Sharp Endpoint Eigenfunction Estimates for the Two-Dimensional Hermite Operator

Cheng Zhang; Guiyu Xie

REVIEW 1 major objections 4 minor 22 references

Sharp Endpoint Eigenfunction Estimates for the Two-Dimensional Hermite Operator

T0 review · 1 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read For the two-dimensional Hermite operator, the spectral projection at eigenvalue λ=2N+2 maps L² to L^{10/3} with norm O(λ^{-1/10}) and no logarithmic factor.

desk verdict Independent proof of a theorem already proved elsewhere; the argument is mostly solid but has an unverified piece-counting claim in Section 2. read the letter →

arxiv 2608.08423 v1 pith:U4BERXR7 submitted 2026-08-09 math.AP math.CAmath.SP

Guiyu Xie , Cheng Zhang This is my paper · ORCID

classification math.APmath.CAmath.SP MSC 42B9942C10

keywords HermiteoperatorspectralprojectionendpointeigenfunctionestimatescriticalexponentLiouville–GreenrepresentationvanderCorputweightedTT*argumentL^{10/3}bounds

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

The paper establishes the sharp endpoint eigenfunction estimate for the Hermite operator H=-Δ+|x|² on R²: for every eigenvalue λ=2N+2, the spectral projection Π_λ satisfies ||Π_λ f||_{$L^{{10/3}}$(R²)} ≤ C $λ^{{-1/10}}$ ||f||_{L²(R²)}. The exponent $λ^{{-1/10}}$ is optimal; the highest-angular-momentum eigenfunction shows that it cannot be improved. The earlier annulus-by-annulus approach gave only $λ^{{-1/10}}$(log λ)^{3/10} at the critical exponent, and the paper removes the logarithmic factor by coupling all radial scales rather than summing dyadic annuli independently. A weighted TT* argument, applied to the radial modes from the polar spectral decomposition, is the mechanism that performs the coupling.

What carries the argument

The central mechanism is the polar spectral decomposition g(r,θ)=(2π)^{-1/2} ∑_{m∈M_N} c_m R_{N,m}(r)$e^{{imθ}}$, with the mode-dependent turning parameter ν_m defined by ν_m(1-ν_m)=(m²-1/4)/λ². For non-glancing modes, a uniform Liouville–Green representation writes each radial mode as an amplitude times $e^{{±i S(u,m)}}$, and the curvature lemma says the second derivative of the phase difference between two radial scales is comparable to |t-s|. Feeding this curvature into a weighted discrete van der Corput estimate and then into a weighted TT* proposition couples all radial scales with kernel |t-s|^{-1/5}, bypassing the logarithmic loss from dyadic summation.

What would settle it

One concrete check: on a single turning-point annulus with μ between C_* $λ^{{-2/3}}$ and u₀, evaluate ||1_{A_in_μ} Π_λ f||_{$L^{{10/3}}$} for highly concentrated data f; if this ratio ever exceeds C $λ^{{-1/10}}$ ||f||₂ for arbitrarily large λ, the localized input fails and the proof cannot close. More directly, any sequence with ||Π_λ f||_{$L^{{10/3}}$} / ||f||₂ ≥ c $λ^{{-1/10+ε}}$ would disprove the theorem.

Watch

Extended reading notes

Core claim

On its own terms, the paper proves Theorem 1.1: there is an absolute constant C such that for every λ in 2N+2 and every f in L²(R²), the projection onto the λ-eigenspace obeys ||Π_λ f||_{$L^{{10/3}}$(R²)} ≤ C $λ^{{-1/10}}$ ||f||_{L²(R²)}. The proof treats the interior collar C_* $λ^{{-2/3}}$ << u << 1, where u=1-|x|²/λ, by expanding g=Π_λ f in the polar Laguerre-Hermite basis. Each angular mode has a turning point at u=ν_m; modes with u >> ν_m are non-glancing and are represented by a Liouville–Green oscillatory phase, then controlled by a weighted exponential-sum estimate. Modes near their turning point or deep in the forbidden region are handled by localized spectral projection bounds and by rapid decay. The five resulting estimates cover R² and combine to give the theorem. Sharpness is shown by the eigenfunction G_N(x₁,x₂)=(x₁+i x₂)^N $e^{{-|x|²/2}}$/√(π N!), whose $L^{{10/3}}$ norm is approximately $N^{{-1/10}}$.

