Non-Intrusive Reduced Basis Codes and Models for Surrogate Modelling and Sensitivity Analyses in Magnetotellurics

This code demonstrate the construction of surrogate models for the magnetotelluric response in geothermal reservoirs using the non-intrusive reduced basis method and gaussian process regression presented in the paper “Sensitivity Analysis using Physics-Based Machine Learning: An Example from Surrogate Modelling for Magnetotellurics“ by N. Lindner, D. Degen, A. Grayver and F. Wellmann. The non-intrusive reduced basis method is a physics-based machine learning technique originating from the field of projection based model order reduction methods, and is an efficient way of performing global sensitivity analysis.本代码演示了如何构建地热储层（geothermal reservoirs）的大地电磁响应（magnetotelluric response）代理模型（surrogate models），所用方法为N. Lindner、D. Degen、A. Grayver与F. Wellmann合著的论文《基于物理驱动机器学习的敏感性分析：以大地电磁代理建模为例》（Sensitivity Analysis using Physics-Based Machine Learning: An Example from Surrogate Modelling for Magnetotellurics）中提出的非侵入式约简基方法（non-intrusive reduced basis method）与高斯过程回归（gaussian process regression）。非侵入式约简基方法是一种源自基于投影的模型降阶方法（projection based model order reduction methods）领域的物理驱动机器学习（Physics-Based Machine Learning）技术，亦是开展全局敏感性分析（global sensitivity analysis）的高效手段。