Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 33: Machine Learning in Dynamics and Statistical Physics I
DY 33.8: Vortrag
Donnerstag, 20. März 2025, 11:30–11:45, H47
Machine learning for prediction of dynamical clustering in granular gases — •Sai Preetham Sata1, Dmitry Puzyrev2,1, and Ralf Stannarius3,2 — 1AMS, Otto von Guericke University, Germany — 2MTRM and MARS, Otto von Guericke University, Germany — 3Department of Engineering, Brandenburg University of Applied Sciences, Germany
Granular gases are sparse ensembles of free-moving macroscopic particles that interact via inelastic collisions. One peculiar property of granular gas is dynamical clustering, i.e. spontaneous increase of local number density. To quantify this effect, microgravity experiments and simulations were performed [1-3] and two gas-cluster transition criteria were established:Kolmogorov-Smirnov test, and caging effect criterion [2]. We perform simulations based on the VIP-GRAN experiment [3] and test these criteria for various combinations of system parameters, revealing their advantages and drawbacks. In addition, we investigate additional criteria that can help to understand the dynamical properties of gas-cluster transition. Based on the simulation data, machine learning can be used to detect dynamical clusters and predict the state of the system for a given set of system parameters. This study is funded by the German Aerospace Center (DLR) within projects VICKI (50WM2252) and EVA II (50WK2348). References: [1] É. Falcon et al., Phys. Rev. Lett., 83:440, 1999. [2] E. Opsomer et al., Europhys. Lett., 99:40001, 2012. [3] S. Aumaître et al., Rev. Sci. Instr., 89:075103, 2018.
Keywords: Granular gas; Microgravity conditions; Dynamical clustering; Kolmogorov-Smirnov Test; Machine learning