This abstract class defining the interface for any type of element assembly rountines in the FEDDLib. More...

#include <AssembleFEBlock_decl.hpp>

Inheritance diagram for FEDD::AssembleFEBlock< SC, LO, GO, NO >:

Public Types
typedef SmallMatrix< SC >	SmallMatrix_Type

typedef Teuchos::RCP< SmallMatrix_Type >	SmallMatrixPtr_Type

typedef AssembleFEBlock< SC, LO, GO, NO >	AssembleFEBlock_Type

Public Types inherited from FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >
typedef SmallMatrix< default_sc >	SmallMatrix_Type

typedef Teuchos::RCP< SmallMatrix_Type >	SmallMatrixPtr_Type

typedef AssembleFE< default_sc, default_lo, default_go, default_no >	AssembleFE_Type

Public Member Functions
virtual void	assembleJacobian ()=0
	Assemble the element Jacobian matrix.

virtual void	assembleRHS ()=0
	Assemble the element right hand side vector.

virtual void	assembleJacobianBlock (LO i)=0
	Assembly Jacobian.

Public Member Functions inherited from FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >
SmallMatrixPtr_Type	getJacobian ()
	Get the currently assembled element Jacobian matrix.

SmallMatrixPtr_Type	getJacobianBlock (default_lo i)
	Get the currently assembled element Jacobian matrix.

vec_dbl_ptr_Type	getRHS ()
	Get the currently assembled right hand side vector.

virtual void	checkParameters ()
	Check the input parameters from the constructor and the ParameterList for completeness and consistency.

virtual void	updateParams (ParameterListPtr_Type params)
	Set or update the parameters read from the ParameterList.

virtual void	updateParameter (std::string type, double value)
	Update the parameter read from the ParameterList.

virtual void	advanceInTime (double dt)
	This function is called every time the FEDDLib proceeds from one to the next time step. The size of the time step will always be provided as input.

double	getTimeStep ()
	Get the time state of the object.

void	advanceNewtonStep ()
	This function is called every time the FEDDLib proceeds from one to the next newton step. The size of the time step will always be provided as input.

int	getNewtonStep ()
	Get the time state of the object.

void	updateSolution (vec_dbl_Type solution)
	Update the solution vector.

vec_dbl_ptr_Type	getSolution ()
	Get the current local solution vector.

void	preProcessing ()
	This function is called in the beginning of each Newton step before actually assmblying anything.

void	postProcessing ()
	This function is called at the end of each Newton step after updating the solution vector.

int	getDim ()
	Get the spatial dimension. (Typically 2 or 3)

vec2D_dbl_Type	getNodesRefConfig ()
	Return the coordnates of the finite element nodes.

void	addRHSFunc (RhsFunc_Type rhsFunc)

tuple_sd_vec_ptr_Type	getTupleElement ()
	Return vector of tupled with element based values. First column per tuple string with description, second column with corresponding value.

double	getTimeIncrement ()
	Returns the time increment. Required by AceGen implementation.

void	setGlobalElementID (default_go goID)

default_go	getGlobalElementID ()

virtual void	computeLocalconstOutputField ()
	E.g. In case of non-newtonian fluids the viscosity is not constant - Compute the viscosity for an element depending on the known velocity solution.

vec_dbl_Type	getLocalconstOutputField ()
	Obtain value of resulting postprocessing field at nodes/ inside an element.

Protected Member Functions
	AssembleFEBlock (int flag, vec2D_dbl_Type nodesRefConfig, ParameterListPtr_Type parameters, tuple_disk_vec_ptr_Type tuple)
	Constructor.

Protected Member Functions inherited from FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >
	AssembleFE (int flag, vec2D_dbl_Type nodesRefConfig, ParameterListPtr_Type parameters, tuple_disk_vec_ptr_Type tuple)
	Constructor.

Friends
class	AssembleFEFactory< SC, LO, GO, NO >

Additional Inherited Members
Protected Attributes inherited from FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >
SmallMatrixPtr_Type	jacobian_

SmallMatrixPtr_Type	jacobianBlock_

vec_dbl_ptr_Type	rhsVec_

RhsFunc_Type	rhsFunc_

int	dim_

tuple_disk_vec_ptr_Type	diskTuple_

tuple_sd_vec_ptr_Type	elementIntormation_

vec2D_dbl_Type	nodesRefConfig_

bool	timeProblem_

int	flag_

double	timeStep_

int	newtonStep_

ParameterListPtr_Type	paramsMaterial_

ParameterListPtr_Type	params_

vec_dbl_ptr_Type	solution_

double	timeIncrement_

default_go	globalElementID_

vec_dbl_Type	constOutputField_

Detailed Description

template<class SC = default_sc, class LO = default_lo, class GO = default_go, class NO = default_no>
class FEDD::AssembleFEBlock< SC, LO, GO, NO >

This abstract class defining the interface for any type of element assembly rountines in the FEDDLib.

