BackwardEulerIvpOdeSolver Class Reference

#include <BackwardEulerIvpOdeSolver.hpp>

Collaboration diagram for BackwardEulerIvpOdeSolver:

Public Member Functions
	BackwardEulerIvpOdeSolver (unsigned sizeOfOdeSystem)
	~BackwardEulerIvpOdeSolver ()
void	SetEpsilonForNumericalJacobian (double epsilon)
void	ForceUseOfNumericalJacobian ()
unsigned	GetSystemSize () const
Protected Member Functions
void	CalculateNextYValue (AbstractOdeSystem *pAbstractOdeSystem, double timeStep, double time, std::vector< double > &rCurrentYValues, std::vector< double > &rNextYValues)
Private Member Functions
template<class Archive >
void	serialize (Archive &archive, const unsigned int version)
void	ComputeResidual (AbstractOdeSystem *pAbstractOdeSystem, double timeStep, double time, std::vector< double > &rCurrentYValues, std::vector< double > &rCurrentGuess)
void	ComputeJacobian (AbstractOdeSystem *pAbstractOdeSystem, double timeStep, double time, std::vector< double > &rCurrentYValues, std::vector< double > &rCurrentGuess)
void	SolveLinearSystem ()
double	ComputeNorm (double *pVector)
void	ComputeNumericalJacobian (AbstractOdeSystem *pAbstractOdeSystem, double timeStep, double time, std::vector< double > &rCurrentYValues, std::vector< double > &rCurrentGuess)
Private Attributes
unsigned	mSizeOfOdeSystem
double	mNumericalJacobianEpsilon
bool	mForceUseOfNumericalJacobian
double *	mResidual
double **	mJacobian
double *	mUpdate
Friends
class	boost::serialization::access

Detailed Description

A concrete one step ODE solver class that employs the backward Euler method. This numerical method is implicit and hence unconditionally stable.

Definition at line 44 of file BackwardEulerIvpOdeSolver.hpp.

Constructor & Destructor Documentation

BackwardEulerIvpOdeSolver::BackwardEulerIvpOdeSolver ( unsigned sizeOfOdeSystem )

Constructor.

Parameters:

sizeOfOdeSystem

the number of state variables in the ODE system

Definition at line 203 of file BackwardEulerIvpOdeSolver.cpp.

References mForceUseOfNumericalJacobian, mJacobian, mNumericalJacobianEpsilon, mResidual, mSizeOfOdeSystem, and mUpdate.

BackwardEulerIvpOdeSolver::~BackwardEulerIvpOdeSolver ( )

Destructor.

Definition at line 222 of file BackwardEulerIvpOdeSolver.cpp.

References mJacobian, mResidual, mSizeOfOdeSystem, and mUpdate.

Member Function Documentation

void BackwardEulerIvpOdeSolver::CalculateNextYValue	(	AbstractOdeSystem *	pAbstractOdeSystem,
		double	timeStep,
		double	time,
		std::vector< double > &	rCurrentYValues,
		std::vector< double > &	rNextYValues
	)			`[protected, virtual]`

Calculate the solution to the ODE system at the next timestep.

A usage example: BackwardEulerIvpOdeSolver mySolver; OdeSolution solution = mySolver.Solve(pMyOdeSystem, yInit, StartTime, EndTime, TimeStep, SamplingTime);

Parameters:

	pAbstractOdeSystem	the ODE system to solve
	timeStep	dt
	time	the current time
	rCurrentYValues	the current (initial) state
	rNextYValues	the state at the next timestep

Implements AbstractOneStepIvpOdeSolver.

Definition at line 160 of file BackwardEulerIvpOdeSolver.cpp.

References ComputeJacobian(), ComputeNorm(), ComputeResidual(), AbstractOdeSystem::GetNumberOfStateVariables(), mSizeOfOdeSystem, mUpdate, and SolveLinearSystem().

void BackwardEulerIvpOdeSolver::ComputeJacobian	(	AbstractOdeSystem *	pAbstractOdeSystem,
		double	timeStep,
		double	time,
		std::vector< double > &	rCurrentYValues,
		std::vector< double > &	rCurrentGuess
	)			`[private]`

Compute the Jacobian of the ODE system.

