pystra.active_learning.learning.learning_eff

learning_eff(mean, std, *, threshold=0.001)

Return expected feasibility, best index (maximum), and stopping flag.

Here G is Gaussian with the supplied mean/std, each (n_candidates,). EFF is symmetric in mean and has the units of the limit state. threshold is an absolute, positive tolerance in those units. Zero spread gives zero EFF, including at the boundary.

Return type:

tuple