Frequently Asked EnergyFlow Questions
- Why should I use Wasserstein instead of the Python Optimal Transport (POT) library?
- How do I cite the Wasserstein package?
- Where can I get the code?
- How do I report an issue or a bug?
Why should I use Wasserstein instead of the Python Optimal Transport (POT) library?
POT offers EMD functionality based on the same algorithm as Wasserstein. You'll find that Wasserstein is faster (up to 2x) for small problems (around 100 particles) due to the more efficient way in which we interface with NumPy. Also, Wasserstein offers greater numerical flexibility, with the ability to control certain floating point epsilons that POT hard codes and can cause failure (especially on macOS, we have found). And then there is the C++ interface that Wasserstein exposes in case you work in that language.
How do I cite the Wasserstein package?
Please cite the relevant papers if they or this package help your research. Here are the BibTeX entries to use:
@article{10.1145/2070781.2024192,
author = "Bonneel, Nicolas and van de Panne, Michiel and Paris, Sylvain and Heidrich, Wolfgang",
title = "{Displacement Interpolation Using Lagrangian Mass Transport}",
year = "2011",
issue_date = "December 2011",
publisher = "Association for Computing Machinery",
address = "New York, NY, USA",
volume = "30",
number = "6",
issn = "0730-0301",
url = "https://doi.org/10.1145/2070781.2024192",
doi = "10.1145/2070781.2024192",
journal = "ACM Trans. Graph.",
month = "dec",
pages = "1–12",
numpages = "12"
}
@article{Komiske:2019fks,
author = "Komiske, Patrick T. and Metodiev, Eric M. and Thaler, Jesse",
title = "{Metric Space of Collider Events}",
eprint = "1902.02346",
archivePrefix = "arXiv",
primaryClass = "hep-ph",
reportNumber = "MIT-CTP 5102",
doi = "10.1103/PhysRevLett.123.041801",
journal = "Phys. Rev. Lett.",
volume = "123",
number = "4",
pages = "041801",
year = "2019"
}
@article{Komiske:2020qhg,
author = "Komiske, Patrick T. and Metodiev, Eric M. and Thaler, Jesse",
title = "{The Hidden Geometry of Particle Collisions}",
eprint = "2004.04159",
archivePrefix = "arXiv",
primaryClass = "hep-ph",
reportNumber = "MIT-CTP 5185",
doi = "10.1007/JHEP07(2020)006",
journal = "JHEP",
volume = "07",
pages = "006",
year = "2020"
}
Where can I get the code?
The code is open source and hosted on GitHub.
How do I report an issue?
Please report any issues you encounter by creating a GitHub Issue.