New Conditional-path Monte Carlo algorithm simulates rare network events

Simulating rare macroscopic events in stochastic networks—such as widespread epidemics, cascading failures, or metastable escapes—has long been stymied by catastrophic rejection rates, weight degeneracy, and critical slowing down. In a new technical paper, Thomas Barthel, Jiazheng Sun, and Jhao-Hong Peng (arXiv:2608.17511) provide the rigorous mathematical backbone for Conditional-path Monte Carlo (CPMC), a method that sidesteps these pitfalls with non-local Swendsen-Wang-like cluster updates operating directly on full system trajectories.

The authors formalize joint path-graph probability weights and derive transition and uniformization sum rules that guarantee detailed balance. Applying CPMC to susceptible-infectious-susceptible (SIS) models, they systematically optimize single-node and edge graph vertex sets to prevent lock avalanches while preserving the mobility of the epidemic trunk. A dynamic programming scheme exactly encodes patient-zero and macroscopic outbreak-size constraints, enabling rejection-free generation of valid epidemic trajectories.

Beyond foundational rigor, the paper details computational complexity, parallelization strategies, and validation against exact solutions on small networks. As the companion to the framework's initial demonstration, this work gives researchers the tools to tackle rare events in epidemiology, infrastructure resilience, and many-body physics—opening the door to practical, large-scale simulations that were previously intractable.