Quantum latent distributions in deep generative models

Code release for:

Bacarreza et al. "Quantum latent distributions in deep generative models." ICML 2026.

This repository currently contains code to reproduce the key result from Figure 1 of the paper: there are significant differences between classical and quantum latent distributions in a GAN trained on a mixture of Gaussians dataset.

Code for the remaining results in the paper will be uploaded soon.

---

Getting started

Requirements: Python 3.10+

Install the package and its dependencies:

pip install -e .

Then open and run qgan_intro.ipynb.

---

Repository structure

src/
  gan.py             # WGAN-GP training loop
  gaussians_utils.py # Seven Gaussians dataset, Generator and Critic networks
  tbi_sampler.py     # Time-bin interferometer sampler (quantum latent distribution)
  utils.py           # Shared utilities
gaussians_example.ipynb     # Notebook reproducing Figure 1
figures/             # Figures used in the notebook

---

Citation

If you use this code in your work, please cite:

@inproceedings{bacarreza2026quantum,
  title     = {Quantum latent distributions in deep generative models},
  author    = {Bacarreza, Omar and Farnsworth, Thorin and Makarovskiy, Alexander and Wallner, Hugo and Hicks, Tessa and Sempere-Llagostera, Santiago and Price, John and Francis-Jones, Robert JA and Clements, William R},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning},
  series    = {Proceedings of Machine Learning Research},
  year      = {2026},
}