☆ 3.8 Article

A DeepONet multi-fidelity approach for residual learning in reduced order modeling

ADVANCED MODELING AND SIMULATION IN ENGINEERING SCIENCES (2023)

Related references

Note: Only part of the references are listed.

Article Computer Science, Interdisciplinary Applications

B-DeepONet: An enhanced Bayesian DeepONet for solving noisy parametric PDEs using accelerated replica exchange SGLD

Guang Lin et al.

Summary: The Deep Operator Network (DeepONet) is a neural network architecture used to approximate operators, including the solution operator of parametric PDEs. In order to handle noisy data, researchers propose a Bayesian DeepONet based on replica exchange Langevin diffusion (reLD) that improves training convergence and uncertainty estimation. They also propose an accelerated training framework and a transfer learning strategy to reduce computational cost and speed up the training process.

JOURNAL OF COMPUTATIONAL PHYSICS (2023)

Add to Collection

Article Engineering, Multidisciplinary

Context-aware learning of hierarchies of low-fidelity models for multi-fidelity uncertainty quantification

Ionut-Gabriel Farcas et al.

Summary: Multi-fidelity Monte Carlo methods leverage low-fidelity and surrogate models for variance reduction to make uncertainty quantification tractable. This work proposes a context-aware multi-fidelity Monte Carlo method that optimizes the balance between training costs and sampling costs. The method applies to hierarchies of different types of low-fidelity models and allows for optimal trade-offs between training and sampling to minimize mean-squared errors of estimators.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2023)

Add to Collection

Article Engineering, Multidisciplinary

A multifidelity approach coupling parameter space reduction and nonintrusive POD with application to structural optimization of passenger ship hulls

Marco Tezzele et al.

Summary: This study presents a structural optimization pipeline for modern passenger ship hulls, focusing on reducing metal raw materials used during manufacturing. Using advanced model order reduction techniques, the dimensionality of input parameters and outputs of interest is reduced. The method incorporates parameter space reduction through active subspaces into the proper orthogonal decomposition with interpolation method in a multi-fidelity setting. Comprehensive testing and error analysis demonstrate the effectiveness and usefulness of the method, especially during the preliminary design phase with high dimensional parameterizations.

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING (2023)

Add to Collection

Article Mathematics, Applied

Towards a Machine Learning Pipeline in Reduced Order Modelling for Inverse Problems: Neural Networks for Boundary Parametrization, Dimensionality Reduction and Solution Manifold Approximation

Anna Ivagnes et al.

Summary: In this work, a model order reduction framework is proposed for dealing with inverse problems in a non-intrusive setting. The computational load of inverse problems, especially in a partial differential equation context, is huge due to the iterative optimization process. To accelerate this procedure, a numerical pipeline involving artificial neural networks is applied to parametrize the boundary conditions, compress the dimensionality of snapshots, and approximate the parametric solution manifold. This approach provides a general framework capable of ad-hoc parametrization of the inlet boundary and quickly converges to the optimal solution thanks to model order reduction. The results obtained by applying these methods to two different CFD test cases are presented in this contribution.

JOURNAL OF SCIENTIFIC COMPUTING (2023)

Add to Collection

Review Engineering, Multidisciplinary

Operator inference for non-intrusive model reduction with quadratic manifolds

Rudy Geelen et al.

Summary: This paper proposes a novel approach for learning a quadratic manifold from high-dimensional data and using it to derive efficient physics-based reduced-order models. The approach involves a polynomial mapping between high-dimensional states and a low-dimensional embedding, consisting of a linear subspace representation and a quadratic component. This non-intrusive method combines the quadratic manifold approximation with projection-based model reduction for scalable learning of reduced-order models.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2023)

Add to Collection

Article Engineering, Multidisciplinary

Hyper-differential sensitivity analysis with respect to model discrepancy: Optimal solution updating

Joseph Hart et al.

