class to solve a linear complementarity problem More...

Inheritance diagram for MBSim::LinearComplementarityProblem:

Public Types
enum	LCPSolvingStrategy { Standard , ReformulatedStandard , ReformulatedNewtonOnly , ReformulatedFixpointOnly , LemkeOnly }

enum	JacobianType { Numerical , LCPSpecial }

enum	CriteriaType { Global , Local }

Public Member Functions
	LinearComplementarityProblem (const fmatvec::SymMat &M_, const fmatvec::Vec &q_, const LCPSolvingStrategy &strategy_=Standard, const JacobianType &jacobianType_=LCPSpecial)
	solves a linear complementarity problem (w = M z + q) More...

void	setSystem (const fmatvec::SymMat &M_, const fmatvec::Vec &q_)

void	setStrategy (const LCPSolvingStrategy &strategy_)

void	setJacobianType (const JacobianType &jacobianType_)

void	setNewtonCriteriaFunction (CriteriaFunction *criteriaFunction_)

void	setFixpointCriteriaFunction (CriteriaFunction *criteriaFunction_)

fmatvec::Vec	solve (const fmatvec::Vec &initialSolution=fmatvec::Vec(0, fmatvec::NONINIT))

Static Public Member Functions
static double	computeMediumEigVal (const fmatvec::SqrMat &M)
	compute the medium eigenvalue of a matrix (for guessing a initial solution for iterative schemes)

static fmatvec::Vec	createInitialSolution (const fmatvec::SymMat &M, const fmatvec::Vec &q, double mediumEigVal=0)
	creates an initial solution for the iterative schemes that use a reformulated system

static fmatvec::Vec	createInitialSolution (const fmatvec::SqrMat &M, const fmatvec::Vec &q, double mediumEigVal=0)
	creates an initial solution for the iterative schemes that use a reformulated system

Protected Member Functions
void	useNewton (fmatvec::Vec &solution, bool &solved)
	change the incoming solution vector using the fixpoint scheme

void	useFixpoint (fmatvec::Vec &solution, bool &solved)
	change the incoming solution vector using the fixpoint scheme

Protected Attributes
fmatvec::SqrMat	M
	linear coupling matrix of the LCP

fmatvec::Vec	q
	constant vector of the LCP

LCPSolvingStrategy	strategy
	algorithm strategy to solve the LCP

LCPSolvingStrategy	oldStrategy
	save old Strategy for switching...

double	mediumEigenValue
	parameter for finding the start solution for the reformulated system

JacobianType	jacobianType
	jacobian function for the solving with thew newton algorithm in the reformulated case

LemkeAlgorithm	lemkeSolver
	LemkeSolver for the direct solution of the LCP.

MBSim::MultiDimensionalNewtonMethod *	newtonSolver
	NewtonSolver for reformulated system.

LCPNewtonReformulationFunction *	newtonFunction
	reformulated LCP suited for a Newton Solver

NewtonJacobianFunction *	jacobianFunction
	Jacobian Function for the reformulated LCP.

CriteriaFunction *	criteriaNewton
	criteria function for Newton solver

MultiDimensionalFixpointSolver *	fixpointSolver
	FixpointSolver for reformulated system.

LCPFixpointReformulationFunction *	fixpointFunction
	reformulated LCP suited for a FixpointSolver

CriteriaFunction *	criteriaFixedpoint
	criteria function for Fixedpoint solver

Detailed Description

class to solve a linear complementarity problem

Todo:: : find a better concept for the properties of the used solving algorithms (maybe single "option" - classes)

Constructor & Destructor Documentation

MBSim::LinearComplementarityProblem::LinearComplementarityProblem	(	const fmatvec::SymMat &	M_,
		const fmatvec::Vec &	q_,
		const LCPSolvingStrategy &	strategy_ = `Standard`,
		const JacobianType &	jacobianType_ = `LCPSpecial`
	)

solves a linear complementarity problem (w = M z + q)

Parameters

M	linear coupling matrix of the LCP
q	constant vector of the LCP
LCPSolvingStrategy	algorithm strategy to solve the LCP
mediumEigVal	parameter for finding the start solution for the reformulated system

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

/mbsim-env/mbsim/kernel/mbsim/numerics/linear_complementarity_problem/linear_complementarity_problem.h
/mbsim-env/mbsim/kernel/mbsim/numerics/linear_complementarity_problem/linear_complementarity_problem.cc