Deep Learning-Based Forward Modeling and Inversion Techniques for Computational Physics Problems

This book explores the emerging deep learning technique in computational physics, assessing its potential to substitute conventional numerical solvers for real-time field calculations. It covers basic DL frameworks, solves heat conduction problems, reconstructs heat flux on curved surfaces, and introduces advanced frameworks and physics applications. It is intended for graduate students, professionals, and researchers interested in DL for computational physics.

Format: Hardback
Length: 180 pages
Publication date: 06 July 2023
Publisher: Taylor & Francis Ltd

This comprehensive book delves into the emerging field of deep learning (DL) in computational physics, exploring its remarkable potential to replace traditional numerical solvers for real-time field calculations. Through rigorous training, the proposed architecture demonstrates remarkable capabilities in both forward computing and inverse retrieval tasks.

The book encompasses a wide range of topics, beginning with an introduction to fundamental DL frameworks. In Chapter 2, the classical U-net is employed to solve the steady heat conduction problem, encompassing both passive and active scenarios. Chapter 3 showcases the reconstruction of sophisticated heat flux on curved surfaces using the Conv-LSTM, exhibiting exceptional accuracy and efficiency. Chapter 4 employs a physics-informed DL structure alongside a nonlinear mapping module to derive the space/temperature/time-related thermal conductivity through transient temperature measurements. Finally, Chapter 5 presents a comprehensive overview of the latest advanced frameworks and their corresponding physics applications.

This book is specifically designed for graduate students, professional practitioners, and researchers who are eager to explore the realm of DL for computational physics. Its detailed explanations, practical examples, and extensive references make it an invaluable resource for anyone seeking to advance their knowledge and expertise in this rapidly evolving field.

Weight: 434g
Dimension: 162 x 241 x 17 (mm)
ISBN-13: 9781032502984