Theoretical Analysis of Hyper-Ordered Structures using Large-Scale DFT Calculations and Machine-Learning Methods

Tsuyoshi Miyazaki

Presentation information

Symposium (Oral)

Symposium » 【Onsite Only】Materials with Hyperordered Structures and their Applications

[16p-A404-1~11] 【Onsite Only】Materials with Hyperordered Structures and their Applications

Thu. Mar 16, 2023 1:30 PM - 6:00 PM A404 (Building No. 6)

Hirokazu Masai(AIST), Hitoshi Wakabayashi(Tokyo Tech), Kazuo Tsutsui(Tokyo Tech), Madoka Ono(Hokkaido Univ./AGC), Koichi Hayashi(Nagoya Inst. Tech.)

5:15 PM - 5:45 PM

[16p-A404-10] Theoretical Analysis of Hyper-Ordered Structures using Large-Scale DFT Calculations and Machine-Learning Methods

〇Tsuyoshi Miyazaki¹ (1.Nat. Inst. Mater. Sci.)

Keywords:First-principles calculations, Machine-learning methods

We have developed a DFT code CONQUEST, which enables us to perform large-scale first-principles calculations of complex materials having hyper-ordered structures. In this talk, we will survey the overview of CONQUEST and its applications on semiconductor nano-structured materials and complex domain patterns in a ferroelectric material. A newly proposed method based on unsupervised machine learning techniques to analyze the local atomic structures observed in large-scale DFT-MD simulations will be also introduced. It will be demonstrated that these methods are useful to clarify the unique properties of materials having hyper-ordered structures.