Machine learning spectral indicators of topology

Workflow

1. Data assembly directory: Scripts for querying, filtering, and processing spectral and label data.
2. pca_kmeans.ipynb: Notebook for conducting an exploratory analysis of the data, including principal component analysis and k-means clustering of XAS spectra.
3. topology_classifier.ipynb: Notebook for training a neural network classifier of band topology from XAS spectral inputs.

Installation

1. Clone the repository:

   git clone https://github.com/ninarina12/XAS_Topo.git

   cd XAS_Topo

2. Create a virtual environment for the project:

   conda create -n xas_topo python=3.7.5

   conda activate xas_topo

3. Install packages:

   pip install -r requirements.txt -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html

   where ${TORCH} and ${CUDA} should be replaced by the specific CUDA version (e.g. cpu, cu102) and PyTorch version (e.g. 1.9.1), respectively. For example:

   pip install -r requirements.txt -f https://pytorch-geometric.com/whl/torch-1.9.1+cu102.html

References

N. Andrejevic*, J. Andrejevic*, B. A. Bernevig, N. Regnault, C. H. Rycroft, and M. Li. (*=equal contributions) Machine learning spectral indicators of topology. arXiv preprint arXiv:2003.00994 (2020).