ripser

What it is and what it does

Ripser.py is a Python wrapper around the fast C++ Ripser library for computing persistent homology, a technique from algebraic topology that tracks how topological features (connected components, holes, voids) persist across different scales in data. It takes point clouds or distance matrices as input and outputs persistence diagrams—visualizations showing which features appear and disappear as you vary a scale parameter. The package provides both a functional interface (ripser function) and a scikit-learn compatible transformer (Rips class) for integration into machine learning pipelines.

The package depends on Cython, numpy, scipy, scikit-learn, and persim for visualization. It compiles to native code, giving it speed advantages over pure-Python implementations. Installation is straightforward on major platforms through precompiled wheels, though Windows users may need MinGW and macOS users should have current Xcode tools. The library is actively maintained and carries no known security vulnerabilities.

Use it for:

• Analyze the topological structure of high-dimensional datasets to discover clusters, holes, or voids without specifying cluster count in advance.
• Visualize persistence diagrams to understand which topological features are robust (signal) versus noise in your data.
• Compute lowerstar filtrations on image data to extract topological features for computer vision tasks.
• Extract representative cochains to identify which data points or features drive topological structure.
• Use as a scikit-learn transformer in unsupervised learning pipelines for topological feature engineering.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Ripser computes persistent homology and persistence diagrams for point clouds and sparse/dense datasets, enabling topological data analysis through an interface wrapping the fast C++ Ripser engine.

Yes. Ripser is a specialized but well-maintained tool for topological data analysis with no security issues, permissive licensing, and broad platform support. Install friction is moderate (compilation required) but manageable via wheels. Suitable if you need persistent homology or topological feature extraction; not necessary for standard clustering or dimensionality reduction.

Install

ripser on PyPI

pip install ripser

uv add ripser

poetry add ripser

Installing ripser

Before you install

Medium install friction due to Cython compilation and compiled dependencies (numpy, scipy, scikit-learn). Wheels available across macOS, Linux, and Windows platforms. Active maintenance with recent release and continuous integration support.

License in practice

MIT license (permissive) allows commercial and private use with minimal restrictions. Derived from the original Ripser C++ library (also MIT), with modifications and Python code copyright to Christopher Tralie and Nathaniel Saul.

Quickstart

pip install ripser

import numpy as np
from ripser import ripser
from persim import plot_diagrams

data = np.random.random((100, 2))
diagrams = ripser(data)['dgms']
plot_diagrams(diagrams, show=True)

Windows users may need MinGW installed; macOS users should update Xcode and command line tools. Compilation from source requires Cython.

Verify before relying

• Whether the optional robin_hood hash table integration (claimed to provide up to 30% speedup) is available in precompiled wheels or only when building from source.
• Specific minimum Python version requirements (requires_python is unspecified in metadata).
• Performance characteristics and scalability limits for large point clouds.

Package facts

Evidence: ripser-0.6.15-cp310-cp310-macosx_10_9_x86_64.whl; ripser-0.6.15-cp310-cp310-macosx_11_0_arm64.whl; ripser-0.6.15-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; ripser-0.6.15-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; ripser-0.6.15-cp310-cp310-musllinux_1_2_aarch64.whl; ripser-0.6.15-cp310-cp310-musllinux_1_2_x86_64.whl; ripser-0.6.15-cp310-cp310-win32.whl; ripser-0.6.15-cp310-cp310-win_amd64.whl; ripser-0.6.15-cp311-cp311-macosx_10_9_x86_64.whl; ripser-0.6.15-cp311-cp311-macosx_11_0_arm64.whl; ripser-0.6.15-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; ripser-0.6.15-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl; ripser-0.6.15-cp311-cp311-musllinux_1_2_aarch64.whl; ripser-0.6.15-cp311-cp311-musllinux_1_2_x86_64.whl; ripser-0.6.15-cp311-cp311-win32.whl; ripser-0.6.15-cp311-cp311-win_amd64.whl; ripser-0.6.15-cp312-cp312-macosx_10_13_x86_64.whl; ripser-0.6.15-cp312-cp312-macosx_11_0_arm64.whl; ripser-0.6.15-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl; ripser-0.6.15-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl

Keywords: topological data analysis, persistent homology, Rips filtration, algebraic topology, unsupervised learning, persistence diagrams

Tags

More Information Analysis packages