Predicting the Behavior of Dynamical and Biological Systems Using Asynchronous Data.

摘要

Physical systems often experience a complexity of behavior which requires many degrees of freedom to model accurately. In practice, it may be impossible to experimentally observe all the necessary state variables in a dynamical model. To make quantitative predictions it becomes necessary to extract information from the observable variables in order to estimate the entire state of the system. I discuss different approaches to making estimating these unobservable state variables. In particular, I explore novel ways of combining data at different times throughout the trajectory of the system to improve the estimate of the system state at a single time.