Originally posted by: Bogdan_Tanasa

Hi guys,

I have a mixed-ILP and the variables are of form (u, v). For each boolean variable u there exists a fractional variable v. I would like to get all integer solutions of the problem. Can populate be tuned to search only for new solutions which change the boolean variable?

Can you please help me with this task?
Thanks,
Bogdan.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Yes. Populate can actually only find solutions that differ in the binary (or integer) variables. Populate will not yield solutions that only differ in the continuous variables.

In order to get all integer solutions to a problem, you need to set the populate parameters accordingly. In the interactive CPLEX shell you would use

set mip pool intensity 4
set mip limits populate 2100000000

The first parameter indicates that populate should collect all feasible solutions.
The second parameter makes sure that populate will only stop after having processed the solution space exhaustively (or if 2.1 billion solutions have been found, which would probably not fit into your memory anyway).
Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Bogdan_Tanasa

Can also one choose a subset out of the total number of boolean variables and ask populate to get solutions which only change this variables?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

No, sorry. This is not possible.

Of course you can for example take some solution, fix some variables to their corresponding solution values, and then let populate enumerate all solutions in the remaining problem space.

Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

While it is not possible to explicitly require that the values for certain variables are the same, you can still use solution pool filters. Both diversity and range filters allow you to prevent certain solutions from entering the solution pool.
Using those filters CPLEX would still enumerate all solutions but it would only store those solutions in the solution pool that satisfy your requirement.
#CPLEXOptimizers
#DecisionOptimization