Generative Path Candidate Sampler for Faster Point-to-Point Ray Tracing

This repository accompanies the paper Transform-Invariant Generative Ray Path Sampling for Efficient Radio Propagation Modeling by Jérome Eertmans, Enrico Maria Vitucci, Vittorio Degli Esposti, Nicola Di Cicco, Laurent Jacques and Claude Oestges.

It provides:

• The source code for the model described in the paper, implemented in JAX, in src/sampling_paths, including a script to train and evaluate the model on synthetic data, in __main__.py.
• Pre-trained model weights, available at this link.
• Tests files in tests/ to verify the correctness of the implementation.
• A tutorial notebook, viewable here, demonstrating how to use the model for path sampling.

Installation

After cloning the repository, run:

pip install .

Alternatively, you can avoid manually cloning the repository by installing directly from GitHub:

pip install git+https://github.com/jeertmans/sampling-paths.git

Usage

After installation, you can train and evaluate the model using:

train-path-sampler --help

Getting Help

For any question about the method or its implementation, make sure to first read the related paper.

If you want to report a bug in this library or the underlying algorithm, please open an issue on this GitHub repository. If you want to request a new feature, please consider opening an issue on DiffeRT's GitHub repository instead.

Citing

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

@misc{eertmans2026,
  title        = {Transform-Invariant Generative Ray Path Sampling for Efficient Radio Propagation Modeling},
  author       = {Jérome Eertmans and Enrico M. Vitucci and Vittorio Degli-Esposti and Nicola Di Cicco and Laurent Jacques and Claude Oestges},
  year         = 2026,
  url          = {https://arxiv.org/abs/2603.01655},
  eprint       = {2603.01655},
  archiveprefix = {arXiv},
  primaryclass = {cs.LG}
}