Materials informatics from first-principles calculation

2020 Fiscal Year Final Research Report

• Project/Area Number: 18H03843
• Research Category: Grant-in-Aid for Scientific Research (A)
• Allocation Type: Single-year Grants
• Section: 一般
• Review Section: Medium-sized Section 26:Materials engineering and related fields
• Research Institution: Kyoto University
• Principal Investigator: TANAKA ISAO 京都大学, 工学研究科, 教授 (70183861)
• Project Period (FY): 2018-04-01 – 2021-03-31
• Keywords: 新規物質探索 / インフォマティクス / 並列合成実験 / 推薦システム / 合成条件予測モデル

Outline of Final Research Achievements

We developed methodologies on parallel synthesis experiments, a synthesis predictor of unexperimented materials, and a feeding back model of synthesis process. They are expected to support rational materials design from the machine learning technique. We constructed a recommender system that predicts the successful or unsuccessful synthesis conditions for unexperimented compounds by applying the tensor decomposition method under the low rank assumption of the tensor. The training dataset was constructed with 1,000 of the 240,000 oxide synthesis conditions. Good prediction ability was demonstrated. These frameworks can accelerate the discovery of unknown materials.

Free Research Field

材料基礎科学

Academic Significance and Societal Importance of the Research Achievements

新規材料の発見を目指して，並列合成実験により1000件規模の酸化物の合成条件データベースを作成し，テンソル分解に基づいた「推薦システム」を構築した。
これまで合成研究者の勘と経験に頼ってきた試行錯誤的な物質合成に、多数のデータに基づいた機械学習手法からの合成成功条件の推薦結果を合わせることで、より効率的に新規物質の探索を行うことが可能になる。