Neumann Data Mass on Perturbed Triangles

Hans Christianson; Jin Xi

REVIEW 3 major objections 3 minor 9 references

Neumann Data Mass on Perturbed Triangles

T0 review · 3 major / 3 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 The exact triangle identity for Neumann data mass is stable: an epsilon-small bump on one side, or an epsilon-small potential inside, changes each side's boundary L2 mass from side-length over area by at most O(epsilon), uniformly in…

desk verdict Solid extension of an exact equidistribution result; acute case works, but the obtuse case has a real gap in the key lemma and Theorem 2 has a spectral assumption issue. read the letter →

arxiv 1908.02863 v1 pith:FFRYUAI4 submitted 2019-08-07 math.AP math.SP

Hans Christianson , Jin Xi This is my paper · ORCID

classification math.APmath.SP MSC 35P2058J5135J25

keywords NeumanndatamassDirichleteigenfunctionssemiclassicalanalysisperturbedtrianglesequidistributiononboundarytracescommutatormethodsmallpotentialperturbations

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 proves that the exact equidistribution of Neumann data on triangles - each side's boundary $L^2$ mass of the semiclassical normal derivative equals the side's length divided by the area - survives small perturbations of both the domain and the operator. If one side of a triangle is replaced by a smooth curve that is $\epsilon$-close to the original side in $C^1$ norm, then every normalized Dirichlet eigenfunction of $-h^2\Delta-1$ on the new domain satisfies $\int_A |h\partial_\nu u|^2\,dS = a/\operatorname{Area}(D)+O(\epsilon)$, with the same statement on the other sides. The same conclusion holds when an $\epsilon$-small potential is added to the Laplacian on a fixed triangle. The error constant is independent of $h$ and of the particular perturbation, so the law is not an artifact of straight edges.

What carries the argument

The central mechanism is the commutator identity $[-h^2\Delta-1,X]=-2h^2\Delta$ for the affine vector field $X=(x+m)\partial_x+(y+n)\partial_y$. Because the eigenfunction is normalized, this identity converts the interior norm into the boundary identity $2=\int_{\partial D}(hXu)(h\partial_\nu u)\,dS$. The Dirichlet condition makes the tangential part of $hXu$ vanish, so on each side $hXu$ becomes a known multiple of $h\partial_\nu u$; matching the three sides yields a $3\times 3$ linear system for the three masses $I_A,I_B,I_{C'}$. The smallness assumptions on $g$ and $g'$ enter as $O(\epsilon)$ remainder terms, and Lemma 2.1, proved by separate vector fields $y\partial_y$ and $\psi(x)\partial_x$, provides the uniform bound on $I_{C'}$ needed to close the estimates.

What would settle it

Fix a right triangle and set $g(x)=\epsilon\sin(\pi x/l)$ on one side, then compute $I_A,I_B,I_{C'}$ for a sequence of normalized Dirichlet eigenfunctions as $h\to 0$ at one small $\epsilon$. If for some eigenfunction $I_{C'}$ grows without bound in $h$, or if any of the three integrals deviates from the corresponding side length divided by the area by more than a fixed multiple of $\epsilon$, then Lemma 2.1 or the uniformity in Theorem 1 is false; a numerical search over low-lying eigenfunctions would settle it.

Watch

Extended reading notes

Core claim

On a triangle with one side $C$ replaced by the graph $f(x)=a_2 x/l+g(x)$, where $|g|,|g'|\le\epsilon$ and $g(0)=g(l)=0$, the paper proves that for the Dirichlet eigenfunction problem $(-h^2\Delta-1)u=0$ with $\|u\|_{L^2}=1$, the Neumann data masses obey $I_A=a/\operatorname{Area}(D)+O(\epsilon)$, $I_B=b/\operatorname{Area}(D)+O(\epsilon)$, and $I_{C'}=\ell(C')/\operatorname{Area}(D)+O(\epsilon)$, where $I_\cdot=\int_\cdot|h\partial_\nu u|^2\,dS$ and the constants are uniform in $h$ and in the perturbation. For a fixed triangle, the same equalities hold up to $O(\epsilon)$ when the Laplacian is replaced by $-h^2\Delta+w_\epsilon$ with $|w_\epsilon|,|\nabla w_\epsilon|\le\epsilon$. Thus the exact triangle identity is stable: small geometric or operator perturbations change each side's boundary mass only by the size of the perturbation, never by a factor that grows as the frequency goes up.

