Magnetic Double-Wells: Absence of Tunneling

Charles L. Fefferman; Jacob Shapiro; Michael I. Weinstein

REVIEW 3 major objections 4 minor 1 cited by

Magnetic Double-Wells: Absence of Tunneling

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read A magnetic double well can have exactly zero tunneling, not just exponentially little.

desk verdict A genuinely new construction for exact zero tunneling in a magnetic double-well, but the proof has a sign error in a central symmetry lemma; the main claim probably survives, but the current write-up is not correct as it stands. read the letter →

arxiv 2509.02857 v1 pith:CZ4BGWFJ submitted 2025-09-02 math-ph cond-mat.mes-hallmath.APmath.CAmath.MPphysics.atm-clus

Charles L. Fefferman , Jacob Shapiro , Michael I. Weinstein This is my paper

classification math-phcond-mat.mes-hallmath.APmath.CAmath.MPphysics.atm-clus MSC 81Q1081Q0535Q40

keywords magnetictunnelingdouble-wellpotentialeigenvaluesplittingtranslationsophonperturbationsstrongfieldparitytransitionquantuminterference

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

This paper tries to show that in a two-dimensional magnetic double-well system the two lowest eigenvalues can be made exactly equal, so a particle sitting in one well never tunnels to the other. The tuning device is a small asymmetric perturbation of an otherwise radial single-well potential: four tiny 'sophon' bumps whose vertical placement changes the sign of a magnetic hopping coefficient. Varying that placement drives the signed even-odd splitting through zero; at the crossing the ground state switches from even to odd and the low-energy eigenspace becomes two-dimensional. If true, this is a genuinely magnetic effect—in non-magnetic double wells the splitting is always positive and tunneling can be suppressed only exponentially, never eliminated. The paper gives a complete proof of the announced construction.

What carries the argument

The engine of the proof is the magnetic hopping coefficient ρ0(λ)=⟨φ^L, (Hλ−eλ)φ^R⟩, which inversion symmetry makes real-valued, and the relation S0(λ) ≈ −2ρ0(λ), Δ0(λ)=|S0(λ)|. The single-well potential is a radial 'planet' λ²v◦ decorated with four 'sophons': tiny translated copies τ W^0(x−ζν) centered at (±D, ±ȳ). Zak's magnetic translation operator attaches oscillating phase factors exp(iλ d∧x) to atomic states, and the dominant planet–sophon interaction carries the phase λ d∧ζν = ±λD^{3/2}ȳ/2, producing the cosine in the expansion. The parameters δ=exp(−MλD^{3/2}) and τ=exp(−λ/4(2D^{5/2}−3D²)) are chosen so that planet–sophon interactions dominate planet–planet and sophon–sophon terms. A

What would settle it

Compute the spectrum of Hλ for the constructed potential with ȳ placed at a zero of cos(λD^{3/2}ȳ/2), e.g. ȳ = π/(λD^{3/2}). If the two lowest eigenvalues are not exactly degenerate—if E1−E0 remains positive with magnitude comparable to the predicted FACρ exp(−λ/4(D³−D²))—or if the even and odd ground states fail to cross, the central claim fails. Alternatively, scan ȳ and check whether the signed splitting S0 changes sign at all.

Watch

Extended reading notes

Core claim

The paper's central result is Theorem 1.1. Fix a smooth radial single-well potential v◦ satisfying hypotheses (V1)–(V5); fix d1 large. Then for all sufficiently large λ there is a family Fλ of smooth, compactly supported, non-positive single-well potentials, each within exp(−Cλ) of λ²v◦ in L∞ and differing from it only on a set of total measure exp(−Cλ), with these properties: some member has zero double-well eigenvalue splitting Δ0(λ)=0, hence no low-energy tunneling; some member has zero magnetic hopping coefficient ρ0(λ)=0; and by deforming within Fλ the signed splitting S0(λ)=E_odd−E_even changes sign, so the ground state switches parity, degenerating in between. The proof works by expan

Load-bearing premise

The whole argument rests on an energy estimate ensuring that the two lowest levels of the double well are exactly the even and odd ground states with a fixed gap to the rest of the spectrum; the sophon perturbations have to stay inside that estimate's regime, and the proof of that estimate is only sketched here.

