starfish.core.solver

Class Solver

java.lang.Object

starfish.core.solver.Solver

Direct Known Subclasses:

ConstantEF, PotentialSolver

public abstract class Solver
extends java.lang.Object

Finite volume Matrix solver for Ax=b using the Gauss-Seidel method (for now)

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	Solver.MeshData
static interface	Solver.NL_Eval

Field Summary

Fields

Modifier and Type	Field and Description
double	den0
double	kTe0
protected int	lin_max_it
protected double	lin_tol
Solver.MeshData[]	mesh_data
protected int	nl_max_it
protected double	nl_tol
double	phi0

Constructor Summary

Constructors

Constructor and Description
Solver()

Method Summary

Modifier and Type	Method and Description
protected double	calculateResidue(Matrix A, double[] x, double[] b) returns L2 norm of R=|Ax-b|
protected void	computeGradient(double[] x, double[] gi, double[] gj, Solver.MeshData md, double scale)
void	exit()
void	init()
int	solveLinearGS(Solver.MeshData[] mesh_data) solves Ax=b for x using the GS method
protected int	solveLinearMultigrid(Matrix A, double[] x, double[] b) solves Ax=b for x using the Multigrid method
protected int	solveLinearPCG(Solver.MeshData[] md) solves Ax=b for x using the PCG metho
protected int	solveLU(Solver.MeshData[] mesh_data)
int	solveNonLinearNR(Solver.MeshData[] md, Solver.NL_Eval nl_eval) solves the nonlinear problem Ax-b=0 using the Newtons method This method is based on the algorithm on page 614 in Numerical Analysis F(x)=Ax-b J(x)=dF/dx=A-db/dx=A-P solve Jy=F update x=x-y also b=b0+b(x) where b0 is constant
abstract void	update() updates potential and calculates new electric field
abstract void	updateGradientField()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

lin_max_it

protected int lin_max_it

lin_tol

protected double lin_tol

nl_max_it

protected int nl_max_it

nl_tol

protected double nl_tol

kTe0

public double kTe0

den0

public double den0

phi0

public double phi0

mesh_data

public Solver.MeshData[] mesh_data

Constructor Detail

Solver

public Solver()

Method Detail

init

public void init()

exit

public void exit()

update

public abstract void update()

updates potential and calculates new electric field

solveNonLinearNR

public int solveNonLinearNR(Solver.MeshData[] md,
                            Solver.NL_Eval nl_eval)

solves the nonlinear problem Ax-b=0 using the Newtons method This method is based on the algorithm on page 614 in Numerical Analysis F(x)=Ax-b J(x)=dF/dx=A-db/dx=A-P solve Jy=F update x=x-y also b=b0+b(x) where b0 is constant

Parameters:

md -

nl_eval -

Returns:

solveLinearGS

public int solveLinearGS(Solver.MeshData[] mesh_data)

solves Ax=b for x using the GS method

Parameters:

mesh_data -

Returns:

solveLinearMultigrid

protected int solveLinearMultigrid(Matrix A,
                                   double[] x,
                                   double[] b)

solves Ax=b for x using the Multigrid method

Parameters:

A -

x -

b -

Returns:

solveLU

protected int solveLU(Solver.MeshData[] mesh_data)

Parameters:

mesh_data -

Returns:

solveLinearPCG

protected int solveLinearPCG(Solver.MeshData[] md)

solves Ax=b for x using the PCG metho

Parameters:

md -

Returns:

d

calculateResidue

protected double calculateResidue(Matrix A,
                                  double[] x,
                                  double[] b)

returns L2 norm of R=|Ax-b|

Parameters:

A -

x -

b -

Returns:

updateGradientField

public abstract void updateGradientField()

computeGradient

protected void computeGradient(double[] x,
                               double[] gi,
                               double[] gj,
                               Solver.MeshData md,
                               double scale)

Parameters:

x -

gi -

gj -

md -

scale -