Design of Simulation Experiments in broad terms refers to the deliberate process of choosing and sequencing the set of scenarios to simulate in order to gain insight into the modelled input-output relationship underlying the phenomenon under study. Analytical objectives include efficiently identifying the key influential factors from the vast array of inputs (screening); characterising and quantifying the statistically significant joint contribution of these factors and their interactions on modelled operational effectiveness (meta-modelling); determining the ideal values of these factors which are predicted to maximise operational effectiveness (optimisation); or comparing a discrete set of specific configurations of input factors (ranking and selection). Key design considerations include the assignment of pseudorandom number streams; optimality with respect to meta-model choice and potential presence of heteroscedasticity and/or correlation in the data; and robust statistical testing procedures. This session particularly seeks submissions on theoretical considerations or the sharing of experiences from defence and other applications.
Key topics: Design of experiments, Statistical efficiency, Meta-model