Editorial extensions

If this is right

For the constructed potential, the low-energy tunneling time 1/Δ0 becomes infinite: a state localized in one well stays there.
At the degeneracy point the ground-state eigenspace is two-dimensional, spanned by one even and one odd state; moving through it flips the parity of the ground state.
The vanishing is non-generic: within the same family, changing ȳ reopens a gap, so exact cancellation occurs at isolated parameter values.
In a periodic array of such magnetic double wells, suppressed nearest-neighbor hopping points toward nearly flat magnetic bands near the atomic energy, as the authors note in their outlook.
The construction shows that exponentially small, spatially placed potential perturbations—not large barriers—can completely eliminate magnetic tunneling.

Reading between the lines

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

If the construction is robust, the same sophon-placement mechanism could cancel hopping in higher-dimensional or lattice settings, possibly with fewer than four bumps when reflection symmetry is not required.
A direct numerical check is available: fix D, M, and λ, scan ȳ, and look for Δ0(λ) to dip to numerical zero at ȳ = (n+1/2)π/(λD^{3/2}); the corresponding tunneling time should diverge.
The exact zero condition λD^{3/2}ȳ/2 = (n+1/2)π gives a clean relation between geometric placement, magnetic field strength, and energy degeneracy—potentially useful as a calibration or switching effect.
The authors' expected result that zeros are non-generic in λ while the phenomenon is open in potential space suggests that exact vanishing is fragile but stably reachable by fine-tuning position, a feature that could matter for engineered quantum devices.

Share X Bluesky LinkedIn Reddit HN

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

Short version: this paper is worth a serious referee, and I think the main result is likely true, but there is a real error in the proof of Proposition 2.13 that the authors need to fix.

What is new: prior lower bounds ([FSW22], [HK24], [Mor24]) were for radial single-wells, where tunneling can be small but never zero. By adding four tiny, exponentially small 'sophon' bumps at positions (±D, ±ȳ) around a radial planet, the authors make the single-well non-radial, and the magnetic phase produces a cosine factor in the hopping coefficient. Tuning ȳ to a zero of the cosine gives exactly zero hopping. The scale separation is handled carefully: planet–planet terms are order exp(−λD^3/4), planet–sophon terms are exp(−λ(D^3−D^2)/4), and sophon–sophon terms are much smaller; the parameter choices for δ and τ do the job. The LCAO reduction and error bounds are seriously done; the comparison between approximate and exact hopping via Appendix C is standard and convincing.

The soft spot: Proposition 2.13, which claims Ω(ν,0) = Ω(0,ν). The proof writes the kernel as e^{+iλ/2 x∧y}, but the definition (2.30c) has e^{−iλ/2 x∧y}. With the correct sign and the substitution x = −y′, y = −x′, you get Ω(ν,0) = \overline{Ω(0,ν)}, not equality. That is not a cosmetic slip: Ω(0,ν) carries a nontrivial phase e^{−iλ d∧ζ_ν}, so equality would contradict reality of ρ. What actually saves the cosine formula is that the sum of a quantity and its conjugate is 2Re, and the formula (2.32) follows if you replace the false equality by the conjugate relation. So the theorem is probably repairable, but the proof as written is wrong at that step. The same false note appears in Proposition 3.7.

Second concern: Proposition 2.16 is labeled a sketch, and the key spectral gap/energy estimate is imported from [FSW25a, Theorem 6.3]. The paper does not verify that the cited theorem's hypotheses cover these λ-dependent potentials with exponentially small sophons. Since the conclusion 'ρ=0 ⇒ Δ=0' depends on that estimate, a referee should check the applicability carefully. That may be a non-issue, but the paper should either prove it or state the hypotheses clearly.

For the right audience—semiclassical analysis and magnetic Schrödinger operators—this is worthwhile and I'd bring it to a reading group. The construction of a tunable sign via non-radial wells is a genuine step beyond the radial results. Recommendation: send to peer review; require the authors to fix Proposition 2.13 and to provide a justified proof (or a precise statement with verification) of the spectral localization used. I would not desk-reject, but I would not accept without the fix.

