Simulation environment with dynamic optimization of fidelity

Abstract

A simulation environment is disclosed. A simulator generates a simulation model corresponding to a sequence of simulation cycles, each cycle including initial conditions, propagatable states, and a level of simulation fidelity. A fidelity optimizer monitors states and conditions through each cycle, changing the fidelity level for the next cycle based on predetermined factors (e.g., a new flight segment). The fidelity optimizer may dynamically increase or decrease the fidelity level for the next cycle (or for one or more nodes within a group) in response to, or in anticipation of, a state change. The fidelity optimizer may sample several corresponding simulation cycles of varying fidelity levels to compare the effects or results, and determine whether the result is dependent on the fidelity level.