Shulou(Shulou.com)12/24 Report--

CTOnews.com December 1 news, Google's DeepMind recently in the journal Nature showed its own AI tool GNoME, and introduced AI in materials science related applications, it is reported that DeepMind using GNoME found 2.2 million new crystals, of which 380,000 crystals belong to stable materials, can be manufactured in the laboratory, is expected to be applied in batteries or superconductors and other aspects.

▲ In DeepMind's current ICSD data, about 20,000 crystals are considered to be in "stable state" in calculation. Previously, teams such as Materials Project found another 28,000 crystals through a series of calculation methods. However, DeepMind believes that although the improved calculation method in the industry can speed up the discovery of new crystal structures, the time and money cost are quite high.

DeepMind's new tool, GNoME, is said to have broken through previous computational methods, accurately predicting a range of stable crystal structures and generating 2.2 million materials from them, which DeepMind claims would take 800 years to calculate manually.

DeepMindCTOnews.com learned from the DeepMind report that the efficiency of GNoME to develop materials is quite high, and the model has designed a total of 52,000 new graphene layered compounds, while before, humans only identified about 1,000 similar materials. In addition, GNoME has identified 528 potential lithium ion conductors with 25 times the conductivity of previous materials. Scientists believe that the above findings alone are expected to improve the energy consumption of batteries currently used in electronic products.

DeepMindDeepMind mentions that GNoME uses two strategies to find materials, the first is to create candidates based on known crystal structures, and the other is to explore candidate structures in a more random way based on chemical companies. The model processes and analyzes the outputs of both methods simultaneously through neural networks, using density functional theory calculations to assess the stability of these candidates. It also uses a method called "Active Learning" to improve the accuracy and efficiency of crystal prediction, thus greatly increasing the speed and success rate of discovering new materials.

DeepMind GNoME model is designed to reduce the cost of discovering new materials. At present, scientists around the world have made 736 new materials predicted by GNoME in the laboratory, which proves the accuracy and feasibility of GNoME crystal prediction in reality. DeepMind has now made the newly discovered crystal database of GNoME public to assist researchers in testing and manufacturing candidate materials.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.