Materials Platform for Data Science: API

The API stands for the application programming interface, a way to get the MPDS scientific data automatically in a high-throughput manner for the machine analysis. The possible applications are high-throughput simulations, machine learning, and other data-intensive techniques in materials science.

Here you will find:

• website developer.mpds.io with the documentation

• issue tracker for the MPDS API (please, report any troubles creating issues)

• kickoff Python scripts:

  • The uranium-oxygen chemical bond length distribution
  • Clustering the band gaps of binary compounds
  • Statistical relationship of physical property and crystalline structure
  • Retrieval of binary systems producing no compounds
  • Calculating the Pilling-Bedworth ratio of metals
  • and more, just see the MPDS API kickoff scripts folder

• MPDS API Jupyter notebooks:

  • Short intro: basic plotting using the periodic table
  • Basic MPDS API usage: machine-learning and peer-reviewed data
  • Advanced MPDS API usage: unusual materials phases from the machine learning
  • Advanced MPDS API usage: pVT-data and EoS fitting

All information here is freely available under the MIT and CC BY 4.0 licenses.

Login at the MPDS if you'd like to use this API with the open MPDS data:

• cell parameters vs. temperature and pressure diagrams (about 6k entries)
• all compounds containing both Ag and K (about 250 entries)
• all binary compounds of oxygen (about 6k entries)
• all data generated by machine-learning (about 900k entries)
• all data generated by first-principles calculations

Contact us at mpds-api@tilde.pro if you'd like to use this API with the all MPDS data.