Analytical Extraction of Conditional Aleatory Sensitivities Across Epistemic Space via a Single PCE Model

Huiyou Tan; Jiangfeng Fu; Shijie Zhong

REVIEW 2 major objections 6 minor 27 references

Analytical Extraction of Conditional Aleatory Sensitivities Across Epistemic Space via a Single PCE Model

T0 review · 2 major / 6 minor · reviewed 2026-07-11 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A single global polynomial chaos expansion yields continuous conditional Sobol' indices across the entire epistemic domain by pure algebra, with Bayesian credible intervals and no double-loop sampling.

desk verdict Clean algebraic trick: one global PCE on the joint latent space yields continuous conditional Sobol’ fields by tensor-product regrouping, with RJMCMC CIs; solid on the analytic check, useful for hybrid UQ practitioners. read the letter →

arxiv 2607.04790 v1 pith:SPYIL66F submitted 2026-07-06 stat.ME

Shijie Zhong , Huiyou Tan , Jiangfeng Fu This is my paper · ORCID

classification stat.ME MSC 62P3065C2060H35

keywords hybriduncertaintyconditionalSobol'indicespolynomialchaosexpansionRJMCMCaleatory-epistemicseparationglobalsensitivityanalysisBayesiansparsePCE

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Hybrid uncertainty problems mix irreducible aleatory randomness with epistemic parameters that may later be refined. Analysts often want to know how the importance of the aleatory inputs changes as those epistemic parameters vary, producing continuous fields of conditional Sobol' indices. The usual route is a double loop: for every fixed epistemic point one rebuilds a surrogate or runs a Monte Carlo campaign, which quickly becomes prohibitive. This paper shows that one global polynomial chaos expansion, built once on the joint latent space of aleatory and epistemic variables, already contains every conditional sensitivity. Because the basis factors into a product of epistemic and aleatory polynomials, the expansion can be rewritten so that the epistemic dependence sits entirely inside coefficient fields; the conditional variances and Sobol' ratios then follow from elementary arithmetic on those fields. An RJMCMC sampler supplies both adaptive sparsity and posterior credible intervals for the resulting maps. On an analytical benchmark and a high-dimensional borehole flow model the algebraic fields match reference surfaces to high accuracy while replacing tens of millions of model evaluations with a few thousand.

What carries the argument

Tensor-product factorization of the orthogonal basis: Ψ_α(θ,ξ) = Ψ_αK(θ) Ψ_αL(ξ), which converts the global expansion into a conditional PCE whose coefficients are continuous functions of the epistemic variables alone (Lemmas 3.1–3.2).

What would settle it

On the analytical polynomial benchmark of Section 4.1, recompute the extracted total-effect fields ST,X1|θ and ST,X2|θ on a dense epistemic grid and check whether the absolute error ever exceeds 10^{-5} relative to the closed-form expressions; any systematic deviation would falsify the claim that the algebraic decomposition is exact for a correctly truncated PCE.

Watch

Extended reading notes

Core claim

From a single global PCE constructed on the joint (aleatory, epistemic) latent space, the conditional Sobol' indices Su|θ are recovered exactly by algebraic post-processing of the epistemic-dependent coefficient fields c_αL(θ) = Σ_αK y_αK,αL Ψ_αK(θ), without any further model evaluations or surrogate reconstructions.

Load-bearing premise

A single sparse global polynomial fit trained on joint samples remains accurate enough everywhere in the epistemic domain that the algebraic conditional variances and ratios stay faithful to the true model.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The one thing worth knowing is that they show how a single global PCE on the PIT-augmented (aleatory, epistemic) latent space can be regrouped, via the usual tensor-product multi-index split, into epistemic-dependent coefficient fields c_αL(θ). Conditional means, variances, and Sobol’ indices then drop out by pure algebra—no per-θ retraining, no nested loops. Lemmas 3.1–3.2 are standard orthonormality + uniqueness of the Sobol’ decomposition; they are correct under the stated assumptions. Algorithm 1 is just the post-processing step over the RJMCMC posterior ensemble.

What is new is the explicit packaging: continuous conditional Sobol’ fields plus surrogate-induced credible intervals from one sparse Bayesian PCE. The ingredients (PIT decoupling, PCE-Sobol’, sparse Bayesian PCE, RJMCMC basis selection) are all in the literature they cite, but the closed-form conditional extraction is not. They do it cleanly.

