Parts | Days | Selection | Search | Updates | Downloads | Help
CPP: Fachverband Chemische Physik und Polymerphysik
CPP 39: Focus Session: Physics of Fluctuating Paths (joint session DY/CPP)
CPP 39.7: Talk
Wednesday, March 29, 2023, 16:45–17:00, ZEU 160
Inferring Fractional Processes Using Path Integrals — •Johannes A. Kassel1, Benjamin Walter2, and Holger Kantz1 — 1Max Planck Institute for the Physics of Complex Systems, Dresden, Germany — 2Imperial College London, London, UK
We present a method for inferring overdamped nonlinear Langevin equations driven by multiplicative fractional Gaussian noise from single-trajectory time series. Constructing a maximum-likelihood estimator, we simultaneously infer the nonlinear deterministic force term and the space-dependent diffusion term. We illustrate our method using artificial time series. We observe that Markovian modeling of long-range correlated data leads to a substantial underestimation of the deterministic force term while for anti-correlated data it leads to an overestimation of the force term.