RSR: Reference-state System Reliability Method

Overview

This repository provides a Python implementation of RSR (Reference-state System Reliability Method) for efficient reliability and resilience analysis of networks. It includes:

• Core package rsr/ with tensor-based algorithms
• Demonstration notebooks under demos/
• Unit tests under tests/

The code is designed for research and education on large-scale system uncertainty quantification.

Publication / Citation

In preparation. Most relevant publication is: Byun, J. E., Ryu, H. & Straub, D. (2024). Branch-and-bound algorithm for efficient reliability analysis of general coherent systems. arXiv preprint arXiv:2410.22363.

Features

• Reference-state system reliability and rule extraction algorithms
• Example benchmark datasets on various systems (e.g., distribution substation, EMA shortest path, toy k-connectivity)
• The network data in the demos are from GitHub repo network-datasets
• PyTorch-friendly implementations for scalable computation

Installation

You can install rsr in one of two ways.

Option 1: Install from PyPI

The package is published on PyPI under the distribution name rsr-duco (the name rsr was already taken). The Python import name is still rsr:

pip install rsr-duco

Option 2: Install from source (developer version)

Clone the repo and install in editable mode (useful for development or when you want to modify the code):

git clone https://github.com/jieunbyun/rsr.git
cd <path/to/rsr>
pip install -e .

Using the package

Either option gives you the same import name:

import rsr
from rsr import rsr, utils

Dependencies are listed in pyproject.toml.

Usage

Refer to the demonstration notebooks in demos/ for example workflows:

License

This project is licensed under the terms of the LICENSE file included in this repository.