pith. machine review for the scientific record. sign in

arxiv: 2605.02871 · v1 · submitted 2026-05-04 · ⚛️ physics.comp-ph · cs.LG

Recognition: unknown

Multi-fidelity surrogates for mechanics of composites: from co-kriging to multi-fidelity neural networks

Haizhou Wen , Elham Kiyani , Gang Li , Srikanth Pilla , George Em Karniadakis , Zhen Li
Authors on Pith no claims yet

Pith reviewed 2026-05-08 01:53 UTC · model grok-4.3

classification ⚛️ physics.comp-ph cs.LG
keywords multi-fidelity surrogate modelingcomposite mechanicsco-Krigingmulti-fidelity neural networksGaussian processessurrogate modelsuncertainty quantificationinverse optimization
0
0 comments X

The pith

Multi-fidelity surrogate models recover reliable high-fidelity predictions for composite mechanics from abundant low-cost data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper reviews how multi-fidelity surrogate modeling combines plentiful inexpensive simulations or tests with scarce high-accuracy data to produce dependable predictions of composite behavior. Composites present particular difficulty because their hierarchical, anisotropic properties arise from coupled mechanisms at multiple scales and from manufacturing history, which drives up the cost of covering design spaces. The authors compare Gaussian-process methods such as co-Kriging and autoregressive formulations against multi-fidelity neural networks, showing distinctions in how each captures cross-fidelity correlations, discrepancies, uncertainty, and computational scaling. They illustrate uses in forward design exploration, inverse parameter fitting, and workflow integration while flagging open issues like regime-dependent fidelity gaps caused by nonlinear damage.

Core claim

Multi-fidelity surrogate modeling addresses this challenge by combining abundant, less expensive data with limited high-accuracy data to recover reliable high-fidelity predictions. The review organizes methods from co-Kriging and coregionalization models through nonlinear autoregressive Gaussian processes and multi-fidelity neural networks, distinguishing them by cross-fidelity correlation handling, discrepancy representation, uncertainty quantification, and scalability. Selected composite applications demonstrate roles in forward prediction for material design spaces, inverse optimization under limited high-fidelity access, and workflow integration that respects heterogeneous data and valid

What carries the argument

Multi-fidelity surrogate models that exploit correlations between data sources of differing accuracy to reduce reliance on expensive high-fidelity evaluations

If this is right

  • Forward prediction becomes feasible for rapid exploration of large material and manufacturing design spaces.
  • Inverse optimization tasks for parameter identification and design search proceed with far fewer high-fidelity evaluations.
  • Workflow integration can accommodate heterogeneous data sources, physical constraints, and validation requirements in a single model.
  • Uncertainty estimates propagate across fidelity levels, supporting risk-aware decisions in composite engineering.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same structure could be tested on other hierarchical materials whose fidelity gaps also vary with damage or processing history.
  • Neural-network variants may scale more readily than Gaussian-process ones when design spaces grow to hundreds of dimensions.
  • Explicit incorporation of manufacturing-history variables as additional inputs could reduce the regime-dependent gaps highlighted as open problems.
  • Hybrid models that embed known physics constraints inside the multi-fidelity framework remain an unexamined extension for reducing extrapolation error.

Load-bearing premise

The reviewed methods can be meaningfully distinguished and applied to composites despite regime-dependent fidelity gaps from nonlinear damage and manufacturing history.

What would settle it

A composite mechanics test case in which adding low-fidelity data produces higher error in high-fidelity predictions than using only the high-fidelity data alone.

Figures

Figures reproduced from arXiv: 2605.02871 by Elham Kiyani, Gang Li, George Em Karniadakis, Haizhou Wen, Srikanth Pilla, Zhen Li.

