glpsol fails to parse a cplex lp file containing constraints that evaluate to zero

[Christophe-Marie Duquesne]

Hi,

Running glpsol on the attached file fails:

 > glpsol --cpxlp debug.lp
GLPSOL: GLPK LP/MIP Solver, v4.43
Parameter(s) specified in the command line:
 --cpxlp debug.lp
Reading problem data from `debug.lp'...
debug.lp:6: missing variable name
CPLEX LP file processing error

Reading some documentation about the cplex lp format [1], I did not
find something that forbids to but "0<=0" constraints, so I am filing
a bug here. Pardon me if I am wrong.
FYI, cbc does not fail on it, which made me suspect a bug in glpsol,
but nothing proves they are right.

[1] http://lpsolve.sourceforge.net/5.5/CPLEX-format.htm

Best regards,
Christophe-Marie Duquesne


_______________________________________________
Bug-glpk mailing list
Bug-glpk@...
https://lists.gnu.org/mailman/listinfo/bug-glpk

[Andrew Makhorin]

> Running glpsol on the attached file fails:
>
>  > glpsol --cpxlp debug.lp
> GLPSOL: GLPK LP/MIP Solver, v4.43
> Parameter(s) specified in the command line:
>  --cpxlp debug.lp
> Reading problem data from `debug.lp'...
> debug.lp:6: missing variable name
> CPLEX LP file processing error
>
> Reading some documentation about the cplex lp format [1], I did not
> find something that forbids to but "0<=0" constraints, so I am filing
> a bug here. Pardon me if I am wrong.
> FYI, cbc does not fail on it, which made me suspect a bug in glpsol,
> but nothing proves they are right.
>
> [1] http://lpsolve.sourceforge.net/5.5/CPLEX-format.htm

Thank you for your report.

I don't think this is a bug, because cplex lp format does not allow
constant terms in the lhs, i.e. the lhs should be a non-empty linear
form. However, something like "0 x1 <= 0" must be valid.


Andrew Makhorin


_______________________________________________
Bug-glpk mailing list
Bug-glpk@...
https://lists.gnu.org/mailman/listinfo/bug-glpk

[Christophe-Marie Duquesne]

On Thu, May 5, 2011 at 5:11 PM, Andrew Makhorin <mao@...> wrote:
> I don't think this is a bug [...]

Given the testimonies of everyone (especially concerning cplex
itself), I don't think it is a bug either. Thank you for helping me
figuring it out.

> However, something like "0 x1 <= 0" must be valid.

No problem with that: I checked and glpsol runs fine in this case.

Thank you.

Best regards,
Christophe-Marie Duquesne

_______________________________________________
Bug-glpk mailing list
Bug-glpk@...
https://lists.gnu.org/mailman/listinfo/bug-glpk

[Christophe-Marie Duquesne]

On Sat, May 7, 2011 at 5:06 PM, Hans D. Mittelmann <MITTELMANN@...> wrote:
> Basically you were confusing "a constraint that evaluates to 0<=0"
> with one that explicity is "0<=0". The first may happen the
> latter should not.
> Hans Mittelmann

Yes, you are right, I was mistaken when submitting this report. FYI my
post here is actually related to a bug in another piece of software
[1], where this time a problem occurs when a constraint actually
evaluates to "0 <= 0" and is not explicitly "0 <= 0". That is why the
confusing title. Sorry.

[1] http://code.google.com/p/pulp-or/issues/detail?id=31

_______________________________________________
Bug-glpk mailing list
Bug-glpk@...
https://lists.gnu.org/mailman/listinfo/bug-glpk