https://github.com/lmcinnes/persistence_wasserstein_benchmarking
Tools for benchmarking implementations of Wassersteing-Kantorovich distance between persistence diagrams
https://github.com/lmcinnes/persistence_wasserstein_benchmarking
Last synced: about 1 month ago
JSON representation
Tools for benchmarking implementations of Wassersteing-Kantorovich distance between persistence diagrams
- Host: GitHub
- URL: https://github.com/lmcinnes/persistence_wasserstein_benchmarking
- Owner: lmcinnes
- License: bsd-3-clause
- Created: 2020-02-25T20:17:56.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-03-03T04:10:34.000Z (over 5 years ago)
- Last Synced: 2025-06-06T23:07:34.269Z (4 months ago)
- Language: C++
- Size: 370 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Persistence Wasserstein Benchmarking Tools
Tools for benchmarking implementations of Wasserstein-Kantorovich distance between persistence diagrams.
This includes Hera, and a new binary for all pairs distances, plus python code for Wasserstein distance and mock diagram generation for benchmarking purposes.
To get started
```bash
$ git clone https://github.com/lmcinnes/persistence_wasserstein_benchmarking
$ cd persistence_wasserstein_benchmarking
$ cmake hera/geom_matching/wasserstein/
$ make
$ python generate_diagrams.py -n 50 -N 100
$
$ time ./wasserstein_dist_all_pairs -p 2 -q 1 -o hera_result_50.txt data_50/*
$ time python python_wasserstein_all_pairs.py -p 2 -q 1 -o py_result_50.txt data_50/*
```
## Requirements
The combination of Hera requirements and python + optimal transport requirements is, at a minimum:
- Cmake
- Boost
- POT
- Scikit-learn
- Numpy