evomo.metrics.hv#

Functions#

hv(objs, ref[, num_sample])

Estimate hypervolume for minimization using Monte Carlo sampling.

Module Contents#

evomo.metrics.hv.hv(objs: torch.Tensor, ref: torch.Tensor, num_sample: int = 100000)#

Estimate hypervolume for minimization using Monte Carlo sampling.

Points outside the reference point contribute no volume. For maximization, negate both the objectives and reference point before calling this function.

Parameters:

  • objs – Objective points of shape (n_points, n_objs).

  • ref – Reference point of shape (n_objs, ).

  • num_sample – Number of Monte Carlo samples.

Returns:

Estimated hypervolume.