The analytic benchmark is the strongest evidence: absolute errors on mean/variance/total-effect maps sit at 1e-6–1e-7, and the 95% CIs cover the exact surfaces. The borehole case is the usual engineering check—relative moment errors ~1e-3 versus DLMC, smooth sensitivity evolution, and a claimed 9000× reduction versus nested sampling. That is enough to show the method works on a standard hybrid problem.

Soft spots are real but proportionate. The load-bearing assumption is that one sparse global PCE remains faithful enough over the whole epistemic domain that the algebraic ratios track the true conditional variances of g; they do not supply approximation theory for that, only the two numerical examples. Hyperparameters for the Poisson–Gamma, modified g-prior, residual variance, and candidate multi-index pool are incompletely specified, and there is no public code or data. Those are ordinary method-paper gaps, not internal contradictions.

This is for people who already do hybrid UQ / GSA with PCE and want continuous conditional maps without double loops. The math and citation pattern look solid; the circularity burden is low because they check against independent closed forms and nested MC. I would send it to peer review. A serious referee can demand the missing hyperparameter table, code, and a clearer statement of the sparse-global fidelity assumption. Worth engaging if you work in this corner.

Referee Report

2 major / 6 minor

Summary. The paper proposes a Bayesian framework that extracts continuous conditional Sobol’ indices (aleatory sensitivities as functions of epistemic parameters) from a single global Polynomial Chaos Expansion on an augmented latent space. After mapping hybrid uncertainty via the inverse probability integral transform, the tensor-product structure of the orthonormal basis is used to regroup the global expansion into epistemic-dependent coefficient fields c_αL(θ). Conditional means, variances, and first-order/total Sobol’ indices then follow by purely algebraic post-processing (Lemmas 3.1–3.2, Algorithm 1), without nested sampling or surrogate reconstruction. An RJMCMC sparse-PCE layer supplies adaptive basis selection and posterior credible intervals. An analytical hybrid polynomial benchmark recovers exact conditional moments and total-effect maps to absolute errors ~10^{-6}–10^{-7}; a borehole groundwater model with three epistemic hyperparameters shows relative moment errors ~10^{-3} versus double-loop Monte Carlo and smooth sensitivity fields, at a claimed 9000× reduction in model evaluations.

Significance. If the sparse global surrogate remains faithful across the epistemic domain, the algebraic extraction removes the classical double-loop bottleneck for conditional Sobol’ fields and supplies continuous maps plus Bayesian uncertainty quantification from one joint design. The derivation is standard orthonormal PCE plus uniqueness of the Sobol’ decomposition under the stated factorization; the analytical benchmark and borehole comparison provide concrete supporting evidence. The combination of a single-surrogate algebraic post-processing step with RJMCMC credible intervals is a useful, practically relevant contribution for hybrid UQ and GSA.

major comments (2)

The central claim rests on the unstated but load-bearing assumption that a single sparse global PCE trained on joint LHS samples remains accurate enough that the algebraic conditional variances and Sobol’ ratios track the true conditional law of g over the full epistemic domain (implicit in §3.1–3.2 and the N=5000 designs of §4). The paper should state this assumption explicitly and supply a practical diagnostic (e.g., leave-one-epistemic-region validation, or comparison of conditional residual variance against a local reference) so that users can detect when truncation/sparsity error systematically distorts the conditional decomposition.
§4.2 reports relative moment errors of order 10^{-3} versus DLMC and visually enlarged credible intervals, but does not quantify how those moment errors propagate into the conditional Sobol’ ratios themselves, nor does it report a direct numerical comparison of Su|θ against nested Monte Carlo on a subset of the epistemic grid. Because the ratios are normalized by the conditional variance, even modest absolute errors can bias the sensitivity fields; a short table or additional panel would make the fidelity claim for the indices (not only the moments) fully load-bearing.

minor comments (6)

