Skip to content

Framework to develop and deploy machine learning models that predict material properties of van der Waal heterostructures

License

Notifications You must be signed in to change notification settings

dwillhelm/HetML

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

58 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

HetML

HetML is a framework to develop and deploy machine learning models that predict material properties of van der Waals heterostructures. Please cite the following reference if you use this repo:

Predicting Van der Waals Heterostructures by a Combined Machine Learning and Density Functional Theory Approach”, Daniel Willhelm, Nathan Wilson, Raymundo Arroyave, Xiaoning Qian, Tahir Cagin, Ruth Pachter, and Xiaofeng Qian, ACS Applied Materials & Interfaces (2022). https://pubs.acs.org/doi/10.1021/acsami.2c04403

Framework

Target Properties MAE RMSE
Band Gap Energy ( $eV$ ) 0.12 $eV$ 0.17 $eV$
Ionization Energy ( $eV$ ) 0.09 $eV$ 0.14 $eV$
Electron Affiniity ( $eV$ ) 0.11 $eV$ 0.17 $eV$
Interlayer Distance ( $\mathring A$ ) 0.11 $\mathring A$ 0.18 $\mathring A$
Interlayer Binding Energy ( $meV/\mathring A^2$ ) 1.4 $meV/\mathring A^2$ 2.3 $meV/\mathring A^2$
Charge Transfer (via Bader Analysis) (Coming Soon!) (Coming Soon!)
Dipole Moment (Coming Soon!) (Coming Soon!)
In-plance lattice constant (Coming Soon!) (Coming Soon!)

Some deep learing models were also tested and can be found at this repo
An active learning (i.e. sequential learning) and baysian optimization demonstration can be found at this repo

Setup:

Python Environment

conda env create -f environment.yml or conda env create -f docs/envs/environment_full.yml (this YAML lists all dependencies and subdependencies) or pip install -r requirements.txt

Install Project Source Code

installs hetml Python package from setup.py pip install . or pip install -e . for a dev install

ToDo:

  • Get dataset

  • Structure library --> compressed.

  • src code and setup into local package

  • figures and interactive figures (in notebook?)

  • Setup public repo to accompany the publication

About

Framework to develop and deploy machine learning models that predict material properties of van der Waal heterostructures

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published