Dynamical Systems Seminars
Fall 2019
The Dynamical Systems seminar is held on Monday afternoons at 4:00 PM in MCS B37. Tea beforehand is at 3:30 PM in MCS 233A.
We will meet this week in MCS B39 at 4:00PM. Come by for refreshments and to meet everyone!
Abstract:Swarming behavior continues to be a subject of immense interest because of its centrality in many naturally occurring systems in physics and biology, as well as its importance in engineering applications such as robotics. Here we examine the effects on coherent swarm pattern formation from aspects of communication, such as latency effects, link topology and environmental uncertainty. With the availability of ever more cheap and powerful computing, interest in the use of mixed-reality and swarm experiments has grown considerably in the physical sciences. Broadly speaking, these experiments consist of a simulated, or virtual model coupled directly to a physical experiment. Within the physical experiment, it is typical to find a good deal of uncertainty and noise since it is connected to the real world, and thus subjected to random perturbations. In contrast, the virtual part of the coupled system represents a somewhat idealized version of reality in which noise can be eliminated entirely. Thus, mixed-reality systems have very skewed sources of uncertainty spread through the entire system.
In this talk, we consider the pattern formation of delay-coupled swarms theoretically and experimentally to illustrate the idea of mixed-reality. Motivated by physical experiments, we then consider a model of a mixed-reality system, and show how noise in the physical part of the system can influence the virtual dynamics through a large fluctuation, even when there is no noise in the virtual components. We quantify the effects of uncertainty by showing how characteristic times of noise induced switching between swarm patterns scale as a function of the coupling between the real and virtual parts of the experiment. This work is done in collaboration with Klimka Szwaykowska, Thomas Carr, Victoria Edwards and Jason Hindes.
Abstract: We introduce two interacting particle system models of vegetation dynamics (one macroscale and one mesoscale) based on the interaction rules from the mean-field Staver-Levin model of forest-savanna-grassland evolution. Using coupling techniques for stochastic jump processes, we prove the convergence of these particle systems to McKean-Vlasov-type jump processes. The generalized Kolmogorov equations which characterize the behavior of these processes are systems of integro-differential equations and are more amenable to analysis than the original particle systems. By analyzing these nonlocal equations, we find that our macroscale model is an elementary example of a process which oscillates in law and that our mesoscale model exhibits a rich array of ecologically relevant spatial dynamics, including traveling waves, breathing fronts and pattern formation.
This is joint work with Carla Staver (Yale), Simon Levin (Princeton) and Jonathan Touboul (Brandeis)