Dear Destin

I was aware of this procedure and tried it in a MIP model with a very strong contraint which usually gives CPLEX a hard time to find a feasible solution. For an example the solver did not find any (feasible solution), I provided an initial solution but it doesn’t seem to make any difference. I thought CPLEX would consider this solution the best integer one but GAMS still reports no feasible solution found.

Below I included some relevante parts of the model:

Variables

X(I,J)

Y(I)

ORDER(I)

CT

MV

Z1;

binary variables X;

integer variables CT, ORDER;

X.l(“0”,“19”) = 1;

X.l(“19”,“20”) = 1;

X.l(“20”,“28”) = 1;

*** and so on…

ORDER.l(“19”) = 1;

ORDER.l(“20”) = 2;

ORDER.l(“28”) = 3;

*** and so on…

CT.l = 0;

## Equations

OBJFUN1

*** and so on…

I would appreciate if you can provide any clarification.

Thanks,

Vicky

Em sexta-feira, 25 de maio de 2012 06h50min31s UTC-3, Destin Zed escreveu:

Hi,

Yea, I also agree that sometimes a warm-start helps in an MIP problem. It is good to tell the solver where to start digging in. I think most solvers by default initialize all variables to zero in the first iteration. Here is what I believe you should do in order that work for you.

Self-initialize your (integer) variables before the SOLVE statement (well, if you more or less know how the solution would look like). You can do something like:

yourVariableName.L(‘yourIndices-SingleQuoted’) = YourPreferredIntegerValue;

Example: Suppose you have a binary variable b over the indices i and j (i=1,2,…10; j=10,11,…, 20). We can initialize b(‘2’,‘13’) as:

b.l(‘2’,‘13’) = 1;

I hope that helps,

Destin

On Thu, May 24, 2012 at 4:25 PM, JapanGo wrote:

Dear all,

I am currently working on a quite big sized MIP problem.

I read that it is possible to offer CPLEX a “warm start”, i.e. an

initial solution.

How can I implement/include this solution?

I assume it is similar to an include file, just with values for my

decision variables…

Thanks a lot and have a great day!

To view this discussion on the web visit https://groups.google.com/d/msg/gamsworld/-/cdDndcSlkfwJ.

To post to this group, send email to gamsworld@googlegroups.com.

To unsubscribe from this group, send email to gamsworld+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/gamsworld?hl=en.