GDA Toolbox (public version)

Public Repository at https://github.com/geomdata/gda-public/

Documentation at https://geomdata.github.io/gda-public/

This package provides easy-to-use tools for manipulating time-series signals (see the timeseries.Signal class), point clouds (see the multidim.PointCloud class) and simplicial complexes (see the multidim.SimplicialComplex class). The goal is to provide intuitive interfaces and fast, practical, and mathematically-correct methods for topological data analysis.

Basic Use Cases

The repository contains many examples at http://nbviewer.jupyter.org/github/geomdata/gda-public/tree/master/examples

To dive into the tools, here are a few good examples to get started:

• Multi-dimensional Data

  • Using the multidim.PointCloud class to compute persistence diagrams of multidimensional data. See example-pointcloud
  • Using the multidim.covertree.CoverTree class to compute Cover Trees with the Friends algorithm. See example-covertree
  • Using the multidim.models.CDER class for supervised machine learning using the CDER algorithm. See example-cder

• Time-Series Data

  • Using the timeseries.Signal class to compute persistence of time-series data. See example-signal-pers0
  • Using the timeseries.SpaceCurve class to build and study the geometry of trajectories. See example-trajectories
  • Using the timeseries.SpaceCurve class to compute signature curves. See example-sigcurves
  • Using mollifiers to clean up time-series data with gaps or jumps. See example-mollification

Startup

• README for GDA-Public
  • Quick Start
  • Building Documentation
  • Running and Using
• Example Worksheets
• README for Advanced Installations
  • Online Setup and Testing with Anaconda on Linux/Mac
  • Online Setup and Testing with Anaconda on Windows
  • Offline Setup and Testing with Anaconda
  • Installation
  • Porting Changes
• Development Guidelines
  • Repository usage
  • Profiling
  • Style
  • Documentation
  • Testing
  • Docker Build Environment
• COPYRIGHT NOTICE
  • Authors
  • License

General Framework

• Multidimensional Data
• Time-Series Data
• Topological Constructions

Fast Algorithms

Performance-sensitive algorithms are written in [Cython]. Usually, only core developers need to access these methods directly, as most users will use the pure Python interfaces for convenience and clarity.

multidim.fast_algorithms	This Cython module contains core algorithms for multidimensional data.
timeseries.fast_algorithms	This Cython module contains core algorithms for timeseries.Signal data.
timeseries.curve_geometry	This Cython module contains the core algorithms for the geometry of curves.
homology.dim0	This Cython module contains the core algorithms for 0-dimension topological algorithms.
homology.dim1	This Cython module contains the core algorithms for 1-dimensional topological algorithms.

Indices and tables

• Index
• Module Index

• All Modules, Classes, and Methods

References