Sample Reuse Techniques of Randomized Algorithms for Control under Uncertainty

TL;DR

This paper extends sample reuse techniques for randomized algorithms in control under uncertainty, demonstrating that the numerical complexity remains bounded for nested sets, even if disconnected, enabling more efficient probabilistic robustness analysis.

Contribution

It generalizes sample reuse to arbitrary nested sets, including disconnected ones, and integrates deterministic and probabilistic analysis for complex decision problems.

Findings

Complexity of i.i.d. experiments is bounded for nested sets.

Sample reuse applies to disconnected and complex sets.

Facilitates integration of deterministic and probabilistic methods.

Abstract

Sample reuse techniques have significantly reduced the numerical complexity of probabilistic robustness analysis. Existing results show that for a nested collection of hyper-spheres the complexity of the problem of performing $N$ equivalent i.i.d. (identical and independent) experiments for each sphere is absolutely bounded, independent of the number of spheres and depending only on the initial and final radii. In this chapter we elevate sample reuse to a new level of generality and establish that the numerical complexity of performing $N$ equivalent i.i.d. experiments for a chain of sets is absolutely bounded if the sets are nested. Each set does not even have to be connected, as long as the nested property holds. Thus, for example, the result permits the integration of deterministic and probabilistic analysis to eliminate regions from an uncertainty set and reduce even further the complexity of some problems. With a more general view, the result enables the analysis of complex decision problems mixing real-valued and discrete-valued random variables.