7月8日 Nicholas Zabaras:Physics-Constrained Learning of Multiscale and Dynamical PDE Systems (曹锡华数学论坛)

来源:中国足球竞彩比分 时间:2019-06-30浏览:91设置


讲座题目:Physics-Constrained Learning of Multiscale and Dynamical PDE Systems

主讲人:Nicholas Zabaras  教授

主持人:朱升峰  副教授

开始来源:中国足球竞彩比分 时间:2019-07-08 10:00:00  结束来源:中国足球竞彩比分 时间:2019-07-08 11:00:00

讲座地址:闵行数学楼102

主办单位:数学科学学院

  

报告人简介:

Prof. Nicholas Zabaras joined Notre Dame in 2016 as the Viola D. Hank Professor of Computational Science and Engineering. He is the Director of the interdisciplinary University of Notre Dame “Center for Informatics and Computational Science (CICS)” that aims to bridge the areas of data-sciences, scientific computing and uncertainty quantification for   complex multiscale/multiphysics problems in science and engineering. Among   the various appointments, Prof. Zabaras was until recently the Hans Fisher   Senior Fellow with the Institute for Advanced Study at the Technical University of Munich where he currently holds the position of TUM Ambassador. He served for nearly 23 years at all academic ranks on the faculty of Cornell University.


报告内容:

Surrogate modeling and uncertainty   quantification tasks for systems governed by PDEs are most often considered as supervised learning problems where input and output data pairs are used for training. The construction of such emulators is by definition a Small Data problem which poses challenges in Deep Learning approaches that have been developed to operate in the Big Data regime. Even in cases where such models have been shown to have good predictive capability in high-input dimensions, they fail to address constraints in the data implied by the PDE model. We will present a methodology in this direction that incorporates the   governing equations of the physical model in the loss/likelihood functions. The resulting   physics-constrained, deep learning models operate by employing only training   input data and provide predictive responses that obey the constraints of the problem at hand just as well as typical deterministic models. This work employs a convolutional encoder-decoder convolutional neural network approach as well as a conditional flow-based generative model for the solution of PDEs, surrogate model construction for PDEs, and uncertainty quantification tasks. We will highlight how these techniques can be extended to learn  Dynamics of complex PDE systems including turbulence models.

The above methods go beyond PDE systems and   are directly applicable to discrete atomistic models in Chemistry, Biology   and elsewhere. We will briefly discuss a data-free approach to   coarse-graining in atomistic simulations using generative models based on   variational auto-encoders.

  


返回原图
/