boundparcongrad Class Reference

#include <boundparcongrad.h>

Public Member Functions
void	add_bound (double buff, double v)
void	add_master_buff (double buff, double cv, long nd)
	boundparcongrad (int np, int mr, long nd, long mes)
void	ilu_mat_petsc (gmatrix gm, FILE out)
void	initialize_diag_prec (gmatrix gm, double precvec, FILE *out)
void	initiate (selnodes selnodschur, partop ptop, gtopology top, FILE out)
void	jacobi_precondition (double precvec, double invec, double outvec, long ndof, FILE out)
void	select_bound (double buff, double v)
void	select_master_buff (double cv, double buff, long nd)
void	solve_fact_mat_petsc (double r, double z, long ndof, FILE *out)
void	solve_system (partop ptop, gtopology top, gmatrix gm, long domproc, double lhs, double rhs, FILE *out, long iv)
double	uxv (double u, double v, long n)
double	v_ixu_i (double u, double v)
	~boundparcongrad ()
Public Attributes
int	Argc
char **	Argv
long **	bcn
double	err
double *	factmat
long *	lbdof
long *	lidof
long	maxnbdof
	maximum number of boundary DOFs on one subdomain
long	mespr
	message printing
int	myrank
	rank of processor
long	nbdof
	number of boundary degrees of freedom on subdomain
long *	nbdofdom
long	nbn
	number of boundary nodes on subdomain
long	ndof
	number of degrees of freedom on subdomain
long	ndofcp
	DATA DEFINED ONLY ON MASTER.
long	ndom
	number of subdomain
long	ni
long	nidof
	number of internal degrees of freedom on subdomain
long	nin
long	nn
int	nproc
	number of processors
PETSC_CONT *	petscpoint
long	prec

Detailed Description

Definition at line 16 of file boundparcongrad.h.

Constructor & Destructor Documentation

boundparcongrad	(	int	np,
		int	mr,
		long	nd,
		long	mes
	)

Constructor of class boundparcongrad JB

Definition at line 12 of file boundparcongrad.cpp.

References bcn, err, lbdof, lidof, maxnbdof, mespr, myrank, nbdof, nbdofdom, ndof, ndom, ni, nidof, nproc, petscpoint, and prec.

~boundparcongrad ( )

Desctructor of class boundparcongrad JB

Definition at line 54 of file boundparcongrad.cpp.

References PETSC_CONT::fact, PETSC_CONT::ix, petscilu, petscpoint, prec, PETSC_CONT::r_rhs, PETSC_CONT::val, and PETSC_CONT::z_lhs.

Member Function Documentation

void add_bound	(	double *	buff,
		double *	v
	)

Definition at line 163 of file boundparcongrad.cpp.

References nbdof, and nidof.

Referenced by solve_system().

void add_master_buff	(	double *	buff,
		double *	cv,
		long	nd
	)

Definition at line 180 of file boundparcongrad.cpp.

References bcn, and nbdofdom.

Referenced by solve_system().

void ilu_mat_petsc	(	gmatrix *	gm,
		FILE *	out
	)

Function computes incomplete factorization of subdomain matrix with PETSC-ILU with the help of PETSc library

Parameters:

	out	- output file for logs
	gm	- pointer to gmatrix class (../GEFEL/gmatrix.cpp)

JB

Definition at line 219 of file boundparcongrad.cpp.

References symcomprow::a, comprow::a, symcomprow::adr, comprow::adr, symcomprow::ci, comprow::ci, compressed_rows, gmatrix::cr, PETSC_CONT::fact, PETSC_CONT::ix, myrank, symcomprow::n, comprow::n, ndof, par_print_err(), petscpoint, PETSC_CONT::r_rhs, gmatrix::scr, symm_comp_rows, gmatrix::ts, PETSC_CONT::val, and PETSC_CONT::z_lhs.

Referenced by solve_system().

void initialize_diag_prec	(	gmatrix *	gm,
		double *	precvec,
		FILE *	out
	)

Function initialises diagonal preconditioning

Parameters:

	gm	- pointer to gmatrix class (../GEFEL/gmatrix.cpp)
	precvec	- vector for preconditionig
	out	- output file for logs

JB

Definition at line 431 of file boundparcongrad.cpp.

References symcomprow::a, comprow::a, symcomprow::adr, comprow::adr, symcomprow::ci, comprow::ci, compressed_rows, gmatrix::cr, mespr, symcomprow::n, comprow::n, gmatrix::scr, symm_comp_rows, and gmatrix::ts.

Referenced by solve_system().

void initiate	(	selnodes *	selnodschur,
		partop *	ptop,
		gtopology *	top,
		FILE *	out
	)

Function initiate boundary version of paralell conjugate gradient solver

Parameters:

	selnodschur	- pointer to selnodes class (selnodes.cpp)
	ptop	- pointer to patrop class (patrop.cpp)
	top	- pointer to gtopology class (../GEFEL/gtopology.cpp)
	out	- output file

JB 14.08.2007

Definition at line 81 of file boundparcongrad.cpp.

References Argc, Argv, bcn, maxnbdof, selnodes::maxsnndof, myrank, partop::nbdof, nbdof, nbdofdom, partop::ndof, ndof, ndofcp, partop::nidof, nidof, petscilu, and prec.

void jacobi_precondition	(	double *	precvec,
		double *	invec,
		double *	outvec,
		long	ndof,
		FILE *	out
	)

Function performs diagonal preconditioning

Parameters:

	precvec	- vector for preconditionig
	invec	- input vector
	outvec	- output vector outvec=precvec*inputvec
	out	- output file for logs JB

Definition at line 488 of file boundparcongrad.cpp.

References mespr.

Referenced by solve_system().

void select_bound	(	double *	buff,
		double *	v
	)

Definition at line 148 of file boundparcongrad.cpp.

References ndof, and nidof.

Referenced by solve_system().

void select_master_buff	(	double *	cv,
		double *	buff,
		long	nd
	)

Definition at line 199 of file boundparcongrad.cpp.

References bcn, and nbdofdom.

Referenced by solve_system().

void solve_fact_mat_petsc	(	double *	r,
		double *	z,
		long	ndof,
		FILE *	out
	)

Function performs back substitution with the help of PETSc library

Parameters:

	r	- residual vector
	z	- preconditoned vector z = PETscILUMAT^-1*r
	ndof	- number of degrees of freedom on subdomain
	out	- output file for logs

JB

Definition at line 382 of file boundparcongrad.cpp.

References PETSC_CONT::fact, PETSC_CONT::ix, nullv(), petscpoint, PETSC_CONT::r_rhs, PETSC_CONT::val, and PETSC_CONT::z_lhs.

Referenced by solve_system().

void solve_system	(	partop *	ptop,
		gtopology *	top,
		gmatrix *	gm,
		long *	domproc,
		double *	lhs,
		double *	rhs,
		FILE *	out,
		long	iv
	)

Function solves system of algebraic equations by boundary version of parallel conjugate gradient method

Parameters:

	top	- topology
	gm	- matrix of the system
	domproc	- domain-processor correspondence
	lhs	- array containing solution of the system
	rhs	- array containing right hand side
	out	- output stream
	iv	- initial values indicator