Load-bearing premise

The proof depends on the localized spectral projection bounds taken as input: on each turning-point annulus of width comparable to $λ^{{-2/3}}$, the L²-to-$L^{{10/3}}$ norm is $λ^{{-1/10}}$, and on interior annuli the L²-to-L² norm is $μ^{{1/4}}$; if those localized bounds were weaker, the dyadic summation and near-mode control would not close.

Editorial extensions

If this is right

At the critical exponent p=10/3 in dimension two, the Hermite spectral projection has the sharp λ^{-1/10} decay with no logarithmic factor.
The highest-angular-momentum eigenfunction G_N is extremal up to constants, so the estimate cannot be improved.
The proof gives a scale-coupling route to the endpoint bound that does not require recursive space-time localization or iterative almost-orthogonality.
The exterior annulus estimate remains summable, and the five-region decomposition shows the full eigenspace projection decays at the endpoint rate.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Inference: Because the TT* kernel estimate only uses curvature of the radial phase difference, the same argument could remove logarithmic losses at critical exponents for radial potentials with a single turning circle, such as |x|^{2k} generalizations of the Hermite operator.
Inference: The sharpness example concentrates on the turning circle, suggesting that improved local inequalities or restriction-type estimates for eigenfunctions may hold on that circle with the same λ^{-1/10} rate.
Inference: The boundedness of the angular Fourier multiplier via the M. Riesz theorem suggests the argument may extend to other orthonormal systems whose angular projections are uniformly bounded in L^p on the circle.
Inference: A testable extension is to perturb the Hermite operator by a bounded potential; if the localized annulus bounds remain true with the same powers, the same coupling would give the endpoint estimate for the perturbed eigenfunctions.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 main thing you should know: the theorem is not new. Jeong–Lee–Ryu already proved the same sharp endpoint bound in a concurrent preprint, and the authors acknowledge this up front. What this paper offers is an independent proof strategy — polar decomposition, a Liouville–Green representation, discrete van der Corput, and a weighted TT* argument that couples radial scales. That is a genuine contribution, even if the result itself is known.

The proof is careful and unusually complete. The Liouville–Green lemma is proved with explicit error bounds, the forbidden-region decay is worked out with the right exponents, and the weighted TT* proposition is stated cleanly and proven. The external Koch–Tataru input is explicitly recorded, and the authors do not hide their dependence on it. All of that is good practice.

The most serious question is in Section 2. Proposition 2.1 converts the Koch–Tataru ℓ∞-over-pieces bound (2.2) into the full-annulus bounds (2.3)–(2.5), justified only by a sentence that each annulus meets a bounded number of pieces. The text never defines the pieces or proves this claim. For the turning-point set {|u|≤Au*}, a natural piece decomposition would give many pieces along the circle, and then an ℓ^p summation would multiply the bound by a positive power of λ — potentially ruining the endpoint. I suspect the claim may be true if the pieces are chosen with polar symmetry or if the ℓ∞_ρ norm is taken over pieces that are the whole annuli, but the paper does not say so. This needs to be fixed before the proof can be accepted.

There is also a minor sign error in Lemma 4.2: the second derivative of the phase difference has the sign of s−t, not t−s. That is harmless because only constant sign is used, but it should be corrected. The definition of r(u) seems clear to me (it is given in Section 3), though repeating it near first use would help the reader.

Who is this for? People working on eigenfunction estimates and harmonic analysis will find the technique valuable. It deserves a serious referee — not because the theorem is new, but because the independent proof is substantive and the Section 2 issue is exactly the kind of thing a knowledgeable referee should check. I would send it to a referee who knows the Koch–Tataru paper and the concurrent Jeong–Lee–Ryu work.

