Synthetic trajectory predictions and rollouts using dACE.jl
Year: 2026
Type: other
Abstract: Supplementary videos for "Interpretable many-body interaction laws for collective motion from few trajectories" (T. S. Gutleb, University of Leeds). Each video shows a simulated collective of self-propelled agents, started from a held-out initial condition, next to the free-running predictions of directional atomic cluster expansion (dACE) models fitted to trajectory data from the same collective and truncated at body order 2, 3 and 4. True paths are grey and predicted paths coloured. The models were fitted to short training simulations and are run here beyond the training time span.Simulation and fitting code, including the rule and the scripts that render these videos, are available at https://github.com/TSGut/dACE.jl
Source: Figshare
Author: Timon S. Gutleb
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7224086383