Code for John's students for delay reconstruction and fast Rips code with "Ripser" (courtesy of Uli Bauer)
To run this code, you will need to checkout and compile Ripser. First, run the following command
git clone --recursive https://github.com/ctralie/Math412S2017.gitThen, go into the ``ripser'' directory, and type the following
make ripser
make ripser-coeffNow you should be able to run all of the examples:
-
TDA.py: Code to compute persistent homology (basic command line wrapper around ripser, with functions to plot diagrams)
-
SlidingWindow.py: Code to compute sliding window embeddings of 1D signals, with an example of cos(t) + cos(pi t), reporting persistent H1 and H2. Also shows how to compute PCA
-
HorseWhinnies.py: Quantifying periodicity and quasiperiodicity of snippets of audio from "horsewhinnie.wav"
-
MusicSpeech.py: Running TDA on audio novelty functions to discover rhythmic periodicity in music. Comparing journey.wav to speech.wav
-
MusicFeatures.py: Quick and dirty music feature computation (MFCC, audio novelty functions) from spectrograms
-
VideoTools.py: Code for doing sliding window videos
-
testVideo.py: Code for testing sliding window videos
-
RipsFiltrationDM.m: A Matlab wrapper around ripser over distance matrices
-
RipsFiltrationPC.m: A Matlab wrapper around ripser over point clouds
-
plotpersistencediagrams.m: A function to plot all persistence diagrams on top of each other
-
TorusExample: An Matlab example of persistent H0, H1, and H2 on a sampled torus
There is also Python code to compare persistence diagrams using a variety of algorithms (DGMTools.py). To compute the Wasserstein Distance, you will need to install a library that implements the Hungarian algorithm for maximum flow bipartite matching. To do this, type
pip install hungarian