Referee Report

1 major / 4 minor

Summary. The paper proves the sharp log-free endpoint estimate for two-dimensional Hermite spectral projections: for λ = 2N+2 and every f in L²(R²), the projection Π_λ satisfies ∥Π_λ f∥_{L^{10/3}} ≲ λ^{-1/10}∥f∥_{L²}. The proof uses a polar-coordinate spectral decomposition, the Koch–Tataru localized eigenfunction bounds, a uniform Liouville–Green representation for radial modes away from turning points, a curvature estimate for the resulting phases, a weighted van der Corput exponential-sum lemma, and a weighted TT* argument over radial scales. The remaining modes are controlled by the Koch–Tataru annulus bounds and by rapid decay in the forbidden region. The paper also shows sharpness via the highest-angular-momentum eigenfunction. The main theorem is the same result already announced in the contemporaneous preprint [10], but the proof strategy here is independent and structurally different.

Significance. If the proof is correct, the paper gives a clean resolution of the two-dimensional endpoint eigenfunction problem by a different mechanism from the asymmetric-localization approach of Jeong–Lee–Ryu, and the weighted TT* argument in Proposition 5.2 is potentially reusable. The proof is largely self-contained beyond the Koch–Tataru input: the Liouville–Green representation, the forbidden-region decay, the exponential-sum estimates, and the TT* coupling are all explicit, with universal constants and no fitted parameters. However, the derivation of the full-annulus bounds from the Koch–Tataru estimate (2.2) is not sufficiently justified as written, and this step is load-bearing for the rest of the paper. The novelty overlap with [10] should also be weighed, though the independent derivation is a point in the manuscript's favor.

major comments (1)

[Section 2, Proposition 2.1 and (2.2)–(2.8)] Proposition 2.1 is derived from the Koch–Tataru estimate (2.2) by asserting that each annulus A_in_μ, A_out_μ, and the set {|u|≤Au*} meets only a bounded number of pieces of the Koch–Tataru turning-point decomposition, but the pieces are never defined. This is not a purely cosmetic issue: if the pieces are the standard ρ^{-1/3} Cartesian cubes (the literal reading of ℓ∞_ρ L^p), then an annulus A_in_μ with μ≫u* has y-width λ^{2/3}μ and contains roughly λ^{5/6}μ cubes across its radial thickness, so the displayed bounds (2.3)–(2.5) would gain a positive power of λ after summation over pieces, and the exterior summability in (2.9) would not close. If instead the pieces are dyadic annuli in the variable y, or are the full annuli themselves, the bounded-overlap claim is true. The manuscript must either quote from [12] the precise theorem that yields the full-annulus bounds directly, or define the pieces and prove the bounded-overlap assertion. Since estimates (2.3)–(2.8) are used in (2.9), in (7.4), and in the final covering argument in Section 8, this gap is load-bearing for Theorem 1.1.

minor comments (4)

[Section 3, after (3.5)] The definition r(u)=√λ(1-u) conflicts with u=1-|x|²/λ and with the factorization (3.5). The correct relation is r(u)=√λ(1-u)^{1/2}; the later computation |rdr|=(λ/2)du in Section 6 is consistent with the square-root definition but not with r(u)=√λ(1-u). This appears to be a typo, but it should be corrected because the phase integrals and the change of variables depend on it.
[Lemma 4.2] The sign assertion in Lemma 4.2 is wrong: for t<s, the second derivative ∂²_ξ(S(u_t,ξ)-S(u_s,ξ)) is positive, not negative, because ∂²_ξ p_ξ<0 and r(u) is decreasing. The magnitude bound |∂²|≈|t-s| is correct, and the sign is immaterial for the subsequent van der Corput application since Lemma 5.1 only requires a constant sign. Still, the statement 'this derivative has the sign of t−s' and the last sentence of the proof should be corrected.
[Section 2, proof of Proposition 2.1] The sentence 'Here the middle estimate follows because the indicated set meets only O(1) turning-point pieces' explicitly justifies bounded overlap only for {|u|≤Au*}. The first and third displayed estimates in the proof are obtained by the same type of restriction, so the manuscript should either give the analogous justification for A_in_μ and A_out_μ or state that these are the Koch–Tataru annulus estimates themselves.
[Throughout] The notation ℓ∞_ρ L^p is introduced but the parameter ρ and the pieces of the decomposition are not precisely defined in the manuscript. A short summary of the relevant Koch–Tataru definitions, or at least a precise quotation of [12, Theorem 3(a)], would remove a significant obstacle for the reader.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: Theorem 1.1 is derived from the external Koch–Tataru localized bounds plus a self-contained weighted TT* argument; no step assumes the target estimate.