Load-bearing premise

The whole proof rests on Lemma 2.1: the integral over the perturbed side $C'$ of $|h\partial_\nu u|^2$ must stay bounded by a constant independent of $h$ and $\epsilon$; if that bound fails, the $O(\epsilon)$ errors in the linear system cannot be controlled and the equidistribution formulas collapse.

Editorial extensions

If this is right

The exact triangle identity extends to a neighborhood of triangles in the $C^1$ topology on the boundary, so the formula is stable under small deformations, not just for polygons.
Because the error is uniform in $h$, the result applies to every eigenfunction in the whole spectrum, including high-frequency modes where boundary concentration might otherwise be expected.
Adding an $\epsilon$-small potential does not break the law, so the same equidistribution holds for perturbed Hamiltonians $-h^2\Delta+w_\epsilon$; the formula is robust for operator perturbations as well as domain perturbations.
The proof supplies explicit bounds from which the $O(\epsilon)$ constants could in principle be extracted from the side lengths and the a priori constant $\Gamma$ of Lemma 2.1.
The authors expect the analogous statement in all dimensions $n\ge3$ for simplices with perturbed faces, following the same commutator scheme.

Reading between the lines

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

A direct consequence the authors leave implicit: because the three boundary masses are measurable, one can recover the area and the three side lengths of an unknown near-triangular domain up to $O(\epsilon)$ from Neumann data alone, a stable inverse problem.
The mechanism is the commutator identity for a linear scaling vector field, so the result should extend to any domain with three boundary pieces on which the Dirichlet condition turns $Xu$ into a tangential multiple; a polygon with more sides would likely give the analogous statement with side lengths weighted by the $C^1$ mismatch.
Only $|g|,|g'|\le\epsilon$ are used outside the cut-off argument in Lemma 2.1, so perturbations with large second derivative but small height and slope may probe whether curvature is hidden in the boundedness lemma.
If the $O(\epsilon)$ error is not just an estimate artifact, the boundary mass distribution on a slightly non-triangular domain could serve as a sensitive probe of the domain's area and side geometry, with no need to know individual 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

Colleague,

This paper takes Christianson's exact equidistribution identity for Dirichlet eigenfunctions on triangles and asks whether it survives two small perturbations: deforming one side by O(ε) and adding an O(ε) potential. The answer is the expected one—the per-side Neumann data mass stays within O(ε) of length/area—and the proof is mostly a direct commutator computation in the spirit of the original. That is a genuine, if modest, new result; I do not see it in the cited papers.

The acute-triangle case is worked out in detail and looks correct. The bookkeeping with the vector field X=(x+m)∂x+(y+n)∂y, the derivation of equations (2.1)–(2.3), and the O(ε) control on the C′ side all hold together. The explicit independence of the error from h is a nice touch.

The trouble is in the obtuse case of Lemma 2.1, which supplies the a priori bound ∫_{C′} |h∂νu|² dS ≤ Γ that everything else leans on. The proof for obtuse triangles uses X=(y−a2/l x)∂y. On side B, y=a1/l x, so Xu·h∂νu = −(a1−a2)x/b |h∂νu|². If a1>a2, that term is negative, while the C′ term is only O(ε) (since Xu on C′ is g(x)∂y, with |g|≤ε). The commutator identity only gives an O(1) bound on the whole boundary sum, so a large positive C′ integral could be cancelled by the negative B term. The remark in the text that "the same argument as in the acute case" works does not address this. A likely fix is to choose X=(y−a1/l x)∂y, which kills the B term and leaves a positive C′ term, but that choice is not made. So the proof of the load-bearing lemma is incomplete as written for at least some obtuse triangles.

A second, smaller issue is Theorem 2. The statement assumes there are normalized eigenfunctions of −h²Δ+w_ε with eigenvalue exactly 1. For a fixed h and generic potential this is not automatic; one would need to work with spectral windows or quasi-modes. The proof uses only |w|<ε and |∇w|<ε, so the gap is probably easy to repair, but it is not just notational.

There are also a few typos and garbled display equations (e.g., in Section 4 the expression for f′ looks off), but nothing that suggests dishonesty or circular reasoning. The paper is an honest, competent extension of an existing program.

