abs

학술발표회초록보기

초록문의 abstract@kcsnet.or.kr

결제문의 member@kcsnet.or.kr

현재 가능한 작업은 아래와 같습니다.
  • 05월 20일 18시 이후 : 초록수정 불가능, 일정확인 및 검색만 가능

제125회 대한화학회 학술발표회 및 총회 Quantum chemical database and prediction of spin coupling constants

등록일
2020년 2월 4일 16시 10분 57초
접수번호
0265
발표코드
PHYS3-4 이곳을 클릭하시면 발표코드에 대한 설명을 보실 수 있습니다.
발표시간
화 16시 : 30분
발표형식
심포지엄
발표분야
Physical Chemistry - Recent Progress in Theoretical and Computational Chemistry
저자 및
공동저자
Sunghwan Choi
Korea Institute of Science and Technology Information, Korea
The field of quantum chemistry is no longer a branch of chemistry but now become an explanatory tool for many interdisciplinary researches. Many properties of chemicals including spectrum and thermodynamic properties can be accurately predicted through simulations. For this wide applicability of quantum chemical simulation, a large number of calculations are performed even at this moment. On the other hand, digitalization makes a transmission, storage, and processing of data less labor-intensive, therefore, various statistical methods to find useful information from existing data become attractive. Here, I want to show some efforts to make traditional quantum chemistry collaborate with new techniques: a construction of highly accurate quantum chemical database and prediction of spin coupling constants of nuclear magnetic resonance spectra. Many of interesting properties can be simultaneously determined by a quantum chemical calculation and those properties are potential resources for machine learning for quantum chemistry. The revised quantum chemical database named QM9-G4MP2 contains all raw output of Gaussian 16 packages. Therefore, anyone can easily find the properties that they want.[1] Second part of this talk focused on a machine learning model and its performance to predict spectroscopic data, spin coupling constants, (∂^2 E)/(∂I_M ∂I_N ). An attention-based model with pseudo-labeling and seed ensemble gives a high prediction performance. Training and validation of the model were performed with the gas-phase organic database. [1] Kim, H., Park, J.Y. & Choi, S. Energy refinement and analysis of structures in the QM9 database via a highly accurate quantum chemical method. Sci Data 6, 109 (2019).

상단으로