qpytorch 0.1.1

An implementation of Q-Exponential Processes in Pytorch based on GPyTorch

Q^ⓔPyTorch

---

Q^ⓔPyTorch is a Python package for Q-exponential process (QEP) implemented using PyTorch and built up on GPyTorch. Q^ⓔPyTorch is designed to facilitate creating scalable, flexible, and modular QPE models.

Different from GPyTorch for Gaussian process (GP) models, Q^ⓔPyTorch focuses on QEP, which generalizes GP by allowing flexible regularization on function spaces through a parameter $q>0$ and embraces GP as a special case with $q=2$. QEP is proven to be superior than GP in modeling inhomogeneous objects with abrupt changes or sharp contrast for $q<2$ [Li et al (2023)]. Inherited from GPyTorch, Q^ⓔPyTorch has an efficient and scalable implementation by taking advantage of numerical linear algebra library LinearOperator and improved GPU utilization.

Tutorials, Examples, and Documentation

See documentation on how to construct various QEP models in Q^ⓔPyTorch.

Installation

Requirements:

• Python >= 3.10
• PyTorch >= 2.0
• GPyTorch >= 1.13

Stable Version

Install Q^ⓔPyTorch using pip or conda:

pip install qpytorch
conda install qpytorch

(To use packages globally but install Q^ⓔPyTorch as a user-only package, use pip install --user above.)

Latest Version

To upgrade to the latest version, run

pip install --upgrade git+https://github.com/lanzithinking/qepytorch.git

from source (for development)

If you are contributing a pull request, it is best to perform a manual installation:

git clone https://github.com/lanzithinking/qepytorch.git
cd qepytorch
# either
pip install -e .[dev,docs,examples,keops,pyro,test]  # keops and pyro are optional
# or
conda env create -f env_install.yaml # installed in the environment qpytorch

Citing Us

If you use Q^ⓔPyTorch, please cite the following paper:

Li, Shuyi, Michael O'Connor, and Shiwei Lan. "Bayesian Learning via Q-Exponential Process." In Advances in Neural Information Processing Systems (2023).

@inproceedings{li2023QEP,
  title={Bayesian Learning via Q-Exponential Process},
  author={Li, Shuyi, Michael O'Connor, and Shiwei Lan},
  booktitle={Advances in Neural Information Processing Systems},
  year={2023}
}

Contributing

See the contributing guidelines CONTRIBUTING.md for information on submitting issues and pull requests.

The Team

Q^ⓔPyTorch is primarily maintained by:

• Shiwei Lan (Arizona State University)

Thanks to the following contributors including (but not limited to)

• Shuyi Li, Guangting Yu, Zhi Chang, Chukwudi Paul Obite, Keyan Wu, and many more!

License

Q^ⓔPyTorch is MIT licensed.