Methods of nonlinear parameter estimation: dynamical coupling and Monte Carlo path integral formulations
By dynamical coupling of data with known models, we determine underlying parameters and unmeasured state variables for a variety of systems, including Lorenz, Colpitts, and Hodgkin-Huxley neurons. The dynamic coupling is mediated by use of a cost function, which is minimized by optimization software (SNOPT, IPOPT) to achieve the desired synchronization. Considering measurement and model noise, we discuss the same problem where we introduced an 'action' in order to estimate states and parameters. This action is used along with Markov Chain Monte Carlo methods in order to sample from a probability distribution.
Tuesday, January 18, 2011
11:00AM AP&M 2402
Center for Computational Mathematics9500 Gilman Dr. #0112La Jolla, CA 92093-0112Tel: (858)534-9813