Implementation of BO

[SamirLamin]

I have an optimization problem with 3 objectives and 2 variables decisions and the functios are a black box. I need to implement BO using Botorch to solve my problem so I need to create a class of my problem in multi_objective.
My question is how I can create my problem like other problems and respect the same way to implement BO.
?
My second question the evaluate_true is the function that evaluates a solution to find the best solution?

[eytan]

Have you tried checking out http://ax.dev —it’s our easy to use interface for practitioners who want to do black box optimization. BoTorch is a lower level framework targeted to people who are developing new Bayesian optimization methods. Ax should be easier to get started.

…

On Wed, Feb 1, 2023 at 12:22 PM SamirLamin ***@***.***> wrote: I have an optimization problem with 3 objectives and 2 variables decisions and the functios are a black box. I need to implement BO using Botorch to solve my problem so I need to create a class of my problem in multi_objective. My question is how I can create my problem like other problems and respect the same way to implement BO. ? My second question the evaluate_true is the function that evaluates a solution to find the best solution? — Reply to this email directly, view it on GitHub <#1649>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAW34LXC2ZW43Z64OAHP7DWVKLVBANCNFSM6AAAAAAUN765OU> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[SamirLamin]

Ok, thank you so much I will check it. but where and how I can add my problem and Implement the BO

[saitcakmak]

Hi @SamirLamin. Ax Service API is probably the easiest way for you to get started. Section 2 of the tutorial shows how tosetup the optimizer with your search space and objectives, and section 3 is where you'd hook up the objective evaluations (could be any arbitrary python code that evaluates your problem): https://ax.dev/tutorials/gpei_hartmann_service.html