BoxLayoutData.H File Reference

#include "LayoutData.H"
#include "Interval.H"
#include "FArrayBox.H"
#include "DisjointBoxLayout.H"
#include "Copier.H"
#include "SPMD.H"
#include "memtrack.H"
#include "NamespaceHeader.H"
#include "NamespaceFooter.H"
#include "BoxLayoutDataI.H"

Include dependency graph for BoxLayoutData.H:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
class	DataFactory< T >
	Factory object to data members of a BoxLayoutData container. More...
class	DefaultDataFactory< T >
	Factory object to data members of a BoxLayoutData container. More...
class	FABAliasDataFactory
class	AliasDataFactory< T >
class	FABAliasFlBxDataFactory
class	FaceFabDataFactory
class	LDOperator< T >
class	BoxLayoutData< T >
	Data on a BoxLayout. More...
Functions
Real	norm (const BoxLayoutData< FArrayBox > &A, const Interval &interval, const int &p=2)
Variables
int	LinearizationTest

---

Function Documentation

Real norm	(	const BoxLayoutData< FArrayBox > &	A,
		const Interval &	interval,
		const int &	p = 2
	)

not actually L-p norms, since it doesn't take into account the dx of the system. A user can take that into account or not.

For p != 0, returns pth root of sum of pth powers over all points in all fabs and all components in the interval:

( sum [ |A[i][pt,var]|^p : FArrayBox A[i], point pt in A[i].box(), var in interval ] )^(1/p)

To turn into an L-p norm, one needs to multiply this by dx^(SpaceDim/p).

For p == 0, returns global max over all points in all fabs and all components in the interval:

max [ |A[i][pt,var]| : FArrayBox A[i], point pt in A[i].box(), var in interval ]

Some people don't like that this norm is not normalized based on number of points in A. Normalization is your problem.

Referenced by PetscSolver< T >::normInfinity(), RelaxSolver< T >::solve(), and BiCGStabSolver< T >::solve().

---

Variable Documentation

int LinearizationTest

Referenced by LDOperator< FArrayBox >::op().