Spatiotemporal quantum advantages for rare-event sampling
π arXiv:2610.03026 Β· π₯ PDF Β· 2026-10-02 Β· cond-mat.stat-mech
Authors: Akshat Gupta [arXiv Β· scholar] , Jianjun Chen [arXiv Β· scholar] , Chengran Yang [arXiv Β· scholar] , Mile Gu [arXiv Β· scholar] , Caterina Doglioni [arXiv Β· scholar] , Jayne Thompson [arXiv Β· scholar] , Thomas J. Elliott [arXiv Β· scholar]
π Abstract
In a world fraught with uncertainty, it behooves us to consider the most disastrous possible outcomes, in order that we can best mitigate against them. Often these are extreme risk events -- of high consequence, but low probability. Yet, the rarity of such events limits our ability to study them; there is a paucity of past occurrences to draw upon, and sampling such rare events from models is typically costly in time. Here, we introduce an approach for rare-event sampling based on quantum stochastic models. This integrates the memory advantages of quantum stochastic modelling -- and its associated efficient means of preparing quantum sample states -- with the `speed-ups' of quantum amplitude amplification, to yield simultaneous reductions in spatial and temporal computational complexity compared to classical Monte Carlo-based approaches.