Development of coupling approach of cluster variation method and phase field method using machine learning

2021 Fiscal Year Final Research Report

• Project/Area Number: 20K22456
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Multi-year Fund
• Review Section: 0401:Materials engineering, chemical engineering, and related fields
• Research Institution: Hokkaido University
• Principal Investigator: Yamada Ryo 北海道大学, 工学研究院, 助教 (60883535)
• Project Period (FY): 2020-09-11 – 2022-03-31
• Keywords: 原子スケール / メゾスケール / マルチスケール / 機械学習

Outline of Final Research Achievements

Detail atomistic information is coarsened and lost in most mesoscopic simulations. This is due to a huge computational burden of atomic simulation. To avoid the issue, I have applied a machine learning to atomistic and mesoscopic simulations. By incorporating atomistic information into mesoscopic calculations, it was found that a time evolution of microstructure can be more reliably predicted than that of a conventional approach. Furthermore, it was made clear that the huge computational burden of local equilibrium calculation between different phases in mesoscopic simulations becomes relatively small by using machine learning.

Free Research Field

計算材料科学

Academic Significance and Societal Importance of the Research Achievements

本研究により、詳細な原子スケールの情報をメゾスケールの計算に適用できれば、より高精度に信頼度の高い材料設計ができることが明らかとなった。その手段として、近年注目を集めている機械学習は非常に有力である。今後は材料物性値の予測のさらなる高精度かや、大規模化が期待されるだろう。その手法の一つとして今回用いたような機械学習の適用法は非常に重要となるだろう。