Figure 1
Figure 1. Figure 1: Example of Gaussian process prediction and associated 95% confidence interval. The la￾tent function f(x) = (x − 0.5) sin x, the observation samples are biased with 10% standard deviation (1% variance). Example uses a GP prior of zero-mean and squared exponential covariance function (Eqn. 2.1.2), and the hyperparameters are learnt by maximizing NLML using Cholesky factorization numerics [73]. 2.1.2 MOGP MOG… view at source ↗
Figure 2
Figure 2. Figure 2: Demonstration of varying dataset types in GP: scalar to scalar sampling in SOGP (left), scalar to vector sampling (a) in MOGP (middle), and scalar to vector sampling (b) in MOGP (right). Sampling (a) indicates the data samples at every § are fully informed, and sampling (b) indicates the data samples at every § are not fully informed. For each output task, a corresponding task function fq (where q ∈ [1, Q]… view at source ↗
Figure 3
Figure 3. Figure 3: Example of one-dimensional input case using Kriging and Co-kriging of Forrester et al. [33]. The fe = (6x − 2)2 sin(12x − 4) is the expensive observation data representing HF data, and the fc = Afe+B(x−0.5)−C with A = 0.5, B = 10, and C = -5 is the cheap observation data representing LF data. The kriging approximation using four expensive data points (ye) has been significantly improved using extensive sam… view at source ↗
Figure 4
Figure 4. Figure 4: Sketch of neural networks: (A) basic architecture of ANN, and (B) typical calculation inside one neuron. 3.1.1 Simple Feedforward Neural Network A feedforward neural network (FNN) [91] is the most basic form of ANN and is widely used to ap￾proximate a direct mapping between a finite-dimensional input vector and output vector. In an FNN, 23 view at source ↗
Figure 5
Figure 5. Figure 5: Predicting latent function f(x) = 0.5(6x − 2)2 sin(12x − 4) + 10(x − 0.5) − 5 with FNN in architecture of [1]+[20]*2+[1]. information flows only in the forward direction, meaning the output of each layer is determined by the activations of the previous layer, without feedback connections. For an L-layer FNN, the forward propagation can be written as h (0) = x, h (ℓ) = A (ℓ) W(ℓ)h (ℓ−1) + b (ℓ)  , ℓ = 1, 2… view at source ↗
Figure 6
Figure 6. Figure 6: Demonstration of PINN: (A) The architecture of PINN and (B) predicted solution T(x, y) to a simple Laplace equation ∂ 2T/∂x2 + ∂ 2T/∂y2 = 0 from PINN and its reference. In many physical and engineering processes, there are principled physical laws or empirically val￾idated rules that govern the dynamics of the system. These laws and rules provide prior knowledge, which can be utilized to inform the trainin… view at source ↗
Figure 7
Figure 7. Figure 7: Demonstration of RNN: (A) the general RNN architecture and (B) computation design of hidden state inside a RNN neuron in Vanilla RNNs and LSTM. with reduced reliance on expensive samples. In addition to GP- and Kriging-based models, neural networks are reliable surrogate models that can be incorporated into the MF framework. Within NN￾based surrogate modeling for composite mechanics, two representative str… view at source ↗
Figure 8
Figure 8. Figure 8: Demonstration of MFNN: (A) the general MFNN architecture and (B) an example to predict a HF latent function fH(x) = (6x − 2)2 sin(12x − 4) with few HF samples and many LF samples of a LF function fL(x) = 0.5(6x − 2)2 sin(12x − 4) + 10(x − 0.5) − 5 with MFNN in architecture of LF network [1]+[20]*2+[1], HF network linear component [2]+[1], and HF network nonlinear component [2]+[10]*2+[1]. 3.2.2 Transfer Le… view at source ↗
Figure 9
Figure 9. Figure 9: Demonstration of (A) a four-scale woven fiber composite with polymer matrix and (B) MF surrogate (metamodel) replaces micro- and meso-simulations in the conventional UQ loop for a woven fiber composite. The images are from the reference [120] with authorization. essentially the same relative L2 errors, such as 0.2208% versus 0.2228% for reaction force and 2.924% versus 3.1477% for matrix damage. Overall, t… view at source ↗
Figure 10
Figure 10. Figure 10: Demonstration of PID assessment in a L-shaped CFRP part with SWGPR surrogate: (A) the trade-off between time/cost and fidelity/accuracy in analysis methods, (B) Flowchart of efficient composites manufacturing analysis method using multi-fidelity simulation, and (C) Comparative results of experimental PIDs and predictions using calibrated SWGPR models. The images are from the reference [123] with authoriza… view at source ↗
Figure 11
Figure 11. Figure 11: Demonstration of fiber arrangement optimization in a VS composite with MF surrogate: (A) LF and HF FE models of the VS cylinder and plate, (B) Flowchart of the optimization procedure, and (C) buckling mode shapes of the both cylinder and plate separately using the QI laminate and optimal VS laminate with the below shown optimal fiber pattern (buckling load in QI plate and optimal VS plate are 14.93 kN and… view at source ↗
Figure 12
Figure 12. Figure 12: Demonstration of composition–modulus triangular diagram: (A) the results from LF source, HF source, and MF-GP surrogate, respectively; (B) the results showing the prediction progression of the MF BO at iterations 4, 24, 35, and 130, respectively (the red dot indicates the best HF data point evaluated so far, while the green squares indicate the other HF data points and the green black dots indicate LF dat… view at source ↗
Figure 13
Figure 13. Figure 13: Demonstration of two U-net surrogate models and their prediction result: (A) block dia￾grams of the two U-net neural networks in the uniform model and informed model; (B) table of maximum test error for prediction methods for varying finite width factor; (C) comparison of simulation time for different stress analysis methods; (D) example stress prediction errors for multiple stress components across the n… view at source ↗
Figure 14
Figure 14. Figure 14: Demonstration of MFNN surrogate applied in learning rheological relation of fiber sus￾pension: (A) sketch of the rough fiber in suspension under the microscale modeling; (B) comparison of the curve fitting accuracy for different constitutive equations for one case with rp = 10, ϵr = 0.01, Bˆ = 0.02, and ϕ = 0.4; (C) Schematic view of the single fidelity DNN and MFNN; (D) two cases showing predictions of D… view at source ↗
Figure 15
Figure 15. Figure 15: Demonstration of MF Triple LSTM surrogate for predicting load–displacement curve of open-hole CFRP laminates: (A) Schematic diagram of open-hole tensile test specimen; (B) the test system and specimen in the tensile experiment; (C) MF Triple LSTM model architecture; (D) comparison of damage morphologies between HF experiment and LF FEA (shown cases use hole diameter of 25mm); (E) the prediction results of… view at source ↗
Figure 16
Figure 16. Figure 16: Demonstration of MORNN surrogate applied in efficient design of process parameters in SPR: (A) the complete flow schematic of the study framework; (B) definition of 9 geometric parameters in a single SPR cross-section; (C) The network architecture of the MORNN; (D) The accuracy comparison of four surrogates (simu-GBM: gradient boosting machine with LF simulation data; simu-NN: single fidelity neural netwo… view at source ↗
Figure 17
Figure 17. Figure 17: Demonstration of probabilistic MFPINN surrogate applied in AE impact localization: (A) schematic representation of proposed mfPINN framework for estimating AE source location and corresponding uncertainty; (B) the top view and 3D schematic view of the experimental setup; (C) True and estimated impact location results using HFANN, LFPINN, and HFPINN. The images are from the reference [132] with authorizati… view at source ↗
Figure 18
Figure 18. Figure 18: Demonstration of GRU-based RNN surrogate with transfer learning applied to predict full stress field of SFRC: (A) Network architecture and the transfer learning approach utilizing a large mean￾field and a small full-field data set; (B) Average mean relative error (MeRE) and maximum relative error (MaRE) results for the test data set (”UD”, ”Planar”, and ”3D” are referred to three fiber orientation cases);… view at source ↗
read the original abstract

