TOKYO – June 1, 2022 – A research paper co-authored by Preferred Networks, Inc. (PFN) and ENEOS Corporation was published in the highly regarded, peer-reviewed scientific journal Nature Communications on May 30, 2022. Today, the paper was included in Nature Communications Editor’s Highlights, a small number of articles selected by the journal’s editors as particularly interesting or important, in two focuses: AI and Machine Learning and Materials Science and Chemistry.

The paper discusses PFP, PFN’s deep learning technology for universal atomistic simulation that accelerates materials discovery. PFP is the core technology of Matlantis™, the universal atomistic simulator jointly developed by PFN and ENEOS, and provided by a joint venture between PFN and ENEOS, Preferred Computational Chemistry. 

Developed with PFN’s deep learning technology and supercomputer, PFP’s accuracy and applicability were enhanced using ENEOS’s expertise in chemistry. PFP is currently used as a base technology for a variety of materials discovery projects for energy conversion, advanced materials and more. The paper’s inclusion in Nature Communications Editor’s Highlights demonstrates that PFP’s academic significance and innovation have been recognized by the international community of researchers.

To learn more about the paper, please read the blog post by So Takamoto, PhD, PFN researcher and the first author of the paper:

Development of Universal Neural Network for Materials Discovery – PFN Tech Blog

The open-access paper is available on Nature Communications here:

Towards universal neural network potential for material discovery applicable to arbitrary combination of 45 elements