Section title “2 Backgroud” is misspelled; correct to “Background”.
Fig. 4 caption states CI widths scaled by 10,000 while the figure legend and surrounding text use #10000 / ×10,000 inconsistently with the ×1,000 wording in the paragraph; align the scaling factor across text, caption, and legend.
Fig. 7 caption and legend claim 95% CIs enlarged by ×10, while the subplot titles write “95%CI#50”; reconcile the enlargement factor.
Notation for the latent variables switches among ξ, uξ, Uξ and for epistemic parameters among θ, Θ, uΘ; a short notational table or consistent choice would improve readability.
The hierarchical prior hyperparameters (aM, bM, ag, bg, aσ, bσ) and the maximum candidate polynomial order are free parameters of the method but are not listed for the numerical experiments; a brief statement of the values used would aid reproducibility.
References [7] (Angus on the probability integral transform) and [19] (Rumsey et al. on Bayesian adaptive PCE) are appropriate; a short pointer to earlier conditional or local Sobol’ work would help situate the contribution for readers outside hybrid UQ.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: conditional Sobol fields are algebraic consequences of a fitted global PCE and are validated against independent closed-form and nested-MC references.

full rationale

The central derivation (Lemmas 3.1–3.2) starts from a single sparse PCE fitted by RJMCMC to joint LHS samples of the true model g in the augmented latent space, then regroups the already-estimated coefficients via the tensor-product factorization Ψα(θ,ξ)=ΨαK(θ)ΨαL(ξ) to obtain parametric fields cαL(θ). The conditional variances and Sobol ratios are therefore exact for the surrogate by orthonormality; they are not defined in terms of the target indices, nor are they fitted to any conditional-Sobol data. External checks against the analytic hybrid polynomial (exact surfaces, 10^{-6} errors) and against double-loop Monte Carlo on the borehole model supply independent corroboration. RJMCMC priors and the modified g-prior are standard hierarchical choices that do not encode the reported sensitivity surfaces. No self-citation supplies a uniqueness theorem or ansatz that forces the result, and no known empirical pattern is merely renamed. The only material assumption is surrogate fidelity over the epistemic domain, which is an accuracy risk rather than a circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim rests on classical PCE orthonormality and Sobol’ uniqueness after an isoprobabilistic map to independent latents, plus a hierarchical Bayesian sparse model for coefficients. Free parameters are the usual RJMCMC/prior knobs and design size; no new physical entities are postulated. Domain assumptions (independence after PIT, finite variance, Gaussian residual) are standard for this literature.

free parameters (5)

Joint LHS sample size N = 5000
N=5000 is chosen for both examples; accuracy and sparsity of the global PCE (hence conditional fields) depend on this budget.
Hierarchical Poisson–Gamma prior on model dimension M (aM, bM)
Controls sparsity of the active basis set in RJMCMC; not uniquely determined by theory.
Modified g-prior hyperparameters (g0^2 ~ Inv-Gamma(ag,bg) and interaction penalties gm)
Shrinkage strength and order-dependent penalties affect which multi-indices survive and the posterior width of Sobol’ indices.
Residual variance prior Inv-Gamma(aσ, bσ) and Gaussian noise model
Observation model for Bayesian PCE; influences posterior coefficient uncertainty and thus CI width.
Maximum total polynomial order / candidate multi-index pool for RJMCMC moves
Truncation of the search space for birth/mutate proposals; not fully specified numerically in the text.

assumptions (5)

standard math Multivariate PCE bases are tensor products of univariate orthonormal polynomials and remain orthonormal under the product measure on independent inputs.
Used throughout §2.2 and Lemma 3.1 to justify variance as sum of squared coefficients and factorization Ψα(θ,ξ)=ΨαK(θ)ΨαL(ξ).
standard math Sobol’ ANOVA decomposition of a square-integrable function of independent inputs is unique; partial variances equal sums of squared PCE coefficients on the corresponding multi-index sets.
Lemma 3.2 and Eqs. 15–17, 34–36.
domain assumption Inverse probability integral transform (and Nataf/Rosenblatt when needed) yields mutually independent latent uniforms decoupled from epistemic parameters.
§2.1 and §3.2; required for tensor-product orthogonality in the joint latent space.
domain assumption Model evaluations equal truncated PCE plus i.i.d. Gaussian noise; hierarchical Poisson and modified g-priors are appropriate for adaptive sparse PCE.
§2.3 Eqs. 18–20; standard Bayesian sparse PCE modeling choice.
ad hoc to paper A single global sparse expansion trained on joint samples is sufficiently accurate that algebraic conditional moments match the true conditional law of g across the epistemic domain of interest.
Implicit load-bearing modeling assumption of §3–4; not proved, only illustrated on two examples.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Analytical Extraction of Conditional Aleatory Sensitivities Across Epistemic Space via a Single PCE Model." pith.science (2026). https://pith.science/paper/SPYIL66F