Composite materials exhibit strongly hierarchical and anisotropic properties governed by coupled mechanisms spanning constituents, plies, laminates, structures, and manufacturing history. This intrinsic complexity makes predictive modeling of composites expensive, because repeated experiments and high-fidelity simulations are needed to cover large design spaces of material, structure, and manufacturing. Multi-fidelity surrogate modeling addresses this challenge by combining abundant, less expensive data with limited high-accuracy data to recover reliable high-fidelity predictions. This review presents a structured overview of multi-fidelity modeling for composite mechanics, covering Gaussian-process or Kriging-based methods, including co-Kriging, coregionalization models, autoregressive formulations, nonlinear autoregressive Gaussian processes, multi-fidelity deep Gaussian processes, and multi-fidelity neural networks. Their distinctions are examined in terms of cross-fidelity correlation, discrepancy representation, uncertainty quantification, and scalability. Selected examples of their applications to composites are introduced according to the roles that multi-fidelity surrogates play in engineering problems, including forward prediction for rapid exploration of material design spaces, inverse optimization for composite parameter identification and design search under limited high-fidelity access, and workflow integration, where heterogeneous data sources, constraints, and validation requirements determine model utility. Open question discussions highlight recurring challenges specific to composites, such as regime-dependent fidelity gaps associated with nonlinear damage and manufacturing history, mismatches between simulations and experiments, and uncertainty propagation across multi-fidelity models.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. This review paper provides a structured overview of multi-fidelity surrogate modeling for composite mechanics. It covers Gaussian-process-based methods (co-Kriging, coregionalization models, autoregressive formulations, nonlinear autoregressive GPs, multi-fidelity deep GPs) and multi-fidelity neural networks, examining their distinctions in cross-fidelity correlation, discrepancy representation, uncertainty quantification, and scalability. Selected examples illustrate applications in forward prediction for material design-space exploration, inverse optimization for parameter identification under limited high-fidelity data, and workflow integration. The paper concludes with open-question discussions on composites-specific challenges including regime-dependent fidelity gaps from nonlinear damage and manufacturing history, simulation-experiment mismatches, and uncertainty propagation.

Significance. If the distinctions among methods are accurately characterized and the applications are representative, the review would be significant for the computational mechanics and composites communities. It synthesizes a progression from classical co-Kriging to modern MFNNs, offers practical guidance on roles in engineering workflows, and flags open challenges that could steer future method development. The emphasis on reducing reliance on expensive high-fidelity simulations in hierarchical, anisotropic materials is timely given the computational demands of composite design.

major comments (2)
  1. [Open question discussions] Open question discussions: The review correctly flags regime-dependent fidelity gaps associated with nonlinear damage and manufacturing history as a recurring challenge. However, it provides no concrete analysis, cited counterexamples, or performance comparisons showing how the reviewed methods (e.g., autoregressive GPs versus MFNNs) handle cases where low-fidelity models omit damage initiation or history effects, leaving the applicability claim for composites resting on an untested extrapolation from linear regimes.
  2. [Selected examples of their applications to composites] Applications section: While examples are organized by engineering role (forward prediction, inverse optimization, workflow integration), the text does not include quantitative metrics (error reduction, computational savings, or UQ calibration) comparing different multi-fidelity approaches on the same composite problem. This weakens the ability to evaluate the claimed distinctions in scalability and uncertainty quantification.
minor comments (2)
  1. Add a summary table listing each method, its key modeling assumptions (correlation/discrepancy form), UQ capabilities, and one representative composite application to improve readability and allow quick comparison.
  2. Ensure consistent terminology and citation for 'multi-fidelity deep Gaussian processes' versus 'multi-fidelity neural networks' throughout the manuscript and abstract.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and positive assessment of the review's potential significance. We address each major comment below and will revise the manuscript to incorporate additional citations and a summary table where feasible within the scope of a review.

read point-by-point responses

  1. Referee: [Open question discussions] Open question discussions: The review correctly flags regime-dependent fidelity gaps associated with nonlinear damage and manufacturing history as a recurring challenge. However, it provides no concrete analysis, cited counterexamples, or performance comparisons showing how the reviewed methods (e.g., autoregressive GPs versus MFNNs) handle cases where low-fidelity models omit damage initiation or history effects, leaving the applicability claim for composites resting on an untested extrapolation from linear regimes.

