def _extract_observed_pareto_2d(
Y: np.ndarray,
reference_point: Optional[Tuple[float, float]],
minimize: Union[bool, Tuple[bool, bool]] = True,
) -> np.ndarray:
if Y.shape[1] != 2:
raise NotImplementedError("Currently only the 2-dim case is handled.")
# If `minimize` is a bool, apply to both dimensions
if isinstance(minimize, bool):
minimize = (minimize, minimize)
Y_copy = deepcopy(torch.from_numpy(Y).to())
if reference_point:
ref_point = torch.tensor(reference_point, dtype=Y_copy.dtype)
for i in range(2):
# Filter based on reference point
Y_copy = (
Y_copy[Y_copy[:, i] < ref_point[i]]
if minimize[i]
else Y_copy[Y_copy[:, i] > ref_point[i]]
)
for i in range(2):
# Flip sign in each dimension based on minimize
Y_copy[:, i] *= (-1) ** minimize[i]
Y_pareto = Y_copy[is_non_dominated(Y_copy)]
Y_pareto = Y_pareto[torch.argsort(input=Y_pareto[:, 0], descending=True)]
for i in range(2):
# Flip sign back
Y_pareto[:, i] *= (-1) ** minimize[i]
assert Y_pareto.shape[1] == 2 # Y_pareto should have two outcomes.
return Y_pareto.detach().cpu().numpy()
def _extract_observed_pareto_2d(Y: torch.Tensor,
reference_point: Tuple[float, float],
minimize: bool = True) -> torch.Tensor:
if Y.shape[1] != 2:
raise NotImplementedError("Currently only the 2-dim case is handled.")
ref_point = torch.tensor(reference_point, dtype=Y.dtype)
Y_pareto = Y[is_non_dominated(-1 * Y if minimize else Y)]
Y_pareto = (Y_pareto[torch.all(Y_pareto < ref_point, dim=1)] if minimize
else Y_pareto[torch.all(Y_pareto > ref_point, dim=1)])
Y_pareto = Y_pareto[torch.argsort(Y_pareto[:, 0])]
if Y_pareto.shape[0] == 0:
better = "below" if minimize else "above"
raise ValueError(
f"No Pareto-optimal points in `Y` were {better} the reference point."
)
assert Y_pareto.shape[1] == 2 # Y_pareto should have two outcomes.
return Y_pareto