Re: [Help-glpk] Upper bound problem

[Andrew Makhorin]

> First of all, excuse my bad english. Im new at mathprog and trying to
> model some problem i get as output some strange thing:
> 
>    No.   Row name        Activity     Lower bound   Upper bound
> ------ ------------    ------------- ------------- -------------
>     .....
>     .....
> 
>      7 constraint[1]
>                                    0                          -0 
>      8 constraint[2]
>                                    0                          -0 
>      9 constraint[3]
>                                   -3                          -0 
>     10 constraint[4]
>                                   -2                          -0 
> 
> I dont get why those values happen at this constraint, because neither
> of the variables or values are negative in the first place. The
> constraint is defined as follow:
> 
> s.t. constraint{j in 1..n}: sum{i in 1..m} d[i,j] <= m * y[j];
> 
> Where m is a param defined in the data, positive integer, y is a
> binary variable and d[i,j] are real variables between 0 and 1
> (0<=d<=1)
> 
> Any help would be greatly appreciated!! thanks in advance

By definition (in MathProg) the auxiliary variable for a row
(constraint) is always the difference between left-hand and right-hand
linear forms of the constraint. Thus, in your case

   constraint[j] = sum{i in 1..m} d[i,j] - m * y[j] <= 0.

Since the solution found is primal feasible, in the row section you see
non-positive values (activities) of auxiliary variables 'constraint[j]'.
For more details please see the MathProg language reference, Subsection
4.4 "Constraint statement".