Steady State Estimation in Water Distribution Systems with Edge-based Diffusion

This repository contains code and resources associated with the paper Accelerating Steady State Analysis of Water Distribution Systems Using GPU-Enhanced Edge-Based Diffusion. The main objective is to demonstrate how an edge-based diffusion mechanism can be used to achieve simultaneous hydraulic simulations (coarse-grained parallelization) by leveraging CUDA.

Roadmap

• Diffusion Model
• Requirements
• Virtual Reservoir Connectors
• CLI option
• Pressure reconstruction
• Generated Data
• Optimization algorithm

Installation

1. Clone the repository:

   git clone https://github.com/erytheis/edge-based-diffusion-for-wds.git
   cd edge-based-diffusion-for-wds

2. Set up a Python environment (conda or venv recommended):

   conda create -n erytheis_env python=3.10
   conda activate erytheis_env

3. Install required packages:

   pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
   pip install torch_geometric
   pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.4.0+cu121.html
   pip install -r requirements.txt

   (Make sure to include requirements.txt or similar in the repo with PyTorch, NumPy, SciPy, and other dependencies.)

Usage

1. Running a Basic Simulation
   Use main.py to run a simple, demonstrative edge-based diffusion simulation:

   python main.py --config config/your_experiment.yaml

   This script will:
   • Parse the configuration file at input/config.yaml
   • Apply the edge-based diffusion solver on a GPU (if available).
   • Calculate $R^2$ with values obtained from WNTR (if available)