Genetic optimization with 2000 integer variables and 800 equality constraints and 300 inequality constraints all involving integer variables using MATLAB R 2013

I want to design a supply chain network's multi-period production and distribution problem using genetic algorithm. I solved this problem using GAMS modeling language with CPLEX MIP solver. I want to compare these results with genetic optimization. The problem involves 2000 integer variables and 800 equality constraints and 300 inequality constraints all involving integer variables. Is it possible to run genetic algorithm using MATLAB R 2013 for this problem . I also want to know whether we can use multidimensional arrays for integer variables in my model for genetic algorithm in MATLAB.

Answer 1

[Matt J]

Technically, ga() does not allow integer problems with equality constraints, but you can eliminate the equality constraints and reformulate the problem as described here,

https://www.mathworks.com/matlabcentral/answers/344574-how-can-i-convert-equality-constraints-into-inequality-constraints

As for multidimensional arrays, it doesn't matter what dimensions the unknown variables x have. The solver will always reshape it into a vector, however, when applying the inequality constraint matrices A*x<=b.

[Walter Roberson]

I would add to this that I would expect ga results for a problem with 2000 variables to be poor.