How to deal with minimization of objectives?

[RossXli]

In my case to minimize one of objectives, do I need to multiply the objective training data by -1 in advance for model initialization since most AF only deal with maximization problems.

[eytan]

Yes.

…

On Sun, Jun 12, 2022 at 8:13 PM RossLi ***@***.***> wrote: In my case to minimize one of objectives, do I need to multiply the objective training data by -1 in advance for model initialization since most AF only deal with maximization problems. — Reply to this email directly, view it on GitHub <#1258>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAW34IMB4N2JNZCBQKBGW3VO2DKZANCNFSM5YSTGV3A> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[saitcakmak]

Alternatively, you can pass a WeightedMCMultiOutputObjective to the acquisition function with weights denoting the optimization direction for each objective (1 for maximization, -1 for minimization`: https://github.com/pytorch/botorch/blob/main/botorch/acquisition/multi_objective/objective.py#L110