1dgord(1)                    PT-Scotch user's manual                   dgord(1)
2
3
4

NAME

6       dgord - compute sparse matrix orderings of graphs in parallel
7

SYNOPSIS

9       dgord [options] [gfile] [ofile] [lfile]
10

DESCRIPTION

12       The  dgord program computes, in a parallel way, an ordering of a Scotch
13       source graph representing the pattern of some symmetric sparse matrix.
14
15       Source graph file gfile is either a centralized graph file, or a set of
16       files  representing  fragments  of  a  distributed graph. The resulting
17       ordering is stored in file ofile. Eventual logging information (such as
18       the  one  produced by option -v) is sent to file lfile. When file names
19       are not specified, data is read from  standard  input  and  written  to
20       standard  output.  Standard streams can also be explicitely represented
21       by a dash '-'.
22
23       When the proper libraries have been included at compile time, dgord can
24       directly  handle  compressed graphs, both as input and output. A stream
25       is treated as compressed whenever its name is  postfixed  with  a  com‐
26       pressed  file  extension, such as in 'brol.grf.bz2' or '-.gz'. The com‐
27       pression formats which can be supported are the bzip2 format  ('.bz2'),
28       the gzip format ('.gz'), and the lzma format ('.lzma', on input only).
29
30       dgord  bases  on  implementations  of  the MPI interface to spread work
31       across the processing elements. It is therefore not likely  to  be  run
32       directly, but instead through some launcher command such as mpirun.
33

OPTIONS

35       -h     Display some help.
36
37       -mmfile
38              Save column block mapping data to file mfile. Mapping data spec‐
39              ifies, for each vertex, the index of the column block  to  which
40              this vertex belongs.
41
42       -ostrat
43              Use parallel graph ordering strategy strat (see PT-Scotch user's
44              manual for more information).
45
46       -rpnum Set root process for centralized files (default is 0).
47
48       -ttfile
49              Save partitioning tree data to  file  tfile.  Partitioning  tree
50              data  specifies,  for each vertex, the index of the first vertex
51              of the parent block of the block to which  the  vertex  belongs.
52              Altogether  with  the  mapping  data  provided in file mfile, it
53              allows one to rebuild the separator tree of the  nested  dissec‐
54              tion process.
55
56       -V     Display program version and copyright.
57
58       -vverb Set  verbose mode to verb. It is a set of one of more characters
59              which can be:
60
61              s      strategy information.
62
63              t      timing information.
64

EXAMPLES

66       Run dgord on 5 processing elements to reorder matrix graph brol.grf and
67       save the resulting ordering to file brol.ord, using the default sequen‐
68       tial graph ordering strategy:
69
70             $ mpirun -np 5 dgord brol.grf brol.ord
71
72       Run dgord on 5 processing elements to reorder  the  distributed  matrix
73       stored on graph fragment files brol5-0.dgr to brol5-4.dgr, and save the
74       resulting ordering to file brol.ord (see dgscat(1) for  an  explanation
75       of  the  '%p'  and  '%r'  sequences in names of distributed graph frag‐
76       ments).
77
78             $ mpirun -np 5 dgord brol%p-%r.dgr brol.ord
79
80

SEE ALSO

82       dgtst(1), dgscat(1), gmk_hy(1), gord(1).
83
84       PT-Scotch user's manual.
85

AUTHOR

87       Francois Pellegrini <francois.pellegrini@labri.fr>
88
89
90
91                              September 08, 2008                      dgord(1)