Bit field counters. More...

Modules
module	mpbits
	Bit field data.

Functions/Subroutines
subroutine	inbits (im, jm, inc)
	Fill bit fields (counters, triangular part). More...

subroutine	irbits (i, j)
	Fill bit fields (counters, rectangular part). More...

subroutine	clbits (in, jreqpe, jhispe, jsngpe, jextnd, idimb, ispc)
	Calculate bit (field) array size, encoding. More...

subroutine	plbits (in, inar, inac, idimb)
	Calculate bit field array size (PARDISO). More...

subroutine	ndbits (npgrp, ndims, nsparr, ihst)
	Analyze bit fields. More...

subroutine	pbsbits (npgrp, ibsize, nnzero, nblock, nbkrow)
	Analyze bit fields. More...

subroutine	pblbits (npgrp, ibsize, nsparr, nsparc)
	Analyze bit fields. More...

subroutine	prbits (npgrp, nsparr)
	Analyze bit fields. More...

subroutine	pcbits (npgrp, nsparr, nsparc)
	Analyze bit fields. More...

subroutine	ckbits (npgrp, ndims)
	Check sparsity of matrix. More...

subroutine	spbits (npgrp, nsparr, nsparc)
	Create sparsity information. More...

subroutine	clbmap (in)
	Clear (additional) bit map. More...

subroutine	inbmap (im, jm)
	Fill bit map. More...

subroutine	gpbmap (ngroup, npgrp, npair)
	Get pairs (statistic) from map. More...

subroutine	ggbmap (ipgrp, npair, npgrp)
	Get paired (parameter) groups from map. More...

Variables
integer(mpl)	mpbits::ndimb
	dimension for bit (field) array More...

integer(mpl)	mpbits::ndimb2
	dimension for bit map More...

integer(mpi)	mpbits::n
	matrix size (counters, sparse, triangular part) More...

integer(mpi)	mpbits::nar
	additional rows (counters, sparse, rectangular part) More...

integer(mpi)	mpbits::nac
	additional columns (counters, sparse, rectangular part) More...

integer(mpi)	mpbits::n2
	matrix size (map) More...

integer(mpi)	mpbits::ibfw
	bit field width More...

integer(mpi)	mpbits::ireqpe
	min number of pair entries More...

integer(mpi)	mpbits::isngpe
	upper bound for pair entry single precision storage More...

integer(mpi)	mpbits::iextnd
	flag for extended storage (both 'halves' of sym. More...

integer(mpi)	mpbits::nspc
	number of precision for sparse global matrix (1=D, 2=D+f) More...

integer(mpi)	mpbits::mxcnt
	max value for bit field counters More...

integer(mpi)	mpbits::nthrd
	number of threads More...

integer(mpi), dimension(:), allocatable	mpbits::bitfieldcounters
	fit field counters for global parameters pairs (tracks) More...

integer(mpi), dimension(:), allocatable	mpbits::bitmap
	fit field map for global parameters pairs (measurements) More...

integer(mpi), parameter	mpbits::bs = BIT_SIZE(1_mpi)
	number of bits in INTEGER(mpi) More...

Detailed Description

Bit field counters.

Author: Volker Blobel, University Hamburg, 2005-2009 (initial Fortran77 version); Claus Kleinwort, DESY (maintenance and developement)

Copyright: Copyright (c) 2009 - 2023 Deutsches Elektronen-Synchroton, Member of the Helmholtz Association, (DESY), HAMBURG, GERMANY

This library is free software; you can redistribute it and/or modify it under the terms of the GNU Library General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Library General Public License for more details.

You should have received a copy of the GNU Library General Public License along with this program (see the file COPYING.LIB for more details); if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

Count pairs of global parameters for sparse storage of global matrix, apply pair entries cut and build (compressed) sparsity structure (row offsets, column lists).

In sparse storage mode for each row the list of column indices with non zero elements (and those elements) are stored. With compression this list is represented by the first column and their number for continous regions (encoded in single INTEGER(mpi) words). Rare elements may be stored in single precision.

An additional bit map is used to monitor the parameter pairs for measurements (or 'equations').

To make use of the Intel oneMKL PARDISO solver for sparse systems the 'CSR3' (or 'BSR3') format (using the 'upper' triangle) is needed. For this a new set of routines have been added (starting with 'P'). In this case the bit array has two parts: A triangular one for pairs of parameter groups and a rectangular one for Lagrange multipliers (parameter constraint pairs).

Definition in file mpbits.f90.

