Abstract This package ports 6 constrained multi-objective test problems from BoTorch’s test_functions.multi_objective module to OptunaHub’s BaseProblem interface: BNH, CONSTR, ConstrainedBraninCurrin, OSY, SRN, and MW7. Each problem is reimplemented with plain Python and numpy only, so it can be evaluated without installing torch or botorch.
Conventions Parameters are named x0, x1, … following each problem’s input vector order. evaluate_constraints keys are named c0, c1, … following each problem’s constraint order. BoTorch’s constraint convention is that a slack c_i(x) >= 0 is feasible; OptunaHub’s BaseProblem convention is that a value <= 0 is feasible.
Abstract Standard Bayesian optimization handles uncertainty by taking random samples of model parameters, but this random sampling creates noise. This noise can cause the optimizer to accidentally rank bad configurations above good ones and steer the search in the wrong direction. This issue is especially noticeable in complex tasks with many variables or when computational limits restrict the number of samples you can take.
This package implements OrthoBO, a framework introduced in this paper from May 2026.