hyperTDA

Piece-wise Linear curves converted to point-clouds, analysed with Persistent Homology, represented as HyperGraphs. Associated with preprint "Hypergraphs for multiscale cycles in structured data" at https://doi.org/10.48550/arXiv.2210.07545

REQUIREMENTS

• Julia (tested on v1.7.3)
• Python (tested on v3.10.6)
• unix shell (tested on zsh v5.8.1)

INSTALL

• ./install.sh which sets a git alias git root, initializes submodules and runs ./install.jl.
• Julia packages are listed in ./install.jl and tested versions in Manifest.toml and Project.toml. Running ./install.sh installs the tested versions to a local environment and ./install.jl --global installs the newest available versions to the global environment.
• Python packages are listed in requirements.txt and can be installed with e.g. pip install -r requirements.txt or conda create --name=hyperTDA --file=requirements.txt requiring either pip (pip3) or conda (e.g. miniconda3 or miniforge). For the exact versions tested on Mac OS with Arm64 architecture use requirements_tested.txt.

EXAMPLE USE

Example input files, commands and output files are provided in subfolders of examples/. See RUNME.sh for each example.

PAPER RESULTS

Results discussed in the paper are provided in results/, see e.g. jupyter notebooks in subfolders.

TIMING

• Git clone ~ 3 min
• Julia package install ~ 3 min
• Python package install ~ 1 min
• running examples
  • default pipeline (two curves of 132 and 101 points) ~ 30 sec
  • with interpolation (two curves each of 500 points) ~ 1 min 30 sec