Introduction
We apply both supervised and unsupervised machine learning algorithms to the study of the string landscape and swampland in 6-dimensions. Our data are the (almost) anomaly-free 6-dimensional N = (1, 0) supergravity models, characterised by the Gram matrix of anomaly coefficients. Our work demonstrates the ability of machine learning algorithms to efficiently learn highly complex features of the landscape and swampland. To learn more about this project we encourage you to follow the link to our Preprint on arXiv in the top left corner. If you're looking for the results from our anomaly inflow classifiers or the data on all models for a particular cluster then please follow the link to our GitHub repository.
This site presents our initial analysis of the results from training an autoencoder (unsupervised) neural network on all 26 million 6d supergravity models. In the Results section below you will find two clickable images of the 2-dimensional latent layer plots, which have been divided up into local clusters (outlined and labeled in black), for all models for your exploration. Simply click on the cluster you'd like to explore and you will be taken to the analysis performed on that cluster. To compare analyses across multiple clusters more easily we created a Comparison View page which we highly recommend you use.
Our plan is to update this site as the project continues to develope so if you would like to see a specific functionality implemented or have any suggestions for improvement we encourage you to reach out to the authors.
Results
Local Clusters
LC-01: Sub-clusters
