Counterexamples to the Landis conjecture in dimensions three and higher

Paata Ivanisvili; Rupert L. Frank

Counterexamples to the Landis conjecture in dimensions three and higher

T0 review · 0 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper proves that the Landis conjecture fails for real-valued potentials in every dimension n≥3, by constructing nonzero solutions with decay at the borderline rate exp(-c|x|^{4/3}).

desk verdict A long, careful proof that the real-valued Landis conjecture fails in every dimension n≥3, with the n≥3 threshold genuinely explained by the tangent-sphere coloring. read the letter →

arxiv 2608.00802 v1 pith:OADUGUCD submitted 2026-08-01 math.AP math-phmath.MP

Rupert L. Frank , Paata Ivanisvili This is my paper · ORCID

classification math.APmath-phmath.MP MSC 35J1035B4035B60

keywords LandisconjectureuniquecontinuationatinfinitySchrödingerequationreal-valuedpotentialexponentialdecayharmonicreplacementtangentialfrequenciesdimensionthreshold

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 Landis conjecture is false for real-valued potentials in every dimension $n\ge 3$. It constructs a nonzero real smooth function $u$ on $\mathbb{R}^n$ and a bounded real potential $V$ such that $-\Delta u+Vu=0$ and $|u(x)|\le C_n\exp(-c_n|x|^{4/3})$. A sympathetic reader would care because this closes a long-standing gap: earlier counterexamples with this decay were complex-valued in two dimensions (and by products, in even dimensions), while a real-valued analogue was known only on a cylinder. The proof's architecture—a radial envelope modulated by tangent plane waves, with harmonic repair in the sparse region where the wave sum is small—locates the obstruction to dimension two in a geometric frequency-separation lemma.

What carries the argument

The load-bearing object is the approximate solution $f=Eq$, where $E=\exp(-\Phi)$ with $\Phi(r)\simeq \frac{3}{4}r^{4/3}$ and $q(x)=\sum_j \chi_j(x)\cos(A_j t_j\cdot(x-x_j))$ is a patched superposition of real plane waves. The frequencies $A_j t_j$ are frozen on macroscopic balls of radius $\kappa_2 A_j$; each $t_j$ is tangent to the sphere through $x_j$, and Lemma 11 assigns these tangent directions to overlapping patches with uniform separation modulo sign, a task possible only when the tangent sphere $S^{n-2}$ has positive dimension, i.e. $n\ge 3$. Two cancellations make the approximate equation work: $\Delta q+A^2q$ is uniformly bounded (radial cancellation) and $(x/A^2)\cdot\nabla q$ is uniformly bounded (tangential cancellation). Balancing the errors produced by the partition of unity and by freezing a tangent vector across a curved patch forces $a(r)\simeq r^{1/3}$, hence decay power $4/3$. The repair mechanism is the safe-quotient step: where $|q|<\tau$, the paper solves $-\Delta w=\Delta f$ in that sparse set and obtains $|w|\le (\tau/20)E L^2/A^2$; a radial mollification then makes $F$ smooth while preserving harmonicity in the small-value region.

What would settle it

Exhibit a nonzero real $C^\infty$ solution in $\mathbb{R}^2$ of $-\Delta u+Vu=0$ with bounded real $V$ and decay $\exp(-c|x|^{4/3})$; this would contradict the two-dimensional positive Landis-type theorem. Short of that, compute the local model for two overlapping patches in $n=2$ with equal weights and opposite tangent directions: $P(z)=\frac12\cos(\xi\cdot z)+\frac12\cos(\xi\cdot z+\pi)=0$ identically, which verifies that the separation condition (37) and the sublevel estimate of Proposition 21 fail exactly where the paper's dimensional restriction enters.

Watch

Extended reading notes

Core claim

The central claim, Theorem 1, is that for every $n\ge 3$ there exists a nonzero real function $u\in C^\infty(\mathbb{R}^n)$, a real function $V\in C^\infty(\mathbb{R}^n)\cap L^\infty(\mathbb{R}^n)$, and positive constants $C_n,c_n$ such that $-\Delta u+Vu=0$ and $|u(x)|\le C_n\exp(-c_n|x|^{4/3})$. The solution is built explicitly: the approximate solution is $f=Eq$, with radial envelope $E=e^{-\Phi}$ and an oscillating factor $q$ formed by patching real plane waves $\cos(A_j t_j\cdot(x-x_j))$ on a covering whose ball radii are proportional to $A(x)\simeq |x|^{1/3}$. The paper proves $|\Delta f|\le C E(1+L^6 A^{-4}|q|)$, then solves a Dirichlet problem on the sparse set $\{|q|<\tau\}$ to replace $f$ by a function $f^*$ that is harmonic there, and finally smooths on the microscopic scale $A^{-1}$ to obtain $F$ with $|\Delta F|\le C|F|$ and $|F|\le C E$. Setting $V=\Delta F/F$ where $F\neq 0$ and $V=0$ where $F$ is harmonic gives the bounded real potential. The key alternative is that $F$ is exactly harmonic near every small value of $q$ and bounded below by $c\tau E$ away from that region, so the ratio $\Delta F/F$ is uniformly bounded.