Summary: A common goal in science and engineering is to optimize computational models, but often high-fidelity numerical emulation is not feasible due to complexity and costs. Instead, lower-fidelity models are used with intrusive algorithms for large-scale optimization. This article introduces a novel approach that updates the optimization solution using post-optimality sensitivities with respect to model discrepancy. Limited high-fidelity data is used to calibrate the model discrepancy in a Bayesian framework. Numerical results show significant improvement in optimal solutions computed using low-fidelity models with limited evaluations of high-fidelity models.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2023)

Add to Collection

Article Engineering, Multidisciplinary

Multi-fidelity regression using artificial neural networks: Efficient approximation of parameter-dependent output quantities

Mengwu Guo et al.

Summary: Multi-fidelity methods incorporate information from other sources to improve model results when time or budget constraints limit the generation of additional data.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2022)

Add to Collection

Article Engineering, Multidisciplinary

The Neural Network shifted-proper orthogonal decomposition: A machine learning approach for non-linear reduction of hyperbolic equations

Davide Papapicco et al.

Summary: In this work, a statistical learning framework is used to automatically detect the correct pre-processing transformation for projection-based reduced order modeling of models with dominant advection. The purely data-driven method allows for the generalization of linear subspace manipulation to nonlinear hyperbolic problems with unknown advection fields.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2022)

Add to Collection

Article Mathematics, Applied

Inadequacy of Linear Methods for Minimal Sensor Placement and Feature Selection in Nonlinear Systems: A New Approach Using Secants

Samuel E. Otto et al.

Summary: Sensor placement and feature selection are crucial steps in engineering, modeling, and data science. Standard techniques often fail in nonlinear systems, and we propose a new data-driven approach to address this issue.

JOURNAL OF NONLINEAR SCIENCE (2022)

Add to Collection

Article Physics, Multidisciplinary

Multifidelity deep neural operators for efficient learning of partial differential equations with application to fast inverse design of nanoscale heat transport

Lu Lu et al.

Summary: This paper proposes a multifidelity neural operator based on deep neural networks, which can reduce the demand for high-fidelity data and achieve smaller errors in solving heat transport problems. By combining with genetic algorithms and topology optimization, it enables fast solvers and inverse design for the phonon Boltzmann transport equation (BTE).

PHYSICAL REVIEW RESEARCH (2022)

Add to Collection

Article Engineering, Multidisciplinary

Multi-fidelity deep neural network surrogate model for aerodynamic shape optimization

Xinshuai Zhang et al.

Summary: The study establishes an effective optimization framework of aerodynamic shape design based on the multi-fidelity deep neural network (MFDNN) model, achieving high accuracy by blending different fidelity information. The proposed framework significantly improves optimization efficiency and outperforms single-fidelity methods.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2021)

Add to Collection

Article Mathematics, Applied

A weighted POD-reduction approach for parametrized PDE-constrained optimal control problems with random inputs and applications to environmental sciences

Giuseppe Carere et al.

Summary: Reduced basis approximations of OCPs governed by steady PDEs with random parametric inputs are analyzed and constructed using a Reduced Order Model based on weighted Proper Orthogonal Decomposition. The well-posedness of such OCPs is demonstrated using the adjoint approach, which is more general than the conventional Lagrangian approach. It is shown that the aggregation step in the construction of Reduced Order Models can be skipped for noncoercive problems, leading to a cheaper online phase.

COMPUTERS & MATHEMATICS WITH APPLICATIONS (2021)

Add to Collection

Article Computer Science, Interdisciplinary Applications

On the comparison of LES data-driven reduced order approaches for hydroacoustic analysis

Mahmoud Gadalla et al.

Summary: In this study, DMD and POD methods were applied to hydroacoustic datasets computed using LES with FWH analogy. Results showed that POD can reduce global spatiotemporal errors compared with DMD, while PODI has a relative superiority in predictive character over DMD when employed in data-driven reduced models.

COMPUTERS & FLUIDS (2021)

Add to Collection

Article Engineering, Marine

Hull Shape Design Optimization with Parameter Space and Model Reductions, and Self-Learning Mesh Morphing

Nicola Demo et al.

Summary: In this study, a data-driven framework was proposed to reduce the computational burden in shape optimization in the field of parametric partial differential equations. The approach involved using multiple reduction techniques such as POD and ASGA for dimensional reduction and efficient genetic optimization. The parameterization of shape was directly applied to the computational mesh, preserving the topology and quality of the original mesh and avoiding the need for additional meshing steps.

