On Wed, 2017-01-04 at 23:43 +0200, Alexey Karakulov wrote:
> Hi, I have this kind of function in the objective:
>
> > crop(s) = max(0, min(1, s))
>
>
> I wonder if it's possible (and how) to reformulate the task to be LP
> problem. I have read this posting [1], but I'm not sure how to apply
> it.
Note that
crop(x) = f(x) - f(x-1)
where
f(x) = 0, if x < 0
= x, if x >= 0
The latter equality can be modeled thru the following linear
constraints:
x = x1 + x2
f = x1
x1, x2 >= 0
where x1, x2 are auxiliary variables.
f(x-1) can be modeled in the same way by taking y = x-1.
(Check all this carefully for errors.)
>
>
> > param maxN default 1000;
> > param maxJ default 10;
> > set N := 1 .. maxN;
> > set J := 1 .. maxJ;
> > param a{N};
> > param w{N};
> > var X0;
> > var X{J};
> > var S{maxJ .. maxN};
>
>
> > maximize Obj: sum {n in N} w[n] * crop(S[n])
>
> > subject to DefineS {n in maxJ .. maxN}: S[n] = X0 + sum {j in J}
> a[n-j+1] * X[j]
>
>
> [1]: http://lists.gnu.org/archive/html/help-glpk/2007-06/msg00005.html
>
_______________________________________________
Help-glpk mailing list
Help-glpk@gnu.org
https://lists.gnu.org/mailman/listinfo/help-glpk
