Berlin 2018 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
BP: Fachverband Biologische Physik
BP 4: Active Matter DY I (joint session DY/CPP/BP)
BP 4.5: Talk
Monday, March 12, 2018, 11:00–11:15, BH-N 243
Learning agents as a model for collective motion — •Katja Ried1, Thomas Müller2, and Hans J. Briegel1,2 — 1Institut für Theoretische Physik, Universität Innsbruck, Technikerstraße 21a, 6020 Innsbruck, Austria — 2Department of Philosophy, University of Konstanz, 78457 Konstanz, Germany
Watching a swarm of fish, birds or insects is mesmerizing, and it inevitably makes one wonder how countless independent individuals can form such a perfectly coordinated whole. A number of theoretical models attempt to answer this question by studying the collective dynamics that arise when individuals interact according to certain rules. However, these rules are often simply postulated ad hoc, and individuals are modelled as featureless points carrying them out. Naturally, such models are unlikely to provide an accurate - or even plausible - account of the individual-level behaviour that ultimately drives the swarm.
I will present a different Ansatz to this problem, wherein individuals are considered as full-fledged agents: distinct entities that can perceive certain (reasonable) features of their surroundings, endowed with a stable internal mechanism for processing these perceptions and deciding how to respond, and capable of modifying these responses as a function of their personal experience. I will illustrate this Ansatz with the example of locusts marching in a one-dimensional arena and discuss what insights agent-based models can offer to the study of collective motion.