JOURNAL OF MARINE SCIENCE AND ENGINEERING (2021)

Add to Collection

Article Engineering, Multidisciplinary

A numerical approach for heat flux estimation in thin slabs continuous casting molds using data assimilation

Umberto Emil Morelli et al.

Summary: In this study, the industrial problem of estimating mold-steel heat flux in continuous casting mold in real-time is considered. The inverse problem is approached through mold modeling and heat flux parameterization, with the parameterization method outperforming the traditional method in terms of performance and computational cost.

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING (2021)

Add to Collection

Article Computer Science, Interdisciplinary Applications

Multi-fidelity Bayesian neural networks: Algorithms and applications

Xuhui Meng et al.

Summary: A novel class of Bayesian neural networks is proposed for training with noisy data of variable fidelity, applied to function approximations and solving inverse problems based on partial differential equations (PDEs). The multi-fidelity BNNs consist of three neural networks, effectively modeling the correlation and uncertainty between different fidelity data and accurately estimating posterior distributions of hyperparameters. The method demonstrates adaptively capturing linear and nonlinear correlation, identifying unknown parameters in PDEs, and quantifying uncertainties in predictions, ultimately enhancing prediction accuracy through active learning.

JOURNAL OF COMPUTATIONAL PHYSICS (2021)

Add to Collection

Article Mathematics, Applied

A SUPERVISED LEARNING APPROACH INVOLVING ACTIVE SUBSPACES FOR AN EFFICIENT GENETIC ALGORITHM IN HIGH-DIMENSIONAL OPTIMIZATION PROBLEMS

Nicola Demo et al.

Summary: This study presents an extension of genetic algorithm utilizing active subspaces, mapping the evolutionary gradient onto a lower-dimensional space to address the complexity of optimizing high-dimensional functions. The novel method ASGA shows differences and similarities with the standard GA method in benchmark function tests, eventually being successfully applied to an aeronautical shape optimization problem.

SIAM JOURNAL ON SCIENTIFIC COMPUTING (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators

Lu Lu et al.

Summary: This study extends the capabilities of neural networks with the introduction of the deep operator network (DeepONet), which can be used to learn various operators, including explicit and implicit operators. Different formulations of the input function space were studied and their effect on generalization error for 16 diverse applications was examined.

NATURE MACHINE INTELLIGENCE (2021)

Add to Collection

Article Computer Science, Interdisciplinary Applications

A composite neural network that learns from multi-fidelity data: Application to function approximation and inverse PDE problems

Xuhui Meng et al.

JOURNAL OF COMPUTATIONAL PHYSICS (2020)

Add to Collection

Article Mathematics, Applied

Lift & Learn: Physics-informed machine learning for large-scale nonlinear dynamical systems

Elizabeth Qian et al.

PHYSICA D-NONLINEAR PHENOMENA (2020)

Add to Collection

Article Mechanics

Leveraging reduced-order models for state estimation using deep learning

Nirmal J. Nair et al.

JOURNAL OF FLUID MECHANICS (2020)

Add to Collection

Article Mathematics, Applied

A Weighted POD Method for Elliptic PDEs with Random Inputs

Luca Venturi et al.

JOURNAL OF SCIENTIFIC COMPUTING (2019)

Add to Collection

Article Engineering, Aerospace

Nonlinear Model Order Reduction via Lifting Transformations and Proper Orthogonal Decomposition

Boris Kramer et al.

AIAA JOURNAL (2019)

Add to Collection

Review Mechanics

A non-intrusive approach for the reconstruction of POD modal coefficients through active subspaces

Nicola Demo et al.

COMPTES RENDUS MECANIQUE (2019)

Add to Collection

Article Mathematics, Applied

Neural network closures for nonlinear model order reduction

Omer San et al.

ADVANCES IN COMPUTATIONAL MATHEMATICS (2018)

Add to Collection

Article Engineering, Multidisciplinary

Multi-fidelity optimization of super-cavitating hydrofoils

L. Bonfiglio et al.

COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING (2018)

Add to Collection

Article Automation & Control Systems

Data-Driven Sparse Sensor Placement for Reconstruction DEMONSTRATING THE BENEFITS OF EXPLOITING KNOWN PATTERNS

Krithika Manohar et al.

