IBI-5 Seminar: Non-Markovian and Collective Search Strategies

Heiko Rieger

Center for Biophysics and Department of Theoretical Physics, Saarland University, Saarbrücken, Germany

Join us in person in Building 04.16, Room 2001

Abstract:

Agents searching for a target can improve their efficiency by memorizing where they have already been searching or by cooperating with other searchers and using strategies that benefit from collective effects. In this talk I will review such concepts: non-Markovian and collective search strategies. I will start with the first passage properties of continuous non-Markovian processes and then proceed to the discrete random walker with 1-step memory (like the persistent random walk) and n-step memory.

Next, I’ll consider the auto-chemotactic walker, a random walker that produces a diffusive chemotactic cue from which the walker tries to avoid (thus formally having infinite memory) and analyze the optimality conditions for an auto-chemotactic search. Then I discuss ensembles of interacting searchers and the conditions that could lead to advantageous collective effects minimizing the average search time – and analyze the chemotactically interacting random walker in detail.

While increasing the number of agents decreases the time at which the first agent finds the target, it usually requires resources to create and sustain more agents. Therefore, I finish with introducing a framework where the collective search cost not only includes the search time but also the cost associated to the creation and the maintenance of an agent – and discuss the optimal number of agents for a collective search, and when to launch them.