SMT | 代理模型Python工具包推荐

1. Introduction

代理模型工具箱 (surrogate modeling toolbox, SMT) 是一个基于Python开发的第三方包，其中包含代理模型方法、采样技术和基准测试函数。此包提供了一个易于使用且便于实施其他方法的代理模型库。

SMT与现有的代理模型库不同，它强调导数，包括用于梯度增强建模的训练导数、预测导数和相对于训练数据的导数。

它还包括其他地方没有的新代理模型：通过偏最小二乘法减少克里金法和能量最小化样条插值。SMT使用自定义工具来记录，用于嵌入自动测试的代码和动态生成的图表，从而以贡献者最少的努力生成高质量的用户指南。

SMT是在新 BSD 许可证下分发的。

To cite SMT 2.0: P. Saves and R. Lafage and N. Bartoli and Y. Diouane and J. H. Bussemaker and T. Lefebvre and J. T. Hwang and J. Morlier and J. R. R. A. Martins. SMT 2.0: A Surrogate Modeling Toolbox with a focus on Hierarchical and Mixed Variables Gaussian Processes. Advances in Engineering Software, 2024.

@article{saves2024smt,
         author = {P. Saves and R. Lafage and N. Bartoli and Y. Diouane and J. Bussemaker and T. Lefebvre and J. T. Hwang and J. Morlier and J. R. R. A. Martins},
         title = {{SMT 2.0: A} Surrogate Modeling Toolbox with a focus on Hierarchical and Mixed Variables Gaussian Processes},
         journal = {Advances in Engineering Sofware},
         year = {2024},
         volume = {188}, 
         pages = {103571},
         doi = {https://doi.org/10.1016/j.advengsoft.2023.103571}}

To cite SMT legacy: M. A. Bouhlel and J. T. Hwang and N. Bartoli and R. Lafage and J. Morlier and J. R. R. A. Martins. A Python surrogate modeling framework with derivatives. Advances in Engineering Software, 2019.

@article{SMT2019,
 Author = {Mohamed Amine Bouhlel and John T. Hwang and Nathalie Bartoli and Rémi Lafage and Joseph Morlier and Joaquim R. R. A. Martins},
 Journal = {Advances in Engineering Software},
 Title = {A Python surrogate modeling framework with derivatives},
 pages = {102662},
 issn = {0965-9978},
 doi = {https://doi.org/10.1016/j.advengsoft.2019.03.005},
 Year = {2019}}

2 Required packages

SMT依赖于以下Python第三方包：

• numpy

• scipy

• scikit-learn

• pyDOE3

• Cython

3. Documentation contents

Getting started

• Surrogate modeling methods

   Least-squares approximation
   
   Second-order polynomial approximation
   
   Gausian process regression
  
       Kriging (KRG)
       KPLS
       KPLSK
       GPX
       GEKPLS
       Marginal Gaussian Process (MGP)
       Sparse Gaussian Process (SGP)
    Radial basis functions
   
   Inverse-distance weighting
  
   Regularized minimal-energy tensor-product splines
   
   GENN
  

Benchmarking problems

	Sphere function
	
	Branin function
	
	Lp norm function
	
	Rosenbrock function
	
	Tensor-product function
	
	Cantilever beam function
	
	Mixed cantilever beam function
	
	Multi-layer perceptron
	
	Robot arm function
	
	Torsion vibration function
	
	Water flow function
	
	Welded beam function
	
	Wing weight function

Sampling methods

	Random sampling
	
	Latin Hypercube sampling

	Full-factorial sampling

	pyDOE sampling methods

Examples

	1-D step-like data set
	
	RANS CRM wing 2-D data set
	
	Boeing 777 engine data set
	
	Learning Airfoil Parameters

Applications

	Mixed Integer and Hierarchical Design Spaces (Variables, Sampling and Context)
	
	Mixed integer surrogate
	
	Mixture of experts (MOE)
	
	Variable-fidelity modeling (VFM)
	
	Multi-Fidelity Kriging (MFK)
	
	Multi-Fidelity Kriging KPLS (MFKPLS)
	
	Multi-Fidelity Kriging KPLSK (MFKPLSK)
	
	Efficient Global Optimization (EGO)
	
	Proper Orthogonal Decomposition + Interpolation (PODI)
	
	Contributing to SMT
	
	Developer API for surrogate models
	
	Developer API for benchmarking problems
	
	Developer API for design spaces

	Developer API for sampling methods

4. Installation

如果您想安装最新版本

pip install smt

或者如果您想从当前主分支安装

pip install git+https://github.com/SMTOrg/smt.git@master

5. Usage

有关如何使用 SMT的示例，您可关注微信公众号"UQLearner"，后期将逐步更新SMT使用教程，或查看SMT网站教程 。