@misc{pith2026260704790,
  author       = {Pith},
  title        = {Pith review of: Analytical Extraction of Conditional Aleatory Sensitivities Across Epistemic Space via a Single PCE Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SPYIL66F}},
  note         = {Machine review of arXiv:2607.04790}
}

read the original abstract

In hybrid uncertainty quantification, evaluating how aleatory sensitivities vary under epistemic uncertainty, referred to as conditional Sobol' indices, is typically hindered by the computationally expensive double-loop procedure. Classical Polynomial Chaos Expansion (PCE) provides efficient access to global sensitivity measures but cannot directly resolve sensitivity variation across the epistemic space without repeated surrogate reconstruction. This study proposes a unified Bayesian framework that extracts continuous conditional Sobol' fields from a single global PCE representation. By exploiting the tensor-product structure of orthogonal polynomial bases in an augmented stochastic space, the global expansion is analytically decomposed into epistemic-dependent coefficient fields, enabling a closed-form variance decomposition. As a result, conditional Sobol' indices can be computed through a purely algebraic post-processing step without additional model evaluations or retraining. In addition, a Reversible Jump Markov Chain Monte Carlo (RJMCMC) scheme is incorporated to perform adaptive basis selection and trans-dimensional inference, while simultaneously providing Bayesian credible intervals for the conditional sensitivity measures. Numerical experiments on a high-dimensional groundwater flow model demonstrate that the proposed method significantly reduces computational cost while maintaining smooth sensitivity fields and statistically consistent uncertainty quantification across the epistemic domain.

Figures

Figures reproduced from arXiv: 2607.04790 by the authors.

**Figure 2.** Comparison of the conditional statistical moments over the epistemic domain [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗

**Figure 3.** Spatial distributions of the conditional total-effect Sobol’ indices over the epistemic domain [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: 1D sensitivity slices of the total-effect Sobol’ indices with 95% Bayesian credible intervals. The widths of the [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]

**Figure 5.** Figure 5: Physical schematic of the Borehole groundwater flow model, illustrating the mapping of hyperparameter [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

**Figure 6.** Figure 6: Comparison of the conditional statistical moments for the dynamic Borehole model over the epistemic domain [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

**Figure 7.** Figure 7: Evolution of conditional total-effect Sobol’ indices driven by epistemic hyperparameters. Solid lines indicate [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references

[1]

Aleatory or epistemic? does it matter?Structural safety, 31(2):105–112, 2009

Armen Der Kiureghian and Ove Ditlevsen. Aleatory or epistemic? does it matter?Structural safety, 31(2):105–112, 2009

2009
[2]

F Owen Hoffman and Jana S Hammonds. Propagation of uncertainty in risk assessments: the need to distinguish between uncertainty due to lack of knowledge and uncertainty due to variability.Risk analysis, 14(5):707–712, 1994

1994
[3]

Uncertainty-based information.NATO SCIENCE SERIES SUB SERIES III COMPUTER AND SYSTEMS SCIENCES, 184:21–52, 2003

George J Klir and M Wierman. Uncertainty-based information.NATO SCIENCE SERIES SUB SERIES III COMPUTER AND SYSTEMS SCIENCES, 184:21–52, 2003

2003
[4]

Number 4015

Scott Ferson, Vladik Kreinovich, Lev Ginzburg, Davis S Myers, and Kari Sentz.Constructing probability boxes and Dempster-Shafer structures. Number 4015. Sandia National Laboratories, 2003

2003
[5]

Hybrid uncertainty propagation based on multi-fidelity surrogate model.Computers & Structures, 293:107267, 2024

Jinxing Liu, Yan Shi, Chen Ding, and Michael Beer. Hybrid uncertainty propagation based on multi-fidelity surrogate model.Computers & Structures, 293:107267, 2024

2024
[6]