Load-bearing premise

The proof depends on Lemma 11, which colors the overlap graph with tangent directions separated modulo sign; if such a coloring did not exist, local superpositions of cosines could cancel identically—as they do in dimension two—and the uniform sublevel estimate that produces the sparse small-value region would fail.

Editorial extensions

If this is right

In every dimension $n\ge 3$, real-valued bounded potentials admit nonzero solutions with decay $\exp(-c|x|^{4/3})$, so the Landis conjecture in its real form is false there.
The exponent $4/3$ is the optimal borderline: a previously known theorem still forces $u=0$ if the solution decays like $\exp(-\tau |x|^{4/3})$ for arbitrarily large $\tau$, so the new example achieves the borderline rate rather than exceeding it.
The two-dimensional positive result (decay $\exp(-c|x|(\ln|x|)^{1/2})$ forces zero) is consistent: the new construction is blocked in $n=2$ by the absence of separated tangent directions on the zero-dimensional tangent sphere.
Because the constructed $u$ is real and nonzero, it is also a complex-valued solution; this settles the Landis question in odd dimensions $n\ge 3$ as well as even dimensions.

Reading between the lines

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

The appendix's scale balance suggests the same envelope-plus-tangential-wave architecture should realize every subcritical decay exponent $p<4/3$, not just the endpoint; the paper proves only $p=4/3$.
The harmonic-repair step is modular: any elliptic equation admitting an approximate solution with a sparse small-value set and a pointwise correction estimate could use the same Dirichlet-replacement-then-mollify recipe; the Schrödinger operator is one instance.
One quantitative question left implicit is how the constants $C_n,c_n$ and the sup-norm of $V$ grow with $n$; the geometric separation lemma works for all $n\ge 3$ but gives no explicit dimensional dependence.
A direct test of the dimensional threshold is to attempt the construction in $n=2$: the model superposition $\frac12\cos(\xi\cdot z)+\frac12\cos(\xi\cdot z+\pi)$ vanishes identically, so the sparse-small-value proposition cannot hold.

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

Here's the one-line: this is a real result, not a routine extension. It constructs, for every n≥3, a nonzero smooth real u and bounded real V satisfying -Δu+Vu=0 with decay exp(-c|x|^{4/3}). That falsifies the real-valued Landis conjecture in all dimensions n≥3, a gap that has been open since Meshkov's complex example. The proof is long but the architecture is clear: a radial envelope E=exp(-Φ), an oscillating factor q built from cosines on a variable-scale atlas, a harmonic repair on the set where q is small, and a final smoothing. I went through the main steps: Proposition 16 gives the approximate equation with the crucial tangential cancellation; Proposition 21 gives the sparse small-value set; Propositions 28, 31 and 37 handle the repair, the harmonic/nonvanishing alternatives, and the global envelope; the gluing of V in Theorem 1 works. I found no circular step: u is constructed first, V=Δu/u afterwards. The debt to Meshkov and to Filonov–Krymskii is explicit and honest.

The genuinely new ingredient is Lemma 11, the tangential-frequency coloring. It uses n≥3 to have a positive-dimensional tangent sphere, so a greedy coloring can separate frequencies on overlapping patches. This is also where n=2 fails—only one frequency pair modulo sign exists, so local terms can cancel identically, which matches the known positive 2D result. The dimensional threshold looks correct, not accidental.

Soft spots? The paper is long and not machine-checked; with this many estimates there is room for a subtle constant or index error. I checked the usual trouble spots—the uniform sublevel estimate, the weighted Poincaré inequality, the pointwise Dirichlet bound, the kernel derivative bounds—and they hold together. The appendix's scale-balance derivation of 4/3 is heuristic, but the proof does not rely on it; the parameters are fixed before the construction, not fit to the target. So the risk is a local typo, not a load-bearing flaw. The AI-tools acknowledgment is transparent, and the citation pattern is appropriate.

