---

Estimating entropy production in a stochastic system with odd-parity variables

Authors: Dong-Kyum Kim^1*, Sangyun Lee^2*, and Hawoong Jeong^1,3
_{* Equal contribution}

¹ _{Department of Physics, KAIST} ² _{School of Physics, KIAS} ³ _{Center for Complex Systems, KAIST}

Introduction

This repo contains source code for the runs in Estimating entropy production in a stochastic system with odd-parity variables

Installation

Supported platforms: MacOS and Ubuntu, Python 3.7

Installation using Miniconda:

git clone https://github.com/kdkyum/odd_neep.git
cd odd_neep
conda create -y --name odd_neep python=3.7
conda activate odd_neep
pip install -r requirements.txt

To enable gpu usage, install gpu version torch package from PyTorch.

Usage

• Training for Underdamped bead-spring model.

python main_ubs.py \
  --save results/ubs/N2_m0-01_Tc1_seed0 \
  --n_layer 2 \
  --n_hidden 256 \
  --N 2 \
  --Tc 1 \
  --Th 10 \
  --m 0.01 \
  --lr 1e-5 \
  --wd 0 \
  --dropout 0 \
  --trj_num 10000 \
  --trj_len 4000 \
  --record_freq 400 \
  --n_iter 100000 \
  --seed 0

• Training for odd-parity Markov jump process.

python main_omj.py \
  --save results/omj/c10_seed0 \
  --n_layer 2 \
  --n_hidden 256 \
  --trj_len 10000 \
  --trj_num 50 \
  --N 10 \
  --c 10 \
  --lr 1e-5 \
  --n_iter 10000 \
  --record_freq 500 \
  --batch_size 4096 \
  --seed 0 

Bibtex

Cite the following Bibtex.

@article{kim2021odd_neep,
  title={Estimating entropy production in a stochastic system with odd-parity variables},
  author={Dong-Kyum Kim and Sangyun Lee and Hawoong Jeong},
  journal={arXiv preprint arXiv:2112.04681},
}

License

This project following the MIT license.