{"id":"036bb026-046a-4ea3-933f-495b36359815","arxiv_id":"2608.04796","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For a 2D potential with one reflection symmetry, a nonzero cubic term, and known transverse derivative data, the first two spectral Birkhoff layers determine the full Taylor series of the potential.","lead":"Low-energy spectra of a two-dimensional quantum system with one mirror symmetry can identify the potential's shape, provided extra derivative data along a transverse line is also known. The result extends inverse spectral methods to a case with fewer symmetries.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The off-diagonal Γ_m terms in the induction step are asserted, not verified; the triangularity of the recursion is load-bearing. A symbolic check of the N=3 step would settle it.","rationale":"The reader identified the external transverse-line data as the weakest assumption; that is a real caveat for the inverse spectral interpretation, but it is explicitly assumed in the theorem and acknowledged in Remark 1.2, so it does not impeach the conditional claim. The more load-bearing point is the unproved triangularity of the off-diagonal contributions in the induction step. The base case is explicit and I checked the sign of (1.6) against the Moyal expansion; the diagonal table in the appendix is internally consistent, and my own index analysis suggests the cross-term claim is probably salvageable. However, since Theorem 1.1's conclusion is exactly this determinacy, and since the paper never computes Γ_m or proves that no cross term couples the current unknown into another equation, the proof has a genuine gap. It is an omission rather than a demonstrated counterexample, and a direct symbolic computation at the first nontrivial order would confirm or refute it. Conditional acceptance remains the appropriate verdict, so no change to the reader's verdict is needed.","tokens_in":12557,"tokens_out":36653,"duration_ms":417585,"concrete_test":"Symbolically compute the full B^2_6 coefficients for the N=3 step (degrees 5 and 6) directly from the complete Moyal expansion of Eq. (3.13), including S^2_5 and all cross-derivative terms in the third-order brackets. Verify that the coefficient matrix relating the unknowns (a32, a50) to the two coefficients of B^2_6 is triangular in the prescribed backward order, with diagonal entries proportional to 64 a30 n(n−1)^2 and with the only off-diagonal entry involving the known transverse coefficient a14. If the matrix is instead not triangular, Theorem 1.1 fails already at this order.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.1 rests on the induction in §3.2, and the decisive step is the claim that the cross-derivative (off-diagonal) contributions to Π({S3,U_{2N-1}}_3 + {S^0_{2N-1},U3}_3) involve only coefficients of U_{2N-1} with larger second index, so that the displayed recurrence is triangular in m. This claim is stated in a single sentence, and the term Γ_m is never computed. I checked the index balance myself: for a degree-3 factor with α1+β1=1, a cross-derivative pattern with p=(2,1) would require p1=2 but has at most α1+β1=1 units of z1-reduction available, so the dangerous term cannot occur. This suggests the claim may be correct, but the proof as written does not establish it. The coefficient 64 a30 n(n−1)^2 is exactly what makes the backward recursion invertible, so if some unaccounted cross term coupled the current unknown a_{2N−1−2m,2m} into a different target monomial, the uniqueness conclusion would fail. The 'known terms' in (3.14) also include ℏ^2-layer Poisson contributions such as −1/2{S^2_{2N−1},U3}, which are not listed explicitly; this is incomplete bookkeeping rather than a demonstrated error. The external transverse-line data in condition (ii) is acknowledged in Remark 1.2 and is a limitation of the inverse spectral interpretation, not an internal inconsistency.