Modern Approaches for Scientific Inference and Uncertainty Quantification
Organiser
Participants
Conference
Proposal Description
Different modeling approaches are utilized in the field of environmetrics, such as differential equations, agent based models, and statistical models. Regardless of the approach, the ability to make rigorous inference of unknown quantities (parameters) and their uncertainty, given a set of data, is critical to understanding the science. This session will present some modern approaches for uncertainty quantification for fractional differential equations (Boone) and agent based models (Westveld), which are naturally stochastic; a scalable Monte Carlo (but not MCMC) approach to examine posterior distributions (Marin), and the use of the Kullback-Leibler divergence in making inference for changes in biodiversity (Shimadzu).