|Friday, March 15th, 2013||3:30pm-4:30pm||MC 103|
Many problems from different areas can be formulated as problems of computing a fixed point of a suitable function. For many of these problems, the function in the fixed point formulation is monotone, and the objects we want to compute are given by a specific fixed point, namely the least fixed point of the function. Many models and problems from a broad variety of areas can be thus expressed as least fixed point computation problems, including in particular the analysis of various probabilistic models, problems in stochastic optimization, and games. It turns out that for some classes of functions we can compute the least fixed point in polynomial time and thus we can analyze efficiently several of these models, while for others there are strong indications that they cannot be solved in polynomial time, and for yet others the question remains open. In this talk we will discuss progress in this area. The talk is based on a series of works with Kousha Etessami and Alistair Stewart.