Robust Optimization with Rhodium

[bre-luise]

Dear David Hadka, Dear Team,

I would like to use the robust_optimize function for my MORDM in rhodium. Unfortunately, I can't find any documentation for it. Is there another possibility than to use the mean setting for the obj_aggregate. If so, which one?

Thanks in advance!

[dhadka]

Apologies for the lack of documentation here. Essentially, the model is getting evaluated for each "state of the world" (SOW) which results in an array of values.

obj_aggregate and constr_aggregate take this array and converts it into a single numeric value summarizing the response across all states of the world. So in that regard, any function that takes a array / list of values and returns a single number can be used. Whether that's the default (mean), other Python Math functions like min, max, sum, etc, or your own function.

Furthermore, you can also supply different aggregate functions for each model response. Simply create a dictionary, where the key is the response name and the value is the aggregate function. For example, obj_aggregate = {'response1': mean, 'response2': max}.

Please let me know if you have any other questions!