{"id":"78c288fd-e3fb-497c-877e-ed0fb32d3449","arxiv_id":"2608.11111","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For generic Z2-symmetric two-dimensional potentials, the first three quantum Birkhoff normal form layers determine the potential's full Taylor series up to spatial inversion.","lead":"This paper proves that for generic two-dimensional Schrödinger potentials with one reflection symmetry and rationally independent well frequencies, the first three layers of the quantum Birkhoff normal form determine the full Taylor series of the potential, up to the spatial inversion x goes to -x. A generalist might read it because it shows how much of a quantum well can be recovered from its low-energy spectrum, the quantum version of hearing the shape of a drum.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the flagged surjectivity in Lemma 6.5/6.6 is the most fragile step, but the Fischer-adjoint argument and the dimension count are sound.","rationale":"The reader's weakest-assumption pick is exactly where I concentrated: Lemma 6.5/6.6 is the most structurally fragile point in the global separation argument. However, my stress-test of that step found the reasoning sound: the Fischer inner product argument gives surjectivity, the tangent vector (G, G) preserves I10 to all orders because only B[0]_10 responds to V10 and it responds identically in both entries, and the resulting incidence dimension count supports the dense-Gδ separation. I therefore do not find a damaging concern, and the ACCEPT verdict should stand. I mark agreement as 'partial' because I share the reader's identification of the fragile step but do not regard it as a failure. The other potential weakness—the reliance on Colin de Verdière's spectral-to-QBNF theorem—is external, published, and not contradicted by anything in the manuscript; it is a legitimate citation rather than an internal gap. No formal verification is present, but the algebraic structure of the proof and the explicit finite-dimensional computations make the argument independently checkable.","tokens_in":38278,"tokens_out":24121,"duration_ms":208282,"concrete_test":"Run a symbolic or high-precision numerical audit of Section 6 at a random point of Ufib: compute the 30 by 32 Jacobian DI10 and verify it has rank 30; for two ten-jets with non-± cubic pairs, compute the 3 by 6 matrix representing 1/2 Π(M3(X*)^2 - M3(X)^2) on E10 and verify it has rank 3; and check by finite differences that Θ in Lemma 6.6 is a submersion. Any rank drop at a generic point would indicate that the separation argument collapses.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I read the argument in good faith and found no internal inconsistency that threatens the central claim. The reader correctly identifies Lemma 6.5/6.6 as the most structurally fragile step: Lemma 6.5 must show that 1/2 Π(M3(X*)^2 - M3(X)^2) maps E10 surjectively onto N[4]_12 whenever the cubic pairs are not ± equal, so that Θ in Lemma 6.6 is a submersion and the exceptional fiber Z has dimension at most 31. I checked this step closely. The proof is valid: T = P(∂) with P ≠ 0 has Fischer adjoint equal to multiplication by P, which is injective on E4 because the polynomial ring has no zero divisors; hence T is surjective, and E2 = Π|_{E4} is an isomorphism by Proposition 3.1. The tangent vector (G, G) in Lemma 6.6 is also legitimate: among the components of I10, only B[0]_10 depends on V10, and it varies by E5G in both entries, while B[2]_10 is V10-independent by (2.28) and Proposition 4.7. The subsequent dimension count dim R = 34, dim Z ≤ 31, and the Sard/compact-exhaustion argument in Proposition 6.7 are internally consistent. The main external reliance is the spectral-to-QBNF equivalence of [CdV09], which is cited rather than reproved; I found no reason to doubt its applicability to the real analytic trapped setting. Overall, the central claim is not undermined by a load-bearing flaw.