Re: [Help-glpk] Error: unable to factorize the basis matrix

[Andrew Makhorin]

There is something wrong in your installation. The error happens due to
nan, which must not appear:

> *   634: obj =              -nan  infeas =  0.000e+00 (0)
> OPTIMAL LP SOLUTION FOUND
> Integer optimization begins...
> +   634: mip =     not found yet >=              -inf        (1; 0)
> Assertion failed: temp1 > 0.0 && temp2 > 0.0
> Error detected in file glpios03.c at line 205
> Aborted (core dumped)
> 
> 

Unfortunately, I'm unable to reproduce the error. I tried glpsol 4.52
(compiled with gcc (Debian 4.4.5-8) 4.4.5) and it solved your mip
successfully:

GLPSOL: GLPK LP/MIP Solver, v4.52
Parameter(s) specified in the command line:
 --lp example.lp --log example.log
Reading problem data from 'example.lp'...
example.lp:1282: warning: missing final end of line
952 rows, 634 columns, 2536 non-zeros
317 integer variables, all of which are binary
1282 lines were read
GLPK Integer Optimizer, v4.52
952 rows, 634 columns, 2536 non-zeros
317 integer variables, all of which are binary
Preprocessing...
952 rows, 634 columns, 2536 non-zeros
317 integer variables, all of which are binary
Scaling...
 A: min|aij| =  1.000e+00  max|aij| =  4.134e+04  ratio =  4.134e+04
GM: min|aij| =  1.655e-01  max|aij| =  6.043e+00  ratio =  3.652e+01
EQ: min|aij| =  2.738e-02  max|aij| =  1.000e+00  ratio =  3.652e+01
2N: min|aij| =  1.514e-02  max|aij| =  1.240e+00  ratio =  8.194e+01
Constructing initial basis...
Size of triangular part is 952
Solving LP relaxation...
GLPK Simplex Optimizer, v4.52
952 rows, 634 columns, 2536 non-zeros
      0: obj =   6.675300000e+03  infeas =  1.388e+02 (0)
    500: obj =   8.180770197e+03  infeas =  7.211e+01 (0)
*   634: obj =   1.185575775e+04  infeas =  0.000e+00 (0)
*   957: obj =   7.589768643e+02  infeas =  0.000e+00 (0)
OPTIMAL LP SOLUTION FOUND
Integer optimization begins...
+   957: mip =     not found yet >=              -inf        (1; 0)
+  1371: >>>>>   8.712560000e+03 >=   1.314220420e+03  84.9% (301; 0)
+  1482: >>>>>   4.443280000e+03 >=   1.358039358e+03  69.4% (304; 6)
+  1511: >>>>>   2.317350000e+03 >=   1.444905676e+03  37.6% (158; 164)
+  1511: mip =   2.317350000e+03 >=     tree is empty   0.0% (0; 619)
INTEGER OPTIMAL SOLUTION FOUND
Time used:   1.6 secs
Memory used: 1.5 Mb (1615554 bytes)



Note that there is some difference on the scaling stage. Your log
reports:

Scaling...
 A: min|aij| =  1.000e+00  max|aij| =  4.134e+04  ratio =  4.134e+04
GM: min|aij| =  1.655e-01  max|aij| =  6.043e+00  ratio =  3.652e+01
EQ: min|aij| =  2.738e-02  max|aij| =  1.000e+00  ratio =  3.652e+01
2N: min|aij| =  3.027e-02  max|aij| =  1.262e+00  ratio =  4.167e+01

while in my case it reports:

 A: min|aij| =  1.000e+00  max|aij| =  4.134e+04  ratio =  4.134e+04
GM: min|aij| =  1.655e-01  max|aij| =  6.043e+00  ratio =  3.652e+01
EQ: min|aij| =  2.738e-02  max|aij| =  1.000e+00  ratio =  3.652e+01
2N: min|aij| =  1.514e-02  max|aij| =  1.240e+00  ratio =  8.194e+01

However, the scaling results should be near the same.

Which version of gcc are you using? (Type 'gcc --version' to get it.)