Who is it for: spectral geometers working on eigenfunction restrictions and boundary traces. With a corrected Lemma 2.1 and a clarified spectral assumption, it would be a solid contribution. I would send it to peer review rather than desk reject, but I would insist on the obtuse case being fixed. For my own work, I would not cite the current version yet.

Referee Report

3 major / 3 minor

Summary. The paper claims two stability results for Christianson's exact Neumann data mass equidistribution on triangles. Theorem 1 replaces one side of a triangle by a smooth curve whose graph distance from the original side is at most epsilon and asserts that, for every normalized Dirichlet eigenfunction of -h^2 Delta - 1, the integral of |h d_nu u|^2 over each side equals the side length divided by the area of the perturbed domain, up to O(epsilon), with constants uniform in h and in the perturbation. Theorem 2 proves the same form of equidistribution, up to O(epsilon), when an epsilon-small potential is added to the Laplacian on a fixed triangle. The proof is based on commutators with affine vector fields and on an a priori bound, Lemma 2.1, for the Neumann data on the perturbed side.

Significance. If the proof is completed, this is a genuinely useful stability statement: the exact triangle identity from [Chr17] is shown to be robust under small boundary and potential perturbations, uniformly in the spectral parameter h and in the perturbation. The argument is elementary, self-contained apart from the zero-perturbation baseline, and no quantity in the conclusion is used as an input, so the derivation is not circular. The main value is as a benchmark for how rigid the triangle equidistribution law is, and the methods could plausibly extend to higher dimensions. The significance is moderate but solid for a specialized analysis paper.

major comments (3)

[Section 4, obtuse case] In the obtuse case of Lemma 2.1, the vector field X=(y-(a2/l)x)d_y does not have the positivity property used in the acute case. On the side B, parametrized by y=(a1/l)x, the boundary contribution in (4.1) is ((a2-a1)x/b)|h d_nu u|^2 dS, which is negative when a1>a2, while the C' contribution f(x) gamma^{-1}|h d_nu u|^2 is positive for x>=delta. The commutator identity bounds only the signed sum of the boundary terms, so the conclusion int_{C' cap {x>=delta}} |h d_nu u|^2 = O(1) does not follow for all obtuse triangles as written. A likely correction is X=(y-(a1/l)x)d_y, which vanishes on B and makes the C' term positive, but this is not the vector field used in the manuscript.
[Section 4, obtuse case] The vector field Y=psi(x)(d_x+(a2/l)d_y) is asserted to be tangential to the side B, but B has slope a1/l, so a tangential field is d_x+(a1/l)d_y. As written, Yu does not vanish on B. The subsequent computation on C' uses hYu=psi(x)(-f'/gamma + a1/(l gamma))h d_nu u, which corresponds to the field d_x+(a1/l)d_y, not the displayed one. This inconsistency must be corrected and the boundary estimates rechecked, since the near-corner control of int_{C' cap {0<=x<=delta}} |h d_nu u|^2 depends on it.
[Section 3, proof of Theorem 2] After equation (3.2), the identity has the form 2+O(epsilon)(1+|m|+|n|) equal to the boundary sum, and the text says 'the rest of the proof proceeds exactly as the proof of Theorem 1.' However, deriving the analogues of (2.2) and (2.3) requires differentiating the identity with respect to m and n. Since the error term depends on m and n, one must justify that differentiation produces only O(epsilon) corrections. This is likely fixable by a direct first-order Taylor expansion in m and n, but it is not written and is load-bearing for Theorem 2.

minor comments (3)

[Section 2.3] The parametrization of C' in the obtuse case is written as f(x)=a2+a1/l x+g(x), which is inconsistent with the later derivative f'(x)=(a1+a2)/l+g'(x). Presumably f(x)=((a1+a2)/l)x+g(x), and the displayed formula should be corrected.
[Section 4, acute case] The notation in the proof of Lemma 2.1 is inconsistent: the slope of C' is a2/l+g'(x), but the text uses a/b in the lower-bound estimates, and the B-boundary coefficient is a1/b, not a2/b. Please define all symbols consistently so that the constants in (4.2) and the near-corner estimate can be verified.
[Remark 1.1] The remark says that g does not have to be of the form epsilon g_tilde, but the domain construction still requires g(0)=g(l)=0 for the perturbed side to close the domain; this endpoint condition should be stated explicitly for the more general g.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation starts from the eigenfunction equation and a commutator identity; previous work enters only as the epsilon=0 baseline.