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the semiclassical inverse spectral problem for P_ℏ = -ℏ²Δ/2 + V on R², with V even in x₂, nonresonant frequencies, and a nondegenerate well at the origin. It proves a generic injectivity theorem at the level of the quantum Birkhoff normal form: for each sign σ there is a dense Gδ subset of the ten-jet space such that if V and W lie in the fixed-sign formal potential space and share the first three QBNF layers B[0], B[2], B[4], then W = V. Equivalently, for such V the first three layers determine the complete Taylor series at the well. A sign-free version gives W = V or W = V(-x); real analytic and trapped versions (Theorems 1.3 and 1.4) pass from QBNF data to low-lying semiclassical spectra via the cited Colin de Verdière spectral-to-QBNF equivalence. The proof is organized as a filtered homological-equation calculation: the first two layers leave exactly one affine defect per degree, whose parameters are shown to be detected generically by the ℏ⁴ layer; low-order compatibility then fixes a14, and a finite-dimensional Sard/incidence argument separates the cubic pair up to sign.","tokens_in":38600,"tokens_out":52506,"duration_ms":424142,"significance":"If correct, this is a substantial advance in the bottom-of-the-well inverse spectral problem in dimension two: prior results required additional data or symmetry beyond a single reflection to recover the Taylor series from one or two QBNF layers, whereas here three layers generically encode the full jet. The proof is unusually concrete: the block operators A_N, the explicit transfer operator T_V, and the derivative computations (e.g. ∂d_N/∂a50 in Proposition 4.11) are directly checkable. I specifically checked the most fragile point identified in the stress test, the surjectivity of 1/2Π(M_3(X*)² - M_3(X)²) in Lemma 6.5 and the resulting submersion Θ in Lemma 6.6; the Fischer-adjoint argument and the dimension count dim R = 34, dim Z ≤ 31 are sound. The paper is also honest about its single external input, the spectral-to-QBNF equivalence cited from [CdV09], and it does not rely on fitted parameters or post-hoc selection of the generic set. If the results are accepted, the paper likely becomes a reference for generic spectral rigidity with Z₂-symmetry.","major_comments":[],"minor_comments":[{"comment":"The statement that z_N(V) depends only on the (2N−2)-jet and B[4](V) omits the dependence of the base point (\\hat V_{2N−1}, \\hat V_{2N}) on B[2]_{2N} and B[0]_{2N}; since those layers are part of the QBNF data, this does not affect the argument, but the wording should be corrected.","section":"Corollary 4.10"},{"comment":"The displayed identities d_N(V + εx_1^5) = d_N(V) + εm_N(V) and d_4(W_ε) = d_4(V^♯) + εm_4 are written as exact equalities although d_N and d_4 are rational functions of the five-jet; they hold only modulo O(ε²). The density conclusions are unaffected because the leading coefficients are nonzero, but the proofs should state first-order expansions.","section":"Proposition 4.11 and Lemma 5.10"},{"comment":"The reference to 'Proposition 6.12' should be to 'Lemma 6.12', which is the proved bijection between common zeros and ten-jets.","section":"Proof of Proposition 6.13"},{"comment":"The phrase 'Sincet∗ V andt ♯ is fixed' contains a typo; it should read 'Since t∗_V and t♯_V are fixed'.","section":"Lemma 5.9"},{"comment":"The references [CSW99] and [ISZ02] are listed in the bibliography but are not cited in the body; they should either be cited where relevant or removed from the bibliography.","section":"References"},{"comment":"There is a stray period in the sentence introducing V(x) and the homogeneity of V_m; it should read '... where V(x) := V_phys(...), and V_m is homogeneous of degree m in x.'","section":"Section 1.1"}],"recommendation":"minor_revision","confidential_remarks":"This is a strong paper. The central theorem is supported by a coherent induction and finite-dimensional transversality arguments; I found no load-bearing error. The minor issues listed (imprecise dependence statements, first-order vs exact expansions, a few typos, and uncited references) are local and easily fixed. The reliance on [CdV09] is standard and explicitly stated, so I do not regard it as a defect. I recommend minor revision rather than outright acceptance only because of the small proof-presentation issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the result is real and the proof is sound. The genuinely new step is using the B[4] layer to kill the one-dimensional affine defect left by B[0] and B[2] at every degree, plus the finite-jet separation of the initial cubic pair. The paper is honest about what is new: Theorem 3.5 is credited to Wang26, and the contribution is the structural reformulation and the B[4] argument built around it.