 |
Center for Nonlinear Studies |
I will discuss a dynamical method for assimilating information in measured data to a model system allowing for generalized synchronization between the physical system and the augmented model+coupling dynamics. This approach then leads to the re-writing of the parameter and state estimation problem into an optimal tracking framework. Solutions to parameters and states of a data producing system are obtained using a constrained nonlinear optimization package (SNOPT). This method will be demonstrated using several systems (Colpitts, Lorenz, Hodgkin-Huxley) with both laboratory and simulated data. An application inferring the cell membrane potential of spiking neurons in a network from fluorescence measurements of intracellular calcium of those neurons will be introduced. This application has the outlook of predicting the (unobserved) membrane potential of many neurons (those within the field-of-view of the fluorescence measurement) simultaneously, assisting in the identification of network topology.
Host: Ilya Nemenman, CCS-3