full rationale

The derivation chain is self-contained relative to a genuine external input. Section 2 imports Koch–Tataru [12, Thm 3(a)-(b)], an independent prior result not by the authors, and Proposition 2.1 converts the ℓ∞-over-pieces bound (2.2) into full-annulus bounds; whether that conversion is correct is a mathematical question, but it is not circular because (2.2) does not contain the endpoint estimate (1.1) and the paper does not fit parameters or rename the conclusion. Sections 4–7 build the endpoint bound from the Liouville–Green representation, Lemma 5.1, and Proposition 5.2; none of these invoke Theorem 1.1. Constants such as C*, u0, U1, U2, κ0, κ1 are fixed universal numbers chosen once, independent of λ and of the target estimate, so there is no fitted-input-called-prediction issue. The only self-citation, Wang–Zhang [22], appears in the introduction as context about sharp local Lp bounds and is not used in any proof; it is not load-bearing. The potentially delicate step flagged by a skeptic — that an annulus {1/2 μ ≤ u ≤ 2μ} meets only O(1) Koch–Tataru pieces — is an inference from the geometry of the turning-point decomposition, not a circular reduction: even if the overlap count were wrong, the failure would be an incorrect application of an external theorem, not an assumption of the conclusion. No uniqueness theorem is invoked, no cosh ansatz is smuggled via citation, and no known empirical result is renamed. Hence the appropriate circularity score is 0.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The proof rests on fixed universal constants selected for structural inequalities (not fitted to the target bound) and on standard theorems. No new entities are invented and no parameters are fit to data; the exponent -1/10 is derived from the L^p calculation, not imposed.

free parameters (1)

Fixed collar and turning-point constants u0, U1, U2, eta0, kappa0, kappa1, c1, C*, lambda* = not fitted; arbitrary positive constants chosen to satisfy structural inequalities (Section 2, Eq. (2.1))
These constants define the annuli, the mode classes, and the eigenvalue threshold. They are chosen by hand to make bounded-overlap and decay estimates uniform, but they never enter the final bound and are not tuned to the target estimate.

assumptions (5)

domain assumption Koch-Tataru localized spectral projection estimates (Section 2, Eq. (2.2); [12, Theorem 3(a,b)])
Control of localized, turning-point, interior, and exterior contributions. The proof uses these as black boxes and does not verify them.
standard math Discrete van der Corput estimate for exponential sums ([8, Chapter 2, Theorem 2.2])
Used in Lemma 5.1 to bound exponential sums with convex phase.
standard math M. Riesz theorem for L^p boundedness of interval Fourier multipliers on the circle ([7, Prop 4.1.6 and Thm 4.1.7])
Used in Proposition 7.1 to control the angular projection operators Q_u.
standard math Liouville-Green (WKB) asymptotics with error bounds ([16, Chapter 6], [17])
Used in Lemma 4.1 to represent radial modes in the allowed region.
standard math Laguerre polynomial orthogonality ([18, (5.1.1)]) and Hecke-Bochner formula ([20, Theorem 3.4.1])
Used in Lemma 3.1 to build the polar orthonormal basis.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Sharp Endpoint Eigenfunction Estimates for the Two-Dimensional Hermite Operator." pith.science (2026). https://pith.science/paper/U4BERXR7