\n\nWhat it does well: the algebraic filtration is organized cleanly. Proposition 2.10 gives a triangular block operator, Theorem 3.4 isolates the affine defect exactly, and Section 4 converts each defect into a scalar observable d_N. I checked the Fischer-adjoint step in Lemma 6.5/6.6: the surjectivity claim is correct, and the dimension count in Proposition 6.7 follows. The main theorems are stated carefully, with the sign/inversion ambiguity handled properly. The generic conditions depend only on the ten-jet, which is the right level of explicitness for this method.\n\nSoft spots, in proportion: the result is generic rather than explicit. The dense G_delta sets are defined by nonvanishing of rational functions plus a Sard/measure-zero exclusion, so you cannot exhibit a concrete admissible jet. That is normal for this kind of argument, not a flaw. Theorem 1.4 relies on the standard CdV09 spectral-to-QBNF equivalence, which is cited rather than reproved; fine, but the spectral conclusion inherits a substantial external theorem. The paper is long and sometimes terse, and the topology appendix is dense. Also, only the three-dimensional resonant projection of B[4]_12 is used in the separation step; that turns out to be enough.\n\nI read the flagged Lemma 6.5/6.6 carefully. The stress-test concern does not land: T = P(∂) has injective Fischer adjoint, hence is surjective; E2 is an isomorphism; the tangent vector (G, G) is legitimate. The fragile step holds.\n\nOverall: a meaningful advance within an established program, not a paradigm shift. It deserves a serious referee. I would send it out, ideally to someone who knows semiclassical normal forms. I would cite it if I worked in inverse spectral theory.","headline":"Solid, honest extension of Wang26: the B[4] layer generically kills the affine defect left by the first two QBNF layers, and the proof checks out.","tokens_in":39081,"tokens_out":2342,"would_cite":true,"duration_ms":22153,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P20","81Q20","35R30","53D50"],"pacs":[],"model":"deepseek-v4-flash","headline":"The low-energy spectrum generically determines a symmetric potential up to inversion.","keywords":["semiclassical inverse spectral problem","quantum Birkhoff normal form","Z2-symmetry","Moyal product","spectral determination","generic injectivity","Schrödinger operator","Taylor series at a well"],"falsifier":"Compute, for a concrete pair of cubic coefficients with $(a_{30},a_{12})\\ne \\pm(a_{30}^*,a_{12}^*)$, the $3\\times 3$ matrix of the map $G\\mapsto \\frac12\\Pi(M_3(X^*)^2-M_3(X)^2)G$ from degree-ten potential polynomials to the resonant span $\\{\\Omega_1^2,\\Omega_1\\Omega_2,\\Omega_2^2\\}$. If the rank is ever less than three, Lemma 6.5 fails and the incidence-dimension argument of Proposition 6.7 loses its codimension; a direct counterexample to Theorem 1.1 would be two sign-compatible solutions of the compatibility system $P_{Y,Z}=0$ arising from one generic ten-jet.","tokens_in":38096,"feed_emoji":"🎵","tokens_out":17327,"duration_ms":132064,"temperature":0.7,"pith_summary":"This paper asks the quantum version of the drum problem: can the low-energy spectrum of a Schrödinger operator on the plane determine the potential? The author proves that, generically, the answer is yes for potentials with one reflection symmetry, up to the unavoidable reflection $V(x)\\mapsto V(-x)$. Specifically, for a nondegenerate well at the origin with rationally independent harmonic frequencies, the first three layers of the quantum Birkhoff normal form---$B^{[0]}$, $B^{[2]}$, $B^{[4]}$---determine the complete Taylor series of the