JARVIS

Official Resources

• Homepage: https://jarvis.nist.gov/
• Documentation: https://jarvis-tools.readthedocs.io/
• Source Repository: https://github.com/usnistgov/jarvis
• License: NIST License (Public Domain)

Overview

JARVIS is an integrated framework and database developed by NIST for data-driven materials design. It consists of the JARVIS-DFT database (DFT calculations), JARVIS-FF (Force fields), and JARVIS-ML (Machine learning). It emphasizes properties relevant to applications (e.g., solar cells, thermoelectrics, dielectrics) and uses high-throughput workflows powered by jarvis-tools.

Scientific domain: Materials database, high-throughput DFT, machine learning
Target user community: Materials scientists, ML researchers

Capabilities (CRITICAL)

• JARVIS-DFT: Database of >80,000 materials with VASP/TB-mBJ calculations.
• Properties: Elastic constants, dielectric tensors, piezoelectricity, topological invariants, solar efficiency, exfoliation energy.
• JARVIS-FF: Database of classical force field properties.
• JARVIS-ML: Machine learning models and descriptors.
• Tools: jarvis-tools python package for automation and analysis.

Sources: JARVIS website, Sci. Data 5, 180082 (2018)

Inputs & Outputs

• Input formats: VASP inputs, LAMMPS inputs
• Output data types: JSON databases, XML files

Interfaces & Ecosystem

• VASP: Primary DFT engine
• LAMMPS: Primary MD engine
• TensorFlow/PyTorch: Used for ML models
• Web API: For programmatic access

Workflow and Usage

1. Web: Search JARVIS-DFT for "MoS2".
2. Python:

   from jarvis.db.figshare import data
   dft_data = data('dft_3d')
   

Performance Characteristics

• High-quality data (TB-mBJ for band gaps)
• Comprehensive coverage of 2D materials

Application Areas

• 2D materials discovery
• Thermoelectric materials
• Topological materials
• Machine learning benchmarking

Community and Support

• Developed by NIST (Kamal Choudhary et al.)
• Public domain

Verification & Sources

Primary sources:

1. Homepage: https://jarvis.nist.gov/
2. GitHub: https://github.com/usnistgov/jarvis
3. Publication: K. Choudhary et al., Sci. Data 5, 180082 (2018)

Confidence: VERIFIED

Verification status: ✅ VERIFIED

• Website: ACTIVE
• Documentation: COMPREHENSIVE
• Source: OPEN (GitHub)
• Development: ACTIVE (NIST)
• Applications: Materials database, ML, DFT