An AI-driven, Cloud-based, Materials Discovery Platform for Nanomaterial Structure: PDFitc

We have developed a cloud-based, AI driven, platform for nanomaterial structure determination: “PDF in the cloud” (PDFitc.org), which consists of various applications for nanostructure determination, including a ML-based classifier for discovering material symmetry from a measured dataset, a high-throughput structure screening tool for predicting the structure of a measured signal, and a data-similarity visualization tool for finding changes in a signal in a time or temperature series.

S. Billinge, Q. Du, D. Hsu (Columbia U.)

Top: (Left) The landing page of PDFitc. (Right) The subpage of the structureMining@PDFitc application. Bottom: (Left) The Structure-mining results page. (Right) The plot of the automated PDF fit returned by structureMining, where the user uploaded PDF is shown in blue with the best-fit calculated PDF of the best guess structure (red).
  • Discovery of novel functional materials relies on a quantitative understanding of nanostructure

  • There is a need for advanced artificial intelligence  (AI) and machine learning (ML) approaches to nanostructure determination

  • We have developed a cloud-based, AI driven, platform for nanomaterial structure determination: “PDF in the cloud” (PDFitc.org).

  • It consists of various applications for nanostructure determination, including a ML-based classifier for discovering material symmetry from a measured dataset, a high-throughput structure screening tool for predicting the structure of a measured signal, and a data-similarity visualization tool for finding changes in a signal in a time or temperature series.

  • Uploaded data may be shared with collaborators and machine learned in the future for new materials discoveries.

The suite of Apps will be augmented over time allowing a wider range and more sophisticated analyses to be carried out and enriching the underlying database of PDFs that can then be used for machine learning

Designing Materials to Revolutionize and Engineer our Future (DMREF)