This repository contains the code for a "Predicting Lattice Parameters from Atomic-Scale Images of Two Dimensional (2D) Materials Using Deep Learning" paper
The basic requirement for using the files is just installing the packages listed in requirements.txt.
It is advisable to create a virtual environment with the correct dependencies.
Sourcecode_general/: Contains files that could be used for all the datasets in general. This contains the files to run the vit+mlp and googlenet. We also have the file to generate the Alex-PBE dataset. We also have a file that does the chemical structure based inference.alex-pbe: Contains the .py or .ipynb files that run all the experiments. The trained models and some plots can be found a zipped file.code/: Contains all code that is same as the other data, but for large size, couldnt be run on jupyter notebooks, so .py files were used.
- `jv2d: Contains the .py or .ipynb files that run all the experiments. The trained models and some plots can be found a zipped file.
c2db: Contains the .py or .ipynb files that run all the experiments. The trained models and some plots can be found a zipped file. For all practical purposes, you can ignore theCLIP vs Densenet.pptx: This was a presentation which was updated during the project.old_code/: We ran some experiments that didnt go into the paper and also regenerated the data, so this is not used.Densenet_plots/: Contains some plots when we were comparing densenet model with vit+mlp abd googlenet. But all important plots went into the paper+SI.GoogleNet plots/: Contains some plots when we were comparing densenet model with vit+mlp abd googlenet. But all important plots went into the paper+SI.vit + mlp plots/: Contains some plots when we were comparing densenet model with vit+mlp abd googlenet. But all important plots went into the paper+SI.
License: Apache-2.0 license
c2db/data: Contains all the images, the splits into train and test, the csv files with the image ids and the labels and all the features.jv2d/data: JContains all the images, the splits into train and test, the csv files with the image ids and the labels and all the features.alex/data: Contains all the images, the splits into train and test, the csv files with the image ids and the labels and all the features.
The data is uploaded on Zenodo and can be found here: https://doi.org/10.5281/zenodo.17775145
Sayak Chakrabarty, Kamal Choudhary, Daniel Wines, Youjia Li, Vishu Gupta, Muhammed Nur Talha Kilic, Alok Choudhary & Ankit Agrawal, “Predicting Lattice Parameters from Atomic-Scale Images of Two Dimensional (2D) Materials Using Deep Learning,” The Journal of Physical Chemistry C, 2025.
This work was performed under the following financial assistance award 70NANB24H136 from U.S. Department of Commerce, National Institute of Standards and Technology as part of the Center for Hierarchical Materials Design (CHiMaD). Partial support is also acknowledged from NSF awards CMMI- 2053929, OAC-2331329, and the Northwestern Center for Nanocombinatorics. This work was performed with funding from the CHIPS Metrology Program, part of CHIPS for America, National Institute of Standards and Technology, U.S. Department of Commerce. Certain commercial equipment, instruments, software, or materials are identified in this paper in order to specify the experimental procedure adequately. Such identifications are not intended to imply recommendation or endorsement by NIST, nor are they intended to imply that the materials or equipment identified are necessarily the best available for the purpose.