    Authors: We agree that the open questions section would benefit from greater specificity. As this is a review, we do not introduce new analyses or direct performance comparisons. In the revision we will add citations to existing composites studies that document limitations of multi-fidelity approaches (including autoregressive GPs and neural networks) when low-fidelity models neglect damage initiation or history-dependent effects. This will supply concrete counterexamples and clarify that the section identifies open challenges rather than asserting untested applicability across regimes. revision: partial

  2. Referee: [Selected examples of their applications to composites] Applications section: While examples are organized by engineering role (forward prediction, inverse optimization, workflow integration), the text does not include quantitative metrics (error reduction, computational savings, or UQ calibration) comparing different multi-fidelity approaches on the same composite problem. This weakens the ability to evaluate the claimed distinctions in scalability and uncertainty quantification.

    Authors: We acknowledge that head-to-head quantitative comparisons on identical problems would strengthen evaluation of the distinctions. Such direct benchmarks are rarely available in the published literature, which motivated our organization by engineering role. We will add a summary table compiling reported error reductions, computational savings, and UQ metrics from the cited composite applications. This will enable readers to assess the distinctions more quantitatively while remaining within the bounds of a review; new side-by-side experiments on a common benchmark lie outside the paper's scope. revision: partial

Circularity Check

0 steps flagged

Review paper catalogs existing multi-fidelity methods with no self-derived predictions or load-bearing self-citations

full rationale

This is a review article that structures an overview of Gaussian-process and neural-network based multi-fidelity surrogate techniques drawn from the broader literature. It examines distinctions in correlation and discrepancy modeling, cites selected composite applications, and flags open challenges such as regime-dependent fidelity gaps without performing any new derivations, parameter fits, or predictions that reduce to the paper's own inputs. No equations or claims in the abstract or described content exhibit self-definition, fitted-input-as-prediction, or uniqueness imported via self-citation chains. The discussion of limitations around nonlinear damage and manufacturing history is presented as unresolved rather than resolved by the review itself, confirming the derivation chain is absent and the content is self-contained as a survey.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a review paper summarizing existing techniques; no new free parameters, axioms, or invented entities are introduced.

pith-pipeline@v0.9.0 · 5574 in / 1015 out tokens · 42019 ms · 2026-05-08T01:53:18.942719+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

166 extracted references · 4 canonical work pages · 1 internal anchor

  1. [1]

    Cambridge uni- versity press, 2019

    Trevor William Clyne and Derek Hull.An introduction to composite materials. Cambridge uni- versity press, 2019

    2019

  2. [2]

    1 - introduction to composite materials

    Haim Abramovich. 1 - introduction to composite materials. In Haim Abramovich, editor,Stability and Vibrations of Thin Walled Composite Structures, pages 1–47. Woodhead Publishing, 2017. 52

    2017

  3. [3]

    Michel, H

    J.C. Michel, H. Moulinec, and P. Suquet. Effective properties of composite materials with pe- riodic microstructure: a computational approach.Computer Methods in Applied Mechanics and Engineering, 172(1):109–143, 1999

    1999

  4. [4]

    The physical properties of composite materials.Journal of Materials Science, 11(11):2105–2141, 1976

    DK Hale. The physical properties of composite materials.Journal of Materials Science, 11(11):2105–2141, 1976

    1976

  5. [5]

    Tsai and Edward M

    Stephen W. Tsai and Edward M. Wu. A general theory of strength for anisotropic materials. Journal of Composite Materials, 5(1):58–80, 1971

    1971

  6. [6]

    C. E. Bakis, Lawrence C. Bank, V. L. Brown, E. Cosenza, J. F. Davalos, J. J. Lesko, A. Machida, S. H. Rizkalla, and T. C. Triantafillou. Fiber-reinforced polymer composites for construction—state-of-the-art review.Journal of Composites for Construction, 6(2):73–87, 2002

    2002

  7. [7]

    Homogenization techniques and micromechanics

    Nicolas Charalambakis. Homogenization techniques and micromechanics. a survey and perspec- tives.Applied Mechanics Reviews, 63(3):030803, 07 2010

    2010

  8. [8]

    Hahn and N.J

    H.T. Hahn and N.J. Pagano. Curing stresses in composite laminates.Journal of Composite Materials, 9(1):91–106, 1975

    1975

  9. [9]

    Erasmo Carrera. Theories and finite elements for multilayered plates and shells:a unified compact formulation with numerical assessment and benchmarking.Archives of Computational Methods in Engineering, 10:215–296, 2003

    2003

  10. [10]

    O. O. Ochoa and J. N. Reddy.Finite Element Analysis of Composite Laminates, pages 37–109. Springer Netherlands, Dordrecht, 1992

    1992

  11. [11]

    Oxford university press New York, 1994

    Isaac M Daniel, Ori Ishai, Issac M Daniel, and Ishai Daniel.Engineering mechanics of composite materials, volume 3. Oxford university press New York, 1994

    1994

  12. [12]

    N.J. Pagano. Stress fields in composite laminates.International Journal of Solids and Structures, 14(5):385–400, 1978

