optuna_integration.botorch.qhvkg_candidates_func

optuna_integration.botorch.qhvkg_candidates_func(train_x, train_obj, train_con, bounds, pending_x)

Quasi MC-based batch Hypervolume Knowledge Gradient (qHVKG).

According to Botorch/Ax documentation, this function may perform better than qEHVI (qehvi_candidates_func). (cf. https://botorch.org/tutorials/decoupled_mobo )

See also

qei_candidates_func() for argument and return value descriptions.

Note

Added in v4.0.0 as an experimental feature. The interface may change in newer versions without prior notice. See https://github.com/optuna/optuna/releases/tag/v4.0.0.

Parameters:

• train_x (torch.Tensor)

• train_obj (torch.Tensor)

• train_con ('torch.Tensor' | None)

• bounds (torch.Tensor)

• pending_x ('torch.Tensor' | None)

Return type:

torch.Tensor