東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Physics-Informed Machine Learning: T...

Wang, Sifan,

Linked to FindBook

Google Book

Amazon

博客來

Physics-Informed Machine Learning: Theory, Algorithms and Applications /

Record Type:	Electronic resources : Monograph/item
Title/Author:	Physics-Informed Machine Learning: Theory, Algorithms and Applications // Sifan Wang.
Author:	Wang, Sifan,
Description:	1 electronic resource (485 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-08, Section: B.
Contained By:	Dissertations Abstracts International85-08B.
Subject:	Applied mathematics. -
Online resource:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30696463
ISBN:	9798381471533

Physics-Informed Machine Learning: Theory, Algorithms and Applications /
Wang, Sifan,

Physics-Informed Machine Learning: Theory, Algorithms and Applications /Sifan Wang. - 1 electronic resource (485 pages)

Source: Dissertations Abstracts International, Volume: 85-08, Section: B.

The remarkable potential of deep learning in areas from computer vision to natural language processing has now found profound implications in modeling and simulating physical systems. Central to these advancements is the emerging field of physics-informed machine learning, a fusion of physical principles with machine learning techniques. There are three predominant strategies to integrate physics: inductive bias, learning bias, and observational bias.Our study delves into the inherent challenges and limitations of physics-informed machine learning, particularly in the physics-informed neural networks (PINNs) and deep operator networks (DeepONet). Our research is driven by overcoming fundamental challenges and enhancing the performance of these frameworks. Firstly, we investigate the gradient flow of PINNs, identifying a training failure stemming from unbalanced back-propagated gradients. This insight motivates us to generalize the neural tangent kernel (NTK) theory to PINNs. With this tool, we theoretically reveal that the training of PINNs suffer from spectral bias, causality violation and discrepancy in convergence rate of loss term. To address these critical issues, we propose several simple yet effective loss re-weighting algorithms and network architecture and validate them across a wide range range of representative benchmarks in computational physics. Besides, we present an extension of PINNs framework for solving free boundary problems.Moreover, we highlight the data-intensive demands of training neural operators and the potential inconsistency of their predictions with the underlying physics. To resolve these challenges, we propose physics-informed DeepONet, introducing a simple and effective regularization mechanism for biasing the outputs of DeepONet models towards ensuring physical consistency. Based on that, we propose a autoreressive training algorithm for performing long-time integration of evolution equations. We also analyze the training dynamics of DeepONets through the lens of NTK theory, uncovering a bias that favors the approximation of functions with larger magnitudes. Therefore, we propose a point-wise loss re-weighting algorithm to correct this bias and a novel network architecture that is more resilient to vanishing gradient pathologies. We leverage the proposed physics-informed DeepONet to build fast and differentiable surrogates for rapidly solving PDE-constrained optimization problems, even in the absence of any paired input-output training data. In summary, this thesis provides in-depth exploration into training, improving and applications aspects of physics-informed machine learning, paving a new way to developing scientific machine learning algorithms with better robustness and accuracy guarantees, as needed for many critical applications in computational science and engineering.

English

ISBN: 9798381471533Subjects--Topical Terms:

2122814
Applied mathematics.
Subjects--Index Terms:

Machine learning

Physics-Informed Machine Learning: Theory, Algorithms and Applications /
LDR:04263nmm a22004213i 4500 001 2428832
005 20250420214332.5
006 m o d
007 cr|nu||||||||
008 261011s2023 miu||||||m |||||||eng d
020 $a 9798381471533
035 $a (MiAaPQD)AAI30696463
035 $a AAI30696463
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Wang, Sifan, $e author. $3 3816295
245 1 0 $a Physics-Informed Machine Learning: Theory, Algorithms and Applications / $c Sifan Wang.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2023
300 $a 1 electronic resource (485 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-08, Section: B.
500 $a Advisors: Perdikaris, Paris Committee members: Lu, Lu; Chaudhari, Pratik.
502 $b Ph.D. $c University of Pennsylvania $d 2023.
520 $a The remarkable potential of deep learning in areas from computer vision to natural language processing has now found profound implications in modeling and simulating physical systems. Central to these advancements is the emerging field of physics-informed machine learning, a fusion of physical principles with machine learning techniques. There are three predominant strategies to integrate physics: inductive bias, learning bias, and observational bias.Our study delves into the inherent challenges and limitations of physics-informed machine learning, particularly in the physics-informed neural networks (PINNs) and deep operator networks (DeepONet). Our research is driven by overcoming fundamental challenges and enhancing the performance of these frameworks. Firstly, we investigate the gradient flow of PINNs, identifying a training failure stemming from unbalanced back-propagated gradients. This insight motivates us to generalize the neural tangent kernel (NTK) theory to PINNs. With this tool, we theoretically reveal that the training of PINNs suffer from spectral bias, causality violation and discrepancy in convergence rate of loss term. To address these critical issues, we propose several simple yet effective loss re-weighting algorithms and network architecture and validate them across a wide range range of representative benchmarks in computational physics. Besides, we present an extension of PINNs framework for solving free boundary problems.Moreover, we highlight the data-intensive demands of training neural operators and the potential inconsistency of their predictions with the underlying physics. To resolve these challenges, we propose physics-informed DeepONet, introducing a simple and effective regularization mechanism for biasing the outputs of DeepONet models towards ensuring physical consistency. Based on that, we propose a autoreressive training algorithm for performing long-time integration of evolution equations. We also analyze the training dynamics of DeepONets through the lens of NTK theory, uncovering a bias that favors the approximation of functions with larger magnitudes. Therefore, we propose a point-wise loss re-weighting algorithm to correct this bias and a novel network architecture that is more resilient to vanishing gradient pathologies. We leverage the proposed physics-informed DeepONet to build fast and differentiable surrogates for rapidly solving PDE-constrained optimization problems, even in the absence of any paired input-output training data. In summary, this thesis provides in-depth exploration into training, improving and applications aspects of physics-informed machine learning, paving a new way to developing scientific machine learning algorithms with better robustness and accuracy guarantees, as needed for many critical applications in computational science and engineering.
546 $a English
590 $a School code: 0175
650 4 $a Applied mathematics. $3 2122814
650 4 $a Computational physics. $3 3343998
653 $a Machine learning
653 $a Neural tangent kernel theory
653 $a Deep learning
653 $a Deep operator networks
690 $a 0364
690 $a 0216
690 $a 0800
710 2 $a University of Pennsylvania. $b Applied Mathematics and Computational Science. $e degree granting institution. $3 3814544
720 1 $a Perdikaris, Paris $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 85-08B.
790 $a 0175
791 $a Ph.D.
792 $a 2023
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30696463