Pith. sign in

REVIEW 2 major objections 2 minor 17 references

A constrained optimization framework modifies polynomial chaos methods to exactly recover mean and variance with low-order terms.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.3

2026-06-27 19:44 UTC pith:SJSYZYYQ

load-bearing objection The constrained optimization tweak for exact mean/variance recovery in low-order PCE is a reasonable incremental fix, but the conditioning and feasibility issues flagged in the stress test look like they need explicit handling. the 2 major comments →

arxiv 2606.07912 v1 pith:SJSYZYYQ submitted 2026-06-06 eess.SY cs.SYmath.OC

On Improved Statistical Accuracy of Low-Order Polynomial Chaos Approximations

Vedang M. Deshpande , Raktim Bhattacharya This is my paper
classification eess.SY cs.SYmath.OC
keywords polynomial chaos expansionstatistical momentsconstrained optimizationsurrogate modelsstochastic systemslow-order approximationsGalerkin projectionleast squares
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Traditional techniques for determining coefficients in polynomial chaos expansions, such as Galerkin projection, stochastic collocation, or least squares, often fail to capture the first two statistical moments accurately unless high-order polynomials are used. This paper introduces a constrained optimization framework that adjusts these standard methods by adding moment-matching constraints so the resulting coefficients precisely recover the mean and variance. A sympathetic reader would care because the change enables accurate statistical predictions from simpler surrogate models for stochastic systems, lowering computational cost while preserving moment fidelity. The approach is tested on algebraic functions of random variables and yields clear gains in moment accuracy at low orders.

Core claim

Polynomial chaos expansions provide surrogate models for stochastic systems, with coefficients typically derived using Galerkin projection, stochastic collocation, or least squares approximation. These traditional approaches often fail to accurately capture statistical moments without resorting to high-order approximations. We propose a constrained optimization framework that modifies standard techniques to determine polynomial chaos coefficients that precisely recover the first two statistical moments. The effectiveness of our approach is demonstrated on several candidate algebraic functions of random variables, showing significant improvements in statistical accuracy even with low-order ap

What carries the argument

The constrained optimization framework that augments standard coefficient calculation methods with explicit constraints enforcing exact recovery of the first two statistical moments.

Load-bearing premise

Adding the moment-matching constraints will not introduce instabilities or degrade accuracy in higher-order statistics or other properties of the approximation.

What would settle it

Apply the constrained method to a simple algebraic test function whose exact mean and variance are known analytically; if the recovered coefficients produce mean or variance values that deviate from the true values, the precise recovery claim does not hold.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Low-order polynomial chaos expansions achieve exact recovery of the mean and variance.
  • Statistical accuracy improves without increasing the polynomial degree.
  • The framework can be applied to common methods including Galerkin projection, collocation, and least squares.
  • Demonstrated gains appear on algebraic functions of random variables.

Where Pith is reading between the lines

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

  • The approach may reduce the expense of uncertainty propagation in engineering models limited by polynomial order.
  • Similar constraints could be tested for recovering additional statistics such as skewness.
  • Because it modifies existing techniques, the framework could be added to current polynomial chaos implementations with modest changes.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The paper proposes a constrained optimization framework that augments standard Galerkin projection, stochastic collocation, or least-squares methods for determining polynomial chaos expansion coefficients so that the resulting surrogate exactly recovers the first two statistical moments (mean and variance) of the target random variable. Effectiveness is shown via numerical experiments on several algebraic test functions, where low-order expansions achieve improved moment accuracy compared with unconstrained baselines.

Significance. If the constrained formulation preserves solvability and numerical stability across bases and dimensions, the method would directly improve the practical utility of low-order PCE surrogates in uncertainty quantification by guaranteeing moment fidelity without increasing polynomial degree. The algebraic test cases provide concrete evidence of moment recovery, which is a clear strength.

major comments (2)
  1. [Section 3 (formulation) or Section 4 (numerical results)] The central claim requires that adjoining the two linear moment-matching equality constraints to the coefficient problem never produces an inconsistent or severely ill-conditioned system. The manuscript should include a brief analysis (or numerical evidence) of the condition number of the resulting saddle-point or equality-constrained system as a function of basis size and input dimension; without this, the practical scope of the method remains unclear.
  2. [Section 4 / Table 1] Table 1 (or equivalent results table) reports moment errors for the test functions but does not tabulate the condition numbers of the constrained versus unconstrained linear systems or the feasibility margin of the equality constraints. Adding these quantities would directly address whether the added constraints introduce the instabilities raised by the reviewer.
minor comments (2)
  1. [Abstract / Introduction] The abstract states that the approach 'modifies standard techniques'; a short sentence in the introduction clarifying whether the constraints are enforced via Lagrange multipliers, null-space projection, or a penalty term would improve readability.
  2. [Section 3] Notation for the constrained optimization problem (e.g., the matrix that encodes the two moment constraints) should be introduced once and used consistently; several places appear to reuse the symbol for the unconstrained projection matrix.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and positive evaluation of the constrained optimization framework. We address each major comment below and will incorporate the suggested numerical evidence in the revised manuscript.

