Computing Sensitivities or Solving Parameter Estimation Problems in ODEs and DDEs

In recent years we have developed reliable order $p$ methods for the approximate solution of general systems of IVPs and DDEs. We have used these methods to implement effective techniques for parameter estimation and sensitivity analysis for IVPs and DDEs. The two techniques we have investigated are the "variational approach" and the "adjoint approach". We will discuss the cost and reliability of each approach and identify several factors that contribute to the performance of each. In particular we will discuss why, for each technique, the underlying numerical IVP or DDE solver must exploit inherent parallelism if the number of state variables, the number of parameters, or the number of data points is large.Non UBCUnreviewedAuthor affiliation: University of TorontoFacult