Skip to main content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.


The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.



Fast screening and mapping energy landscape with chemo-structural descriptors


Technological advancement requires discovery of new materials, the possible materials can span up to 10100. Conventional methods such as high-throughput (HT) experiment and HT-computational methods such as DFT and FF are too expensive. One of the possible approaches to deal with this problem is using artificial-intelligence (AI) tools such as machine-learning. Key challenges in employing AI techniques to materials are:

  • having a  high-fidelity large (>104) database
  • choosing effective descriptors for materials
  • choosing appropriate algorithm/work-flow during AI design

Relation to AI

In this program, we are

  • developing a high-fidelity DFT database (JARVIS-DFT)
  • developing chemo-structural descriptors (total 1557 right now for one material)
  • applying highly-explainable and acclaimed algorithm: gradient boosting decision trees (GBDT)

Both our database and machine learning models are publicly available at


  • Computational expertise and NIST facility
  • Continuous collaboration with DFT and machine learning experts in NIST’s Material Measurement Lab and Statistical Engineering divisions
  • Experimental validation with NIST collaborators

Major Accomplishments

  • >30000 materials (>105 calculated properties) in JARVIS-DFT
  • Developed completely new set of structural descriptors
  • Obtained highly-accurate machine learning models for formation energy, band gap, refractive index, exfoliation energy, and magnetic moment
  • Integration of genetic algorithm with machine learning to map energy landscape
  • New website for predicting properties with machine learning on the fly:

Ongoing Challenges

  • Computing more data, especially 2D materials
  • Visualization of large datasetsBetter descriptors for magnetic properties
  • Machine learning on topological materials properties

Created March 26, 2019, Updated April 3, 2019