REVIEW 4 major objections 5 minor 2 cited by
Spectral Obstructions to Contracting Transport Maps on Curved Spaces
T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash
Pith's one-line read In dimensions four and higher, smooth metrics on spheres and Gaussian-weighted spaces can have Laplace spectra below those of the round sphere and Gaussian model, which rules out any 1-Lipschitz optimal transport map between them.
desk verdict The d≥4 counterexamples are solid and worth taking seriously, but the abstract promises results the body doesn't contain. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The engine is the product structure of the model spaces. The sphere is presented as a warped product over an S¹ fiber, and Euclidean space as a rotationally symmetric warped product over S^{d−1}; both fibers carry many low-lying modes. The paper perturbs the fiber metric — enlarging the S¹ factor on the sphere, and making the sphere-slices of the Euclidean metric cylindrical with radius √(d−2) — and uses the Ricci formula for multiply warped products to control the lower curvature bound. The 'Contraction Principle' (an L-Lipschitz map forces λ_k(target) ≥ L^{−2} λ_k(source)) is the named theorem that converts a spectral drop into a transport obstruction. In the weighted case, a torpedo radiu
What would settle it
For d=4, directly compute the Rayleigh quotient R_{g_ε}(f_k) from the explicit formula and check whether it stays below (ρ_ε/(d−1))k(k+d−1) for some k≥k_0; if not, Theorem 1's spectral drop fails. For Theorem 2, evaluate the Dirichlet-to-mass ratios of the d+2 test functions U_{ε,i} and B_ε on the explicit torpedo metric; if any ratio reaches 2, the inequality λ_{d+2}<2 fails. Both are finite calculations using the formulas in Sections 3 and 4.
Extended reading notes
Core claim
At the core is an explicit spectral computation. On S^d, the warped metric g_ε = dt² + sin²t(1+ε sin⁴t)dθ² + cos²t g_{S^{d−2}} has Ricci curvature at least ρ_ε = d−1−25ε/12; test functions f_k = Re(z^k) give Rayleigh quotients below the round sphere's eigenvalue ρ_ε/(d−1)·k(k+d−1). On R^d, a rotationally symmetric 'torpedo' metric with cross-section radius √(d−2) and a Gaussian weight on the cylinder satisfies CD(1,∞), yet d+2 test functions have Rayleigh quotient below 2, forcing λ_{d+2}<2 while the Gaussian model has λ_{d+2}=2. The Contraction Principle turns each spectral drop into an obstruction to 1-Lipschitz transport maps.
Load-bearing premise
The counterexamples stand or fall on the Ricci-curvature lower bound of the constructed warped metrics (and the curvature-dimension condition in the weighted case); if either curvature computation is off, the eigenvalue comparison against the round sphere or Gaussian model has no valid target.
Editorial extensions
If this is right
- The spherical spectral comparison conjectures are false in every dimension d≥4: the constructed metric has a strictly smaller Laplace eigenvalue than the round sphere with the same Ricci lower bound.
- The Gaussian spectral comparison on weighted spaces is false in every dimension d≥4: the constructed CD(1,∞) space has λ_{d+2}<2, whereas the standard Gaussian has λ_{d+2}=2.
- Consequently, no 1-Lipschitz map pushing the round-sphere volume or the standard Gaussian forward onto these targets can exist; the Contraction Principle would force the opposite eigenvalue inequality.
- The product-structure mechanism (many small eigenvalues arising from fiber coordinates) is a general source of spectral obstructions, not a peculiarity of the Clifford torus.
Reading between the lines
- The same warped-product mechanism should transfer to other symmetric model spaces, such as complex projective space or hyperbolic space: a small fiber perturbation should create low eigenvalue clusters below the symmetric-space threshold.
- The weighted construction verifies the stronger conditions Ric_g≥0 and ∇²V≥g only for d≥5; whether d=4 can also satisfy both simultaneously is an open seam exposed by this paper.