potential at the well, provided the ten-jet of the potential avoids a meager exceptional set. For real analytic potentials with a unique isolated well, the same conclusion follows from equality of the low-lying semiclassical spectra. The payoff is that a finite, explicitly checkable jet condition turns the first three quantum layers into a complete Taylor-series bookkeeper, so the generic symmetric well is spectrally unique up to spatial inversion.","feed_headline":"Low-energy spectrum pins down symmetric wells generically","feed_subtitle":"For most 2D wells with one reflection axis, three Birkhoff layers encode the full Taylor series.","key_machinery":"The quantum Birkhoff normal form (QBNF) is the unique resonant normal form obtained by a formal Moyal automorphism (a formal change of variables compatible with the noncommutative product of quantum observables); its layers $B^{[2r]}$ are the coefficients of $\\hbar^{2r}$ in the normalized symbol, and 'resonant' means commuting with the harmonic-oscillator part. Three linked mechanisms carry the argument. First, the normal-form map is triangular in the Weyl grading (with $\\deg x_j=\\deg\\xi_j=1$ and $\\deg\\hbar=2$): at each degree $2N$ the block operator $A_N$ that sends the new pair $(V_{2N-1},V_{2N})$ to the recorded $B^{[0]}$ and $B^{[2]}$ coefficients has exactly a one-dimensional kernel spanned by $(\\Phi_N,\\Psi_N)$, so the first two layers leave an affine line of possibilities parametrized by the edge coefficient $a_{1,2N-2}$. Second, the $\\hbar^4$ layer enters through a transfer operator $T_V$; the scalar edge functions $d_N$ couple to that kernel, making each edge coefficient observable, while $B^{[4]}_8$ reduces the first ambiguous edge coefficient $a_{14}$ to at most two candidates and a compatibility determinant rejects the false branch. Third, global separation of the initial cubic pair is a finite-dimensional incidence statement: a polynomial map $I_{10}$ on the 32-dimensional ten-jet space has generic rank 30, and the additional $\\hbar^4$ data $B^{[4]}_{12}$ cut the incidence set below the dimension of the jet space, with Lemma 6.5's surjectivity making the cut effective.","core_discovery":"At the level of formal Taylor series, the central claim is a generic injectivity theorem for the quantum Birkhoff normal form map. For each sign $\\sigma$ of the leading cubic coefficient $a_{30}$, there is a dense $G_\\delta$ set of ten-jets such that any two $\\mathbb{Z}_2$-symmetric formal potentials whose jets lie in that set and whose QBNF layers $B^{[0]}$, $B^{[2]}$, $B^{[4]}$ agree must be identical; without fixing the sign, they differ by at most the spatial inversion $\\iota V(x)=V(-x)$. At the spectral level, for real analytic trapped potentials, equality of the semiclassical spectrum---or equality of low-energy fixed-window spectra for all sufficiently small $\\hbar$---forces $W=V$ or $W=\\iota V$. The reconstruction is effective: after the initial cubic pair is separated by a finite compatibility system, the remaining Taylor coefficients are recovered recursively from the three layers. Thus the first three QBNF layers generically encode the complete Taylor series at the well, and the only generic spectral ambiguity is the reflection symmetry inherent in the problem.","pith_inferences":["Beyond the paper: the filtered block structure with finite-dimensional kernels and a higher-layer coupling is a plausible template for other symmetry groups; one would expect fewer layers to suffice with more symmetry and more layers with none.","Beyond the paper: because the compatibility data are encoded in a finite polynomial system, the reconstruction is in principle algorithmic; the paper does not address numerical stability or the effect of finite-precision spectral data.","Beyond the paper: the theorem is generic, so it leaves open whether any ten-jet in the exceptional set actually gives rise to a distinct potential with identical first three layers; constructing or ruling out such an example would clarify whether the genericity hypothesis is intrinsic or only a proof device."],"forward_implications":["For a