    1978

  13. [13]

    R. Talreja. A continuum mechanics characterization of damage in composite materials.Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 399(1817):195–216, 06 1985

    1985

  14. [14]

    W. A. Curtin and N. Takeda. Tensile strength of fiber-reinforced composites: Ii. application to polymer matrix composites.Journal of Composite Materials, 32(22):2060–2081, 1998

    2060

  15. [15]

    Hutchinson and Z

    J.W. Hutchinson and Z. Suo. Mixed mode cracking in layered materials. volume 29 ofAdvances in Applied Mechanics, pages 63–191. Elsevier, 1991

    1991

  16. [16]

    Manufacturing-induced imperfections in composite parts manufactured via automated fiber placement.Journal of Composites Science, 3(2), 2019

    Falk Heinecke and Christian Willberg. Manufacturing-induced imperfections in composite parts manufactured via automated fiber placement.Journal of Composites Science, 3(2), 2019

    2019

  17. [17]

    S. Dey, T. Mukhopadhyay, and S. Adhikari. Metamodel based high-fidelity stochastic analysis of composite laminates: A concise review with critical comparative assessment.Composite Structures, 171:227–250, 2017. 53

    2017

  18. [18]

    Aanchna Sharma, Tanmoy Mukhopadhyay, Sanjay Mavinkere Rangappa, Suchart Siengchin, and Vinod Kushvaha. Advances in computational intelligence of polymer composite materials: ma- chine learning assisted modeling, analysis and design.Archives of Computational Methods in Engineering, 29(5):3341–3385, 2022

    2022

  19. [19]

    Efficient optimization design method using kriging model.Journal of Aircraft, 42(2):413–420, 2005

    Shinkyu Jeong, Mitsuhiro Murayama, and Kazuomi Yamamoto. Efficient optimization design method using kriging model.Journal of Aircraft, 42(2):413–420, 2005

    2005

  20. [20]

    Efficient reliability analysis of laminated composites using advanced kriging surrogate model.Composite Structures, 149:26–32, 2016

    Ali Haeri and Mohammad Javad Fadaee. Efficient reliability analysis of laminated composites using advanced kriging surrogate model.Composite Structures, 149:26–32, 2016

    2016

  21. [21]

    Tanmoy Mukhopadhyay, S Chakraborty, S Dey, S Adhikari, and Rajib Chowdhury. A critical assessment of kriging model variants for high-fidelity uncertainty quantification in dynamics of composite shells.Archives of Computational Methods in Engineering, 24(3):495–518, 2017

    2017

  22. [22]

    3d-printing and machine learning control of soft ionic polymer-metal composite actuators.Scientific Reports, 9(1):17482, 2019

    James D Carrico, Tucker Hermans, Kwang J Kim, and Kam K Leang. 3d-printing and machine learning control of soft ionic polymer-metal composite actuators.Scientific Reports, 9(1):17482, 2019

    2019

  23. [23]

    Almeida Jr., Luc St-Pierre, Zhonglai Wang, and Saullo G.P

    Zhihua Wang, Jos´ e Humberto S. Almeida Jr., Luc St-Pierre, Zhonglai Wang, and Saullo G.P. Cas- tro. Reliability-based buckling optimization with an accelerated kriging metamodel for filament- wound variable angle tow composite cylinders.Composite Structures, 254:112821, 2020

    2020

  24. [24]

    Jose Vicente Aguado, Domenico Borzacchiello, Chady Ghnatios, Fran¸ cois Lebel, Ram Upadhyay, Christophe Binetruy, and Francisco Chinesta. A simulation app based on reduced order mod- eling for manufacturing optimization of composite outlet guide vanes.Advanced Modeling and Simulation in Engineering Sciences, 4(1):1, 2017

    2017

  25. [25]

    A reduced-order mod- elling for real-time identification of damages in multi-layered composite materials.Inverse Prob- lems in Science and Engineering, 29(1):73–94, 2021

    Yu Liang, Xiao-Wei Gao, Bing-Bing Xu, Miao Cui, and Bao-Jing Zheng. A reduced-order mod- elling for real-time identification of damages in multi-layered composite materials.Inverse Prob- lems in Science and Engineering, 29(1):73–94, 2021

    2021

  26. [26]

    Artificial convolution neural network techniques and applications for lung nodule detection.IEEE Transactions on Medical Imaging, 14(4):711–718, 1995

    S-CB Lo, S-LA Lou, Jyh-Shyan Lin, Matthew T Freedman, Minze V Chien, and Seong Ki Mun. Artificial convolution neural network techniques and applications for lung nodule detection.IEEE Transactions on Medical Imaging, 14(4):711–718, 1995

    1995

  27. [27]

    Vassilopoulos, Efstratios F

    Anastasios P. Vassilopoulos, Efstratios F. Georgopoulos, and Vasileios Dionysopoulos. Artificial neural networks in spectrum fatigue life prediction of composite materials.International Journal of Fatigue, 29(1):20–29, 2007

    2007

  28. [28]

    Karnik, V.N

    S.R. Karnik, V.N. Gaitonde, J. Campos Rubio, A. Esteves Correia, A.M. Abr˜ ao, and J. Paulo Davim. Delamination analysis in high speed drilling of carbon fiber reinforced plastics (cfrp) using artificial neural network model.Materials & Design, 29(9):1768–1776, 2008