Function/Subroutine Documentation

◆ ckbits()

subroutine ckbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpl), dimension(4), intent(out)	ndims
	)

Check sparsity of matrix.

Parameters

[in]	npgrp	parameter groups
[out]	ndims	(1): number of non-zero elements; (2): size of column lists; (3/4): number of (double/single precision) off diagonal elements;

Definition at line 1111 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::ibfw, mpbits::iextnd, mpbits::ireqpe, mpbits::isngpe, and mpbits::n.

Referenced by loop2().

◆ clbits()

subroutine clbits	(	integer(mpi), intent(in)	in,
		integer(mpi), intent(in)	jreqpe,
		integer(mpi), intent(in)	jhispe,
		integer(mpi), intent(in)	jsngpe,
		integer(mpi), intent(in)	jextnd,
		integer(mpl), intent(out)	idimb,
		integer(mpi), intent(out)	ispc
	)

Calculate bit (field) array size, encoding.

Parameters

[in]	in	matrix size
[in]	jreqpe	min number of pair entries
[in]	jhispe	upper bound for pair entry histogrammimg
[in]	jsngpe	upper bound for pair entry single precision storage
[in]	jextnd	flag for extended storage
[out]	idimb	dimension for bit (field) array
[out]	ispc	number of precision for sparse global matrix

Definition at line 178 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::ibfw, mpbits::iextnd, mpbits::ireqpe, mpbits::isngpe, mpbits::mxcnt, mpbits::n, mpbits::nac, mpbits::nar, mpbits::ndimb, mpbits::nspc, and mpbits::nthrd.

Referenced by loop2().

◆ clbmap()

subroutine clbmap ( integer(mpi), intent(in) in )

Clear (additional) bit map.

Parameters

[in] in matrix size

Definition at line 1341 of file mpbits.f90.

References mpbits::bitmap, mpbits::bs, mpbits::n2, and mpbits::ndimb2.

Referenced by loop2().

◆ ggbmap()

subroutine ggbmap	(	integer(mpi), intent(in)	ipgrp,
		integer(mpi), intent(out)	npair,
		integer(mpi), dimension(:), intent(out)	npgrp
	)

Get paired (parameter) groups from map.

Parameters

[in]	ipgrp	parameter group
[out]	npair	number of paired parameters
[in]	npgrp	paired parameter groups (for ipgrp)

Definition at line 1453 of file mpbits.f90.

References mpbits::bitmap, mpbits::bs, and mpbits::n2.

Referenced by loop2(), and prtstat().

◆ gpbmap()

subroutine gpbmap	(	integer(mpi), intent(in)	ngroup,
		integer(mpi), dimension(:,:), intent(in)	npgrp,
		integer(mpi), dimension(:), intent(out)	npair
	)

Get pairs (statistic) from map.

Parameters

[in]	ngroup	number of parameter groups
[in]	npgrp	parameter groups
[out]	npair	number of paired parameters

Definition at line 1407 of file mpbits.f90.

References mpbits::bitmap, and mpbits::bs.

Referenced by loop2().

◆ inbits()

subroutine inbits	(	integer(mpi), intent(in)	im,
		integer(mpi), intent(in)	jm,
		integer(mpi), intent(in)	inc
	)

Fill bit fields (counters, triangular part).

Parameters

[in]	im	first index
[in]	jm	second index
[in]	inc	increment (usually 1)

Definition at line 69 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::ibfw, mpbits::iextnd, mpbits::mxcnt, and mpbits::n.

Referenced by loop2().

◆ inbmap()

subroutine inbmap	(	integer(mpi), intent(in)	im,
		integer(mpi), intent(in)	jm
	)

Fill bit map.

Parameters

[in]	im	first index
[in]	jm	second index

Definition at line 1373 of file mpbits.f90.

References mpbits::bitmap, mpbits::bs, and mpbits::n2.

Referenced by loop2().

◆ irbits()

subroutine irbits	(	integer(mpi), intent(in)	i,
		integer(mpi), intent(in)	j
	)

Fill bit fields (counters, rectangular part).

Parameters

[in]	i	row (parameter)
[in]	j	column (constraint)

Definition at line 145 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::nac, mpbits::nar, and mpbits::ndimb.

Referenced by loop2().

◆ ndbits()

subroutine ndbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpl), dimension(4), intent(out)	ndims,
		integer(mpl), dimension(:,:), intent(out)	nsparr,
		integer(mpi), intent(in)	ihst
	)

Analyze bit fields.