Who should read it: anyone in elliptic PDE or unique continuation. It deserves a serious referee. My recommendation: send it to peer review. If I were refereeing, I would ask for a careful check of Sections 4–7, but I would not desk-reject.

Referee Report

0 major / 5 minor

Summary. The paper constructs, for every n≥3, a nonzero smooth real-valued function u on R^n and a bounded smooth real potential V satisfying -Δu+Vu=0, with |u(x)|≤C_n exp(-c_n|x|^{4/3}). The construction builds an approximate solution f=Eq from a radial envelope E and a patched superposition of tangential plane waves whose frequencies are separated on overlapping patches; this separated-coloring step is where n≥3 enters. The set {|q|<τ} is shown to be sparse on the microscopic scale, a harmonic-repair step replaces f there, and a variable-radius radial smoothing yields a smooth F satisfying |ΔF|≤C|F| and |F|≤CE. The potential is then defined as V=ΔF/F, glued with V=0 on the region where F is harmonic. The proof is self-contained, the parameter ordering is explicit, and the main theorem gives a negative answer to the Landis conjecture for real-valued potentials in every dimension n≥3.

Significance. If correct, the theorem settles the long-standing real-valued Landis problem in all dimensions n≥3 and shows that the dimensional restriction is genuinely geometric, consistent with the known two-dimensional positive results. The paper is a substantial technical achievement: the proof is detailed and well structured, with the key mechanisms isolated as sparse sublevel estimates, harmonic repair, and variable-kernel smoothing. I found no circularity or post-hoc fitting; the 'safe quotient' gluing of V is legitimate. The potential weak point identified by the reader, Lemma 11 (the tangential-frequency coloring), withstands scrutiny: the forbidden-cap estimate and greedy coloring are sound, and the separation is inherited by the frozen local frequencies in Lemma 20. The paper is appropriate for a leading analysis journal.

minor comments (5)

[Outline] In the outline, the phrase 'Φ(r)∼3/4r^{4/3} for larger' appears to be a typo for 'for large r'; please correct it.
[Lemma 27] In the proof of Lemma 27, just after the display 'Lemma 26 yields the bounds', the constant is written 'Cloc_n' in the first integral with the subscript 'loc' dropped; the notation should be made consistent.
[Lemma 19] The phrase 'a real-analytic function that is not identically zero on any connected open set' is imprecise; the zero-set theorem should be quoted for a function that is not identically zero on a nonempty open set, or on a connected component of its domain.
[Lemma 26] In the proof of Lemma 26, the subharmonicity of H in B(x,R) is used without explicitly noting that the term g1_{D\setminus B(x,R)} vanishes on B(x,R); a one-sentence justification would improve readability, although the argument is correct.
[Sections 2-10] Given the number of parameters (L, ρ, θ, τ, σ) and the order in which they are fixed, a short summary table of dependencies would help the reader; the current ordering is consistent, but it is hard to track.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the construction is self-contained and no prediction reduces to a fitted input.

full rationale

The paper constructs u first as a smooth function via an explicit radial envelope E and a localized plane-wave superposition q, and only afterwards defines the potential by V = ΔF/F after proving pointwise bounds |ΔF| ≤ C*|F| and |F| ≲ E. This is a standard reverse-engineered potential, not a fitted input called a prediction: no parameter is calibrated to the target decay rate, and the exponent 4/3 is fixed by explicit definitions (A = (L^6 + r^2)^{1/6}, Φ = (3/4)(A^4 − L^4)) and then verified, rather than fitted to the conclusion. The only external input is the Filonov–Krymskii construction on R × T^d, cited as a precursor whose scheme is re-implemented with new proofs (Lemmas 22–36); no load-bearing claim is imported through self-citation, and no uniqueness theorem from the authors' prior work is invoked. The dimensional separation step (Lemma 11) is proved in-paper from the measure estimate Lemma 10, and the sparseness estimate (Proposition 21) follows from compactness and linear independence of separated frequencies, all established directly. The harmonic-repair, smoothing, and gluing steps are carried out with explicit estimates rather than by assuming the conclusion. Thus no step in the derivation reduces to its own inputs, and there is no circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The theorem is an existence proof, so no empirical constants are fitted to data. The auxiliary parameters L, rho, theta, tau, sigma, and delta_n are chosen to satisfy explicit inequalities and depend only on dimension. The recurring input from outside the paper is a set of standard theorems from elliptic PDE and real analysis, plus the geometric assumption n>=3 that makes the tangential coloring possible.

free parameters (6)