Template Parameters

SC	The scalar type. So far, this is always double, but having it as a template parameter would allow flexibily, e.g., for using complex instead
LO	The local ordinal type. The is the index type for local indices
GO	The global ordinal type. The is the index type for global indices

Todo: This should actually be removed since the class should operate only on element level)

Template Parameters

NO	The Kokkos Node type. This would allow for performance portibility when using Kokkos. Currently, this is not used.

Any new assembly routine on element level should implemented following the interface provided in this class. During the setup of a specific boundary value problem one AssembleFEBlock object will be constructed using the AssembleFEBlockFactory for each finite element. This is can be understood roughly as follows:

for (int i=1; i<numElements; i++) {
    AssembleFEBlock assmeblyFe[i] = AssembleFEBlockFactory<>::build("problemType",flag,nodesRefConfig,params,tuple);
}

It is not possible to construct an AssembleFEBlock object without using the AssembleFEBlockFactory since the constructor is protected and hence not directly accessible.

Similar to constructing the AssembleFEBlock, all other member functions will be called automatically by the FEDDLib during the program flow. For instance, the assembly of the element Jacobian matrices will be performed:

for (int i=1; i<numElements; i++) {
    assmeblyFe[i].assembleJacobian();
    Matrix_Type elementJacobian[i] = assmeblyFe[i].getJacobian();
}

A specific implementation of a class derived from AssembleFEBlock can only interact with the FEDDLib by implementing the public member functions in AssembleFEBlock for

Construction
Assmebly of the Jacobian and right hand side
Getting the Jacobian and right hand side
Upating the solution
...

They will be automatically executed as the construction and assembly of the Jacobian; see above.

If additional public member functions are added, they will not be executed from the FEDDLib. Therefore, we only allow for adding additional protected or private functions.

Upon construction, the FEDDLib will provide some information, such as

The element flag
The coordinates of the finite element nodes
...

Additional parameters, such as material parameters, can provided through a Teuchos::ParameterList object which will contain all the parameters specified in the input file ABC.xml. The structure of the input file and, hence, of the resulting parameter list can be chosen freely depending on the specific implementation of an element assembly. The FEDDLib will take care of reading the parameters from the file and making them available to every AssembleFEBlock object.

Constructor & Destructor Documentation

◆ AssembleFEBlock()

template<class SC, class LO, class GO, class NO>

FEDD::AssembleFEBlock< SC, LO, GO, NO >::AssembleFEBlock	(	int	flag,
		vec2D_dbl_Type	nodesRefConfig,
		ParameterListPtr_Type	parameters,
		tuple_disk_vec_ptr_Type	tuple )

protected

Constructor.

Parameters

[in]	flag	Flag of element
[in]	nodesRefConfig	Nodes of element in reference configuration
[in]	params	Parameterlist for current problem
[in]	tuple	vector of element information tuples.

Element Numbering for triangular elements:

Here is the call graph for this function:

Member Function Documentation

◆ assembleJacobian()

template<class SC, class LO, class GO, class NO>

void FEDD::AssembleFEBlock< SC, LO, GO, NO >::assembleJacobian ( )

pure virtual

Assemble the element Jacobian matrix.

Assembly Jacobian.

Returns: the element Jacobian matrix

Implements FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >.

◆ assembleJacobianBlock()

template<class SC, class LO, class GO, class NO>

void FEDD::AssembleFEBlock< SC, LO, GO, NO >::assembleJacobianBlock ( LO i )

pure virtual

Assembly Jacobian.

Parameters

[in] &elementMatrix

Implements FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >.

◆ assembleRHS()

template<class SC = default_sc, class LO = default_lo, class GO = default_go, class NO = default_no>

virtual void FEDD::AssembleFEBlock< SC, LO, GO, NO >::assembleRHS ( )

pure virtual

Assemble the element right hand side vector.

Returns: the element right hand side vector

Implements FEDD::AssembleFE< default_sc, default_lo, default_go, default_no >.

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

feddlib/core/AceFemAssembly/AssembleFEBlock_decl.hpp
feddlib/core/AceFemAssembly/AssembleFEBlock_def.hpp

Public Types

Public Member Functions

Protected Member Functions

Friends

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ AssembleFEBlock()

Member Function Documentation

◆ assembleJacobian()

◆ assembleJacobianBlock()

◆ assembleRHS()