[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Bug-glpk] Re: Glpk bug?
From: |
Andrew Makhorin |
Subject: |
[Bug-glpk] Re: Glpk bug? |
Date: |
Thu, 5 Jul 2007 22:46:10 +0400 |
> I have a problem to solve the given MIP with Glpk. It aborts with the
> following error message:
>> glpsol --intopt --cpxlp optconfig-solve.lp-glpk.lp
> GLPK internal error: x >= lb; file glpmip02.c, line 228
> I am using version 4.18, but the error is the same for 4.4 and 4.15.
> ILOG Cplex solves the given MIP successfully, so I guess it is OK.
> Is there any fix for this bug?
This is not a bug, but a defect of the b&b solver.
The error happens due to excessive round-off errors, which, in turn,
happen due to huge bounds of some integer variables. For example:
0 <= gmodeli_p2_p5 <= 2147483647
It would be better not to specify the upper bound at all.
Nevertheless, adding the option '--cgr' I managed to solve your
instance to optimality:
lpx_read_cpxlp: reading problem data from `bug.lp'...
lpx_read_cpxlp: 3226 rows, 1932 columns, 8937 non-zeros
lpx_read_cpxlp: 1932 integer columns, 1878 of which are binary
lpx_read_cpxlp: 7626 lines were read
ipp_basic_tech: 418 row(s) and 60 column(s) removed
ipp_reduce_bnds: 2 pass(es) made, 488 bound(s) reduced
ipp_basic_tech: 452 row(s) and 450 column(s) removed
ipp_reduce_coef: 2 pass(es) made, 20 coefficient(s) reduced
lpx_intopt: presolved MIP has 2356 rows, 1422 columns, 5973 non-zeros
lpx_intopt: 1422 integer columns, 1377 of which are binary
lpx_adv_basis: size of triangular part = 2356
Solving LP relaxation...
0: objval = 1.754620000e+05 infeas = 1.000000000e+00 (0)
200: objval = 5.682940000e+05 infeas = 4.678937555e-04 (0)
400: objval = 5.683725057e+05 infeas = 4.676103586e-04 (0)
600: objval = 6.758156656e+05 infeas = 2.033088516e-05 (0)
619: objval = 6.768716656e+05 infeas = 2.525222006e-20 (0)
* 619: objval = 6.768716656e+05 infeas = 1.987299214e-14 (0)
* 800: objval = 6.645538361e+05 infeas = 2.999919154e-14 (0)
* 1000: objval = 5.714568399e+05 infeas = 1.393510204e-15 (0)
* 1054: objval = 5.684413548e+05 infeas = 1.009830884e-13 (0)
OPTIMAL SOLUTION FOUND
Integer optimization begins...
Objective function is integral
+ 1054: mip = not found yet >= -inf (1; 0)
+ 1590: mip = not found yet >= 5.684420000e+05 (9; 0)
+ 1656: mip = not found yet >= 5.684420000e+05 (18; 0)
+ 1666: mip = not found yet >= 5.684420000e+05 (28; 0)
+ 1677: mip = not found yet >= 5.684420000e+05 (39; 0)
+ 1687: mip = not found yet >= 5.684420000e+05 (49; 0)
+ 1698: mip = not found yet >= 5.684420000e+05 (60; 0)
+ 1710: mip = not found yet >= 5.684420000e+05 (72; 0)
+ 1721: mip = not found yet >= 5.684420000e+05 (83; 0)
+ 1732: mip = not found yet >= 5.684420000e+05 (94; 0)
+ 2217: mip = not found yet >= 5.684420000e+05 (108; 0)
+ 2277: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (167; 0)
+ 3231: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (90; 146)
+ 4364: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (54; 193)
+ 4742: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (83; 193)
+ 5529: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (99; 195)
+ 6404: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (124; 197)
+ 7562: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (138; 207)
+ 7980: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (131; 242)
+ 8227: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (155; 247)
+ 8475: mip = 5.696190000e+05 >= 5.684420000e+05 0.2% (174; 255)
+ 8526: mip = 5.685620000e+05 >= 5.684420000e+05 < 0.1% (224; 255)
+ 9348: mip = 5.685620000e+05 >= 5.685330000e+05 < 0.1% (65; 591)
+ 9611: mip = 5.685620000e+05 >= tree is empty 0.0% (0; 855)
INTEGER OPTIMAL SOLUTION FOUND
Time used: 113.0 secs
Memory used: 5.1 Mb (5371216 bytes)
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [Bug-glpk] Re: Glpk bug?,
Andrew Makhorin <=