Referee Report

3 major / 4 minor

Summary. The paper constructs a magnetic double-well Schrödinger Hamiltonian on R^2, in a strong constant magnetic field, whose two lowest eigenvalues are exactly degenerate, so that tunneling between the wells vanishes. The single-well potential is a radially symmetric 'planet' λ²v◦ perturbed by four very small, very localized 'sophon' potentials. The key object is the hopping coefficient ρ(λ), which is expanded into planet-planet, planet-sophon, and sophon-sophon interaction terms. The authors show that the dominant planet-sophon contribution is proportional to cos((λ/2)D^{3/2}ȳ) times an exponentially small positive factor, so by varying the vertical offset ȳ one can make ρ vanish or change sign. A linear-combination-of-atomic-orbitals (LCAO) reduction then identifies the eigenvalue splitting Δ(λ) with |−2ρ(λ)+error|, yielding the main theorem: a family of potentials with exactly zero tunneling, including a parity flip of the ground state at the degeneracy point.

Significance. If the proof is correct, this is a striking result: exact cancellation of tunneling in a magnetic double well, with no analogue in the non-magnetic case. The paper gives a concrete, parameter-dependent construction and a precise asymptotic formula for the hopping coefficient, with all exponential error terms controlled. The mechanism—using exponentially small symmetry-breaking perturbations to seed magnetic phase oscillations—is original and likely to influence later work on flat bands and tight-binding reductions. The claims are falsifiable through the explicit cosine formula and the predicted degeneracy at specific values of ȳ. The main limitation is that the proof relies on several prior results, and one load-bearing identity in the derivation of the cosine formula is stated with the wrong phase; this issue appears fixable but must be addressed before the central claim is established.

major comments (3)

[§2.6.2, Eq. (2.30c), Prop. 2.13] The proof of Proposition 2.13 starts by writing Ω(ν,0) with the phase e^{+iλ/2 x∧y}, but the definition (2.30c) has e^{-iλ/2 x∧y}. With the definition, substitution x=−ỹ, y=−x̃ gives Ω(ν,0) = conjugate of Ω(0,ν), not Ω(0,ν). The asserted equality is used in §3.1.2 (the line 'We note that Ω◦(ν,0)=Ω◦(0,ν)' and Proposition 3.7) to produce the real cosine in (3.44). Moreover, if Ω(ν,0)=Ω(0,ν) as stated, then Ω(0,1)+Ω(0,2) is not automatically real because the factors Rν in Proposition 3.4 are real and may differ, and no proof of R1=R2 is given. Replacing the equality by the conjugate relation, together with a proof that Ω(0,2)=conj(Ω(0,1)), can still lead to the cosine formula, so the gap is probably repairable; as written, however, the derivation of the sign-changing cosine is not established.
[§2.8, Prop. 2.16] The identification {E0,λ,E1,λ} = {Eeven,λ,Eodd,λ} is essential: it converts a zero of the hopping coefficient into a zero eigenvalue splitting, Δ=|S|. The proof is only a sketch and invokes an energy estimate from [FSW25a, Theorem 6.3] without stating the result or checking its hypotheses for the present λ-dependent potentials with exponentially small, distant sophons. Since this is the bridge between ρ and the actual spectrum of Hλ, the manuscript should either state a precise version of the cited estimate and verify its applicability, or give a complete proof. This is not a mere presentation issue; it is load-bearing for Theorems 2.21 and 1.1.
[§3.1.3] The transition from the approximate hopping −ρ◦ to the full −ρ is dispatched in a short paragraph: 'The results of Appendix C imply ... terms in Ω(μ,ν)−Ω◦(μ,ν) all have at least one additional factor of τ ... bounded in a manner similar to (3.20).' Given that the parameter choices for τ and δ are delicate and are what makes the planet-sophon terms dominate, this step needs a more explicit estimate. In particular, the error should be tracked through all terms of (2.29), including the phase factors and the overlap integrals, so the asserted Proposition 2.14 can be checked without rederiving the argument.

minor comments (4)

[§2.9, Eq. (2.41)] Typo: 'λmiin' should be 'λmin'.
[§2.6.2, Prop. 2.13 proof] The phase in the first displayed equation of the proof appears to be a typo; with the definition (2.30c) it should read e^{−iλ/2 x∧y}. Please correct and adjust the surrounding text accordingly.
[References] The citation [3BP] in Section 2.3 appears to be a work of fiction rather than a mathematical reference; consider replacing it with a standard perturbative/effective-Hamiltonian reference or removing the informal shorthand.
[Theorem 1.1 statement] The statement says 'Let d1>0 be given and sufficiently large', but in the construction d1 is set to D^{3/2}/2 with D itself a large parameter chosen later. Please rephrase to avoid the impression that arbitrary large d1 are covered.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: zero-tunneling conclusion emerges from the construction, not from an input or self-citation chain.

