A loop inside another loop: problems with fixed variables

Archiver · April 30, 2014, 5:39pm

Dear GAMS users,

I have implemented a model with one loop in GAMS that works very well. Now, I would like to use this model in a sensitivity analysis framework and for this I need to include another loop in the model: loop j, please see the simplified version of the model below. The loop j includes the loop dd. At the beginning we enter loop j (j=1, d=1) and solve problem1 with default initial values. Next, we enter loop dd (j=1, dd=2) and solve problem1 with X(‘1’) and Y(‘1’) fixed based on values init_x and init_y derived from the previous solution. The problem occurs when j=2 and d=1: the problem1 is solved using the fixed values X(‘1’) and Y(‘1’) that are taken from the solution of problem1 in dd loop (j=1, dd=365) . Then I get an infeasible solution since the Constraint1(‘1’) is not satisfied (X(‘1’)+Y(‘1’) is not equal to A(‘2’,‘1’,‘1’)).
Is there a possibility to unfix the variables X and Y when j increases by 1?
I mean, how to remove the fixation of X(‘1’) and Y(‘1’) obtained in the dd loop with dd=365?
So that, when j increases by 1 and d=1, problem1 can be solved without any fixation (X and Y are free just like at the beginning when j=1 and d=1).

I would be very thankful for any help on this issue.
Thank you,
Best regards,
Alicja,
PhD candidate
TU Delft

Simplified model with two loops (one loop inside another loop):

Sets
j /1100/
d /1365/
dd /2365/
h /136/

Parameters
init_x
init_y
a
b
A(j,d,h)

Variables
Y(h)
X(h)
TotalCost

Equations
Constraint1(h);
LeastCost;

Constraint1(h)…X(h)+Y(h)=e=A(j,d,h);
LeastCost… TotalCost=e=asum(h,X(h))+bsum(h,Y(h))

Model problem1 /all/;

loop(j,
d=1;
Solve problem1;
init_x=X.l(‘25’);
init_y=Y.l(‘25’);

loop(dd,
X.fx(‘1’)=init_x;
Y.fx(‘1’)=init_y;
Solve problem1
init_x=X.l(‘25’);
init_y=Y.l(‘25’);
);

);

\

To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/d/optout.

Archiver · May 2, 2014, 9:40am

Hi Alicja

You could use an if-condition and the ord of the outer loop:

if ord(j) eq 1 then fix the variables, otherwise unfix the variables by setting lower and upper bounds for these variables equal to +INF and â€“INF.

Cheers

Renger

\

Modelworks

Gewerbestrasse 15

3600 Thun â€“ Switzerland

+41 79 818 53 73

Info@modelworks.ch

blog.modelworks.ch

From: gamsworld@googlegroups.com [mailto:gamsworld@googlegroups.com] On Behalf Of Alicja Lojowska
Sent: Mittwoch, 30. April 2014 17:39
To: gamsworld@googlegroups.com
Subject: A loop inside another loop: problems with fixed variables

Dear GAMS users,

I have implemented a model with one loop in GAMS that works very well. Now, I would like to use this model in a sensitivity analysis framework and for this I need to include another loop in the model: loop j, please see the simplified version of the model below. The loop j includes the loop dd. At the beginning we enter loop j (j=1, d=1) and solve problem1 with default initial values. Next, we enter loop dd (j=1, dd=2) and solve problem1 with X(‘1’) and Y(‘1’) fixed based on values init_x and init_y derived from the previous solution. The problem occurs when j=2 and d=1: the problem1 is solved using the fixed values X(‘1’) and Y(‘1’) that are taken from the solution of problem1 in dd loop (j=1, dd=365) . Then I get an infeasible solution since the Constraint1(‘1’) is not satisfied (X(‘1’)+Y(‘1’) is not equal to A(‘2’,‘1’,‘1’)).
Is there a possibility to unfix the variables X and Y when j increases by 1?
I mean, how to remove the fixation of X(‘1’) and Y(‘1’) obtained in the dd loop with dd=365?
So that, when j increases by 1 and d=1, problem1 can be solved without any fixation (X and Y are free just like at the beginning when j=1 and d=1).