[Flopcpp] Antw: [Coin-discuss] GAMS to FlopC++-Translation

[Tim Hultberg]

Hi Christian,

1) your idea does not seem right to me. You want to limit the number of
constrainst NOT the left hand side terms. So you
should use

AP_AlleEinplan(c,d,g).such_that(DC(d,c)).such_that(DG(d,g))  =  ...

actually  
AP_AlleEinplan(c,d,g).such_that(DC(d,c)*DG(d,g))  =  ...
should really work as well, but doesnt seem to at the moment, I ll look
into this


2) your idea seems fine to me. But you will run into trouble because
FLOPC++ only allows for a maximum of 5 indices (see the homepage). I ll
take an action to increase the max number of indices to 8, but dont
expect this change to happen soon since I am very busy with other
things. In the meanwhile you can create a workaround using cartesean
MP_subsets

Cheers, Tim 


Tim Hultberg

>>> WoelliUSM at aol.com 22.02.06 19.13 >>>
Dear all,
 
I got two questions about how to "translate" Gams-equations to
FlopC++-constraints.
 
1)
 
The problem is, that I don't know how to translate "$(DC(d,c) and
DG(d,g))" to "such_that" in this special case:
 
GAMS:
..
Equations
AP_AlleEinplan(c,d,g)
..
AP_AlleEinplan(c,d,g)$(DC(d,c) and DG(d,g)).. sum(r,sum(t,
x(r,t,d,g)))=e=1;
 
My idea about this:
 
FlopC++:
..
MP_constraint
 AP_AlleEinplan(c,d,g)
..
AP_AlleEinplan(c,d,g) = sum(d.such_that(DC(c,d)>-2),
sum(g.such_that(DG(d,g)>-2), sum(r,sum(t, x(r,t,d,g)))))==1;

2)
 
The second problem is about "Alias" in a similar case as (1):
 
GAMS:
..
Alias(d, dtilde)
Alias(g, gtilde)
..
Equations
AP_GruppenLink(c,l,g,gtilde,d,dtilde,r,t) 
..
AP_GruppenLink(c,l,g,gtilde,d,dtilde,r,t)..
x(r,t,d,g)=e=x(r,t+1,dtilde,gtilde);
 
FlopC++:
..
MP_index dtilde, gtilde;
..
MP_constraint
AP_GruppenLink(c,l,g,g(gtilde),d,d(dtilde),r,t) 
..
AP_GruppenLink(c,l,g,g(gtilde),d,d(dtilde),r,t)..
x(r,t,d,g)==x(r,t+1,d(dtilde),g(gtilde));
 
 
Thank you in advance,
 
Greetings from Cologne,
 
Christian
_______________________________________________
Coin-discuss mailing list
Coin-discuss at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-discuss