@misc{pith2026260808423,
  author       = {Pith},
  title        = {Pith review of: Sharp Endpoint Eigenfunction Estimates for the Two-Dimensional Hermite Operator},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U4BERXR7}},
  note         = {Machine review of arXiv:2608.08423}
}

abstract

Let $\mathcal H=-\Delta+|x|^2$ be the Hermite operator on $\mathbb R^2$, and let $\Pi_\lambda$ denote the spectral projection corresponding to $\lambda=2N+2$. We prove the sharp log-free endpoint estimate $||\Pi_\lambda||_{L^2(\mathbb R^2)\to L^{10/3}(\mathbb R^2)}\lesssim\lambda^{-1/10}$. The proof uses a spectral decomposition in polar coordinates and combines Koch-Tataru localized spectral projection bounds with a Liouville-Green representation, van der Corput estimates for exponential sums, and a weighted $TT^*$ argument across radial scales.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 21 canonical work pages

[10]

The Endpoint Eigenfunction Bound for the Hermite Operator in Two Dimensions

E. Jeong, S. Lee and J. Ryu,The endpoint eigenfunction bound for the Hermite operator in two dimensions, arXiv:2607.25859

work page Pith review arXiv
[12]

Koch and D

H. Koch and D. Tataru,Lp eigenfunction bounds for the Hermite operator, Duke Math. J.128(2005), no. 2, 369–392

work page 2005
[1]

P. Chen, X. T. Duong, D. He, S. Lee and L. Yan,Almost everywhere convergence of Bochner–Riesz means for the Hermite operators, Adv. Math.392(2021), Paper No. 108042, 42 pp

work page 2021
[2]

P. Chen, S. Lee, A. Sikora and L. Yan,Bounds on the maximal Bochner–Riesz means for elliptic operators, Trans. Amer. Math. Soc.373(2020), no. 6, 3793–3828

work page 2020
[3]

P. Chen, J. Li, L. Ward and L. Yan,Weak-type endpoint bounds for Bochner–Riesz means for the Hermite operator, C. R. Math. Acad. Sci. Paris360(2022), 111–126

work page 2022
[4]

P. Chen, X. Lin and L. Yan,The commutators of Bochner–Riesz operators for Hermite operator, J. Geom. Anal.33(2023), no. 3, Paper No. 87, 19 pp

work page 2023
[5]

Escauriaza,Carleman inequalities and the heat operator, Duke Math

L. Escauriaza,Carleman inequalities and the heat operator, Duke Math. J.104(2000), no. 1, 113–127

work page 2000
[6]

Escauriaza and L

L. Escauriaza and L. Vega,Carleman inequalities and the heat operator. II, Indiana Univ. Math. J.50(2001), no. 3, 1149–1169

work page 2001

Show all 22 references

[7]

Grafakos,Classical Fourier Analysis, third edition, Graduate Texts in Mathematics, Vol

L. Grafakos,Classical Fourier Analysis, third edition, Graduate Texts in Mathematics, Vol. 249, Springer, New York, 2014

2014
[8]

S. W. Graham and G. Kolesnik,Van der Corput’s Method of Exponential Sums, LMS Lecture Note Series, Vol. 126, Cambridge University Press, 1991

1991
[9]

Jeong, S

E. Jeong, S. Lee and J. Ryu,Endpoint eigenfunction bounds for the Hermite operator, J. Eur. Math. Soc.28(2026), no. 3, 1313–1352

2026
[11]

G. B. Karadzhov,Riesz summability of multiple Hermite series inLp spaces, C. R. Acad. Bulgare Sci.47(1994), 5–8

1994
[13]

Koch and D

H. Koch and D. Tataru,Carleman estimates and unique continuation for second order parabolic equations with nonsmooth coefficients, Comm. Partial Differential Equations 34(2009), no. 4–6, 305–366

