On fitting differential-equation models to data
Simon Preston (University of Nottingham)
Thursday 26th January, 2012 14:00-15:00 Math & Stats Bldg, LT 325
Fitting ordinary differential equation (ODE)-based models to data and making inferential statements about, say, the values of model parameters can be scientifically valuable. But this is generally a difficult problem, because for most ODE models there is a strongly non-linear relationship between the parameters and the model solution. I will give an introduction to aspects of fitting linear and non-linear (e.g. ODE) models to data, and talk about classical and Bayesian approaches to parameter inference. I will also discuss some recent developments, including the use of Markov Chain Monte Carlo (MCMC) methods, tempering approaches, and the use of fitting criteria different to the default "likelihood"/"least-squares" criterion.