arxiv: 2605.06854 · v1 · submitted 2026-05-07 · 🧮 math.OC

Recognition: 2 theorem links

· Lean Theorem

Simplicial Regularizability of the Pseudo-Moment Cone and Carath\'eodory-Type Atomic Decomposition of Moment Matrices

Heng Yang, Shucheng Kang

Pith reviewed 2026-05-11 02:06 UTC · model grok-4.3

classification 🧮 math.OC

keywords pseudo-moment conesimplicial facesatomic decompositionCarathéodory theoremmoment matricesspectrahedral conesextreme rays

0 comments

The pith

Moment matrices from O(n^d) generic atoms have simplicial minimal faces in the pseudo-moment cone generated by their atoms.

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

The paper shows that for fixed d at least 2, a moment matrix built from O(n^d) generically chosen weighted atoms has a minimal face in the pseudo-moment cone that is simplicial and spanned exactly by the planted rank-one atoms. This geometric fact supports a new Carathéodory-type algorithm that decomposes such matrices back into atoms by finding extreme rays of spectrahedral cones. A sympathetic reader would care because the result gives a concrete, efficient recovery procedure for moment matrices in the regime where the number of atoms scales as n to the power d, with numerical tests indicating it works well even slightly beyond the proven range.

Core claim

If a moment matrix is formed by O(n^d) generically chosen weighted atoms, then its minimal face in the matrix realization of the pseudo-moment cone is simplicial and generated by the planted rank-one atoms. Based on this geometric result, a Carathéodory-type extreme-ray decomposition algorithm is developed for spectrahedral cones and specializes to an efficient atomic decomposition method for generically generated moment matrices in the same regime.

What carries the argument

The simplicial minimal face of the moment matrix inside the pseudo-moment cone, which is spanned by the planted rank-one atoms and serves as the basis for the extreme-ray decomposition algorithm.

If this is right

The algorithm recovers the original atoms efficiently for generic moment matrices in the O(n^d) regime.
Numerical implementation shows strong recovery performance on tested instances.
Outside the proven regime the same procedure can still sample high-rank extreme rays of the cone.

Where Pith is reading between the lines

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

The simpliciality property may extend to other spectrahedral cones with similar facial structure.
The decomposition method could improve practical solvers for polynomial optimization problems that rely on moment relaxations.
Stabilization techniques used in the numerics suggest the approach remains usable for mildly non-generic data.

Load-bearing premise

The atoms must be chosen generically and their number must remain at most O(n^d) for fixed d at least 2.

What would settle it

A single explicit example of O(n^d) generic atoms whose moment matrix has a minimal face that is not simplicial or not generated solely by those atoms would falsify the claim.

Figures

Figures reproduced from arXiv: 2605.06854 by Heng Yang, Shucheng Kang.

**Figure 1.** Figure 1: Response curves of ew and ez as functions of s for different d and n. d 2 2 2 2 2 2 2 2 3 3 3 3 n 3 4 5 6 7 8 9 10 2 3 4 5 Theoretical Upper Bound from (7) 2 3 5 7 9 12 15 18 2 6 12 20 Empirical Upper Bound from Experiments 6 10 15 20 26 35 43 53 7 16 29 47 [PITH_FULL_IMAGE:figures/full_fig_p019_1.png] view at source ↗

**Figure 2.** Figure 2: Numerical ranks of extreme rays for d = 2, n = 3 when s is increased from 3 to 10. Each point on x-axis represents a random seed. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_2.png] view at source ↗

**Figure 3.** Figure 3: Numerical ranks of extreme rays for different [PITH_FULL_IMAGE:figures/full_fig_p021_3.png] view at source ↗

read the original abstract

We study the facial geometry of the homogeneous pseudo-moment cone \(\Sigma_{n,2d}^*\) and its implications for atomic decomposition of moment matrices. For fixed \(d \ge 2\), we show that if a moment matrix is formed by \(O(n^d)\) generically chosen weighted atoms, then its minimal face in the matrix realization of the pseudo-moment cone is \emph{simplicial} and generated by the planted rank-one atoms. Based on this geometric result, we develop a Carath\'eodory-type extreme-ray decomposition algorithm for spectrahedral cones and show that, when specialized to the pseudo-moment cone, it yields an efficient atomic decomposition method for generically generated moment matrices in the same regime. A stabilized numerical implementation demonstrates strong recovery performance and suggests that, outside the guaranteed regime, the algorithm may serve as a practical sampler of high-rank extreme rays.

