/*$Id: matio.c,v 1.66 2000/04/09 04:36:57 bsmith Exp bsmith $*/
/*
This file contains simple binary read/write routines for matrices.
*/
#include "petsc.h"
#include "src/mat/matimpl.h" /*I "mat.h" I*/
#include "sys.h"
static int MatLoadersSet = 0,(*MatLoaders[MAX_MATRIX_TYPES])(Viewer,MatType,Mat*) =
{0,0,0,0,0,0,0,0,0,0,0,0};
#undef __FUNC__
#define __FUNC__ /**/"MatLoadRegister"
/*@C
MatLoadRegister - Allows one to register a routine that reads matrices
from a binary file for a particular matrix type.
Not Collective
Input Parameters:
+ type - the type of matrix (defined in include/mat.h), for example, MATSEQAIJ.
- loader - the function that reads the matrix from the binary file.
Level: developer
.seealso: MatLoadRegisterAll(), MatLoad()
@*/
int MatLoadRegister(MatType type,int (*loader)(Viewer,MatType,Mat*))
{
PetscFunctionBegin;
MatLoaders[type] = loader;
MatLoadersSet = 1;
PetscFunctionReturn(0);
}
#undef __FUNC__
#define __FUNC__ /**/"MatLoadPrintHelp_Private"
static int MatLoadPrintHelp_Private(Mat A)
{
static PetscTruth called = PETSC_FALSE;
MPI_Comm comm = A->comm;
int ierr;
PetscFunctionBegin;
if (called) {PetscFunctionReturn(0);} else called = PETSC_TRUE;
ierr = (*PetscHelpPrintf)(comm," Options for MatLoad:\n");CHKERRQ(ierr);
ierr = (*PetscHelpPrintf)(comm," -matload_block_size :Used for MATBAIJ, MATBDIAG\n");CHKERRQ(ierr);
ierr = (*PetscHelpPrintf)(comm," -matload_bdiag_diags : Used for MATBDIAG\n");CHKERRQ(ierr);
PetscFunctionReturn(0);
}
#undef __FUNC__
#define __FUNC__ /**/"MatLoad"
/*@C
MatLoad - Loads a matrix that has been stored in binary format
with MatView(). The matrix format is determined from the options database.
Generates a parallel MPI matrix if the communicator has more than one
processor. The default matrix type is AIJ.
Collective on Viewer
Input Parameters:
+ viewer - binary file viewer, created with ViewerBinaryOpen()
- outtype - type of matrix desired, for example MATSEQAIJ,
MATMPIROWBS, etc. See types in petsc/include/mat.h.
Output Parameters:
. newmat - new matrix
Basic Options Database Keys:
The following options will work if you first call MatGetTypeFromOptions()
and pass the resulting type to MatLoad().
These options use MatCreateSeqXXX or MatCreateMPIXXX,
depending on the communicator, comm.
+ -mat_aij - AIJ type
. -mat_baij - block AIJ type
. -mat_dense - dense type
. -mat_bdiag - block diagonal type
. -mat_complex - indicates the matrix has complex entries
- -mat_double - indicates the matrix has double entries
More Options Database Keys:
+ -mat_seqaij - AIJ type
. -mat_mpiaij - parallel AIJ type
. -mat_seqbaij - block AIJ type
. -mat_mpibaij - parallel block AIJ type
. -mat_seqbdiag - block diagonal type
. -mat_mpibdiag - parallel block diagonal type
. -mat_mpirowbs - parallel rowbs type
. -mat_seqdense - dense type
- -mat_mpidense - parallel dense type
More Options Database Keys:
Used with block matrix formats (MATSEQBAIJ, MATMPIBDIAG, ...) to specify
block size
. -matload_block_size
Used to specify block diagonal numbers for MATSEQBDIAG and MATMPIBDIAG formats
. -matload_bdiag_diags
Level: beginner
Notes:
MatLoad() automatically loads into the options database any options
given in the file filename.info where filename is the name of the file
that was passed to the ViewerBinaryOpen(). The options in the info
file will be ignored if you use the -matload_ignore_info option.
In parallel, each processor can load a subset of rows (or the
entire matrix). This routine is especially useful when a large
matrix is stored on disk and only part of it existsis desired on each
processor. For example, a parallel solver may access only some of
the rows from each processor. The algorithm used here reads
relatively small blocks of data rather than reading the entire
matrix and then subsetting it.
Notes for advanced users:
Most users should not need to know the details of the binary storage
format, since MatLoad() and MatView() completely hide these details.
But for anyone who's interested, the standard binary matrix storage
format is
$ int MAT_COOKIE
$ int number of rows
$ int number of columns
$ int total number of nonzeros
$ int *number nonzeros in each row
$ int *column indices of all nonzeros (starting index is zero)
$ Scalar *values of all nonzeros
Note for Cray users, the int's stored in the binary file are 32 bit
integers; not 64 as they are represented in the memory, so if you
write your own routines to read/write these binary files from the Cray
you need to adjust the integer sizes that you read in, see
PetscReadBinary() and PetscWriteBinary() to see how this may be
done.
In addition, PETSc automatically does the byte swapping for
machines that store the bytes reversed, e.g. DEC alpha, freebsd,
linux, nt and the paragon; thus if you write your own binary
read/write routines you have to swap the bytes; see PetscReadBinary()
and PetscWriteBinary() to see how this may be done.
.keywords: matrix, load, binary, input
.seealso: ViewerBinaryOpen(), MatView(), VecLoad(), MatLoadRegister(),
MatLoadRegisterAll(), MatGetTypeFromOptions()
@*/
int MatLoad(Viewer viewer,MatType outtype,Mat *newmat)
{
int ierr;
PetscTruth set,isbinary,flg;
MatType type;
MPI_Comm comm;
PetscFunctionBegin;
if (outtype > MAX_MATRIX_TYPES || outtype < 0) {
SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,1,"Not a valid matrix type");
}
PetscValidHeaderSpecific(viewer,VIEWER_COOKIE);
*newmat = 0;
if (!MatLoadersSet) {
ierr = MatLoadRegisterAll();CHKERRQ(ierr);
}
ierr = PetscTypeCompare((PetscObject)viewer,BINARY_VIEWER,&isbinary);CHKERRQ(ierr);
if (!isbinary) {
SETERRQ(PETSC_ERR_ARG_WRONG,0,"Invalid viewer; open viewer with ViewerBinaryOpen()");
}
ierr = PetscObjectGetComm((PetscObject)viewer,&comm);CHKERRQ(ierr);
ierr = MatGetTypeFromOptions(comm,0,&type,&set);CHKERRQ(ierr);
if (!set) type = outtype;
PLogEventBegin(MAT_Load,viewer,0,0,0);
if (!MatLoaders[outtype]) {
SETERRQ(PETSC_ERR_ARG_WRONG,1,"Invalid matrix type, or matrix load not registered");
}
ierr = (*MatLoaders[outtype])(viewer,type,newmat);CHKERRQ(ierr);
ierr = OptionsHasName(PETSC_NULL,"-help",&flg);CHKERRQ(ierr);
if (flg) {ierr = MatLoadPrintHelp_Private(*newmat);CHKERRQ(ierr); }
PLogEventEnd(MAT_Load,viewer,0,0,0);
PetscFunctionReturn(0);
}