- Numerically evaluating the explicit Rayleigh quotients for d=4 would yield how small ε must be for the K-th eigenvalue to drop below the round value, turning the existence argument into a quantitative threshold.
- If the abstract's dimension-two inverse mean curvature flow construction and dimension-three literature counterexample are later supplied in full, the 'every dimension' conclusion would connect these d≥4 results to a complete picture; the d≥4 counterexamples are independent of those additions.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper constructs counterexamples to two spectral-comparison conjectures of Milman in the style of Caffarelli's contraction theorem. For every d≥4, Theorem 1 gives a smooth metric g on S^d with Ric_g ≥ ρg but λ_K(S^d,g) < λ_K(S^d,g^ρ_can), and Theorem 2 gives a CD(1,∞) weighted manifold (R^d,g,μ) with λ_{d+2}(R^d,g,μ)<2. The proofs use warped-product Ricci computations, explicit test functions, and min–max arguments. The abstract additionally claims a dimension-two contracting transport map constructed by inverse mean curvature flow, a dimension-three counterexample by Lin–Wang–Xu, and a resolution of the Beck–Jerison hemisphere conjecture; these ingredients do not appear in the submitted text (Sections 1–4 only).
Significance. If the d≥4 results are correct, they disprove Milman's Conjecture 3 and Conjecture 1* in general, and via the contraction principle also the corresponding transport-map conjectures. The proofs in Sections 3–4 are detailed and internally consistent: the Ricci lower bounds, Rayleigh-quotient expansions, and orthogonality arguments all check out. This is a substantial contribution. However, the paper's headline claim—that the spherical spectral comparison is settled in every dimension d≥2—is not supported by the submitted manuscript, which proves only the d≥4 counterexamples. The d=2 and d=3 ingredients are absent, so the strongest advertised conclusions are unverified.
major comments (4)
- [Abstract and Introduction] The abstract states that the work 'settles the spherical spectral comparison in every dimension d≥2' and answers the Beck–Jerison conjecture. The submitted full text contains only Sections 1–4, proving Theorems 1 and 2 for d≥4. There is no Section 5, no inverse mean curvature flow argument for dimension two, and no mention or proof of the dimension-three counterexample [LWX26]. The bibliography omits [BJ21], [CM98], [LWX26], and [FFGZ26], all cited in the abstract. The 'settles every dimension' claim therefore collapses unless the missing material is supplied. This is a load-bearing overstatement; the manuscript's actually supported contribution is the d≥4 counterexamples.
- [Abstract and §4] The abstract further claims: 'In dimensions d≥5, the weighted counterexamples can be chosen to satisfy Ric_g≥0 and ∇_g^2 V ≥ g separately.' Theorem 2 in the body establishes only the CD(1,∞) condition, i.e. Ric_g + ∇^2 V ≥ g. No result in Sections 1–4 proves or even states the separate inequalities Ric_g≥0 and ∇^2 V ≥ g. If this refinement is true, it must be stated and proved; otherwise it should be removed from the abstract.
- [§3.1 (Theorem 1, Step (iii))] The inequality for Rε(h) on E_{k−1} in (3.5) is asserted with the factor sqrt(1+ε). It is correct: gε ≥ g0, so |∇h|²_{gε} ≤ |∇h|²_{g0}, and dvol_gε = sqrt(1+ε q²) dvol_g0 ≤ sqrt(1+ε) dvol_g0, while the denominator is ≥ ∫h² dvol_g0. The text would benefit from stating this justification; currently the reader must supply the metric comparison. This is a presentation issue rather than an error.
- [§4 (Theorem 2)] The proof of Theorem 2 uses a non-C¹ test function b_A and then treats its derivative in the weak sense. This is legitimate because b_A ∈ W^{1,2}(R^d,gε,νε), as noted in the text. However, the manuscript should explicitly indicate that the Rayleigh quotient and the Dirichlet-energy orthogonality are understood with weak derivatives; the current wording (e.g. 'all appearances of B'_ε = b'_A below are understood in the weak sense') is adequate but could be stated earlier and more prominently, since it is essential for the admissibility of Bε.
minor comments (5)
- [Title/header] The title on the first page contains typos: 'TRANSPOR T' and 'CUR VED SP ACES' should be 'TRANSPORT' and 'CURVED SPACES'.