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.

Desk Editor's Note private letter to a colleague

The paper shows that generic O(n^d)-atom moment matrices have simplicial minimal faces in the pseudo-moment cone and turns that into a Carathéodory-type decomposition algorithm.

read the letter

The central claim is that a moment matrix formed by O(n^d) generically chosen weighted atoms has a simplicial minimal face in the matrix realization of the pseudo-moment cone, spanned exactly by the planted rank-one atoms. They prove this for fixed d at least 2 by a dimension-counting argument: the set of configurations that admit an extraneous extreme ray is a proper algebraic subvariety, so it is avoided generically. From the resulting facial structure they derive a general extreme-ray decomposition procedure for spectrahedral cones and specialize it to recover the atoms in this regime. The numerical section is only an empirical check, not part of the proof, and shows decent recovery inside the bound plus some heuristic utility outside it. This is the main new piece: a direct geometric route from facial simpliciality to an atomic decomposition method that stays inside the cone rather than relying on generic SDP solvers. The argument looks internally consistent and avoids circularity; the genericity assumption is stated up front and the dimension count matches the ambient space dimension binom(n+d,d). One soft spot is the restriction to generic atoms and the O(n^d) count. If the atoms are algebraically dependent or you exceed that number, the simpliciality can fail and the algorithm becomes a heuristic sampler with no recovery guarantee. The paper does not overclaim on that point. Another minor point is that the real-algebraic geometry steps are sketched at a level that assumes familiarity with semi-algebraic sets and dimension theory; a reader without that background will need to fill in details. This work is for people already working on moment problems, polynomial optimization, and atomic recovery in convex relaxations. A reader who needs a concrete geometric handle on when low-rank moment matrices can be decomposed without extra rays will get direct value. The thinking is clear and the evidence for the main theorem is a standard algebraic-geometry counting argument rather than ad-hoc fitting. It deserves a serious referee because the geometric statement is precise and the algorithmic consequence follows without hidden assumptions. I would send it out for review.

Referee Report

0 major / 4 minor

Summary. The paper studies the facial geometry of the homogeneous pseudo-moment cone Σ_{n,2d}^*. For fixed d ≥ 2, it proves that if a moment matrix is formed by O(n^d) generically chosen weighted atoms, then its minimal face in the matrix realization of the cone is simplicial and generated by the planted rank-one atoms. This geometric result is used to derive a Carathéodory-type extreme-ray decomposition algorithm for spectrahedral cones; when specialized to the pseudo-moment cone, the algorithm yields an efficient atomic decomposition method for generically generated moment matrices in the same regime. A stabilized numerical implementation is presented to demonstrate recovery performance and to suggest utility as a sampler of high-rank extreme rays outside the guaranteed regime.

Significance. If the central claim holds, the work provides a rigorous geometric foundation for atomic decompositions of moment matrices via the facial structure of the pseudo-moment cone, advancing understanding in polynomial optimization, semidefinite programming, and the moment problem. The dimension-counting argument establishing that configurations admitting extraneous extreme rays form a proper algebraic subvariety of the configuration space is a clear strength, as is the direct derivation of the decomposition algorithm from the resulting simplicial facial structure. The empirical illustration of performance outside the O(n^d) regime adds practical value. These elements together offer both theoretical insight and an algorithmic tool with potential applications in high-dimensional atomic recovery.

minor comments (4)

