Originally posted by: gtoptimizer

Hey all,

I read a MIP from a .lp file. The MIP has some of its integer variables fixed to specific values.

Before solving the MIP, I am interested in eliminating the fixed variables from the model.

For example, given the following MIP as input:

Minimize
 obj: X1 - X2 + X3
Subject To
 A1: 3 X1 + X2 = 44 
 A2: X2 = 23
 A3: X3 + X2 >= 24
Bounds
 12 <= X1 <= 55
       X2 <= 23
  0 <= X3 <= 1
End

I would like to obtain the following MIP:

Minimize
 obj: X3
Subject to
 B1: X3 >= 1
Bounds
  0 <= X3 <= 1
End

Are there existing functions in CPLEX that can eliminate fixed variables from a MIP model?

I thought CPXpresolve would do this, but the model it produces is identical to the original one (i.e. the fixed variables are still there in the constraints and objective).

I can think of a way of performing the elimination by iterating over the model components and modifying them, but I would rather use a standard CPLEX call if possible.

Thanks!

Originally posted by: gtoptimizer

Solved: CPXgetredlp should be called after CPXpresolve to get the presolved MIP.