generic symmetric well, the complete Taylor series at the well and, for real analytic potentials, the potential itself are determined by the low-lying spectrum up to the reflection $x\\mapsto -x$; no additional normalisation or transverse family is needed.","The exceptional potentials form a meager, nowhere dense set in the ten-jet space, so any non-uniqueness is a rare phenomenon rather than an open family of counterexamples.","The first two QBNF layers leave one scalar ambiguity at every degree, and the $\\hbar^4$ layer resolves it; this explains why three layers suffice and where each layer's information resides.","Equality of the exact low-energy fixed-window spectra for all sufficiently small $\\hbar$ gives the same conclusion as equality of the semiclassical spectrum, so the result applies to a finite energy window rather than only to an asymptotic family."],"supporting_citations":[{"why":"It supplies the semiclassical Birkhoff normal form whose asymptotic expansions connect the QBNF to the low-lying eigenvalues.","marker":"[CVN08]"},{"why":"It proves that the semiclassical spectrum determines the complete QBNF, the bridge from spectral data to the formal reconstruction.","marker":"[CdV09]"},{"why":"It establishes the bottom-of-the-well trace invariants and the QBNF-spectrum comparison used in the spectral part.","marker":"[GPU07]"},{"why":"It gives the prior two-dimensional QBNF reconstruction under one coordinate reflection plus an additional symmetry, which the paper extends to a single reflection generically.","marker":"[GU11]"},{"why":"It proves the recursive edge-coefficient theorem that the paper re-derives structurally and then upgrades to the global generic setting.","marker":"[Wang26]"},{"why":"It provides the one-dimensional semiclassical inverse result at QBNF level that frames the layer-by-layer reconstruction strategy.","marker":"[CdVG11]"},{"why":"It derives explicit bottom-of-the-well wave invariants and a reconstruction for special potential forms, serving as a comparison baseline.","marker":"[Hez09]"}],"fun_headline_variants":["Three Birkhoff layers reveal full Taylor series generically","Generic 2D wells: low-energy spectrum fixes potential up to reflection","Semiclassical spectrum determines Z2-symmetric wells generically","First three quantum Birkhoff layers encode entire well up to flip"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that genuinely different cubic terms always generate enough independent degree-twelve $\\hbar^4$ data to rule out a coincidence; if the map $\\frac12\\Pi(M_3(X^*)^2-M_3(X)^2)$ failed to be surjective for some pair of cubic terms, the exceptional set in Proposition 6.7 could grow from a meager set to a positive-dimensional family.","fun_headline_variants_meta":{"raw":{"variants":["Three Birkhoff layers reveal full Taylor series generically","Generic 2D wells: low-energy spectrum fixes potential up to reflection","Semiclassical spectrum determines Z2-symmetric wells generically","First three quantum Birkhoff layers encode entire well up to flip"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000563,"raw_usage":{"total_tokens":2665,"prompt_tokens":932,"completion_tokens":1733,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":548,"completion_tokens_details":{"reasoning_tokens":1661}},"tokens_in":548,"tokens_out":1733,"duration_ms":12858,"temperature":1.0,"reasoning_tokens":1661,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T06:00:41.782108+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a concrete pair of cubic coefficients with $(a_{30},a_{12})\\ne \\pm(a_{30}^*,a_{12}^*)$, the $3\\times 3$ matrix of the map $G\\mapsto \\frac12\\Pi(M_3(X^*)^2-M_3(X)^2)G$ from degree-ten potential polynomials to the resonant span $\\{\\Omega_1^2,\\Omega_1\\Omega_2,\\Omega_2^2\\}$. If the rank is ever less than three, Lemma 6.5 fails and the incidence-dimension argument of Proposition 6.7 loses its codimension; a direct counterexample to Theorem 1.1 would be two sign-compatible solutions of the compatibility system $P_{Y,Z}=0$ arising from one generic ten-jet.","supporting_citations":[],"review_version":1}