Data-Efficient Discovery of Hyperelastic TPMS Metamaterials

This repository contains the official implementation for the paper: Data-Efficient Discovery of Hyperelastic TPMS Metamaterials with Extreme Energy Dissipation.

Paper Link: https://doi.org/10.1145/3721238.3730759

Overview

This project uses a data-efficient method driven by physical experiments to discover novel metamaterial structures with exceptional energy-dissipation capabilities.

Our method uses a batch Bayesian optimization framework to guide the discovery process:

1. A Deep Ensemble model is trained on physical experiment data to predict stress-strain behavior and uncertainty.

2. A penalized Upper Confidence Bound (UCB) acquisition function selects new candidates by balancing exploration (improving model accuracy) and exploitation (maximizing energy dissipation).

3. This iterative loop efficiently discovers high-performing structures within a limited experimental budget.

Setup and Installation

1. Clone Repository

git clone https://github.com/maxineAPS/data-efficient-metamaterial-discovery
cd data-efficient-metamaterial-discovery

2. Create Virtual Environment & Install Dependencies

# Create and activate a virtual environment
python3 -m venv venv
source venv/bin/activate

# Install requirements
pip install -r requirements.txt

3. Download Dataset

Download the experimental dataset from https://github.com/maxineAPS/energy-dissipation-metamaterial-dataset.git. Unzip and place the contents in the data/ directory.

Usage

1. Train the Ensemble

Run train.py to train the deep ensemble on the experimental data in data/. The models will be saved to the models/ directory.

python train.py

2. Select Next Batch

Run bayesian.py to identify the next batch of promising structures to fabricate. This script uses the trained ensemble to select the top 40 candidates.

python bayesian.py

The selected parameters will be saved to bayesian_selected_points.json.

3. Visualize Predictions (Optional)

Run the visualization script to check the performance of your trained ensemble against test data.

python visualize_predictions.py

Citation

If you use this dataset in your research, please cite our paper:

@inproceedings{perroni-scharf2025data,
  author = {Perroni-Scharf, Maxine and Ferguson, Zachary and Butruille, Thomas and Portela, Carlos M. and Konakovi\'{c} Lukovi\'{c}, Mina},
  title = {Data-Efficient Discovery of Hyperelastic TPMS Metamaterials with Extreme Energy Dissipation},
  year = {2025},
  isbn = {979-8-4007-1540-2},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  url = {[https://doi.org/10.1145/3721238.3730759](https://doi.org/10.1145/3721238.3730759)},
  doi = {10.1145/3721238.3730759},
  booktitle = {Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Papers},
  series = {SIGGRAPH '25}
}