Dynamic Mode Decomposition for approximating spectral Information of Koopman operators: overview and application
이승준
27동 116호
0
5063
2025.02.25 13:58
| 구분 | 기타 |
|---|---|
| 일정 | 2025-04-01(화) 17:00~19:30 |
| 세미나실 | 27동 116호 |
| 강연자 | 조항준 (Department of Mechanical Engineering, University of Washington) |
| 담당교수 | 하승열 |
| 기타 | HYKE-Hwarang Seminar |
※ 시간: 2025년 4월 1일 (화) 17:00 - 17:50
초록: Dynamic Mode Decomposition (DMD) is a well-known data-driven technique for analyzing time-series data. In this talk, we review recent developments in DMD, focusing on its connection to the spectral properties of Koopman operators and its application to model interpretation.
Part 1: We review various DMD schemes, emphasizing their relationship to Koopman operators.
Part 2: We apply Extended DMD, a variant of DMD that approximates Koopman spectra, to study glassy behavior in networks of coupled Kuramoto oscillators. Our simulations reveal a signature in the eigenvalue distribution associated with the emergence of glassy features.
