Scheduler Dependencies in Agent-Based Models: A Case-Study Using a Contagion Model

Scheduler dependencies in Agent-Based Models (ABM) simulations are known to exist, but often remain sidelined in practice, and continue to be theoretically and technically opaque. Understanding whether the differences introduced by scheduling choices yield significant deviations from both analytic results and phenomenological observation can provide insights on whether those discrepancies may lead to meaningful differences of scientific interpretation or decision making. Here, we demonstrate their presence and non-trivial nature in a variant of the standard agent-based SIRD contagion model, one in which an agent’s local decisions are informed by both their local environment and the consensus of their spatially non-local social network, where event priorities are computed from the network centrality of each agent. Preliminary simulation outcomes suggest that priority scheduling introduces resonant stochastic fluctuations modulated by social mimicry.