read point-by-point responses

  1. Referee: [Section 3 (formulation) or Section 4 (numerical results)] The central claim requires that adjoining the two linear moment-matching equality constraints to the coefficient problem never produces an inconsistent or severely ill-conditioned system. The manuscript should include a brief analysis (or numerical evidence) of the condition number of the resulting saddle-point or equality-constrained system as a function of basis size and input dimension; without this, the practical scope of the method remains unclear.

    Authors: We agree that a discussion of conditioning is valuable for clarifying the method's scope. The equality constraints are linear and the underlying moment-matching problem is always consistent by construction when the constraints are enforced via the optimization formulation. In the revised manuscript we will add numerical evidence in Section 4 showing condition numbers of the constrained systems (and their unconstrained counterparts) for increasing basis sizes and input dimensions on the algebraic test problems. This will demonstrate that the added constraints do not produce severe ill-conditioning within the regimes examined. revision: yes

  2. Referee: [Section 4 / Table 1] Table 1 (or equivalent results table) reports moment errors for the test functions but does not tabulate the condition numbers of the constrained versus unconstrained linear systems or the feasibility margin of the equality constraints. Adding these quantities would directly address whether the added constraints introduce the instabilities raised by the reviewer.

    Authors: We will augment the numerical results (either by extending Table 1 or adding a companion table) to report the 2-norm condition numbers of the constrained and unconstrained coefficient problems together with the feasibility margins, measured as the Euclidean norms of the two equality-constraint residuals. These quantities will be provided for all test cases and will confirm that the constraints are satisfied to near machine precision without degrading numerical stability. revision: yes

Circularity Check

0 steps flagged

No significant circularity; method enforces moments by explicit constraints rather than deriving them from inputs

full rationale

The paper proposes adding explicit moment-matching equality constraints to standard projection or regression problems for polynomial chaos coefficients. This is a direct modification of existing techniques whose purpose is to enforce the first two moments by construction; the central claim does not reduce any derived result or prediction to a self-referential fit or self-citation. No equations, uniqueness theorems, or ansatzes from prior self-work are invoked in the abstract or described approach to justify the framework itself. The derivation chain is therefore self-contained as an optimization formulation rather than a tautological renaming or forced prediction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; assessment is limited to the high-level description of modifying standard PCE techniques via constraints.

pith-pipeline@v0.9.1-grok · 5608 in / 958 out tokens · 13433 ms · 2026-06-27T19:44:59.682850+00:00 · methodology

0 comments
read the original abstract

Polynomial chaos expansions provide surrogate models for stochastic systems, with coefficients typically derived using Galerkin projection, stochastic collocation, or least squares approximation. These traditional approaches often fail to accurately capture statistical moments without resorting to high-order approximations. We propose a constrained optimization framework that modifies standard techniques to determine polynomial chaos coefficients that precisely recover the first two statistical moments. The effectiveness of our approach is demonstrated on several candidate algebraic functions of random variables, showing significant improvements in statistical accuracy even with low-order approximations.

Figures

Figures reproduced from arXiv: 2606.07912 by Raktim Bhattacharya, Vedang M. Deshpande.

Figure 1
Figure 1. Figure 1: Absolute errors in estimated moments of f(∆) = ∆8 using LS, SC, and GP formulations for different approximation orders (κ), where ∆ is uniformly distributed over [−1, 1]. (GP) yields deterministic evolution equations for the expansion coefficients, solvable with standard numerical methods, and the effectiveness of generalized PC (gPC) in such problems is well established [6], [7]. GP requires inner-product… view at source ↗
Figure 2
Figure 2. Figure 2: Accuracy of PC approximations of different orders ( [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Accuracy of PC approximations of different orders ( [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

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

17 extracted references · 1 canonical work pages · 1 internal anchor

  1. [1]

    The wiener-askey polynomial chaos for stochastic differential equations

    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

  2. [2]