full rationale

The derivation chain is explicit: the potential vλ is constructed as λ²v◦ plus exponentially small 'sophon' perturbations; the hopping coefficient ρ is expanded via the resolvent kernel into planet/sophon interaction terms (2.28)-(2.30); Proposition 2.14 gives the oscillatory formula ρ = -[cos(λD^{3/2}ȳ/2)+ERR]·FAC·exp(-λ(D³-D²)/4); Proposition 2.20 connects the signed splitting to -2ρ; then the sign change and zero of ρ are obtained by varying the free parameter ȳ. The cosine phase is computed from the magnetic phases of the localized states, not inserted as an ansatz, and no parameter is fitted to the target Δ=0. The choice of τ and δ enforces dominance of the planet-sophon terms but does not force the sign of ρ. The paper does rely on the authors' own prior results — [FSW22] for the radial-well tunneling bounds and [FSW25a, Thm 6.3] for the spectral-gap/energy estimate used in Proposition 2.16 — but these are independent theorems with stated assumptions that do not include the vanishing-tunneling conclusion, so they are legitimate external support rather than circular self-citation. The 'sketch' label on Prop 2.16 and the possible phase/sign inconsistency in the proof of Prop 2.13 are mathematical correctness concerns, not reductions of the conclusion to its inputs. No equation is defined in terms of the target result, and no fitted quantity is renamed as a prediction.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim is an existence result for a constructed family of potentials. The construction depends on the geometric parameters D, M, ybar; the target (zero splitting) is achieved by tuning ybar. The spectral analysis relies on two prior theorems from the authors' own research program.

free parameters (4)

ybar = ybar*_n(lambda) = (n+1/2)pi / (lambda/2 D^{3/2}), n chosen so 0 < ybar < 1
Vertical displacement of the sophons; tuned continuously to make the cosine in (2.32) vanish, producing zero hopping and zero splitting. This is the design parameter that realizes the theorem.
D = sufficiently large (D >= Dmin)
Horizontal scale: well separation 2|d| = D^{3/2} and sophon distance D from planet; chosen large to satisfy D^{3/2} >> D and the ordering of interaction terms.
M = sufficiently large
Governs the exponential smallness delta = exp(-M lambda D^{3/2}) of sophon support; chosen in the proof to dominate error terms.
lambda_min = sufficiently large depending on v0, D, M
Asymptotic threshold; all estimates require lambda >= lambda_min.

assumptions (3)

domain assumption Radial ground state and uniform spectral gap for the radial single-well Hamiltonian h0_lambda (Assumptions (V1)-(V5), Section 2.1)
The construction and all estimates assume v0 is radial, compactly supported, with radial ground state and a uniform O(1) spectral gap; Remark 2.3 notes these are satisfied for potentials with a unique non-degenerate minimum.
standard math [FSW22, Theorem 1.7] lower bound on the radial magnetic hopping coefficient rho0,lambda(Y)
Lemma 3.2 uses this published result for the lower bound in (3.11); the current paper proves only the upper bound.
standard math [FSW25a, Theorem 6.3] energy estimate for low-lying double-well magnetic spectra
Used in Proposition 2.9 and Proposition 2.16 to locate the two lowest H_lambda eigenvalues in even/odd sectors and to ensure the spectral gap; not proved in this paper.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Magnetic Double-Wells: Absence of Tunneling." pith.science (2026). https://pith.science/paper/CZ4BGWFJ

@misc{pith2026250902857,
  author       = {Pith},
  title        = {Pith review of: Magnetic Double-Wells: Absence of Tunneling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CZ4BGWFJ}},
  note         = {Machine review of arXiv:2509.02857}
}

read the original abstract

We present a magnetic double-well Hamiltonian where the tunneling between the two wells vanishes, as recently announced.

Figures

Figures reproduced from arXiv: 2509.02857 by the authors.

