1WARNING: this documentation is (at least partly) obsolete.
2----------------------------------------------------------
3
4duplicates -nrels <n> <rel_file>  >  <rel_file2>
5  Remove duplicates. The output format is the same as the input format
6  for relations.
7
8purge -poly <poly_file> -nrels <n> <rel_file>  >  <purged_file>
9  Remove singletons. The format of the purged file is as follows:
10     nrows ncols
11     index_in_rel_file nb_of_entries ncol_1 ncol_2 ...
12     index_in_rel_file nb_of_entries ncol_1 ncol_2 ...
13     ...
14
15merge -prune <p> -merge <m> -mat <purged_file> -cwmax <c> -rwmax <r> [-v] > <history_file>
16  Perform the merging. Keep the info in the history file, that contains on each
17  line the indices of two rows to be added.
18
19replay <purged_file> <history_file> <small_matrix_file> <index_file> [-v]
20  Replays the history given by merge in order to produce a sparse matrix. The
21  correspondance with the initial data is kept in index_file.
22  The format of the sparse matrix is:
23    nrows ncols
24    nb_entries ncol_1 ncol_2 ...
25    nb_entries ncol_1 ncol_2 ...
26    nb_entries ncol_1 ncol_2 ...
27    ...
28  The format of the index file is:
29    nrows ncols    ## same as in sparse matrix
30    nb_of_ab_pairs index_in_rel_file_1 index_in_rel_file_2 ...
31    nb_of_ab_pairs index_in_rel_file_1 index_in_rel_file_2 ...
32    ...
33  The lines of sparse matrix and index files are in correspondance.
34
35linalg <small_matrix_file> <ker_file>
36  Produces vectors in the the kernel of the matrix.
37
38characters / allsqrt : do not really belong to linalg subdir, and will change
39anyway, so we don't explain more.
40
41