    Polynomial chaos expansion for parametric problems in engineering systems: A review

    Danfeng Shen, Hao Wu, Bingqing Xia, and Deqiang Gan. Polynomial chaos expansion for parametric problems in engineering systems: A review. IEEE Systems Journal, 14(3):4500–4514, 2020

    2020

  3. [3]

    Deshpande, and Raktim Bhattacharya

    Sunsoo Kim, Vedang M. Deshpande, and Raktim Bhattacharya. Robust kalman filtering with probabilistic uncertainty in system parameters. IEEE Control Systems Letters, 5(1):295–300, 2021

    2021

  4. [4]

    Hover and Michael S

    Franz S. Hover and Michael S. Triantafyllou. Application of polynomial chaos in stability and control. Automatica, 42(5):789–795, 2006

    2006

  5. [5]

    Robust lqr design for systems with probabilistic uncertainty

    Raktim Bhattacharya. Robust lqr design for systems with probabilistic uncertainty. International Journal of Robust and Nonlinear Control, 29(10):3217–3237, 2019

    2019

  6. [6]

    Modeling uncertainty in flow simulations via generalized polynomial chaos

    Dongbin Xiu and George Em Karniadakis. Modeling uncertainty in flow simulations via generalized polynomial chaos. Journal of Computational Physics, 187(1):137–167, 2003

    2003

  7. [7]

    Supersensitivity due to uncertain boundary conditions

    Dongbin Xiu and George Em Karniadakis. Supersensitivity due to uncertain boundary conditions. International Journal for Numerical Methods in Engineering, 61(12):2114–2138, 2004

    2004

  8. [8]

    Evaluation of non-intrusive approaches for wiener-askey generalized polynomial chaos

    Michael Eldred, Clayton Webster, and Paul Constantine. Evaluation of non-intrusive approaches for wiener-askey generalized polynomial chaos. In 49th AIAA/ASME/ASCE /AHS/ASC Structures, Structural Dynamics, and Materials Conference, 2008

    2008

  9. [9]

    Comparison of non-intrusive polynomial chaos and stochastic collocation methods for uncertainty quantification

    Michael Eldred and John Burkardt. Comparison of non-intrusive polynomial chaos and stochastic collocation methods for uncertainty quantification. In 47th AIAA Aerospace Sciences Meeting including The New Horizons Forum and Aerospace Exposition, 2009

    2009

  10. [10]

    Cameron and W

    R. Cameron and W. Martin. The orthogonal development of non-linear functionals in series of fourier-hermite functionals. Annals of Mathematics, 48(2):385–392, 1947

    1947

  11. [11]

    Semidefinite relaxations for quadratically constrained quadratic programming: A review and comparisons

    Xiaowei Bao, Nikolaos V Sahinidis, and Mohit Tawarmalani. Semidefinite relaxations for quadratically constrained quadratic programming: A review and comparisons. Mathematical programming, 129(1):129, 2011

    2011

  12. [12]

    On convex relaxations for quadratically constrained quadratic programming

    Kurt M Anstreicher. On convex relaxations for quadratically constrained quadratic programming. Mathematical programming, 136(2):233–251, 2012

    2012

  13. [13]

    General Heuristics for Nonconvex Quadratically Constrained Quadratic Programming

    Jaehyun Park and Stephen Boyd. General heuristics for nonconvex quadratically constrained quadratic programming. arXiv preprint arXiv:1703.07870, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  14. [14]

    CVXPY: A Python-embedded modeling language for convex optimization

    Steven Diamond and Stephen Boyd. CVXPY: A Python-embedded modeling language for convex optimization. Journal of Machine Learning Research, 17(83):1–5, 2016

    2016

  15. [15]

    A rewriting system for convex optimization problems

    Akshay Agrawal, Robin Verschueren, Steven Diamond, and Stephen Boyd. A rewriting system for convex optimization problems. Journal of Control and Decision, 5(1):42–60, 2018

    2018

  16. [16]

    Towards stochastic fluid mechanics via polynomial chaos

    Robert Walters. Towards stochastic fluid mechanics via polynomial chaos. In 41 st AIAA Aerospace Sciences Meeting & Exhibit, Reno, NV, 2003

    2003

  17. [17]

    Efficient sampling for non-intrusive polynomial chaos applications with multiple uncertain input variables

    Serhat Hosder, Robert Walters, and Michael Balch. Efficient sampling for non-intrusive polynomial chaos applications with multiple uncertain input variables. In 48th AIAA/ASME/ ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, page 1939, 2007

    1939