Best Approach for Optimisation using both doubles and integers as variables

I have an optimisation problem where the inputs are both integer based (take from a table of options say 50) and double based (dimensions). Is there a way of including both types of variables in an optimisation using the optimisation toolbox or other method?

What I have been doing is limiting the GA between 0 and 1 then scaling the values. For the integer I have ben assigning them ranges. For example for 10 options 0 to 0.1 would be 1, 0.1 to 0.2 is 2 and so on. I know this is not very efficient

Any suggestions would be appreciated

Ross

Answer 1

[Alan Weiss]

If you have Global Optimization Toolbox R2011b or later, then mixed integer programming is a built-in feature. See the documentation:

http://www.mathworks.com/help/gads/mixed-integer-optimization.html

http://www.mathworks.com/help/gads/solving-a-mixed-integer-engineering-design-problem-using-the-genetic-algorithm.html

If you have R2011a or earlier, you can try this tech support solution, which will require you to do some programming.

Good luck,

Alan Weiss

MATLAB mathematical toolbox documentation

[Ross]

Thanks for you response

Do you know if there is a local search algorithm as well. The GA has problems finding the optimum fully it is very effective in finding the correct local area. This is not unexpected. Is there a local search algorithm that i can give the GA optimum to improve it further?

ross