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

Benjamin Peherstorfer, Karen Willcox, Max Gunzburger

Research output: Contribution to journalReview article

Abstract

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 and varying fidelities. Typically, a computationally expensive high-fidelity model describes the system with the accuracy required by the current application at hand, while lower-fidelity models are less accurate but computationally cheaper than the high-fidelity model. Outer-loop applications, such as optimization, inference, and uncertainty quantification, require multiple model evaluations at many different inputs, which often leads to computational demands that exceed available resources if only the high-fidelity model is used. This work surveys multifidelity methods that accelerate the solution of outer-loop applications by combining high-fidelity and low-fidelity model evaluations, where the low-fidelity evaluations arise from an explicit low-fidelity model (e.g., a simplified physics approximation, a reduced model, a data-fit surrogate) that approximates the same output quantity as the high-fidelity model. The overall premise of these multifidelity methods is that low-fidelity models are leveraged for speedup while the high-fidelity model is kept in the loop to establish accuracy and/or convergence guarantees. We categorize multifidelity methods according to three classes of strategies: adaptation, fusion, and filtering. The paper reviews multifidelity methods in the outer-loop contexts of uncertainty propagation, inference, and optimization.

Original languageEnglish (US)
Pages (from-to)550-591
Number of pages42
JournalSIAM Review
Volume60
Issue number3
DOIs
StatePublished - Jan 1 2018

Fingerprint

Uncertainty Propagation
Fidelity
Optimization
Model Evaluation
Multiple Models
Model
Uncertainty
Uncertainty Quantification
Computational Science
Reduced Model
Evaluation
Computational Model
Accelerate

Keywords

  • Model reduction
  • Multifidelity
  • Multifidelity optimization
  • Multifidelity statistical inference
  • Multifidelity uncertainty propagation
  • Multifidelity uncertainty quantification
  • Surrogate models

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computational Mathematics
  • Applied Mathematics

Cite this

Survey of multifidelity methods in uncertainty propagation, inference, and optimization. / Peherstorfer, Benjamin; Willcox, Karen; Gunzburger, Max.

In: SIAM Review, Vol. 60, No. 3, 01.01.2018, p. 550-591.

Research output: Contribution to journalReview article

Peherstorfer, Benjamin ; Willcox, Karen ; Gunzburger, Max. / Survey of multifidelity methods in uncertainty propagation, inference, and optimization. In: SIAM Review. 2018 ; Vol. 60, No. 3. pp. 550-591.
@article{9dcd56c04a4649fcbc27aeee345d3f07,
title = "Survey of multifidelity methods in uncertainty propagation, inference, and optimization",
abstract = "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 and varying fidelities. Typically, a computationally expensive high-fidelity model describes the system with the accuracy required by the current application at hand, while lower-fidelity models are less accurate but computationally cheaper than the high-fidelity model. Outer-loop applications, such as optimization, inference, and uncertainty quantification, require multiple model evaluations at many different inputs, which often leads to computational demands that exceed available resources if only the high-fidelity model is used. This work surveys multifidelity methods that accelerate the solution of outer-loop applications by combining high-fidelity and low-fidelity model evaluations, where the low-fidelity evaluations arise from an explicit low-fidelity model (e.g., a simplified physics approximation, a reduced model, a data-fit surrogate) that approximates the same output quantity as the high-fidelity model. The overall premise of these multifidelity methods is that low-fidelity models are leveraged for speedup while the high-fidelity model is kept in the loop to establish accuracy and/or convergence guarantees. We categorize multifidelity methods according to three classes of strategies: adaptation, fusion, and filtering. The paper reviews multifidelity methods in the outer-loop contexts of uncertainty propagation, inference, and optimization.",
keywords = "Model reduction, Multifidelity, Multifidelity optimization, Multifidelity statistical inference, Multifidelity uncertainty propagation, Multifidelity uncertainty quantification, Surrogate models",
author = "Benjamin Peherstorfer and Karen Willcox and Max Gunzburger",
year = "2018",
month = "1",
day = "1",
doi = "10.1137/16M1082469",
language = "English (US)",
volume = "60",
pages = "550--591",
journal = "SIAM Review",
issn = "0036-1445",
publisher = "Society for Industrial and Applied Mathematics Publications",
number = "3",

}

TY - JOUR

T1 - Survey of multifidelity methods in uncertainty propagation, inference, and optimization

AU - Peherstorfer, Benjamin

AU - Willcox, Karen

AU - Gunzburger, Max

PY - 2018/1/1

Y1 - 2018/1/1

N2 - 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 and varying fidelities. Typically, a computationally expensive high-fidelity model describes the system with the accuracy required by the current application at hand, while lower-fidelity models are less accurate but computationally cheaper than the high-fidelity model. Outer-loop applications, such as optimization, inference, and uncertainty quantification, require multiple model evaluations at many different inputs, which often leads to computational demands that exceed available resources if only the high-fidelity model is used. This work surveys multifidelity methods that accelerate the solution of outer-loop applications by combining high-fidelity and low-fidelity model evaluations, where the low-fidelity evaluations arise from an explicit low-fidelity model (e.g., a simplified physics approximation, a reduced model, a data-fit surrogate) that approximates the same output quantity as the high-fidelity model. The overall premise of these multifidelity methods is that low-fidelity models are leveraged for speedup while the high-fidelity model is kept in the loop to establish accuracy and/or convergence guarantees. We categorize multifidelity methods according to three classes of strategies: adaptation, fusion, and filtering. The paper reviews multifidelity methods in the outer-loop contexts of uncertainty propagation, inference, and optimization.

AB - 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 and varying fidelities. Typically, a computationally expensive high-fidelity model describes the system with the accuracy required by the current application at hand, while lower-fidelity models are less accurate but computationally cheaper than the high-fidelity model. Outer-loop applications, such as optimization, inference, and uncertainty quantification, require multiple model evaluations at many different inputs, which often leads to computational demands that exceed available resources if only the high-fidelity model is used. This work surveys multifidelity methods that accelerate the solution of outer-loop applications by combining high-fidelity and low-fidelity model evaluations, where the low-fidelity evaluations arise from an explicit low-fidelity model (e.g., a simplified physics approximation, a reduced model, a data-fit surrogate) that approximates the same output quantity as the high-fidelity model. The overall premise of these multifidelity methods is that low-fidelity models are leveraged for speedup while the high-fidelity model is kept in the loop to establish accuracy and/or convergence guarantees. We categorize multifidelity methods according to three classes of strategies: adaptation, fusion, and filtering. The paper reviews multifidelity methods in the outer-loop contexts of uncertainty propagation, inference, and optimization.

KW - Model reduction

KW - Multifidelity

KW - Multifidelity optimization

KW - Multifidelity statistical inference

KW - Multifidelity uncertainty propagation

KW - Multifidelity uncertainty quantification

KW - Surrogate models

UR - http://www.scopus.com/inward/record.url?scp=85053270641&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85053270641&partnerID=8YFLogxK

U2 - 10.1137/16M1082469

DO - 10.1137/16M1082469

M3 - Review article

AN - SCOPUS:85053270641

VL - 60

SP - 550

EP - 591

JO - SIAM Review

JF - SIAM Review

SN - 0036-1445

IS - 3

ER -