This is a link to the code and numerical experiments used to estimate the intrinsic dimension of datasets as well as the minimal diffusion maps algorithm (MDM).

The code and numerical experiments have been produced by Johannes von Lindheim in the "Computational Molecular Design" Group headed by Marcus Weber at Zuse-Institut Berlin, Takustr. 7, D-14195 Berlin, Germany.
The zip-file contains Python code with the Minimal Diffusion Maps algorithm (MDM), the Gaussian Annulus ID heuristics (GAV and GAP) and code for generating example datasets. Additionally, two Jupyter notebooks are provided for presentation purposes, as well as a csv-file containing the intrinsic dimension estimation results of the ID heuristics.

This code was used for the master thesis "On Intrinsic Dimension Estimation and Minimal Diffusion Maps Embeddings of Point Clouds" by Johannes von Lindheim. The thesis, which contains detailed descriptions of the aforementioned algorithms, can be downloaded via the following link.

If you have any questions, do not hesitate to send an email to: