Assignment of probabilities to events for combat simulation

Multi trajectory simulation allows explicit management of random events by allowing particular events to be resolved by random draw, by a deterministic choice, or by creating new states to allow following multiple trajectories. The policy for resolution method is under the control of the analyst, and may depend on event type, trajectory probability, or even some metric indicating the trajectory importance. However, taking advantage of the technique for exploring possible outcome spaces requires a probabilistic modeling of events that, in simulations of ground combat, are often treated as deterministic, such as decisionmaking. Even for events such as attrition, which have long been modeled as stochastic, how should the full event outcome set be sampled if one is only to keep two or three samples? This paper explores these issues with the goal of outlining what kinds of data would be needed to fully exploit multi trajectory methods in a combat simulation context.