Two-Dimensional Magnetotelluric Modeling Based on Stochastic Path Integral: TE Case

In this study, we investigate the application of the stochastic path integral (SPI) for 2-D magnetotelluric (MT) modeling. Our goal is to assess whether electromagnetic (EM) numerical algorithms can mechanistically adapt to parallel heterogeneous computing architectures and maximize the conversion o...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IEEE transactions on geoscience and remote sensing Ročník 63; s. 1 - 11
Hlavní autori: Zhou, Hongyu, Li, Maokun, Yang, Fan, Xu, Shenheng
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: New York IEEE 2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Predmet:
Algorithms
Computational electromagnetics
Computational modeling
Conductivity
Finite difference method
Helmholtz equations
Integral equations
magnetotellurics (MTs)
Mathematical models
Modelling
Monte Carlo (MC) simulation
Monte Carlo methods
numerical modeling
Numerical models
parallel computing
Parallel processing
Random walk
Shape
stochastic algorithm
Stochastic processes
Vectors
ISSN:0196-2892, 1558-0644
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this study, we investigate the application of the stochastic path integral (SPI) for 2-D magnetotelluric (MT) modeling. Our goal is to assess whether electromagnetic (EM) numerical algorithms can mechanistically adapt to parallel heterogeneous computing architectures and maximize the conversion of computational power into efficient EM field simulations. Using the Feynman-Kac formula, we derive a path integral representation of the MT Helmholtz equation in a stochastic framework. We can then investigate the online-offline two-stage MT-SPI algorithm that includes Monte Carlo (MC) random walks, mapping matrix construction, and large-scale matrix-vector multiplication. Numerical experiments verify the correctness and stability of the proposed SPI algorithm. At the relative error of 1.2%, SPI simultaneously achieves over <inline-formula> <tex-math notation="LaTeX">550\times </tex-math></inline-formula> acceleration and approximately 30% memory reduction on GPU under a MATLAB-based implementation, compared with the finite difference method (FDM).
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0196-2892
1558-0644
DOI:10.1109/TGRS.2025.3618304