The definition and basic properties of the homogeneous pseudo-moment cone Σ_{n,2d}^* and its matrix realization should be recalled or referenced explicitly in the introduction or the section containing the main theorem to improve accessibility for readers outside the immediate subfield.
In the section deriving the Carathéodory-type algorithm, the specialization step from general spectrahedral cones to the pseudo-moment cone would benefit from a short explicit statement of how the simpliciality implies the decomposition terminates with exactly the planted atoms.
The numerical experiments section presents the implementation as an empirical illustration; adding a brief discussion of the stabilization technique (e.g., regularization parameter choice or pivoting strategy) would enhance reproducibility without altering the theoretical focus.
A small number of typographical inconsistencies appear in the notation for the atom weights and the dimension count O(n^d); a final consistency pass on symbols would improve readability.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive and constructive review of our manuscript on the facial geometry of the pseudo-moment cone and the associated Carathéodory-type atomic decomposition algorithm. The recommendation for minor revision is appreciated, as is the recognition of the dimension-counting argument and the practical value of the numerical implementation. Since the report contains no specific major comments requiring point-by-point response, we have no substantive revisions to propose at this time.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The central claim is a geometric theorem: for fixed d ≥ 2 and O(n^d) generically chosen atoms, the minimal face of the moment matrix in the pseudo-moment cone is simplicial and generated by the planted rank-1 atoms. This rests on a dimension-counting argument that the locus of configurations admitting an extraneous extreme ray is a proper algebraic subvariety of the configuration space. The Carathéodory-type algorithm is then derived directly from the resulting facial structure. No self-definitional steps, fitted parameters renamed as predictions, load-bearing self-citations, uniqueness theorems imported from the authors' prior work, or ansätze smuggled via citation appear in the derivation. The numerical section is presented only as empirical illustration. The argument is self-contained against external algebraic-geometry benchmarks and does not reduce to its inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based solely on the abstract; the work relies on standard convex geometry of spectrahedral cones and facial structure, with no free parameters, invented entities, or ad-hoc axioms explicitly introduced in the summary.

axioms (1)

standard math Standard properties of spectrahedral cones and their facial geometry in convex optimization
The paper invokes known results on faces of convex cones to establish the simplicial property.

pith-pipeline@v0.9.0 · 5455 in / 1228 out tokens · 43862 ms · 2026-05-11T02:06:02.062307+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean (J-uniqueness, Aczél classification) washburn_uniqueness_aczel unclear
Theorem 1 (Simplicial regularizability...) and its proof via Theorem 2 (Generic mixed-term avoidance [20, Theorem 15]) on span{v_i ∨ v_j | i<j} ∩ K = {0} for generic atoms
IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean reality_from_one_distinction unclear
Algorithm 1 (Carathéodory-type extreme-ray decomposition) and facial lemmas (Lemma 6–7, 9–11) on minface(X, S_+^N) and simplicial cones

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

[1]

MOSEK optimization toolbox for MATLAB.User’s Guide and Reference Manual, Version, 4(1), 2019

Mosek ApS. MOSEK optimization toolbox for MATLAB.User’s Guide and Reference Manual, Version, 4(1), 2019

work page 2019
[2]

General tensor decompo- sition, moment matrices and applications.Journal of Symbolic Computation, 52:51–71, 2013

Alessandra Bernardi, Jerome Brachat, Pierre Comon, and Bernard Mourrain. General tensor decompo- sition, moment matrices and applications.Journal of Symbolic Computation, 52:51–71, 2013

work page 2013
[3]

Nonnegative polynomials and sums of squares.Journal of the American Mathe- matical Society, 25(3):617–635, 2012

Grigoriy Blekherman. Nonnegative polynomials and sums of squares.Journal of the American Mathe- matical Society, 25(3):617–635, 2012

work page 2012
[4]

Positive gorenstein ideals.Proceedings of the American Mathematical Society, 143(1):69–86, 2015

Grigoriy Blekherman. Positive gorenstein ideals.Proceedings of the American Mathematical Society, 143(1):69–86, 2015

work page 2015
[5]

Algebraic boundaries of hilbert’s sos cones.Compositio Mathematica, 148(6):1717–1735, 2012

Grigoriy Blekherman, Jonathan Hauenstein, John Christian Ottem, Kristian Ranestad, and Bernd Sturmfels. Algebraic boundaries of hilbert’s sos cones.Compositio Mathematica, 148(6):1717–1735, 2012

work page 2012
[6]

SIAM, 2012

Grigoriy Blekherman, Pablo A Parrilo, and Rekha R Thomas.Semidefinite optimization and convex algebraic geometry. SIAM, 2012

work page 2012
[7]

Extreme rays of hankel spectrahedra for ternary forms.Journal of Symbolic Computation, 79:23–42, 2017

Grigoriy Blekherman and Rainer Sinn. Extreme rays of hankel spectrahedra for ternary forms.Journal of Symbolic Computation, 79:23–42, 2017

work page 2017
[8]

