Active work

Yann-Edwin Keta — DAMTP, University of Cambridge — 2019

Introduction

This repository contains scripts, for simulation and analysis purposes, developed for a research project, detailed in this wiki, concerned with active Brownian particles (ABPs) and the large deviations of the work of self-propelling forces (active work). [Nemoto et al., Phys. Rev. E 99, 022605 (2019)]

Simulation and cloning scripts are written in C++. Wrapper scripts to launch the latter are written in Python, and other Python classes and functions are available to read and analyse the generated data.

While C++ files can be quite cumbersome, Python wrappers are hopefully more readable and commented enough so that their functioning can be easily understood.

Requirements

All code was developed and tested on 64-bit linux. C++ cloning scripts necessitate OpenMP. Python scripts are written for python3.*, import the active_work package which necessitates the directory containing this repository to be added to the $PYTHONPATH, e.g. by executing

echo "export PYTHONPATH=\$PYTHONPATH:${PWD}/.." >> ~/.bashrc

from this directory, and rely on the following packages:

• matplotlib: plotting,
• seaborn: color palettes,
• numpy: mathematical functions and array manipulation,
• scipy: various optimisation methods and special functions,
• fastkde: kernel density estimation (scde.py),

which can be installed by running pip.sh, provided that pip is installed.

Production of movies, via frame.py, necessitates ffmpeg — though other functionalities of the former can be used without the latter.

Memory error detection and profiling, using make memcheck and make massif (see Makefile), necessitates valgrind.

Execution

Compilation of all relevant executables, using g++, is possible by running compile.sh — which essentially performs all relevant make commands (see Makefile).

Given these have been compiled, they can be executed with the Python scripts listed below.

Simulations of ABPs

ABP model and simulation procedure is detailed in this tiddler.

• Simulations with custom relations between parameters are launched using launch.py.
• Simulations with custom relations between parameters and for different values of the torque parameter are launched using launchG.py.
• Simulations of general ABPs are launched using launch0.py.

Simulations of interacting Brownian rotors

Interacting Brownian rotors model is detailed in this tiddler.

• Simulations are launched using launchR.py.

Cloning of ABPs

Principle and computation scheme of the scaled cumulant generating function (SCGF) of the active work and corresponding averages in the biased ensemble are detailed in this tiddler.

• Cloning of trajectories of ABPs systems with custom relations between parameters, and biased with respect to either the polarisation or the active work, are launched using cloning.py.

Cloning of non-interacting Brownian rotors

Principle and computation scheme of the scaled cumulant generating function (SCGF) of the (squared) polarisation and corresponding averages in the biased ensemble are detailed in this tiddler.

• Cloning of trajectories of Brownian rotors are launched using cloningR.py.