Aligning Active Particles Py Package

TL;DR

This paper introduces an open-source Python package for simulating aligning self-propelled particles, such as the Vicsek model, using agent-based methods that are not compatible with traditional molecular dynamics frameworks.

Contribution

The package provides the first accessible simulation tool for non-Hamiltonian, agent-based models of self-propelled particles, filling a gap in existing computational resources.

Findings

Enables simulation of models not supported by existing MD frameworks.

Supports both discrete interaction rules and Brownian dynamics.

Provides a Python interface with C-based simulation core.

Abstract

The package performs molecular-dynamics-like agent-based simulations for models of aligning self-propelled particles in two dimensions such as e.g. the seminal Vicsek model or variants of it. In one class of the covered models, the microscopic dynamics is determined by certain time discrete interaction rules. Thus, it is no Hamiltonian dynamics and quantities such as energy are not defined. In the other class of considered models (that are generally believed to behave qualitatively the same) Brownian dynamics is considered. However, also there, the forces are not derived from a Hamiltonian. Furthermore, in most cases, the forces depend on the state of all particles and can not be decomposed into a sum of forces that only depend on the states of pairs of particles. Due to the above specified features of the microscopic dynamics of such models, they are not implemented in major molecular dynamics simulation frameworks to the best of the authors knowledge. Models that are covered by this package have been studied with agent-based simulations by dozens of papers. However, no simulation framework of such models seems to be openly available. The program is provided as a Python package. The simulation code is written in C. In the current version, parallelization is not implemented.