deephyper.skopt.moo.non_dominated_set_ranked#

deephyper.skopt.moo.non_dominated_set_ranked(y, fraction, return_mask=True)[source]#

Find the set of top-fraction x 100% of non-dominated points. The number of points returned is min(n_points, ceil(fraction * n_points)) where n_points is the number of points in the input array. Function assumes minimization.

Parameters:

  • y (array or list) – Array or list of size (n_points, n_objectives)

  • fraction (float or int) – Fraction of points to return.

  • return_mask (bool, optional) – Whether to return a mask or the actual indices of the non-dominated set. Defaults to True.

Raises:

ValueError – Raised if fraction is not a non-negative number or if y is not an array of size (n_points, n_objectives).

Returns:

If return_mask is True, this will be an (n_points, ) boolean array. Else it will be a 1-d integer array of indices indicating which points are in the top non-dominated set.

Return type:

array