Parameters:

	pAbstractOdeSystem	the ODE system to solve
	timeStep	dt
	time	the current time
	rCurrentYValues	the current (initial) state
	rCurrentGuess	current guess for the state at the next timestep

Definition at line 47 of file BackwardEulerIvpOdeSolver.cpp.

References AbstractOdeSystemWithAnalyticJacobian::AnalyticJacobian(), ComputeNumericalJacobian(), AbstractOdeSystem::GetUseAnalyticJacobian(), mForceUseOfNumericalJacobian, mJacobian, and mSizeOfOdeSystem.

Referenced by CalculateNextYValue().

double BackwardEulerIvpOdeSolver::ComputeNorm ( double * pVector ) [private]

Compute the infinity/maximum norm of a vector. Used by the method CalculateNextYValue.

Parameters:

pVector

a pointer to a vector

Returns:: the vector's norm.

Definition at line 105 of file BackwardEulerIvpOdeSolver.cpp.

References mSizeOfOdeSystem.

Referenced by CalculateNextYValue().

void BackwardEulerIvpOdeSolver::ComputeNumericalJacobian	(	AbstractOdeSystem *	pAbstractOdeSystem,
		double	timeStep,
		double	time,
		std::vector< double > &	rCurrentYValues,
		std::vector< double > &	rCurrentGuess
	)			`[private]`

Compute the Jacobian of the ODE system numerically.

Parameters:

	pAbstractOdeSystem	the ODE system to solve
	timeStep	dt
	time	the current time
	rCurrentYValues	the current (initial) state
	rCurrentGuess	current guess for the state at the next timestep

Definition at line 118 of file BackwardEulerIvpOdeSolver.cpp.

References ComputeResidual(), mJacobian, mNumericalJacobianEpsilon, mResidual, and mSizeOfOdeSystem.

Referenced by ComputeJacobian().

void BackwardEulerIvpOdeSolver::ComputeResidual	(	AbstractOdeSystem *	pAbstractOdeSystem,
		double	timeStep,
		double	time,
		std::vector< double > &	rCurrentYValues,
		std::vector< double > &	rCurrentGuess
	)			`[private]`

Compute the current residual.

Parameters:

	pAbstractOdeSystem	the ODE system to solve
	timeStep	dt
	time	the current time
	rCurrentYValues	the current (initial) state
	rCurrentGuess	current guess for the state at the next timestep

Definition at line 33 of file BackwardEulerIvpOdeSolver.cpp.

References AbstractOdeSystem::EvaluateYDerivatives(), mResidual, and mSizeOfOdeSystem.

Referenced by CalculateNextYValue(), and ComputeNumericalJacobian().

void BackwardEulerIvpOdeSolver::ForceUseOfNumericalJacobian ( )

Force the solver to use the numerical Jacobian even if the ODE system object provides an analytical Jacobian.

Definition at line 241 of file BackwardEulerIvpOdeSolver.cpp.

References mForceUseOfNumericalJacobian.

unsigned BackwardEulerIvpOdeSolver::GetSystemSize ( ) const [inline]

Public method used in archiving.

Returns:: the size of the system

Definition at line 208 of file BackwardEulerIvpOdeSolver.hpp.

References mSizeOfOdeSystem.

template<class Archive >

void BackwardEulerIvpOdeSolver::serialize	(	Archive &	archive,
		const unsigned int	version
	)			`[inline, private]`

Archive the abstract IVP Solver, never used directly - boost uses this.

Parameters:

	archive	the archive
	version	the current version of this class

Reimplemented from AbstractOneStepIvpOdeSolver.

Definition at line 56 of file BackwardEulerIvpOdeSolver.hpp.

References mForceUseOfNumericalJacobian, and mNumericalJacobianEpsilon.

void BackwardEulerIvpOdeSolver::SetEpsilonForNumericalJacobian ( double epsilon )

Set the epsilon to use in calculating the numerical Jacobian of the ODE system.

Parameters:

epsilon

Definition at line 235 of file BackwardEulerIvpOdeSolver.cpp.

References mNumericalJacobianEpsilon.

void BackwardEulerIvpOdeSolver::SolveLinearSystem ( ) [private]

Solve a linear system of equations to update the current guess for the solution to the ODE system at the next timestep. Used by the method CalculateNextYValue.

Definition at line 78 of file BackwardEulerIvpOdeSolver.cpp.

References mJacobian, mResidual, mSizeOfOdeSystem, and mUpdate.

Referenced by CalculateNextYValue().