    2008

  29. [29]

    McCrory, Safaa Kh

    John P. McCrory, Safaa Kh. Al-Jumaili, Davide Crivelli, Matthew R. Pearson, Mark J. Eaton, Carol A. Featherston, Mario Guagliano, Karen M. Holford, and Rhys Pullin. Damage classification in carbon fibre composites using acoustic emission: A comparison of three techniques.Composites Part B: Engineering, 68:424–430, 2015

    2015

  30. [30]

    Hamdia, Tom Lahmer, Trung Nguyen-Thoi, and Timon Rabczuk

    Khader M. Hamdia, Tom Lahmer, Trung Nguyen-Thoi, and Timon Rabczuk. Predicting the fracture toughness of pncs: A stochastic approach based on ann and anfis.Computational Materials Science, 102:304–313, 2015. 54

    2015

  31. [31]

    Probabilistic predictions of process- induced deformation in carbon/epoxy composites using a deep operator network, 2025

    Elham Kiyani, Amit Makarand Deshpande, Madhura Limaye, Zhiwei Gao, Sai Aditya Pradeep, Srikanth Pilla, Gang Li, Zhen Li, and George Em Karniadakis. Probabilistic predictions of process- induced deformation in carbon/epoxy composites using a deep operator network, 2025

    2025

  32. [32]

    Predicting the output from a complex computer code when fast approximations are available.Biometrika, 87(1):1–13, 03 2000

    MC Kennedy and A O’Hagan. Predicting the output from a complex computer code when fast approximations are available.Biometrika, 87(1):1–13, 03 2000

    2000

  33. [33]

    Multi-fidelity optimization via surrogate modelling.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 463(2088):3251–3269, 2007

    Alexander I.J Forrester, Andr´ as S´ obester, and Andy J Keane. Multi-fidelity optimization via surrogate modelling.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 463(2088):3251–3269, 2007

    2088

  34. [34]

    Multi-fidelity design of stiffened com- posite panel with a crack.Structural and Multidisciplinary Optimization, 23(5):347–356, 2002

    Roberto Vitali, Raphael T Haftka, and Bhavani V Sankar. Multi-fidelity design of stiffened com- posite panel with a crack.Structural and Multidisciplinary Optimization, 23(5):347–356, 2002

    2002

  35. [35]

    Survey of multifidelity methods in uncertainty propagation, inference, and optimization.Siam Review, 60(3):550–591, 2018

    Benjamin Peherstorfer, Karen Willcox, and Max Gunzburger. Survey of multifidelity methods in uncertainty propagation, inference, and optimization.Siam Review, 60(3):550–591, 2018

    2018

  36. [36]

    arXiv preprint arXiv:1609.07196 , year=

    M Giselle Fern´ andez-Godino. Review of multi-fidelity models.arXiv preprint arXiv:1609.07196, 2016

    work page arXiv 2016

  37. [37]

    Paris Perdikaris, Daniele Venturi, Johannes O Royset, and George Em Karniadakis. Multi-fidelity modelling via recursive co-kriging and gaussian–markov random fields.Proceedings of the royal society a: mathematical, physical and engineering sciences, 471(2179), 2015

    2015

  38. [38]

    Paris Perdikaris, Maziar Raissi, Andreas Damianou, Neil D Lawrence, and George Em Karniadakis. Nonlinear information fusion algorithms for data-efficient multi-fidelity modelling.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 473(2198), 2017

    2017

  39. [39]

    Oberkampf and Frederick G

    William L. Oberkampf and Frederick G. Blottner. Issues in computational fluid dynamics code verification and validation.AIAA Journal, 36(5):687–695, 1998

    1998

  40. [40]

    Oberkampf and Timothy G

    William L. Oberkampf and Timothy G. Trucano. Verification and validation in computational fluid dynamics.Progress in Aerospace Sciences, 38(3):209–272, 2002

    2002

  41. [41]

    Wang, Krzysztof Fidkowski, R´ emi Abgrall, Francesco Bassi, Doru Caraeni, Andrew Cary, Herman Deconinck, Ralf Hartmann, Koen Hillewaert, H.T

    Z.J. Wang, Krzysztof Fidkowski, R´ emi Abgrall, Francesco Bassi, Doru Caraeni, Andrew Cary, Herman Deconinck, Ralf Hartmann, Koen Hillewaert, H.T. Huynh, Norbert Kroll, Georg May, Per-Olof Persson, Bram van Leer, and Miguel Visbal. High-order cfd methods: current status and perspective.International Journal for Numerical Methods in Fluids, 72(8):811–845, 2013

    2013

  42. [42]

    Saunders and Gregory A

    Marissa G. Saunders and Gregory A. Voth. Coarse-graining methods for computational biology. Annual Review of Biophysics, 42(Volume 42, 2013):73–93, 2013

    2013

  43. [43]

    Springer Netherlands, Dordrecht, 2016

    Xuejin Li, Zhen Li, Xin Bian, Mingge Deng, Changho Kim, Yu-Hang Tang, Alireza Yazdani, and George Em Karniadakis.Dissipative Particle Dynamics, Overview, pages 1–8. Springer Netherlands, Dordrecht, 2016

    2016

  44. [44]

    N. Saba, M. Jawaid, and M.T.H. Sultan. 1 - an overview of mechanical and physical testing of composite materials. In Mohammad Jawaid, Mohamed Thariq, and Naheed Saba, editors,Me- chanical and Physical Testing of Biocomposites, Fibre-Reinforced Composites and Hybrid Compos- ites, Woodhead Publishing Series in Composites Science and Engineering, pages 1–12....

