Parameter estimation for deterministic simulators with time series output
LE3 .A278 2010
2010
Ranjan, Pritam
Acadia University
Bachelor of Science
Honours
Mathematics and Statistics
Mathematics & Statistics
Computer simulators provide an easy way to emulate complex real world phenomena using statistical models. Modeling real life phenomena is done because conducting real life experiments will be difficult or even infeasible. Typically simulators are deterministic and computationally expensive, and as a result modeling simulators with statistical emulators has become popular. It is often beneficial to find inputs for the simulator corresponding to a particular output and solving the inverse problem for computer simulators with scalar output is a non- trivial but known procedure. Solving the inverse problem for computer simulators with functional output can be difficult. We will devise a new method for solving the inverse problem for simulators with functional responses based on the method for scalar responses. We will show that this method works for finding inverse solutions efficiently. We will attempt to find the inverse solution to a model that simulates the population dynamics of European red mites on apple trees in the Annapolis Valley, NS and the inverse solution to the Canadian Integrated Modeling Systems ( CIMS) model, an energy- economy model for predicting technology changes.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.
https://scholar.acadiau.ca/islandora/object/theses:670