DeepONet Template Library

This is a template library for conducting research on DeepONet, a deep learning architecture for learning operators and solving partial differential equations. The library provides a modular and extensible framework for defining, training, and evaluating DeepONet models.

Directory Structure

The template library has the following directory structure:

./
├── README.md
├── analyze.py
├── deeponet
│   ├── __init__.py
│   ├── data.py
│   ├── model.py
│   ├── train.py
│   └── utils.py
├── freeze.sh
├── kan.sh
├── requirements.txt
└── run.py

README.md: This file, providing an overview and documentation of the library.
analyze.py: A script for analyzing trained models and generating visualizations.
deeponet/: The main package containing the core components of the library.
- __init__.py: Package initialization file.
- data.py: Module for loading and preprocessing data.
- model.py: Module defining the DeepONet model architectures.
- train.py: Module for training DeepONet models.
- utils.py: Module containing utility functions and classes.
freeze.sh: A shell script for freezing the library's dependencies into requirements.txt.
kan.sh: A shell script for cloning and setting up the Kolmogorov-Arnold Network (KAN) repository.
requirements.txt: A file listing the library's dependencies.
run.py: The main script for running experiments and training models.

Prerequisites

Python 3.7 or higher
Git
uv

Installation

To install the library and its dependencies, follow these steps:

Clone the repository:

git clone https://github.com/Axect/DeepONet_Template

Navigate to the library directory:
```
cd DeepONet_Template
```

Set up a virtual environment via uv:

uv venv
uv pip sync requirements.txt
source .venv/bin/activate

(Optional) Run the kan.sh script to download efficient-kan (Kolmogorov-Arnold Network):
```
sh kan.sh
```

Usage

To use the DeepONet template library, follow these steps:

Prepare your data:
- Organize your data into the appropriate format required by the library.
- Modify the data.py module to load and preprocess your data.
Define your model:
- Choose the appropriate DeepONet model architecture from the model.py module or create a new one.
- Modify the model architecture and hyperparameters as needed.
Train your model:
- Use the run.py script to train your model.
- Adjust the training parameters and hyperparameters in the run.py script.
Analyze and visualize results:
- Use the analyze.py script to analyze the trained models and generate visualizations.
- Customize the analysis and visualization code in analyze.py based on your requirements.
(Optional) Freeze the dependencies:
- If you make changes to the library's dependencies, run the freeze.sh script to update the requirements.txt file.

Contributing

Contributions to the DeepONet template library are welcome! If you find any issues or have suggestions for improvements, please open an issue or submit a pull request on the GitHub repository.

License

This project is licensed under the MIT License.

Acknowledgments

This template library was inspired by various research papers and implementations of DeepONet. We would like to acknowledge the contributors and researchers who have made significant contributions to the field of DeepONet and related areas.