- [Bibliography] The references [BJ21], [CM98], [LWX26], and [FFGZ26] are cited in the abstract but are missing from the reference list. All cited works must appear in the bibliography.
- [§3 (Theorem 1)] In the definition of ε₀, the denominator is written without parentheses: 'ε_0 := 2k+d−2 / 1/2 (k−1)(k+d−2)+ 25/(12(d−1)) k(k+d−1)'. This should be formatted unambiguously, e.g. ε₀ = (2k+d−2) / [ (1/2)(k−1)(k+d−2) + (25/(12(d−1))) k(k+d−1) ].
- [§4 (Theorem 2)] The constants δ_U and δ_b are introduced and then used immediately; the reader must track that d≥4 is used to make them positive. The text does this, but the display for δ_U has several missing superscripts and spacing: 'δd−2', 'δd+1' etc. should be written as δ^{d−2}, δ^{d+1} for readability.
- [§4 (Lemma 4.1)] In Lemma 4.1 the proof that η(−s)=1−η(s) is correct, but the sentence 'Since ψ is even' should also note that this is what makes the integral of η over [−1,1] equal 1. This is a minor clarity point.
Circularity Check
No circularity: the d≥4 counterexamples are explicitly constructed and spectrally compared by variational estimates; the abstract's all-dimension claim is an incompleteness gap, not a circular argument.
full rationale
I walked the derivation chain. Theorem 1 explicitly constructs the metric g_ε and test functions f_k; the Ricci lower bound is derived from the cited warped-product Ricci formula (Lemma 3.1), the Rayleigh quotient is bounded by an explicit expansion in ε using beta-distribution moments, the orthogonality of f_k to lower spherical harmonics is checked by a U(1)-rotation symmetry argument, and the min–max comparison is made against the explicitly rescaled round-sphere eigenvalue. The parameters ε and k are chosen after the estimates, not fitted to force the target inequality. Theorem 2 explicitly constructs the torpedo metric and potential V_ε, verifies CD(1,∞) directly from the derived pointwise Ricci bounds, and shows that the d+2 test functions have Rayleigh quotient below 2; the comparison target λ_{d+2}(R^d,|·|,γ^d)=2 is quoted from the known Gaussian spectrum (2.1). The contraction principle (Theorem 2.1) is cited from Milman and is independent external support, not a self-citation. No step reduces to its own input by definition, and no fitted parameter is renamed as a prediction. The abstract's headline claim that the spherical spectral comparison is 'settled in every dimension d≥2' depends on a dimension-two inverse-mean-curvature-flow construction and on [LWX26] that do not appear in the submitted full text; this is an omitted-support / proof-gap issue in the advertised conclusion, but it is not circularity. The d≥4 theorems as written are self-contained and non-circular.
Assumptions & free parameters
free parameters (3)
- ε (metric perturbation / torpedo transition scale)
- k (degree of spherical test function)
- A (slope of radial tail b_A) =
δ/d
assumptions (5)
- standard math Contraction Principle (Theorem 2.1): an L-Lipschitz map pushing µ1 onto µ2 gives λ_k(M2) ≥ (1/L²) λ_k(M1).
- standard math Ricci tensor formula for multiply warped products (Lemma 3.1).
- standard math Bakry–Emery criterion: Ric_g + ∇²W ≥ ρg implies CD(ρ,∞) for smooth weighted manifolds.
- standard math Gaussian spectrum: λ_{d+2}(R^d,|·|,γ^d)=2 (equation (2.1)).
- standard math Beta-distribution moments of |z|² on S^d and the recurrence Q_{k+2}/Q_k.