2009
[14]

Lee and J

S. Lee and J. Ryu,Bochner–Riesz means for the Hermite and special Hermite expan- sions, Adv. Math.400(2022), Paper No. 108260, 52 pp

2022
[15]

Markett,Norm estimates for(C,δ )means of Hermite expansions and bounds for δeff, Acta Math

C. Markett,Norm estimates for(C,δ )means of Hermite expansions and bounds for δeff, Acta Math. Hungar.43(1984), 187–198

1984
[16]

F. W. J. Olver,Asymptotics and Special Functions, Academic Press, New York, 1974

1974
[17]

F. W. J. Olver,Error bounds for the Liouville–Green (or WKB) approximation, Math. Proc. Cambridge Philos. Soc.57(1961), no. 4, 790–810

1961
[18]

Szegő,Orthogonal Polynomials, fourth edition, AMS Colloquium Publications, Vol

G. Szegő,Orthogonal Polynomials, fourth edition, AMS Colloquium Publications, Vol. 23, 1975

1975
[19]

Thangavelu,Multipliers for Hermite expansions, Rev

S. Thangavelu,Multipliers for Hermite expansions, Rev. Mat. Iberoam.3(1987), 1–24

1987
[20]

Thangavelu,Lectures on Hermite and Laguerre Expansions, Mathematical Notes, Vol

S. Thangavelu,Lectures on Hermite and Laguerre Expansions, Mathematical Notes, Vol. 42, Princeton University Press, 1993

1993
[21]

Thangavelu,Hermite and special Hermite expansions revisited, Duke Math

S. Thangavelu,Hermite and special Hermite expansions revisited, Duke Math. J.94 (1998), no. 2, 257–278

1998
[22]

Wang and C

X. Wang and C. Zhang,Sharp localLp estimates for the Hermite eigenfunctions, Trans. Amer. Math. Soc.378(2025), no. 11, 8223–8258. Yau Mathematical Sciences Center, Tsinghua University, Beijing, 100084, P.R. China Email address:xiegy25@mails.tsinghua.edu.cn; czhang98@tsinghua.edu.cn

2025

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [10]

The Endpoint Eigenfunction Bound for the Hermite Operator in Two Dimensions

E. Jeong, S. Lee and J. Ryu,The endpoint eigenfunction bound for the Hermite operator in two dimensions, arXiv:2607.25859

work page Pith review arXiv

[2] [12]

Koch and D

H. Koch and D. Tataru,Lp eigenfunction bounds for the Hermite operator, Duke Math. J.128(2005), no. 2, 369–392

work page 2005

[3] [1]

P. Chen, X. T. Duong, D. He, S. Lee and L. Yan,Almost everywhere convergence of Bochner–Riesz means for the Hermite operators, Adv. Math.392(2021), Paper No. 108042, 42 pp

work page 2021

[4] [2]

P. Chen, S. Lee, A. Sikora and L. Yan,Bounds on the maximal Bochner–Riesz means for elliptic operators, Trans. Amer. Math. Soc.373(2020), no. 6, 3793–3828

work page 2020

[5] [3]

P. Chen, J. Li, L. Ward and L. Yan,Weak-type endpoint bounds for Bochner–Riesz means for the Hermite operator, C. R. Math. Acad. Sci. Paris360(2022), 111–126

work page 2022

[6] [4]

P. Chen, X. Lin and L. Yan,The commutators of Bochner–Riesz operators for Hermite operator, J. Geom. Anal.33(2023), no. 3, Paper No. 87, 19 pp

work page 2023

[7] [5]

Escauriaza,Carleman inequalities and the heat operator, Duke Math

L. Escauriaza,Carleman inequalities and the heat operator, Duke Math. J.104(2000), no. 1, 113–127

work page 2000

[8] [6]

Escauriaza and L

L. Escauriaza and L. Vega,Carleman inequalities and the heat operator. II, Indiana Univ. Math. J.50(2001), no. 3, 1149–1169

work page 2001