Bayesian optimisation (coupled?) constraints - Bayesopt

consider an optimization with two variables x1, x2 with the following bounds:

a<x1<b

a<x2<x1

What is the best approach to implement this using BayesOpt? I am using the following setup, but not sure if it's the most efficient way:

----------------------------------------------

x1 = optimizableVariable('x1',[a,b]);

x2 = optimizableVariable('x2',[a,b]);

bayesopt(fun,[x1, x2],'XConstraintFcn',@xconstraint)

--------------

function tf = xconstraint(X)

tf = X.x1 > X.x2;

end

----------------------------------------------

Answer 1

[Alan Weiss]

I think that you have set up the problem correctly, and in the most efficient way.

Alan Weiss

MATLAB mathematical toolbox documentation