Dear users,
I have two questions.
Q1: I have a MIP model with the below variables:
.
.
.
positive variable a,b,c,d;
binary variable e,f,g,h;
.
.
.
I want to solve the model with relaxation just e and f, not all. I
know for relaxation all variables RMIP can be used. Moreover, I know I
can consider positive e and f. However, I solve the model sometimes
with relaxation not always. How can I do it?
Q2: How can I extract from GAMS the constraints lead the model to infeasibility?
\
Hi,
Your first question can be easily answered. GAMS discrete variable have the priority attribute. Setting this to inf makes the variable relaxed even if you solve as a MIP:
e.prior(i,j,k) = inf;
Answering Q2 is harder. You could add slacks you your constraints and penalize them in the objective. To see where the infeasibilities are. Some solvers have an options for this (e.g. Cplex feasopt). If you know a feasible solution to your problem you can preload (using var.l(i,j,k) = …) this and look at the equation listing (increase limrow to a large number to see all equations in a block) to see which equations are infeasible to isolate your problem with the model/data/logic/…
Hope this helps,
Michael
On Sunday, October 27, 2013 7:53:03 AM UTC-4, hossein karimi wrote:
Dear users,
I have two questions.
Q1: I have a MIP model with the below variables:
.
.
.
positive variable a,b,c,d;
binary variable e,f,g,h;
.
.
.
I want to solve the model with relaxation just e and f, not all. I
know for relaxation all variables RMIP can be used. Moreover, I know I
can consider positive e and f. However, I solve the model sometimes
with relaxation not always. How can I do it?
Q2: How can I extract from GAMS the constraints lead the model to infeasibility?
–
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/groups/opt_out.