PFCC Launches New Feature LightPFP for Large-Scale Materials Simulation on Matlantis
LightPFP enables a more than tenfold increase in the maximum number of atoms that can be simulated on Matlantis
TOKYO – January 21, 2025 – Preferred Computational Chemistry, Inc. (PFCC) today launched a new feature named LightPFP for Matlantis™, the universal atomistic simulator jointly developed by Preferred Networks (PFN) and ENEOS Corporation (ENEOS) for AI-aided materials discovery.
PFP*, Matlantis’s core technology, is a universal machine learning potential that enables users to perform molecular dynamics simulations without limiting the type of the material. PFP does not require fine tuning for each material type, allowing a single model to simulate a wide variety of materials. There has been, however, a demand among Matlantis users to perform calculations at a larger scale for more realistic simulations of more complex phenomena than they currently can.
Developed in response to such demand, LightPFP allows users to limit the category of simulated material and use PFP as the training data for a light machine learning potential. This enables them to perform molecular dynamics calculations for hundreds of thousands of atoms, which is over ten times larger than PFP can currently process. Conventionally, a category-specific machine learning potential requires a vast amount of quantum chemical calculations to build the training data, making the training phase a major bottleneck. By using PFP for the training data, Matlantis users can reduce the time required for building the training data and testing the inference results.
In reality, materials for research are seldom perfect crystals but often have complex structures, and the subjects often include large-scale surfaces and interfaces. PFCC expects LightPFP will meet such requirements and further accelerate materials discovery for batteries, semiconductors, glass, polymer and other categories.

About Preferred Computational Chemistry
Preferred Computational Chemistry, Inc. (PFCC) was established in 2021 in Tokyo as a joint venture between Preferred Networks, Inc. (PFN) and ENEOS Corporation (ENEOS) with the mission to accelerate innovation and support companies and organizations to discover innovative materials for a sustainable future. PFCC provides Matlantis™, a cloud-based universal atomistic simulator for elucidating phenomena at an atomic level and discovering new materials. In developing Matlantis, PFN and ENEOS have incorporated a deep learning model into a conventional physical simulator to increase the simulation speed by tens of thousands of times and to support a wide variety of materials. Matlantis was built by combining PFN’s AI expertise and computing infrastructure with ENEOS’s expertise in chemistry. Matlantis is used by over 100 companies and organizations for discovering various materials including e-fuel catalysts, next-generation batteries, exhaust gas purifying catalysts, anti-friction lubricants, semiconductors and more. International sales of Matlantis started in 2023, and current international clients include Massachusetts Institute of Technology and Hyundai Motor Company. For more information, please visit: https://matlantis.com
*PFP is the name of the unique neural network potential (NNP) that PFN and ENEOS co-developed for universal atomistic simulation for materials discovery using deep learning. An NNP expresses molecular dynamics using a neural network. Source: S. Takamoto, et al. Nat Commun 13, 2991 (2022). doi: 10.1038/s41467-022-30687-9