    2019

  45. [45]

    Giselle Fern´ andez-Godino

    M. Giselle Fern´ andez-Godino. Review of multi-fidelity models.Advances in Computational Science and Engineering, 1(4):351–400, 2023

    2023

  46. [46]

    Survey of multifidelity methods in uncertainty propagation, inference, and optimization.SIAM Review, 60(3):550–591, 2018

    Benjamin Peherstorfer, Karen Willcox, and Max Gunzburger. Survey of multifidelity methods in uncertainty propagation, inference, and optimization.SIAM Review, 60(3):550–591, 2018

    2018

  47. [47]

    Overview of gaussian process based multi- fidelity techniques with variable relationship between fidelities, application to aerospace systems

    Lo¨ ıc Brevault, Mathieu Balesdent, and Ali Hebbal. Overview of gaussian process based multi- fidelity techniques with variable relationship between fidelities, application to aerospace systems. Aerospace Science and Technology, 107:106339, 2020

    2020

  48. [48]

    T. J. Dodwell, C. Ketelsen, R. Scheichl, and A. L. Teckentrup. A hierarchical multilevel markov chain monte carlo algorithm with applications to uncertainty quantification in subsurface flow. SIAM/ASA Journal on Uncertainty Quantification, 3(1):1075–1108, 2015

    2015

  49. [49]

    M. B. Scheichl R. Teckentrup A. L. Cliffe, K. A. Giles. Multilevel monte carlo methods and applications to elliptic pdes with random coefficients.Computing and Visualization in Science, 14(1):3, 2011

    2011

  50. [50]

    Hoeting, David Madigan, Adrian E

    Jennifer A. Hoeting, David Madigan, Adrian E. Raftery, and Chris T. Volinsky. Bayesian model averaging: a tutorial (with comments by M. Clyde, David Draper and E. I. George, and a rejoinder by the authors.Statistical Science, 14(4):382 – 417, 1999

    1999

  51. [51]

    H˚ akon Hoel, Kody J. H. Law, and Raul Tempone. Multilevel ensemble kalman filtering.SIAM Journal on Numerical Analysis, 54(3):1813–1839, 2016

    2016

  52. [52]

    Haftka, and Nam H

    Chanyoung Park, Raphael T. Haftka, and Nam H. Kim. Remarks on multi-fidelity surrogates. Structural and Multidisciplinary Optimization, 55(3):1029–1050, 2017

    2017

  53. [53]

    Peter Z. G. Qian and C. F. Jeff Wu. Bayesian hierarchical modeling for integrating low-accuracy and high-accuracy experiments.Technometrics, 50(2):192–204, 2008

    2008

  54. [54]

    Fischer and Ramana V

    Christopher C. Fischer and Ramana V. Grandhi.A Surrogate-based Adjustment Factor Approach to Multi-Fidelity Design Optimization

  55. [55]

    Pais, Raphael T

    Alexandra Coppe, Matthew J. Pais, Raphael T. Haftka, and Nam H. Kim. Using a simple crack growth model in predicting remaining useful life.Journal of Aircraft, 49(6):1965–1973, 2012

    1965

  56. [56]

    Supplementary details on bayesian calibration of com- puter models

    Marc C Kennedy and Anthony O’Hagan. Supplementary details on bayesian calibration of com- puter models. Technical report, Internal Report. URL http://www. shef. ac. uk/˜ st1ao/ps/calsup. ps, 2001

    2001

  57. [57]

    J. D. Koch P. N. Allen J. K. Simpson, T. W. Poplinski. Metamodels for computer-based engineering design: Survey and recommendations.Engineering with Computers, 17(2):129–150, 2001

    2001

  58. [58]

    Perdikaris, D

    P. Perdikaris, D. Venturi, J. O. Royset, and G. E. Karniadakis. Multi-fidelity modelling via recursive co-kriging and gaussian–markov random fields.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 471(2179):20150018, 2015

    2015

  59. [59]

    Physics- informed neural networks (pinns) for fluid mechanics: a review.Acta Mechanica Sinica, 37(12):1727–1738, 2021

    Zhiping Wang Zhicheng Yin Minglang Karniadakis George Em Cai, Shengze Mao. Physics- informed neural networks (pinns) for fluid mechanics: a review.Acta Mechanica Sinica, 37(12):1727–1738, 2021. 56

    2021

  60. [60]

    Xuhui Meng and George Em Karniadakis. A composite neural network that learns from multi- fidelity data: Application to function approximation and inverse pde problems.Journal of Com- putational Physics, 401:109020, 2020

    2020

  61. [61]

    A stochastic collocation algorithm with mul- tifidelity models.SIAM Journal on Scientific Computing, 36(2):A495–A521, 2014

    Akil Narayan, Claude Gittelson, and Dongbin Xiu. A stochastic collocation algorithm with mul- tifidelity models.SIAM Journal on Scientific Computing, 36(2):A495–A521, 2014

