Probability and Math Finance Seminar
### Soummya Kar

ECE, Carnegie Mellon

** Sample-path Large Deviations of Jump Markov Processes: Revisited**

**Abstract: ** In this talk, we introduce a framework for analyzing large deviation (LD) properties of jump Markov processes with time-inhomogeneous, Lipschitz continuous jump rates. The key idea consists of a path-space representation of such processes in terms of appropriately time-changed Poisson processes and analyzing regularity properties of an associated pathwise mapping. Under broad conditions, we obtain explicit variational representations of the LD rae fucntions. An application to Jackson networks of many-server queues is discussed. (This is joint work with Kavita Ramanan.) BIOGRAPHY: Soummya Kar is a fourth year PhD Student in the Department of Electrical and Computer Engineering, Carnegie Mellon University. Prior to this, he obtained his Bachelor of Technology (BTech) degree from the Indian Institute of Technology, Kharagpur, India in May, 2005. His research interests include distributed computations in sensor networks, stochastic approximations, stochastic networks, and stochastic filtering in the context of sensor networks.

*MONDAY, April 20, 2009*

**Time:** 5:00 P.M.

**Location:** WeH 6423