Facialreductionforacone-convexprogrammingproblem.Journal of the Australian Mathematical Society, 30(3):369–380, 1981

JonMBorweinandHenryWolkowicz. Facialreductionforacone-convexprogrammingproblem.Journal of the Australian Mathematical Society, 30(3):369–380, 1981

work page 1981
[9]

Improving the threshold for finding rank-1 matrices in a subspace.arXiv preprint arXiv:2504.17947, 2025

Jeshu Dastidar, Tait Weicht, and Alexander S Wein. Improving the threshold for finding rank-1 matrices in a subspace.arXiv preprint arXiv:2504.17947, 2025

work page arXiv 2025
[10]

A link between the canonical decomposition in multilinear algebra and simul- taneous matrix diagonalization.SIAM journal on Matrix Analysis and Applications, 28(3):642–666, 2006

Lieven De Lathauwer. A link between the canonical decomposition in multilinear algebra and simul- taneous matrix diagonalization.SIAM journal on Matrix Analysis and Applications, 28(3):642–666, 2006. 22

work page 2006
[11]

The multidimensional truncated moment problem: Carathéodory numbers from hilbert functions.Mathematische Annalen, 380(1):267–291, 2021

Philipp J di Dio and Mario Kummer. The multidimensional truncated moment problem: Carathéodory numbers from hilbert functions.Mathematische Annalen, 380(1):267–291, 2021

work page 2021
[12]

The multidimensional truncated moment problem: atoms, determinacy, and core variety.Journal of Functional Analysis, 274(11):3124–3148, 2018

Philipp J di Dio and Konrad Schmüdgen. The multidimensional truncated moment problem: atoms, determinacy, and core variety.Journal of Functional Analysis, 274(11):3124–3148, 2018

work page 2018
[13]

The multidimensional truncated moment problem: The moment cone.arXiv preprint arXiv:1809.00584, 2018

Philipp J di Dio and Konrad Schmüdgen. The multidimensional truncated moment problem: The moment cone.arXiv preprint arXiv:1809.00584, 2018

work page arXiv 2018
[14]

A moment matrix approach to multivariable cubature.Integral Equations and Operator Theory, 52(1):85–124, 2005

Lawrence Fialkow and Srdjan Petrovic. A moment matrix approach to multivariable cubature.Integral Equations and Operator Theory, 52(1):85–124, 2005

work page 2005
[15]

Springer, Cham, 2020

Jean Gallier and Jocelyn Quaintance.Differential Geometry and Lie Groups: A Computational Per- spective, volume 12 ofGeometry and Computing. Springer, Cham, 2020

work page 2020
[16]

Springer, New York, 1977

Robin Hartshorne.Algebraic Geometry, volume 52 ofGraduate Texts in Mathematics. Springer, New York, 1977

work page 1977
[17]

Positively not sos: pseudo-moments and extreme rays in exact arithmetic.arXiv preprint arXiv:2509.01382, 2025

Didier Henrion. Positively not sos: pseudo-moments and extreme rays in exact arithmetic.arXiv preprint arXiv:2509.01382, 2025

work page arXiv 2025
[18]

Detecting global optimality and extracting solutions in gloptipoly

Didier Henrion and Jean-Bernard Lasserre. Detecting global optimality and extracting solutions in gloptipoly. InPositive polynomials in control, pages 293–310. Springer, 2005

work page 2005
[19]

A bound on the carathéodory number.Linear Algebra and its Applications, 532:347–363, 2017

Masaru Ito and Bruno F Lourenço. A bound on the carathéodory number.Linear Algebra and its Applications, 532:347–363, 2017

work page 2017
[20]

Computing linear sections of varieties: quantum entanglement, tensor decompositions and beyond

Nathaniel Johnston, Benjamin Lovitz, and Aravindan Vijayaraghavan. Computing linear sections of varieties: quantum entanglement, tensor decompositions and beyond. In2023 IEEE 64th Annual Sym- posium on Foundations of Computer Science (FOCS), pages 1316–1336. IEEE, 2023

work page 2023
[21]

Minimizer extraction in polynomial optimization is robust

Igor Klep, Janez Povh, and Jurij Volcic. Minimizer extraction in polynomial optimization is robust. SIAM Journal on Optimization, 28(4):3177–3207, 2018

work page 2018
[22]

