Project files provided as supporting information to the manuscript "The bliss of dimensionality: how an unsupervised criterion identifies optimal low-resolution representations of high-dimensional datasets"

README file to the project files provided as supporting information to the manuscript “The bliss of dimensionality: how an unsupervised criterion identifies optimal low-resolution representations of high-dimensional datasets” March 4, 2026 Authors: Margherita Mele and Raffaello Potestio ================================== The dataset contains the following files: DATA/ Alanine/ # Alanine dipeptide MD + derived metrics (RMSD, angle) MNIST/ # MNIST data and derived metrics (mean, covariance) RESULTS/ Alanine/ # Outputs for "production.ipynb" on Alanine Dypeptide system MNIST/ clean/ # Outputs for "production.ipynb" on NOT weighed class data in produced via "sample_mnist.py" clean_weight/ # Outputs for "production.ipynb" on weighed class data in produced via "sample_mnist.py" raw/ # Outputs for “sample_mnist.py" for NOT weighed class raw_weight/ # Outputs for “sample_mnist.py" for NOT weighed class Structured/ # Outputs for structured synthetic-mixture experimentsUnstructured/ # Outputs for unstructured synthetic-mixture experiments lib_res_rel.py # Lybrary containing utilities for computing and processing Res-Rel curvemain.ipynb # Main notebook (Visualization of the results)production.ipynb # Notebook used to produce the data, and process the Res-Rel curvesample_mnist.py # Script for MNIST mixture experimentssample_mixture.py # Script for synthetic mixture experiments