Bayesopt library extracts data which violate the given deterministic constraints.

I utilized Bayesopt Matlab library to implement bayesian optimization and have two questions about bayesopt.

First, I found an issue related to deterministic constraints.

Though I set the deterministic constraints, the bayesopt extracts points which violated the deterministic constraints.

can bayesopt library extract data which violate the given deterministic constraints?

or if not, how can I correct this matter? (Additionally, there is no error about failing to find sufficient data for deterministic constraints evaluation)

Second, though I can set the deterministic constraint, time consumption for evaluation of constraint is too much.

So, I want to make surrogate model for the constraint which will be multiplied to the acquisition function(EI).

How can I generate the code for it?

Answer 1

[Umang Pandey]

Hi,

Bayesian optimization in MATLAB using the “bayesopt” function respects the deterministic constraints if they are correctly defined.

To know more about how these constraints are handled, you can refer to the following documentation on constraints handling in Bayesian Optimization:

https://www.mathworks.com/help/stats/constraints-in-bayesian-optimization.html

The above documentation will also come in handy for knowing about implementational aspects of constraints and designing your own surrogate model.

You can also refer to the following documentation on “bayesopt” for more information:

https://www.mathworks.com/help/stats/bayesopt.html?s_tid=answers_rc2-2_p5_MLT

Hope this helps!

Best,

Umang