IEEE CONTROL SYSTEMS MAGAZINE (2018)

Add to Collection

Article Mathematics, Applied

THE SHIFTED PROPER ORTHOGONAL DECOMPOSITION: A MODE DECOMPOSITION FOR MULTIPLE TRANSPORT PHENOMENA

J. Reiss et al.

SIAM JOURNAL ON SCIENTIFIC COMPUTING (2018)

Add to Collection

Article Mathematics, Applied

Survey of Multifidelity Methods in Uncertainty Propagation, Inference, and Optimization

Benjamin Peherstorfer et al.

SIAM REVIEW (2018)

Add to Collection

Article Engineering, Marine

Improving SWATH Seakeeping Performance using Multi-Fidelity Gaussian Process and Bayesian Optimization

Luca Bonfiglio et al.

JOURNAL OF SHIP RESEARCH (2018)

Add to Collection

Article Computer Science, Interdisciplinary Applications

Inferring solutions of differential equations using noisy multi-fidelity data

Maziar Raissi et al.

JOURNAL OF COMPUTATIONAL PHYSICS (2017)

Add to Collection

Article Multidisciplinary Sciences

Nonlinear information fusion algorithms for data-efficient multi-fidelity modelling

P. Perdikaris et al.

PROCEEDINGS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES (2017)

Add to Collection

Article Engineering, Aerospace

Surrogate Modeling Approach to Support Real-Time Structural Assessment and Decision Making

L. Mainini et al.

AIAA JOURNAL (2015)

Add to Collection

Article Engineering, Multidisciplinary

Supremizer stabilization of POD-Galerkin approximation of parametrized steady incompressible Navier-Stokes equations

Francesco Ballarin et al.

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING (2015)

Add to Collection

Article Engineering, Multidisciplinary

Progressive construction of a parametric reduced-order model for PDE-constrained optimization

Matthew J. Zahr et al.

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING (2015)

Add to Collection

Article Computer Science, Interdisciplinary Applications

Design optimization using hyper-reduced-order models

David Amsallem et al.

STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION (2015)

Add to Collection

Article Engineering, Multidisciplinary

Multifidelity approaches for optimization under uncertainty

Leo W. T. Ng et al.

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING (2014)

Add to Collection

Article Mathematics, Applied

Compressive Sensing and Low-Rank Libraries for Classification of Bifurcation Regimes in Nonlinear Dynamical Systems

Steven L. Brunton et al.

SIAM JOURNAL ON APPLIED DYNAMICAL SYSTEMS (2014)

Add to Collection

Article Mechanics

Compressive sensing based machine learning strategy for characterizing the flow around a cylinder with limited pressure measurements

Ido Bright et al.

PHYSICS OF FLUIDS (2013)

Add to Collection

Article Engineering, Multidisciplinary

Nonlinear model order reduction based on local reduced-order bases

David Amsallem et al.

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING (2012)

Add to Collection

Article Thermodynamics

A comparative study of POD interpolation and POD projection methods for fast and accurate prediction of heat transfer problems

Yi Wang et al.

INTERNATIONAL JOURNAL OF HEAT AND MASS TRANSFER (2012)

Add to Collection

Article Meteorology & Atmospheric Sciences

Efficient sensor placement for ocean measurements using low-dimensional concepts

B. Yildirim et al.

OCEAN MODELLING (2009)

Add to Collection

Article Multidisciplinary Sciences

Multi-fidelity optimization via surrogate modelling

Alexander I. J. Forrester et al.

PROCEEDINGS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES (2007)

Add to Collection

Article Computer Science, Interdisciplinary Applications

Unsteady flow sensing and estimation via the gappy proper orthogonal decomposition

K Willcox

COMPUTERS & FLUIDS (2006)

Add to Collection

Article Engineering, Aerospace

Aerodynamic data reconstruction and inverse design using proper orthogonal decomposition

T Bui-Thanh et al.

AIAA JOURNAL (2004)

Add to Collection

© Peeref 2019-2024. All rights reserved.