Global optimization with polynomials and the problem of moments.SIAM Journal on Optimization, 11(3):796–817, 2001

Jean B Lasserre. Global optimization with polynomials and the problem of moments.SIAM Journal on Optimization, 11(3):796–817, 2001

work page 2001
[23]

A decomposition for three-way arrays.SIAM Journal on Matrix Analysis and Applications, 14(4):1064–1083, 1993

Sue E Leurgans, Robert T Ross, and Rebecca B Abel. A decomposition for three-way arrays.SIAM Journal on Matrix Analysis and Applications, 14(4):1064–1083, 1993

work page 1993
[24]

Eigenvalue-constrained faces.Linear Algebra and its Applications, 269(1-3):159– 181, 1998

Adrian Stephen Lewis. Eigenvalue-constrained faces.Linear Algebra and its Applications, 269(1-3):159– 181, 1998

work page 1998
[25]

The a-truncated k-moment problem.Foundations of Computational Mathematics, 14(6):1243–1276, 2014

Jiawang Nie. The a-truncated k-moment problem.Foundations of Computational Mathematics, 14(6):1243–1276, 2014

work page 2014
[26]

Generating polynomials and symmetric tensor decompositions.Foundations of Computa- tional Mathematics, 17(2):423–465, 2017

Jiawang Nie. Generating polynomials and symmetric tensor decompositions.Foundations of Computa- tional Mathematics, 17(2):423–465, 2017

work page 2017
[27]

Sum-of-squares optimization without semidefinite programming.SIAM Journal on Optimization, 29(1):822–851, 2019

Dávid Papp and Sercan Yildiz. Sum-of-squares optimization without semidefinite programming.SIAM Journal on Optimization, 29(1):822–851, 2019

work page 2019
[28]

Semidefinite programming relaxations for semialgebraic problems.Mathematical programming, 96(2):293–320, 2003

Pablo A Parrilo. Semidefinite programming relaxations for semialgebraic problems.Mathematical programming, 96(2):293–320, 2003

work page 2003
[29]

Strong duality for semidefinite program- ming.SIAM Journal on Optimization, 7(3):641–662, 1997

Motakuri V Ramana, Levent Tunçel, and Henry Wolkowicz. Strong duality for semidefinite program- ming.SIAM Journal on Optimization, 7(3):641–662, 1997. 23

work page 1997
[30]

Cambridge University Press, Cambridge, 2nd edition, 2013

Rolf Schneider.Convex Bodies: The Brunn–Minkowski Theory, volume 151 ofEncyclopedia of Mathe- matics and its Applications. Cambridge University Press, Cambridge, 2nd edition, 2013

work page 2013
[31]

Efficient tensor decomposition via moment matrix extension

Bobby Shi, Julia Lindberg, and Joe Kileel. Efficient tensor decomposition via moment matrix extension. arXiv preprint arXiv:2506.22564, 2025

work page arXiv 2025
[32]

Semidefinite programming.SIAM review, 38(1):49–95, 1996

Lieven Vandenberghe and Stephen Boyd. Semidefinite programming.SIAM review, 38(1):49–95, 1996

work page 1996
[33]

Tssos: A moment-sos hierarchy that exploits term sparsity.SIAM Journal on Optimization, 31(1):30–58, 2021

Jie Wang, Victor Magron, and Jean-Bernard Lasserre. Tssos: A moment-sos hierarchy that exploits term sparsity.SIAM Journal on Optimization, 31(1):30–58, 2021

work page 2021
[34]

Multi-subspace power method for decom- posing all tensors.arXiv preprint arXiv:2510.18627, 2025

Kexin Wang, João M Pereira, Joe Kileel, and Anna Seigal. Multi-subspace power method for decom- posing all tensors.arXiv preprint arXiv:2510.18627, 2025

work page arXiv 2025
[35]

Kluwer Academic Publishers, Boston, MA, 2000

Henry Wolkowicz, Romesh Saigal, and Lieven Vandenberghe, editors.Handbook of Semidefinite Pro- gramming: Theory, Algorithms, and Applications, volume 27 ofInternational Series in Operations Research & Management Science. Kluwer Academic Publishers, Boston, MA, 2000. 24

work page 2000