mathcomp.assemble
Class ErrorCalculator

java.lang.Object
  mathcomp.monitor.DefaultReportable
      mathcomp.assemble.ErrorCalculator

All Implemented Interfaces:

Reportable

Direct Known Subclasses:

DGErrorCalculator, ErrorCalculatorCompare, ErrorCalculatorOverkill, ErrorCalculatorStep, MSErrorCalculator

public class ErrorCalculator
extends DefaultReportable

A class to calculate errors for a given FEM solution. If we are able to evaluate the solutions at any Node, the error calculation routine remains the same for any elements. If thats proper enough...?

Author:

ch.wust

Constructor Summary

ErrorCalculator()

ErrorCalculator(Assembler assembler)

Method Summary

Triangle	findTriangle(Node n)
double	getH1Error()
double	getL2Error() Returns the L2 error to the FEM solution x.
double	getMaxError() Gets the pointwise maximal error compared to the exact solution.
void	printSol()
void	report()
void	setAssembler(Assembler assembler)
void	setFineSolution(Vector fineSolution)
void	setSolution(Vector solution) Sets a solution to this ErrorCalculator.

Methods inherited from class mathcomp.monitor.DefaultReportable

getReport, setReport

Methods inherited from class java.lang.Object

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

Constructor Detail

ErrorCalculator

public ErrorCalculator(Assembler assembler)

ErrorCalculator

public ErrorCalculator()

Method Detail

setAssembler

public void setAssembler(Assembler assembler)

setSolution

public void setSolution(Vector solution)

Sets a solution to this ErrorCalculator.

Parameters:

solution - the solution to be set (adds the Dirichlet Nodes to that solution)

getMaxError

public double getMaxError()

Gets the pointwise maximal error compared to the exact solution. The points that are considered here are all the global dofs and all boundary nodes (we really would not need them, but I used the paintNodes for this, and they're in here. Remember that the getExactValue mathod has to be implemented correctly for this.

Parameters:

x - the solution vector

Returns:

getL2Error

public double getL2Error()

Returns the L2 error to the FEM solution x. Remember that the getExactValue mathod has to be implemented correctly for this.

Parameters:

x - the solution vector

Returns:

the L2 error

getH1Error

public double getH1Error()

report

public void report()

Specified by:

report in interface Reportable

Overrides:

report in class DefaultReportable

printSol

public void printSol()

findTriangle

public Triangle findTriangle(Node n)

setFineSolution

public void setFineSolution(Vector fineSolution)