Originally posted by: SystemAdmin

[eothein said:]

Hi everybody, I am new to this forum and I have some questions.

The first questions deals with the output of the optimizer (I use the java callable library).


Clique table members: 335086.
Reduced MIP has 23512 rows and 445865 columns


What do these parameters mean : does this mean my ILP has 23512*445865 variables?

My second questions deals with the solution space and time to find a solution.

I have two (almost identical) ILP formulations where the number of variables only differs a factor from each other, but still the ILP with the larger search space (the most variables) instantly finds a solution while the other can work for more than several hours trying to find the solution... How can this be explained..?


#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

[notdarkyet said:]

You're right, your LP - after presolve reductions, which were stated above the line you quoted - has this number of vars and constraints.

Regarding your second question have a look at e.g. the book by Wolsey in the chapter about the quality of formulations. It might well be that by introducing
more variables the formulation (the polyhedron of the relaxed problem) is tighter - and easier to solve. There are examples given in the textbooks to demonstrate
this.


#CPLEXOptimizers
#DecisionOptimization