At this colloquium, we are happy to welcome:

Sabine Klapp (TU Berlin)

Nonreciprocal active and decision-making systems: Connections between particle and continuum scale

Nonreciprocal systems have recently received much interest in the statistical physics community and beyond.  One intriguing feature of such systems is their unusual collective behavior including, e.g., spontaneous time-symmetry breaking. These phenomena have been widely studied using hydrodynamic theories, whereas the corresponding particle-level behavior if often less understood. Here we discuss the connection between field-theoretical and corresponding particle-level results for two types of systems: polar active matter and the shepherding problem.

Nico van der Vegt (TU Darmstadt)

Applications of Generalized Langevin Equations in Molecular and Coarse-Grained Dynamics

I will present applications of generalized Langevin equations (GLEs) to two distinct problems. First, I introduce a systematic approach for parameterizing GLE thermostats in molecular dynamics simulations of single-site models of molecular liquids, with the goal of accurately capturing diIusive transport properties. Second, I discuss the use of GLEs as coarse-grained dynamical models for one-dimensional collective coordinates. Applications include hydrophobic polymer collapse and contact line dynamics in wetting phenomena.

Benjamin Hery (FU Berlin)

Generalized Langevin equations with a non-linear potential of mean force

The Mori-Zwanzig projection operator formalism constitutes a powerful and robust theoretical framework for deriving generalized Langevin equations (GLEs) for a given observable of interest using evolution and projection operators. We present four GLEs, derived for a scalar observable of interest using different projection operators, which consist in a Markovian force deriving from a potential, a running integral over time of a non-Markovian linear friction force, and the orthogonal force that is often interpreted as a random force. We highlight that all four GLEs are characterized by different properties, such as observable-dependent effective masses and observable-dependent memory kernels, and capture accurately the joint-statistics of the observable of interest when it can be modeled exactly or approximately by an effective Boltzmann-like distribution.