Parameters

[in]	npgrp	parameter groups
[out]	ndims	(1): (reduced) size of bit array; (2): size of column lists; (3/4): number of (double/single precision) off diagonal elements;
[out]	nsparr	row offsets
[in]	ihst	>0: histogram number

Definition at line 301 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, hmpent(), mpbits::ibfw, mpbits::iextnd, mpbits::ireqpe, mpbits::isngpe, mpbits::n, mpbits::ndimb, mpbits::nspc, and mpbits::nthrd.

Referenced by loop2().

◆ pblbits()

subroutine pblbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpi), intent(in)	ibsize,
		integer(mpl), dimension(:), intent(in)	nsparr,
		integer(mpl), dimension(:), intent(out)	nsparc
	)

Analyze bit fields.

Calculate BSR3 column list (for PARDISO)

Parameters

[in]	npgrp	parameter groups
[in]	ibsize	block size
[in]	nsparr	row offsets
[out]	nsparc	column list

Definition at line 751 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::n, mpbits::nac, mpbits::ndimb, and mpbits::nthrd.

Referenced by loop2().

◆ pbsbits()

subroutine pbsbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpi), intent(in)	ibsize,
		integer(mpl), intent(out)	nnzero,
		integer(mpl), intent(out)	nblock,
		integer(mpi), dimension(:), intent(out)	nbkrow
	)

Analyze bit fields.

Check block structure (for PARDISO BSR3 storage)

Parameters

[in]	npgrp	parameter groups
[in]	ibsize	block size
[out]	nnzero	number of non-zero elements
[out]	nblock	number of blocks used
[out]	nbkrow	number of (column) blocks in row (blocks)

Definition at line 574 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::n, mpbits::nac, mpbits::ndimb, and mpbits::nthrd.

Referenced by loop2().

◆ pcbits()

subroutine pcbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpl), dimension(:), intent(in)	nsparr,
		integer(mpl), dimension(:), intent(out)	nsparc
	)

Analyze bit fields.

Calculate CSR3 column list (for PARDISO)

Parameters

[in]	npgrp	parameter groups
[in]	nsparr	row offsets
[out]	nsparc	column list

Definition at line 1017 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::n, mpbits::nac, mpbits::ndimb, and mpbits::nthrd.

Referenced by loop2().

◆ plbits()

subroutine plbits	(	integer(mpi), intent(in)	in,
		integer(mpi), intent(in)	inar,
		integer(mpi), intent(in)	inac,
		integer(mpl), intent(out)	idimb
	)

Calculate bit field array size (PARDISO).

Count pairs of parameter groups in triangular part, pairs of constraints and parameters in rectangular part.

Parameters

[in]	in	matrix size (triangular part, number of parameter groups)
[in]	inar	rows for rectangular part (number of parameters)
[in]	inac	cols for rectangular part (number of constraints)
[out]	idimb	dimension for bit (field) array

Definition at line 251 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::ibfw, mpbits::iextnd, mpbits::ireqpe, mpbits::isngpe, mpbits::n, mpbits::nac, mpbits::nar, mpbits::ndimb, mpbits::nspc, and mpbits::nthrd.

Referenced by loop2().

◆ prbits()

subroutine prbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpl), dimension(:), intent(out)	nsparr
	)

Analyze bit fields.

Calculate CSR3 row offsets (for PARDISO)

Parameters

[in]	npgrp	parameter groups
[out]	nsparr	row offsets (preset with number of Lagrange multipliers (at row+1))

Definition at line 918 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::n, mpbits::nac, mpbits::ndimb, and mpbits::nthrd.

Referenced by loop2().

◆ spbits()

subroutine spbits	(	integer(mpi), dimension(:), intent(in)	npgrp,
		integer(mpl), dimension(:,:), intent(in)	nsparr,
		integer(mpi), dimension(:), intent(out)	nsparc
	)

Create sparsity information.

Parameters

[in]	npgrp	parameter groups
[in]	nsparr	row offsets
[out]	nsparc	column indices

Definition at line 1204 of file mpbits.f90.

References mpbits::bitfieldcounters, mpbits::bs, mpbits::ibfw, mpbits::iextnd, mpbits::n, mpbits::nspc, and mpbits::nthrd.

Referenced by loop2().

Modules

Functions/Subroutines

Variables

Detailed Description

Function/Subroutine Documentation

◆ ckbits()

◆ clbits()

◆ clbmap()

◆ ggbmap()

◆ gpbmap()

◆ inbits()

◆ inbmap()

◆ irbits()

◆ ndbits()

◆ pblbits()

◆ pbsbits()

◆ pcbits()

◆ plbits()

◆ prbits()

◆ spbits()