Speaker: Michael Joswig, TU Darmstadt 

Title: Polytope propagation

Abstract: 

By assigning vectors to the arcs of an acyclic directed graph 
one can recursively define a convex polytope at each sink.  
Dynamic programming techniques yield algorithms for the computation 
of such propagated polytopes.  Applications, pioneered by Pachter and Sturmfels, 
include the parametric inference for hidden Markov processes, which are popular in
computational biology.  In addition to a presentation of theoretical
results the talk will include an overview of an implementation in
polymake.