mixed integer(all are binary) problem

I solved mixed integer problem. Specification are as follows

• 38 variables where 23 are integer variables and rest of them are real variables.
• 70 constrains 10 of them are equality constrians.
• lower limit 0 and upper limit 1 of all variables.
• I write the equality constrian in the form 3x1 – 2x2 ≤ 5,–3x1 + 2x2 ≤ –5 as give in matlab.
• Equaltiy constrain are on real variables.
• All constrains are linear.

It is not satisfying linear constrains. I run ga even with 2000 generation population size of 500 cross over fraction .8 stall generation 2000 elite count 50.

GA cant solved mixed integer programming?

I have solved same problem from lingo also.It was satisfying constrian.

Thanks you in advance .

Answer 1

[Johan Löfberg]

If you only have linear constraints, and the problem is as small as it is here, you should really use a dedicated mixed-integer linear solver. In my opinion, GA is the wrong tool here.

There are many free MILP solvers with interfaces to MATLAB, such as scip, cbc, lpsolve, glpkmex etc, and many industry-strength solvers with free licenses for academia (cplex, gurobi, mosek)

[suvrat]

Thank you very much for your reply.

My Actual problem is mixed integer linear programming with 346 variable and 336 constrain.

Actually I have to compare result obtain from ga and traditional approach.

And if GA is violating the constrain what is the reason behind it.