**Figure 1.** Support of the “atomic well”, vλ. One radially symmetric planet with four “sophons”; see Section 2.3. 2 The proof of Theorem 1.1 We first provide a detailed roadmap for the proof of Theorem 1.1 and then, in Section 3, we implement the steps of the roadmap, providing the technical arguments. 2.1 Roadmap, Step 1: The single-well radial potential, λ 2 v ◦ (x), the magnetic atomic Hamiltonian, h ◦ λ , and its ground sta… view at source ↗

**Figure 2.** Double-well configuration consisting of two atomic wells ( [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

Magnetic Double-Wells: Lower Bounds on Tunneling
math-ph 2025-11 conditional novelty 8.0 of 10

For generic coupling λ in a strongly magnetic double well, the tunneling splitting and hopping coefficient are bounded below by exp(-λ^{1+ε}) outside a zero-density exceptional set.

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages · cited by 1 Pith paper

[2]

arXiv:2506.12650 [math-ph]. [FSW25b] Fefferman,C.L.,Shapiro,J.,andWeinstein,M.I.:Quantumtunnelingandits absence in deep wells and strong magnetic fields Proceedings of the National Academy of Sciences.122 (8), e2420062122 (2025) [HS84] Helffer, B. and Sjostrand, J.: Multiple wells in the semi-classical limit I Com- munications in Partial Differential Equa...

work page Pith review arXiv 2025
[157]

Providence, RI: American Mathematical Society, 2014.doi: 10.1090/ gsm/157

Graduate Studies in Mathe- matics. Providence, RI: American Mathematical Society, 2014.doi: 10.1090/ gsm/157. [Zak64] Zak, J.: Magnetic Translation Group Phys. Rev. 134, A1602–A1606 (1964) 40

work page 2014
[2014]

[Mat94] Matsumoto,H.:SemiclassicalasymptoticsofeigenvaluedistributionsforSchrödinger operators with magnetic fields Communications in Partial Differential Equa- tions. 19 (5-6), 719–759 (1994) [Mor24] Morin, L.: Tunneling effect between radial electric wells in a homogeneous magnetic field Letters in Mathematical Physics.114 (1), 29 (2024) [SW22] Shapiro,...

work page 1994
[2025]

The Flea on the Magnetic Elephant

arXiv: 2505.12846 [math-ph]. [FSW22] Fefferman, C., Shapiro, J., and Weinstein, M. I.: Lower Bound on Quantum Tunneling for Strong Magnetic Fields SIAM Journal on Mathematical Anal- ysis. 54 (1), 1105–1130 (2022) [FLW18] Fefferman,C.L.,Lee-Thorp,J.P.,andWeinstein,M.I.:HoneycombSchrödinger Operators in the Strong Binding Regime Communications on Pure and A...

work page Pith review arXiv 2022

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [2]

arXiv:2506.12650 [math-ph]. [FSW25b] Fefferman,C.L.,Shapiro,J.,andWeinstein,M.I.:Quantumtunnelingandits absence in deep wells and strong magnetic fields Proceedings of the National Academy of Sciences.122 (8), e2420062122 (2025) [HS84] Helffer, B. and Sjostrand, J.: Multiple wells in the semi-classical limit I Com- munications in Partial Differential Equa...

work page Pith review arXiv 2025

[2] [157]

Providence, RI: American Mathematical Society, 2014.doi: 10.1090/ gsm/157

Graduate Studies in Mathe- matics. Providence, RI: American Mathematical Society, 2014.doi: 10.1090/ gsm/157. [Zak64] Zak, J.: Magnetic Translation Group Phys. Rev. 134, A1602–A1606 (1964) 40

work page 2014

[3] [2014]

[Mat94] Matsumoto,H.:SemiclassicalasymptoticsofeigenvaluedistributionsforSchrödinger operators with magnetic fields Communications in Partial Differential Equa- tions. 19 (5-6), 719–759 (1994) [Mor24] Morin, L.: Tunneling effect between radial electric wells in a homogeneous magnetic field Letters in Mathematical Physics.114 (1), 29 (2024) [SW22] Shapiro,...

work page 1994

[4] [2025]

The Flea on the Magnetic Elephant

arXiv: 2505.12846 [math-ph]. [FSW22] Fefferman, C., Shapiro, J., and Weinstein, M. I.: Lower Bound on Quantum Tunneling for Strong Magnetic Fields SIAM Journal on Mathematical Anal- ysis. 54 (1), 1105–1130 (2022) [FLW18] Fefferman,C.L.,Lee-Thorp,J.P.,andWeinstein,M.I.:HoneycombSchrödinger Operators in the Strong Binding Regime Communications on Pure and A...

work page Pith review arXiv 2022