[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-glpk] [Help-glpk] 4.54 sgf_reduce_nuc triggers assertion on lin
From: |
Andrew Makhorin |
Subject: |
Re: [Bug-glpk] [Help-glpk] 4.54 sgf_reduce_nuc triggers assertion on line 165 of sgf.c |
Date: |
Sun, 27 Apr 2014 16:27:37 +0400 |
> Consider the simple problem
>
> p lp min 1 1 0
> i 1 d -1 1
> n i 1 constraint
> j 1 f
> n j 1 variable
> e o f
>
> Standard, the variable is non-basic fixed at lower bound and the
> constraint is basic. When I set the variable as basic and constraint as
> non-basic, I get, when warming up the basis:
>
> Assertion failed: k1 < k2
> Error detected in file <path>/glpk-4.54/src/bflib/sgf.c at line 165
>
> Is this expected? (I was expecting an error code signaling an invalid
> basis at the worst.)
>
Thank you for the bug report.
The bug appears when the basis to be factorized is structurally singular
(more than one singleton in the same row or column), in which case
glp_factorize (called from glp_warm_up and glp_simplex) should return
GLP_ESING.
To avoid the bug please change file glpk/src/bflib/sgf.c, function
sgf_factorize, as follows:
. . .
1300 int i, j, k, k1, k2, p, q, nnz;
++++ singl = 0; /* disable singleton phase */
1301 /* build matrix V = A in row-wise format */
1302 luf_build_v_rows(luf, rs_prev);
. . .
I will fix this bug in a next release.
Andrew Makhorin
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Re: [Bug-glpk] [Help-glpk] 4.54 sgf_reduce_nuc triggers assertion on line 165 of sgf.c,
Andrew Makhorin <=