L = fixed large, depending only on n; no numeric value
Enters A(r)=(L^6+r^2)^{1/6} to regularize the radial scale at the origin. It is chosen large in Proposition 28 so that covering, freezing, and correction estimates hold.
rho = dimension-dependent in (0,1]; chosen so C_n C_n rho^2 <= 1/40
Microscopic ball radius fraction used in the sparseness estimate and Dirichlet estimates. Fixed in Proposition 28, Step 1(ii).
theta = theta_n = 1/2 min{1, theta_{n,0}, theta_{n,4}}
Volume fraction of the small-q set in every microscopic ball. Chosen small enough to make Lemma 27 applicable.
tau = positive, 0<tau<=3/4, from Proposition 21
Defines Omega={|q|<tau}. It must be small enough for the sparseness estimate and bounded away from zero for the final lower bound on F.
sigma = small, satisfying 2 C_n sigma <= tau/8
Smoothing kernel radius fraction fixed by (69) so that q changes by at most tau/8 on the smoothing support.
delta_n = small positive; forbidden caps have total measure less than the tangent sphere
Angular separation constant in Lemma 11. Chosen so that each new patch has an admissible tangential direction left after excluding finitely many small caps.

assumptions (5)

standard math Standard elliptic regularity and Weyl's lemma: weak harmonic functions are smooth.
Used in Sections 6 and 7 to ensure the harmonic replacement f* is smooth inside Omega and that the constructed F is smooth.
standard math Lax-Milgram solvability of Poisson-Dirichlet problems on open subsets of R^n.
Used in Proposition 24 to solve -Delta w = Delta f in Omega and in Lemma 27 to control the solution.
standard math Nonzero real-analytic functions have zero sets of Lebesgue measure zero.
Used in Lemma 19 to convert compactness and non-identically-zero local models into a uniform sublevel estimate.
standard math Green's function bounds and mean-value inequalities for the Laplacian in n>=3.
Used in Lemma 26 to derive the pointwise boundary estimate that feeds Lemma 27.
domain assumption For n>=3, the tangent sphere S^{n-2} is positive-dimensional, so finitely many small forbidden caps cannot cover it.
This is the geometric fact behind Lemma 11 and Remark 12. The entire construction relies on the possibility of separated tangential directions in dimensions three and higher.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Counterexamples to the Landis conjecture in dimensions three and higher." pith.science (2026). https://pith.science/paper/OADUGUCD

@misc{pith2026260800802,
  author       = {Pith},
  title        = {Pith review of: Counterexamples to the Landis conjecture in dimensions three and higher},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OADUGUCD}},
  note         = {Machine review of arXiv:2608.00802}
}

abstract

In every dimension three and higher we construct a nonzero function that satisfies the stationary Schr\"odinger equation with bounded, real-valued potential and decays like $\exp(-c|x|^{4/3})$. This gives a counterexample to the Landis conjecture in dimension three and higher.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Quantitative Unique Continuation on Simplex and $\mathbb{Z}^n$
math.AP 2026-08 conditional novelty 8.0 of 10

For every n≥3, non-trivial solutions on the simplex lattice Δ_N^{(n)} and on Z^n have at least c N^{⌈n/2⌉} (resp. c L^{⌈n/2⌉}/log L) sites where the value is not exponentially small.
Unique continuation at infinity for potentials with arbitrary radial growth
math.AP 2026-08 conditional novelty 6.0 of 10

For radial potentials with arbitrary growth, any real-valued solution of Δu=Vu with u(0)≠0 has a sequence of points where |u(x)|≥c e^{-β(|x|)}, with β explicit in terms of the growth bound.

Reference graph

Works this paper leans on

12 extracted references · 11 canonical work pages · cited by 2 Pith papers

[1]

Bourgain and C

J. Bourgain and C. E. Kenig, On localization in the continuous Anderson–Bernoulli model in higher dimension,Invent. Math.161(2005), no. 2, 389–426. 33

work page 2005
[2]

Cruz-Sampedro, Unique continuation at infinity of solutions to Schrödinger equations with complex-valued potentials, Proc

J. Cruz-Sampedro, Unique continuation at infinity of solutions to Schrödinger equations with complex-valued potentials, Proc. Edinburgh Math. Soc. (2)42(1999), no. 1, 143–153

work page 1999
[3]

Thomas Duyckaerts, Xu Zhang, Enrique Zuazua, On the optimality of the observability inequalities for parabolic and hyperbolic systems with potentials,Ann. I. H. Poincaré – AN 25(2008), 1–41

work page 2008
[4]