Sensitivity estimates for nonlinear mathematical models.Math

Ilya M Sobol. Sensitivity estimates for nonlinear mathematical models.Math. Model. Comput. Exp., 1(4):407–414, 1993

1993
[7]

The probability integral transform and related results.SIAM review, 36(4):652–654, 1994

John E Angus. The probability integral transform and related results.SIAM review, 36(4):652–654, 1994

1994
[8]

Efficient global sensitivity analysis framework and approach for structures with hybrid uncertainties.Computer Methods in Applied Mechanics and Engineering, 436:117726, 2025

Jinxing Liu, Yan Shi, Chen Ding, and Michael Beer. Efficient global sensitivity analysis framework and approach for structures with hybrid uncertainties.Computer Methods in Applied Mechanics and Engineering, 436:117726, 2025

2025

Show all 27 references

[9]

Yi Luo, Meng-Ze Lyu, Matteo Broggi, Marko Thiele, Vasileios C Fragkoulis, and Michael Beer. Unified framework for hybrid aleatory and epistemic uncertainty propagation via decoupled multi-probability density evolution method.Reliability Engineering & System Safety, page 112171, 2025

2025
[10]

Courier Corporation, 2003

Roger G Ghanem and Pol D Spanos.Stochastic finite elements: a spectral approach. Courier Corporation, 2003

2003
[11]

Extending polynomial chaos to include interval analysis

Antonello Monti, Ferdinanda Ponci, and Marco Valtorta. Extending polynomial chaos to include interval analysis. IEEE Transactions on Instrumentation and Measurement, 59(1):48–55, 2009

2009
[12]

Mohd Yusuf and Sourajeet Roy. A polymorphic polynomial chaos formulation for mixed epistemic-aleatory uncertainty quantification of rf/microwave circuits.IEEE Transactions on Microwave Theory and Techniques, 70(1):926–937, 2021

2021
[13]

A novel dimension fusion based polynomial chaos approach for mixed aleatory-epistemic uncertainty quantification of carbon nanotube interconnects

Aditi Krishna Prasad and Sourajeet Roy. A novel dimension fusion based polynomial chaos approach for mixed aleatory-epistemic uncertainty quantification of carbon nanotube interconnects. In2017 IEEE International Symposium on Electromagnetic Compatibility & Signal/Power Integr...

2017
[14]

The wiener–askey polynomial chaos for stochastic differential equations.SIAM journal on scientific computing, 24(2):619–644, 2002

Dongbin Xiu and George Em Karniadakis. The wiener–askey polynomial chaos for stochastic differential equations.SIAM journal on scientific computing, 24(2):619–644, 2002

2002
[15]

Separating the contributions of variability and parameter uncertainty in probability distributions.Reliability Engineering & System Safety, 112:187–199, 2013

Shankar Sankararaman and Sankaran Mahadevan. Separating the contributions of variability and parameter uncertainty in probability distributions.Reliability Engineering & System Safety, 112:187–199, 2013

2013
[16]

Signal recovery from partial information via orthogonal matching pursuit.IEEE Trans

Joel Tropp, Anna C Gilbert, et al. Signal recovery from partial information via orthogonal matching pursuit.IEEE Trans. Inform. Theory, 53(12):4655–4666, 2007

2007
[17]

Adaptive sparse polynomial chaos expansion based on least angle regression

Géraud Blatman and Bruno Sudret. Adaptive sparse polynomial chaos expansion based on least angle regression. J. Comput. Phys., 230:2345–2367, 2011

2011
[18]

Bayesian sparse polynomial chaos expansion for global sensitivity analysis.Computer Methods in Applied Mechanics and Engineering, 318:474–496, 2017

Qian Shao, Anis Younes, Marwan Fahs, and Thierry A Mara. Bayesian sparse polynomial chaos expansion for global sensitivity analysis.Computer Methods in Applied Mechanics and Engineering, 318:474–496, 2017

2017
[19]

Bayesian adaptive polynomial chaos expansions.Stat, 15(1):e70151, 2026

Kellin N Rumsey, Devin Francom, Graham Gibson, J Derek Tucker, and Gabriel Huerta. Bayesian adaptive polynomial chaos expansions.Stat, 15(1):e70151, 2026

2026
[20]