full rationale

The paper's central claims (Theorems 1 and 2) are proved by taking the commutator of the semiclassical operator with the linear vector field X=(x+m)∂_x+(y+n)∂_y, integrating by parts, and expressing 2 (or 2+O(epsilon)) as a sum of boundary integrals over A, B, and C'. Differentiation in the free parameters m,n yields three independent equations, (2.1)-(2.3) in the acute case and (2.7)-(2.8) in the obtuse case, which are then solved for I_A, I_B, and I_C'. No fitted parameter is introduced and no target quantity is used as an input. The exact triangle identity from [Chr17] is invoked only to identify the constant 2/l with a/Area(T); the perturbed equalities are obtained by direct O(epsilon) estimates using |g|≤epsilon and |g'|≤epsilon, and the small-potential theorem is handled by the same commutator with additional O(epsilon) terms from w and Xw. The a priori bound in Lemma 2.1 is proved, not assumed, via commutators with y∂_y (acute case) and with (y-(a2/l)x)∂_y plus a cutoff (obtuse case). The skeptic note identifies a possible sign/cancellation gap in the obtuse-case proof of Lemma 2.1; even if that gap were real, it is a correctness defect in an auxiliary estimate, not a circularity, because the estimate is established independently of the theorem it supports. There is no self-citation chain used to forbid alternatives and no known result repackaged under new names.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The proof relies on standard commutator and Green identities plus the explicit smallness assumptions on g and w. There are no free parameters fitted to data. The main unstated input is the existence of eigenvalue-1 eigenfunctions after adding the potential, and the uniform a priori bound of Lemma 2.1.

assumptions (4)

standard math The paper uses Green's identity and integration by parts with Dirichlet boundary conditions throughout Sections 2 and 3.
Standard spectral theory background, invoked when converting the commutator integral into boundary integrals.
standard math For the linear vector field X = (x+m)d_x + (y+n)d_y, the commutator identity [-h^2 Delta - 1, X] = -2h^2 Delta holds.
A direct calculation used before equation (2.1) and in the proof of Theorem 2.
domain assumption The perturbed side function g satisfies g(0)=g(l)=0, |g| <= epsilon, and |g'| <= epsilon, and the potential w satisfies |w| <= epsilon and |grad w| <= epsilon.
These smallness assumptions are stated in Theorems 1 and 2 and are used for every O(epsilon) estimate in the proofs.
domain assumption The theorem assumes the existence of normalized Dirichlet eigenfunctions u_epsilon solving -h^2 Delta u_epsilon = u_epsilon in D_epsilon or (-h^2 Delta + w_epsilon) u_epsilon = u_epsilon in T.
For the perturbed domain this is standard for appropriate h. For the potential-perturbed operator, existence of eigenfunctions with eigenvalue exactly 1 is asserted but not justified; this is a gap in the statement.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Neumann Data Mass on Perturbed Triangles." pith.science (2026). https://pith.science/paper/FFRYUAI4

@misc{pith2026190802863,
  author       = {Pith},
  title        = {Pith review of: Neumann Data Mass on Perturbed Triangles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FFRYUAI4}},
  note         = {Machine review of arXiv:1908.02863}
}

abstract

Based on a previous paper [Chr17] on Neumann data for Dirichlet eigenfunctions on triangles, we extend the study in two ways. First, we investigate the (semi-classical) Neumann data mass on perturbed triangles. Specifically, we replace one side of a triangle by adding a smooth perturbation, and assume that the disparity between the perturbation and the original side is bounded by a small value $\epsilon$. Second, we add a small $\epsilon$ sized potential to the (semi-classical) Laplacian and see how the results change on triangles. In both cases, we find that the $L^2$ norm of Neumann data on each side is close to the length of the side divided by the area of the triangle, and the difference is dominated by $\epsilon$.

Figures

Figures reproduced from arXiv: 1908.02863 by the authors.

