ML_FiniteElements::AbstractProblem Class Reference

Abstract interface to define linear problems. More...

#include <ml_AbstractProblem.h>

Inheritance diagram for ML_FiniteElements::AbstractProblem:

[legend]List of all members.

Public Member Functions
virtual	~AbstractProblem ()
	Destructor.
virtual Epetra_RowMatrix &	A ()=0
	Returns a reference to the linear system matrix.
virtual Epetra_MultiVector &	RHS ()=0
	Returns a reference to the multi-vector of right-hand side.
virtual Epetra_MultiVector &	LHS ()=0
	Returns a reference to the multi-vector of starting solution.
virtual const AbstractGrid &	Grid () const =0
	Returns a reference to the grid object.
virtual const AbstractVariational &	Variational () const =0
	Returns a reference to the variational object.
virtual void	Compute ()=0
	Computes the linear system matrix, LHS and RHS.
virtual void	ComputeNorms (Epetra_MultiVector &RowMatrixField, int(*ExactSolution)(double, double, double, double *), const bool verbose=true, double *SolutionNorm=0, double *ExactNorm=0, double *DiffNorm=0)=0
	Computes the norm of computed solution, exact solution, and error.

---

Detailed Description

Abstract interface to define linear problems.

AbstractProblem defines a set of abstract interfaces, used to construct the linear system corresponding to the finite element discretization of a scalar PDE problem. Concrete implementation will require an AbstractGrid and an AbstractVariational object; the former is used to query for the grid elements, the latter to integrate the variational form over such elements. The role of AbstractProblem is to take the elemental matrices, given by AbstractVariational, and insert them into the global, distributed matrix (whose RowMatrixRowMap() is given by Grid().RowMap()).

Author:

Marzio Sala, SNL 9214.

Date:

Last updated on Apr-05.

---

Member Function Documentation

virtual void ML_FiniteElements::AbstractProblem::ComputeNorms (  Epetra_MultiVector &  RowMatrixField, int(*)(double, double, double, double *)  ExactSolution, const bool  verbose = true, double *  SolutionNorm = 0, double *  ExactNorm = 0, double *  DiffNorm = 0 )  [pure virtual]

Computes the norm of computed solution, exact solution, and error. Parameters: RowMatrixField  - (In) Multi-vector defined on Grid().RowMap() which contains the numerical solution. ExactSolution  - (In) Function defined as in the following example: ExactSolution(double x, double y, double x, double* sol) will contain the value of the solution in sol[0], the x-derivative in sol[1], the y-derivative in sol[2], and the z-derivative in sol[2]. verbose  - (In) If true, prints out the results. SolutionNorm  - (Out) a double array of size 3, which will contain the L2 norm, the semi-H1 norm and the H1-norm of the numerical solution. ExactNorm  - (Out) a double array of size 3, which will contain the L2 norm, the semi-H1 norm and the H1-norm of the exact solution. DiffNorm  - (Out) a double array of size 3, which will contain the L2 norm, the semi-H1 norm and the H1-norm of the error. Implemented in ML_FiniteElements::LinearProblem.

---

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

• ml_AbstractProblem.h

---