Fernández-Bertolin, D

A. Fernández-Bertolin, D. Stan, L. Roncal, Landis’ conjecture: a survey, Preprint (2024), arXiv:2412.00788

arXiv 2024
[5]

N. D. Filonov and S. T. Krymskii, On the Landis conjecture in a cylinder,Russian J. Math. Phys.31(2024), no. 4, 645–665

work page 2024
[6]

Froese, I

R. Froese, I. Herbst, M. Hoffmann-Ostenhof, T. Hoffmann-Ostenhof,L2-lower bounds to solutions of one-body Schrödinger equations,Proc. Roy. Soc. Edinburgh Sect. A95(1983), no. 1-2, 25–38

work page 1983
[7]

C. E. Kenig, L. Silvestre, and J.-N. Wang, On Landis’ conjecture in the plane,Comm. Partial Differential Equations40(2015), no. 4, 766–789

work page 2015
[8]

V. A. Kondrat’ev and E. M. Landis, Qualitative theory of second order linear partial differential equations (Russian), Partial differential equations, 3 (Russian), 99–215, 220, Itogi Nauki i Tekhniki, Sovrem. Probl. Mat. Fund. Naprav., 32, Akad. Nauk SSSR, Vsesoyuz. Inst. Nauchn. i Tekhn. Inform., Moscow, 1988

work page 1988

Show all 12 references

[9]

V. A. Kondrat’ev and E. M. Landis, Qualitative theory of second order linear partial differential equations, inPartial Differential Equations III, Encyclopaedia Math. Sci. 32, Springer, Berlin, 1991, 87–192

1991
[10]

Logunov, E

A. Logunov, E. Malinnikova, N. Nadirashvili, and F. Nazarov, The Landis conjecture on exponential decay,Invent. Math.241(2025), 465–508

2025
[11]

V. Z. Meshkov, On the possible rate of decay at infinity of solutions of second order partial differential equations,Mat. Sb.182(1991), no. 3, 364–383; English transl.,Math. USSR-Sb. 72(1992), no. 2, 343–361

1991
[12]

Differential Equations and their Applications

O. A. Oleinik, M. A. Shubin, International conference of graduates of Moscow State University “Differential Equations and their Applications”, Uspekhi Mat. Nauk,37(1982), no. 6, 261–285. 34

1982

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Bourgain and C

J. Bourgain and C. E. Kenig, On localization in the continuous Anderson–Bernoulli model in higher dimension,Invent. Math.161(2005), no. 2, 389–426. 33

work page 2005

[2] [2]

Cruz-Sampedro, Unique continuation at infinity of solutions to Schrödinger equations with complex-valued potentials, Proc

J. Cruz-Sampedro, Unique continuation at infinity of solutions to Schrödinger equations with complex-valued potentials, Proc. Edinburgh Math. Soc. (2)42(1999), no. 1, 143–153

work page 1999

[3] [3]

Thomas Duyckaerts, Xu Zhang, Enrique Zuazua, On the optimality of the observability inequalities for parabolic and hyperbolic systems with potentials,Ann. I. H. Poincaré – AN 25(2008), 1–41

work page 2008

[4] [4]

Fernández-Bertolin, D

A. Fernández-Bertolin, D. Stan, L. Roncal, Landis’ conjecture: a survey, Preprint (2024), arXiv:2412.00788

arXiv 2024

[5] [5]

N. D. Filonov and S. T. Krymskii, On the Landis conjecture in a cylinder,Russian J. Math. Phys.31(2024), no. 4, 645–665

work page 2024

[6] [6]

Froese, I

R. Froese, I. Herbst, M. Hoffmann-Ostenhof, T. Hoffmann-Ostenhof,L2-lower bounds to solutions of one-body Schrödinger equations,Proc. Roy. Soc. Edinburgh Sect. A95(1983), no. 1-2, 25–38

work page 1983

[7] [7]

C. E. Kenig, L. Silvestre, and J.-N. Wang, On Landis’ conjecture in the plane,Comm. Partial Differential Equations40(2015), no. 4, 766–789

work page 2015

[8] [8]

V. A. Kondrat’ev and E. M. Landis, Qualitative theory of second order linear partial differential equations (Russian), Partial differential equations, 3 (Russian), 99–215, 220, Itogi Nauki i Tekhniki, Sovrem. Probl. Mat. Fund. Naprav., 32, Akad. Nauk SSSR, Vsesoyuz. Inst. Nauchn. i Tekhn. Inform., Moscow, 1988

work page 1988