Statistical reasoning with imprecise probabilities

Peter Walley. Statistical reasoning with imprecise probabilities. 1991

1991
[21]

Global sensitivity analysis using polynomial chaos expansions.Reliability engineering & system safety, 93(7):964–979, 2008

Bruno Sudret. Global sensitivity analysis using polynomial chaos expansions.Reliability engineering & system safety, 93(7):964–979, 2008

2008
[22]

Princeton university press, 2010

Dongbin Xiu.Numerical methods for stochastic computations: a spectral method approach. Princeton university press, 2010. 14 APREPRINT- JULY7, 2026

2010
[23]

Bass: An r package for fitting and performing sensitivity analysis of bayesian adaptive spline surfaces.Journal of Statistical Software, 94:1–36, 2020

Devin Francom and Bruno Sansó. Bass: An r package for fitting and performing sensitivity analysis of bayesian adaptive spline surfaces.Journal of Statistical Software, 94:1–36, 2020

2020
[24]

A two-component g-prior for variable selection.Bayesian Analysis, 11(2):353, 2015

Hongmei Zhang, Xianzheng Huang, Jianjun Gan, Wilfried Karmaus, and Tara Sabo-Attwood. A two-component g-prior for variable selection.Bayesian Analysis, 11(2):353, 2015

2015
[25]

On assessing prior distributions and bayesian regression analysis with g-prior distributions

Arnold Zellner. On assessing prior distributions and bayesian regression analysis with g-prior distributions. Bayesian inference and decision techniques, 1986

1986
[26]

Springer, 2009

Peter D Hoff.A first course in Bayesian statistical methods, volume 580. Springer, 2009

2009
[27]

Delayed rejection in reversible jump metropolis–hastings.Biometrika, 88(4):1035–1053, 2001

Peter J Green and Antonietta Mira. Delayed rejection in reversible jump metropolis–hastings.Biometrika, 88(4):1035–1053, 2001. 15

2001

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.

[1] [1]

Aleatory or epistemic? does it matter?Structural safety, 31(2):105–112, 2009

Armen Der Kiureghian and Ove Ditlevsen. Aleatory or epistemic? does it matter?Structural safety, 31(2):105–112, 2009

2009

[2] [2]

F Owen Hoffman and Jana S Hammonds. Propagation of uncertainty in risk assessments: the need to distinguish between uncertainty due to lack of knowledge and uncertainty due to variability.Risk analysis, 14(5):707–712, 1994

1994

[3] [3]

Uncertainty-based information.NATO SCIENCE SERIES SUB SERIES III COMPUTER AND SYSTEMS SCIENCES, 184:21–52, 2003

George J Klir and M Wierman. Uncertainty-based information.NATO SCIENCE SERIES SUB SERIES III COMPUTER AND SYSTEMS SCIENCES, 184:21–52, 2003

2003

[4] [4]

Number 4015

Scott Ferson, Vladik Kreinovich, Lev Ginzburg, Davis S Myers, and Kari Sentz.Constructing probability boxes and Dempster-Shafer structures. Number 4015. Sandia National Laboratories, 2003

2003

[5] [5]

Hybrid uncertainty propagation based on multi-fidelity surrogate model.Computers & Structures, 293:107267, 2024

Jinxing Liu, Yan Shi, Chen Ding, and Michael Beer. Hybrid uncertainty propagation based on multi-fidelity surrogate model.Computers & Structures, 293:107267, 2024

2024

[6] [6]

Sensitivity estimates for nonlinear mathematical models.Math

Ilya M Sobol. Sensitivity estimates for nonlinear mathematical models.Math. Model. Comput. Exp., 1(4):407–414, 1993

1993

[7] [7]

The probability integral transform and related results.SIAM review, 36(4):652–654, 1994

John E Angus. The probability integral transform and related results.SIAM review, 36(4):652–654, 1994

1994

[8] [8]

Efficient global sensitivity analysis framework and approach for structures with hybrid uncertainties.Computer Methods in Applied Mechanics and Engineering, 436:117726, 2025

Jinxing Liu, Yan Shi, Chen Ding, and Michael Beer. Efficient global sensitivity analysis framework and approach for structures with hybrid uncertainties.Computer Methods in Applied Mechanics and Engineering, 436:117726, 2025

2025