::fem::TimestepMatrices::dJdx ( ) const

inline

dKdx_times_x() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::dKdx_times_x ( )

inline

dKdx_times_x() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::dKdx_times_x ( ) const

inline

dMdx_times_h() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::dMdx_times_h ( )

inline

dMdx_times_h() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::dMdx_times_h ( ) const

inline

dMdx_times_x() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::dMdx_times_x ( )

inline

dMdx_times_x() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::dMdx_times_x ( ) const

inline

f() [1/2]

Vector< real > & belfem::fem::TimestepMatrices::f ( )

inline

f() [2/2]

const Vector< real > & belfem::fem::TimestepMatrices::f ( ) const

inline

flags()

const Bitset< 8 > & belfem::fem::TimestepMatrices::flags ( ) const

inline

Direct access to flags bitset.

has_flag()

bool belfem::fem::TimestepMatrices::has_flag ( const MatrixFlag aFlag ) const

inline

Check if a matrix has been populated.

initialize()

void belfem::fem::TimestepMatrices::initialize

(

const index_t

aNumDofs

)

Initialize all matrices to given size.

Parameters

aNumDofs

Number of element DOFs (typically 4-27 for common elements)

J()

const Matrix< real > & belfem::fem::TimestepMatrices::J ( ) const

inline

K() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::K ( )

inline

K() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::K ( ) const

inline

M() [1/2]

Matrix< real > & belfem::fem::TimestepMatrices::M ( )

inline

M() [2/2]

const Matrix< real > & belfem::fem::TimestepMatrices::M ( ) const

inline

reset()

void belfem::fem::TimestepMatrices::reset

(

)

reset_flags()

void belfem::fem::TimestepMatrices::reset_flags ( )

inline

Clear all flags ( unused; the IWG sets its flags once in the constructor ).

set_flag()

void belfem::fem::TimestepMatrices::set_flag ( const MatrixFlag aFlag )

inline

Set a flag after populating corresponding matrix.

---

The documentation for this class was generated from the following files:

• src/fem/iwg/cl_TimestepMatrices.hpp
• src/fem/iwg/cl_TimestepMatrices.cpp