How to find the index of a particular child when using use parallel = True option in genetic algorithm optimizer?

I am using genetic algorithm in global optimization toolbox.

I have an objective function within which

i) I create directories

ii) run an executable using system() command

iii) write and read files to evaluate the objective function for a given child

I would like to evaluate the objective function for different childs in parallel using the UseParallel=True option. However, I would need to know the index of the particular child in the population so that I can create a directory with the name corresponding to the index of the particular child. This would mean that I will need to use the index of the child within my objective function.

So, my question is how to find the index of a particular child when using the parallel option in genetic algorithm optimizer?

It would also be great if there is a solution to find the index during parallel operation of other optimizers (eg: patternsearch, etc.,) in MATLAB as well. Thank you.

Answer 1

[Matt J]

If you have the Parallel Computing Toolbox, set UseParallel=false and UseVectorized=true. Then, inside your objective function and constraints, you will have access to all poopulation members and their indices. You can use parfor or other PCT tools to parallel-split the operations within those functions, but with full control over the splitting.

[Sravan Kumar Kanchari Bavajigari]

Thanks Matt for the suggestion. I thought only one child (the optimizer guess) is passed to the objective function during the optimization. Can you please let me know how I can access the population members in the objective function with UseVectorized=true option? thanks

[Matt J]

When UseVectorized=true, the fitness and constraint functions will be passed an MxN matrix in each generation, where each row of the matrix is a population member. In this case, your functions are responsible for processing the entire population in a single call, see also,

https://www.mathworks.com/help/gads/vectorizing-the-fitness-function.html

[Sravan Kumar Kanchari Bavajigari]

This works. Thank you.