[Coin-discuss] GLPK AND SYMPHONY not getting the same solutions

[Leo Lopes]

I am forced to mention that if you use LPFML or OSiL none of these
problems will happen... In addition your files will be much smaller,
and compress much better, they will be easier to generate, they can
contain the problem as well as the solution, etc... For the details
see:

Robert Fourer, Leo Lopes, and Kipp Martin, "LPFML: A W3C XML Schema
for Linear Programming ", INFORMS Journal on Computing, 17, pages
139-158.

On 10/19/05, Ted Ralphs <tkralphs at lehigh.edu> wrote:
> Hi Andrew,
>
> Under the heading "BOUNDS Section" and the subheading "Default Bounds,"
> the OSL manual says the following:
>
> "For integer variables, the default bounds on columns are 0 and 1.
> (WARNING, if you want a very large bound for an integer variable,
> explicitly set it to 1020. DO NOT specify a value larger than 1020, or
> leave the value unspecified, thinking that the result will be "unbounded.")"
>
> Hope this clarifies things. Perhaps John Forrest could comment.
>
> Cheers,
>
> Ted
>
> Andrew Makhorin wrote:
> >>Yes, this is exactly the problem. According to the MPS standard, any
> >>integer variable without bounds is assumed to be binary. The COIN MPS
> >>reader complies with this standard, but not all readers do (GLPKs reader
> >>does not, for instance).
> >
> >
> > OSL documentation says (chapter "Passing Your Model Using Mathematical
> > Programming System (MPS) Format", section "Integer Data"):
> >
> >         General Integer Variables
> >
> >         For general integer variables, the following indicates the
> >         various values you may use for the fields:
> >
> >         If you are defining general integer variables, field 1 must
> >         be blank and field 5 must contain the value 'INTORG'
> >         (including the quotation marks) in the record that denotes
> >         the start of integer variables. In the record that denotes
> >         the end of integer variables, field 1 must be blank and
> >         field 5 must contain the value 'INTEND'.
> >
> >         You can select any name for field 2 as long as it is
> >         different from the preceding and following column names.
> >
> >         Field 4, although ignored, must be blank in free format.
> >
> >         All variables between the INTORG and INTEND markers are
> >         taken as integer variables if the intunit parameter in
> >         the call to EKKMPS is nonzero.
> >
> > And this is clarified by the following example:
> >
> >         The following is an example of defining general integer
> >         variables:
> >
> > *234567890123456789012345678901234567890123456789012345678901234567890
> > COLUMNS
> >     COL01     OBJ                1.0
> >     COL01     ROW01              3.0   ROW05              5.6
> >     COL02     ROW01              1.0   ROW02              2.0
> > *
> > *  Mark COL03 and COL04 as integer
> > *
> >     INT1      'MARKER'                 'INTORG'
> >     COL03     ROW02              1.1   ROW03              1.0
> >     COL04     ROW01             -2.0   ROW04              2.8
> >     INT1END   'MARKER'                 'INTEND'
> > *
> >     COL05     OBJ                2.0
> >     COL05     ROW01             -1.0   ROW05              1.0
> >     COL06     ROW03              1.0
> >     COL07     ROW04             -1.2
> >     COL08     OBJ               -1.0
> >     COL08     ROW01             -1.0   ROW05              1.9
> >
> >         Here, the columns named COL03 and COL04 must take on integer
> >         values.
> >
> > I didn't find any points that "any integer variable without bounds is
> > assumed to be binary".
> >
> > Andrew Makhorin,
> > maintainer of glpk
> >
> > (Sorry, if my message sent twice due to incorrect clock settings.)
>
>
> --
> Dr. Ted Ralphs
> Assistant Professor
> Industrial and Systems Engineering
> Lehigh University
> (610)758-4784
> tkralphs at lehigh.edu
> www.lehigh.edu/~tkr2
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-discuss
>


--
========================================================================
Leonardo B. Lopes                                    leo at sie.arizona.edu
Assistant Professor                                        (520)621-2342
SIE - University of Arizona  http://www.sie.arizona.edu/faculty/leolopes