Where a simulation is stochastic (such as agent-based simulations), the simulation should be run n number of times (n can be established using Technique 1 - Aleatory Analysis), after which the first method of this technique will generate median values for each output measure over the n runs. The technique above provides the means to sample the parameter space using a latin hypercube approach, and this technique allows for the analysis of simulation results generated using these parameter sets. Thus we have included this method described by Read et al, Saltelli et al, and others (Reference's below). Though Technique 2 does elucidate the effects of perturbations of one parameter, it cannot show any non-linear effects which occur when two or more are adjusted simultaneously. Technique 3: Latin-Hypercube: Perform Analysis of Results: LHC: Perform Analysis of Results Description
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |