Stochastic Convergence of Sobol-Based Mahalanobis Shell Sampling Collision Probability Computation

What is it about?

Under certain assumptions, the Mahalanobis Shell Sampling (MSS) algorithm allows to directly obtain sample points that are unlikely to be drawn in a regular random sample, which is helpful when trying to study rare events. In this paper, we present the MSS algorithm, and we measure its statistical performance against the Monte Carlo benchmark, specifically, in the context of measuring the instantaneous risk of collisions between two agents that have bounded orbits relative to one another.

Featured Image