dgssor(3P) Sun Performance Library dgssor(3P)NAME
dgssor - General sparse solver ordering and symbolic factorization.
SYNOPSIS
SUBROUTINE DGSSOR ( ORDMTHD, HANDLE, IER )
CHARACTER*3 ORDMTHD
INTEGER IER
DOUBLE PRECISION HANDLE(150)PURPOSEDGSSOR - Orders and symbolically factors a sparse matrix.
PARAMETERS
ORDMTHD - CHARACTER*3
On entry, ORDMTHD specifies the fill-reducing ordering to be
used by the sparse solver. Valid options are:
'nat' or 'NAT' - natural ordering (no ordering)
'mmd' or 'MMD' - multiple minimum degree (default)
'gnd' or 'GND' - general nested dissection
When SuperLU is selected in the initialization routine DGSSIN,
valid options are:
'nat' or 'NAT' - natural ordering (no ordering)
'mmd' or 'MMD' - minimum degree ordering on A'*A (default)
'spm' or 'SPM' - Minimum degree ordering on A'+A
'sam' or 'SAM' - Approximate minimum degree column ordering
In the case of an invalid option, the default method is used.
A user-specified ordering can be used by calling subroutine
DGSSUO() in place of DGSSOR(). Unchanged on exit.
HANDLE(150) - DOUBLE PRECISION array
On entry, HANDLE(*) is an array containing information needed
by the solver, and must be passed unchanged to each sparse
solver subroutine. Modified on exit.
IER - INTEGER
Error number. If no error encountered, unchanged on exit. If
error encountered, it is set to a non-zero integer. Error num‐
bers set by this subroutine:
-200 : Invalid calling sequence - need to call DGSSIN first.
-201 : Failure to dynamically allocate memory.
-202 : Need to call DGSSUO instead of DGSSOR to set user-specified ordering
-666 : Internal error.
3rd Berkeley Distribution 6 Mar 2009 dgssor(3P)