Survey of multifidelity methods in uncertainty propagation, inference, and optimization

By Benjamin Peherstorfer, Karen E. Willcox, Max D. Gunzburger. 2018

938 citations recorded by the source corpus.

In many situations across computational science and engineering, multiple computational models are available that describe a system of interest. These different models have varying evaluation costs...

Subfields: Computational Theory and Mathematics, Statistical and Nonlinear Physics, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Statistical and Nonlinear Physics

In Statistics and Probability

Citations in the corpus

Conditional deep surrogate models for stochastic, high-dimensional, and multi-fidelity systems (2019, methodology): “As discussed in section 1, these methods have been successful in a wide spectrum of applications including design optimization [1, 2, 3, 4, 5], model calibration [7, 8, 9], and uncertainty quantification [10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21].”
Deep UQ: Learning deep neural network surrogate models for high dimensional uncertainty quantification (2018, methodology): “DNNs are also naturally suited for tasks for multilevel/multifidelity UQ [73, 74, 75].”
Solving Irregular and Data-enriched Differential Equations using Deep Neural Networks (2019, methodology): “To mitigate the expense of running forward models, surrogate modeling [51] and multifidelity hierarchies [52] are often conceived in order to develop cost-effective ways of examining the response of a PDE relative to its parametric responses.”
Data-driven reduced order modeling for time-dependent problems (2019, background): “One can refer to [33] for some multi-fidelity techniques and corrections for obtaining unbiased solutions.”
Machine Learning for High-entropy Alloys: Progress, Challenges and Opportunities (2022, methodology): “For instance, for a 250-atom CuZn supercell, the first-principles Monte-Carlo simulation requires the self-consistent (SCF) calculations of 600,000 DFT energies [46], which can easily consume a million CPU core-hours.”
Multi-Fidelity Physics-Constrained Neural Network and Its Application in Materials Modeling (2019, methodology): “By integrating the data from HF and low-ﬁ delity (LF) simulations, the trade-off between ef ﬁ ciency and accuracy for metamodeling can be made [38 – 40].”
A comprehensive review of digital twin — part 1: modeling and twinning enabling technologies (2022, neutral): “The field of multi-fidelity modeling is rich with literature, and interested readers should read several of the dedicated review papers (Fernández-Godino et al. 2016; Peherstorfer et al. 2018; Giselle Fernández-Godino et al. 2019).”
Managing computational complexity using surrogate models: a critical review (2020, methodology): “Other approaches to uncertainty propagation include fuzzy set approaches and worst-case scenario analysis (Sadaghiani et al. 2014; Aliza-deh et al. 2016a, b; Peherstorfer et al. 2018).”
Review of multi-fidelity models (2016, methodology): “In their 2016 survey, Peherstorfer et al. 2016 classiﬁed techniques for integrating multiple ﬁdelities into three categories: adaptation, fusion, and ﬁltering.”
A survey of adaptive sampling for global metamodeling in support of simulation-based complex engineering design (2017, background): “Some reviews and comparison studies regarding MFM can be found in (Toal 2015; Fernández-Godino et al. 2016; Peherstorfer et al. 2016; Park et al. 2017).”
Machine Learning in Aerodynamic Shape Optimization (2022, methodology): “Lower-fidelity CFD models are also available through coarsening high-fidelity CFD meshes [111, 112].”