    2014

  62. [62]

    Parametric/stochastic model reduction: Low-rank representation, non-intrusive bi-fidelity approximation, and conver- gence analysis, 2017

    Jerrad Hampton, Hillary Fairbanks, Akil Narayan, and Alireza Doostan. Parametric/stochastic model reduction: Low-rank representation, non-intrusive bi-fidelity approximation, and conver- gence analysis, 2017

    2017

  63. [63]

    Copula processes

    Andrew G Wilson and Zoubin Ghahramani. Copula processes. In J. Lafferty, C. Williams, J. Shawe-Taylor, R. Zemel, and A. Culotta, editors,Advances in Neural Information Process- ing Systems, volume 23. Curran Associates, Inc., 2010

    2010

  64. [64]

    Gaussian processes for regression

    Christopher Williams and Carl Rasmussen. Gaussian processes for regression. In D. Touretzky, M.C. Mozer, and M. Hasselmo, editors,Advances in Neural Information Processing Systems, volume 8. MIT Press, 1995

    1995

  65. [65]

    A tutorial on gaussian process regression: Modelling, exploring, and exploiting functions.Journal of Mathematical Psychology, 85:1–16, 2018

    Eric Schulz, Maarten Speekenbrink, and Andreas Krause. A tutorial on gaussian process regression: Modelling, exploring, and exploiting functions.Journal of Mathematical Psychology, 85:1–16, 2018

    2018

  66. [66]

    Remarks on multi-output gaussian process regression

    Haitao Liu, Jianfei Cai, and Yew-Soon Ong. Remarks on multi-output gaussian process regression. Knowledge-Based Systems, 144:102–121, 2018

    2018

  67. [67]

    Springer Berlin Heidelberg, Berlin, Heidelberg, 2004

    Carl Edward Rasmussen.Gaussian Processes in Machine Learning, pages 63–71. Springer Berlin Heidelberg, Berlin, Heidelberg, 2004

    2004

  68. [68]

    Systematic errors in two-dimensional digital image correlation due to lens distortion.Optics and Lasers in Engineering, 51(2):140–147, 2013

    Bing Pan, Liping Yu, Dafang Wu, and Liqun Tang. Systematic errors in two-dimensional digital image correlation due to lens distortion.Optics and Lasers in Engineering, 51(2):140–147, 2013

    2013

  69. [69]

    Navid Nasajpour-Esfahani, Saba Karimi, Simin Nasseri, Homa Borna, Ali Fadavi Boostani, Ruoqi Gao, Wei Huang, Hamid Garmestani, and Steven Y. Liang. Advancements and applications of digital image correlation to characterize residual stress: A review.Materials Characterization, 228:115416, 2025

    2025

  70. [70]

    A unified theory for constitutive modeling of composites.Journal of Mechanics of Materials and Structures, 11(4):379–411, 2016

    Wenbin Yu. A unified theory for constitutive modeling of composites.Journal of Mechanics of Materials and Structures, 11(4):379–411, 2016

    2016

  71. [71]

    Coexist- ing sub-zero temperature and relative humidity influences on sensors and composite material

    Katarzyna Majewska, Magdalena Mieloszyk, Michal Jurek, and Wieslaw Ostachowicz. Coexist- ing sub-zero temperature and relative humidity influences on sensors and composite material. Composite Structures, 260:113263, 2021

    2021

  72. [72]

    David J. C. MacKay. Bayesian interpolation.Neural Computation, 4(3):415–447, 05 1992

    1992

  73. [73]

    Nicholas J. Higham. Cholesky factorization.WIREs Computational Statistics, 1(2):251–254, 2009

    2009

  74. [74]

    Helterbrand and Noel Cressie

    Jeffrey D. Helterbrand and Noel Cressie. Universal cokriging under intrinsic coregionalization. Mathematical Geology, 26(2):205–226, 1994

    1994

  75. [75]

    The origins of kriging.Mathematical Geology, 22(3):239–252, 1990

    Noel Cressie. The origins of kriging.Mathematical Geology, 22(3):239–252, 1990. 57

    1990

  76. [76]

    D.G. Krige. A statistical approach to some basic mine valuation problems on the witwatersrand. Journal of the Southern African Institute of Mining and Metallurgy, 52(6):119–139, 1951

    1951

  77. [77]

    Principles of geostatistics.Economic Geology, 58(8):1246–1266, 1963

    Matheron Georges. Principles of geostatistics.Economic Geology, 58(8):1246–1266, 1963

    1963

  78. [78]

    M. A. OLIVER and R. WEBSTER. Kriging: a method of interpolation for geographical informa- tion systems.International journal of geographical information systems, 4(3):313–332, 1990

    1990

  79. [79]

    Tutorial on maximum likelihood estimation.Journal of Mathematical Psychology, 47(1):90–100, 2003

    In Jae Myung. Tutorial on maximum likelihood estimation.Journal of Mathematical Psychology, 47(1):90–100, 2003

    2003

  80. [80]

    R. R. Corbeil and S. R. Searle. Restricted maximum likelihood (reml) estimation of variance components in the mixed model.Technometrics, 18(1):31–38, 1976

    1976

Showing first 80 references.