**Figure 2.** Setup for obtuse triangles. 1.1. History. The study of restrictions of eigenfunctions and the study of boundary traces is an old subject. In this very abbreviated history we just focus on some of the recent developments particularly relevant to the present work. Previous results on restrictions primarily focused on upper bounds. In general, it is difficult to separate the behaviour of the Dirichlet and Neumann dat… view at source ↗

**Figure 3.** The function ψ Thus we have [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

9 extracted references · 9 canonical work pages

[1]

N. Burq, P. G \'e rard, and N. Tzvetkov. Restrictions of the L aplace- B eltrami eigenfunctions to submanifolds. Duke Math. J. , 138(3):445--486, 2007

work page 2007
[2]

Equidistribution of N eumann data mass on triangles

Hans Christianson. Equidistribution of N eumann data mass on triangles. Proc. Amer. Math. Soc. , 145(12):5247--5255, 2017

work page 2017
[3]

Equidistribution of neumann data mass on simplices and a simple inverse problem

Hans Christianson. Equidistribution of neumann data mass on simplices and a simple inverse problem. Math. Res. Lett. to appear , 2018

work page 2018
[4]

Hans Christianson, Andrew Hassell, and John A. Toth. Exterior mass estimates and L^2 -restriction bounds for N eumann data along hypersurfaces. Int. Math. Res. Not. IMRN , (6):1638--1665, 2015

work page 2015
[5]

Toth, and Steve Zelditch

Hans Christianson, John A. Toth, and Steve Zelditch. Quantum ergodic restriction for C auchy data: interior que and restricted que. Math. Res. Lett. , 20(3):465--475, 2013

work page 2013
[6]

Ergodic properties of eigenfunctions for the D irichlet problem

Patrick G \'e rard and \'E ric Leichtnam. Ergodic properties of eigenfunctions for the D irichlet problem. Duke Math. J. , 71(2):559--607, 1993

work page 1993
[7]

Quantum ergodicity of boundary values of eigenfunctions

Andrew Hassell and Steve Zelditch. Quantum ergodicity of boundary values of eigenfunctions. Comm. Math. Phys. , 248(1):119--168, 2004

work page 2004
[8]

Toth and S

J.A. Toth and S. Zelditch. Quantum ergodic restriction theorems, i: interior hypersurfaces in domains with ergodic billiards. Annales Henri Poincar\' e , 13:599--670, 2012

work page 2012

Show all 9 references

[9]

Toth and Steve Zelditch

John A. Toth and Steve Zelditch. Quantum ergodic restriction theorems: manifolds without boundary. Geom. Funct. Anal. , 23(2):715--775, 2013

2013

Pith tools

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

[1] [1]

N. Burq, P. G \'e rard, and N. Tzvetkov. Restrictions of the L aplace- B eltrami eigenfunctions to submanifolds. Duke Math. J. , 138(3):445--486, 2007

work page 2007

[2] [2]

Equidistribution of N eumann data mass on triangles

Hans Christianson. Equidistribution of N eumann data mass on triangles. Proc. Amer. Math. Soc. , 145(12):5247--5255, 2017

work page 2017

[3] [3]

Equidistribution of neumann data mass on simplices and a simple inverse problem

Hans Christianson. Equidistribution of neumann data mass on simplices and a simple inverse problem. Math. Res. Lett. to appear , 2018

work page 2018

[4] [4]

Hans Christianson, Andrew Hassell, and John A. Toth. Exterior mass estimates and L^2 -restriction bounds for N eumann data along hypersurfaces. Int. Math. Res. Not. IMRN , (6):1638--1665, 2015

work page 2015

[5] [5]

Toth, and Steve Zelditch

Hans Christianson, John A. Toth, and Steve Zelditch. Quantum ergodic restriction for C auchy data: interior que and restricted que. Math. Res. Lett. , 20(3):465--475, 2013

work page 2013

[6] [6]

Ergodic properties of eigenfunctions for the D irichlet problem

Patrick G \'e rard and \'E ric Leichtnam. Ergodic properties of eigenfunctions for the D irichlet problem. Duke Math. J. , 71(2):559--607, 1993

work page 1993

[7] [7]

Quantum ergodicity of boundary values of eigenfunctions

Andrew Hassell and Steve Zelditch. Quantum ergodicity of boundary values of eigenfunctions. Comm. Math. Phys. , 248(1):119--168, 2004

work page 2004

[8] [8]

Toth and S

J.A. Toth and S. Zelditch. Quantum ergodic restriction theorems, i: interior hypersurfaces in domains with ergodic billiards. Annales Henri Poincar\' e , 13:599--670, 2012

work page 2012