BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Department of Computational and Data Sciences - ECPv5.3.1//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Department of Computational and Data Sciences
X-ORIGINAL-URL:https://cds.iisc.ac.in
X-WR-CALDESC:Events for Department of Computational and Data Sciences
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
BEGIN:STANDARD
TZOFFSETFROM:+0530
TZOFFSETTO:+0530
TZNAME:IST
DTSTART:20210101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20210805T180000
DTEND;TZID=Asia/Kolkata:20210805T190000
DTSTAMP:20220516T145307
CREATED:20210803T034642Z
LAST-MODIFIED:20210803T034649Z
UID:6725-1628186400-1628190000@cds.iisc.ac.in
SUMMARY:{Seminar} @ CDS: 05th August 2021: "Neural Closure Models for Dynamical Systems."
DESCRIPTION:Department of Computational and Data Sciences\n\nDepartment Seminar\n\nSpeaker : Dr. Abhinav Gupta\, MIT \n\nTitle : “Neural Closure Models for Dynamical Systems”\nDate & Time : August 05\, 2021\, 06:00 PM\nVenue : Online\n\n\nABSTRACT\nComplex dynamical systems are used for predictions in many domains. Because of computational costs\, models are truncated\, coarsened\, or aggregated. As the neglected and unresolved terms become important\, the utility of model predictions diminishes. We develop a novel\, versatile\, and rigorous methodology to learn non-Markovian closure parameterizations for known-physics/low-fidelity models using data from high-fidelity simulations. The new “neural closure models” augment low-fidelity models with neural delay differential equations (nDDEs)\, motivated by the Mori-Zwanzig formulation and the inherent delays in complex dynamical systems. We demonstrate that neural closures efficiently account for truncated modes in reduced-order-models\, capture the effects of subgrid-scale processes in coarse models\, and augment the simplification of complex biological and physical-biogeochemical models. We find that using non-Markovian over Markovian closures improves long-term prediction accuracy and requires smaller networks. We derive adjoint equations and network architectures needed to efficiently implement the new discrete and distributed nDDEs\, for any time-integration schemes and allowing nonuniformly-spaced temporal training data. The performance of discrete over distributed delays in closure models is explained using information theory\, and we find an optimal amount of past information for a specified architecture. Finally\, we analyze computational complexity and explain the limited additional cost due to neural closure models.\n\n \n\n\nPreprint: https://arxiv.org/abs/2012.13869 \n\n\n \n\nBIOGRAPHY\n\n\n\nAbhinav is a 5th year PhD candidate in Mechanical Engineering and Computation at MIT. He received his Bachelor’s degree and Master’s degree in Mechanical Engineering from the Indian Institute of Technology\, Kanpur. At MIT he was a fellow of the MIT-Tata Center for Technology & Design from 2018-20\, and recipient of the 2020-21 MathWorks Mechanical Engineering Fellowship. \n\nAbhinav is currently developing state-of-the-art scientific machine learning algorithms\, with applications to predictive ocean modeling. Apart from his present work\, he has specifically worked on uncertainty quantification\, data assimilation\, Bayesian model learning\, and optimal sampling for high-dimensional systems. The algorithms he develops are problem agnostic and can be widely applied. He believes that his unique background in mechanical engineering\, applied mathematics\, machine learning\, and computing position him to identify and implement cross-disciplinary solutions to problems. \n \n\n\n\nHost Faculty: Prof. Deepak Subramani
URL:https://cds.iisc.ac.in/calendar/seminar-cds-05th-august-neural-closure-models-for-dynamical-systems/
END:VEVENT
END:VCALENDAR