Cite this review
Pith. "Pith review of Spectral Obstructions to Contracting Transport Maps on Curved Spaces." pith.science (2026). https://pith.science/paper/WPQMVRIP
@misc{pith2026260524705,
author = {Pith},
title = {Pith review of: Spectral Obstructions to Contracting Transport Maps on Curved Spaces},
year = {2026},
howpublished = {\url{https://pith.science/paper/WPQMVRIP}},
note = {Machine review of arXiv:2605.24705}
}
abstract
Caffarelli's contraction theorem states that the Brenier optimal transport map from the standard Gaussian measure to a more log-concave probability measure is $1$-Lipschitz. Motivated by this theorem, Milman [Mil18] formulated several conjectures for the round sphere and for weighted manifolds satisfying the curvature-dimension condition $\operatorname{CD}(\rho,\infty)$. Recently, Beck and Jerison [BJ21] raised related questions on the round hemisphere. The existence of a contracting transport map implies a corresponding spectral comparison. In the spherical setting, this comparison was also conjectured by Colding and Minicozzi [CM98] for compact manifolds with Ricci curvature lower bounds. In this work, we construct counterexamples to the corresponding spectral comparisons on spheres and on weighted manifolds satisfying $\operatorname{CD}(1,\infty)$ in dimensions $d\geq4$, yielding obstructions to contracting transport maps. In dimensions $d\geq5$, the weighted counterexamples can be chosen to satisfy $\operatorname{Ric}_g\geq 0$ and $\nabla_g^2 V \geq g$ separately. In dimension two, we use inverse mean curvature flow to construct a contracting transport map from the suitably rescaled round sphere to every closed connected Riemannian surface satisfying the same positive Ricci curvature lower bound. This implies the spectral comparison in dimension two. Together with the recent counterexample in dimension three by Lin, Wang, and Xu [LWX26], this settles the spherical spectral comparison in every dimension $d\geq2$. Using the same method, we also construct a contracting transport from the uniform probability measure on a hemisphere onto the normalized uniform measure on any geodesically convex subset of positive volume, thereby answering affirmatively the remaining case of a conjecture by Beck and Jerison [BJ21] following the work of Fathi, Fradelizi, Gozlan, and Zugmeyer [FFGZ26].
Forward citations
Cited by 2 Pith papers
-
Contraction Maps Generated by Inverse Mean Curvature Flow
Inverse mean curvature flow produces 1-Lipschitz measure-preserving maps from the round sphere onto every smooth two-sphere of Gaussian curvature at least one, resolving E. Milman's contraction conjecture in dimension two.
-
Geometric obstructions to Lipschitz transport between weighted Hessian $\mathrm{CD}(\kappa,\infty)$ manifolds
Constructs CD(1/2,∞) manifold on R² without Lipschitz transport from centered Gaussian and proves its weighted Laplacian eigenvalues are asymptotically negligible compared to the Gaussian case.
Reference graph
Works this paper leans on
-
[1]
Homotopy groups of the moduli space of metrics of positive scalar curvature.Geometry & Topology, 14(4):2047–2076,
[BHSW10] Boris Botvinnik, Bernhard Hanke, Thomas Schick, and Mark Walsh. Homotopy groups of the moduli space of metrics of positive scalar curvature.Geometry & Topology, 14(4):2047–2076,
-
[2000]
On optimal transport maps between 1/d-concave densities.arXiv preprint arXiv:2404.05456,
[CFS24] Guillaume Carlier, Alessio Figalli, and Filippo Santambrogio. On optimal transport maps between 1/d-concave densities.arXiv preprint arXiv:2404.05456,
-
[2004]
[GS25] Yuxin Ge and Jordan Serres. A generalization of caffarelli’s contraction theorem to nearly spherical manifolds.arXiv preprint arXiv:2512.01496,
-
[2024]
Moment sequences of beta distribution.arXiv preprint arXiv:2112.13420,
[Sza21] Pawe l J Szab lowski. Moment sequences of beta distribution.arXiv preprint arXiv:2112.13420,
Reviewed August 4, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.