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a formal local inverse result for a two-dimensional semiclassical Schrödinger operator whose potential is even in one coordinate only. After a linear symplectic normalization and under a nonresonance assumption v1/v2 not in Q, the quantum Birkhoff normal form is computed. Theorem 1.1 states that the first two layers of the QBNF coefficients (b_{0,k,ell} and b_{1,k,ell}), together with the sign of a30 and the transverse-line data {a_{1,2k}}, uniquely determine the full Taylor series of V at the bottom of the well. The proof proceeds by induction on the homogeneous degree: the base case U3 and U4 is worked out and yields the explicit relation a30^2 = 2v1 b_{1,0,0} + (v2^2/(v1^2-4v2^2)) a12^2; the induction step reconstructs the odd part U_{2N-1} by a backward recursion on the second index and then the even part U_{2N} from its resonant projection.","tokens_in":12869,"tokens_out":10094,"duration_ms":111401,"significance":"If the induction step is made rigorous, this is a meaningful advance: it weakens the symmetry hypotheses in two-dimensional semiclassical inverse spectral theory from full evenness or two-coordinate symmetry to a single reflection symmetry, and it generalizes Hezari's setting by allowing nonzero transverse-line data rather than requiring a_{1,2k}=0. The paper also gives an explicit, parameter-free formula for the leading reconstruction and a worked base case, including a concrete computation of b_{1,0,0}. The authors are candid that the result is a Cauchy problem with prescribed line data rather than a pure spectral rigidity theorem, and Remark 1.2 states this limitation clearly. However, the central induction step rests on an unproved triangularity assertion; the main diagonal computation in the appendix is not enough to establish the claimed uniqueness without a complete treatment of the cross-derivative terms.","major_comments":[{"comment":"The triangularity claim is the load-bearing step of the induction, but it is not proved. The paragraph after (3.14) asserts that cross-derivative contributions to Π({S3,U_{2N-1}}_3 + {S^0_{2N-1},U3}_3) involve only coefficients of U_{2N-1} with larger second index, and Appendix A explicitly computes only the pure z1/bar z1 derivative terms, writing '+ cross terms' without giving them. The displayed recurrence is not a well-formed equation: it has no right-hand side, and Γ_m is never defined or computed. Since the invertibility of the backward recursion depends on the coefficient 64 a30/v1 2^{2N-1} binom(2m,m) binom(2n-1,n) n(n-1)^2 being the only coupling to the current unknown a_{2N-1-2m,2m}, the manuscript needs either a complete index-balance argument for all cross-derivative patterns or a symbolic verification of the N=3 and N=4 steps. This is required for the uniqueness claim in Theorem 1.1.","section":"§3.2, Eq. (3.14) and Appendix A"},{"comment":"The 'known terms' in (3.14) are not enumerated. They include the ℏ^2-layer Poisson contribution -1/2 {S^2_{2N-1}, U3}, the ℏ^2 component of R_{2N}, and potentially similar terms arising from the BCH expansion. Without an explicit list or a lemma accounting for all ℏ^2 contributions from (3.13), the claimed formula for B^2_{2N} cannot be checked independently. This is bookkeeping, but it is essential because the recurrence is used to solve for each unknown coefficient in the backward induction.","section":"§3.2, Eq. (3.14)"},{"comment":"The reconstruction of U_{2N-1} uses the prescribed coefficient a_{1,2N-2} as the starting value of the backward recursion. The text correctly identifies this as transverse-line data, but the induction statement should make explicit that the entire sequence {a_{1,2k}} is an external input, not recovered by the method. This is a limitation of the inverse spectral interpretation rather than an internal inconsistency, but it should be stated in the theorem statement itself and not only in Remark 1.2.","section":"§3.2, paragraph after (3.12)"}],"minor_comments":[{"comment":"The table reporting algebraic contributions should show how the common factor a30/v1 2^{2N-1} binom(2m,m) a_{2N-1-2m,2m} is extracted, and at least one row of the table should be derived in detail so the reader can verify the signs and combinatorial factors.","section":"Appendix A"},{"comment":"The term Dℏ^2 in (3.3) is later identified with b_{1,0,0}; this identification should be made explicitly at the point where D is introduced.","section":"Eq. (3.3)"},{"comment":"The multi-index notation z^{(...)} bar z^{(...)} is used without a prior definition; define it or state that it means the product of the two monomials.","section":"Eq. (3.6)"},{"comment":"Reference [1] is listed with only an arXiv identifier and no publication data; complete the bibliographic information if available.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The result is plausible and the gap is localized to the triangularity argument in the induction step. I would be willing to review a revision that supplies a complete proof of the cross-term index balance or a verified small-N computation. The external-line-data assumption is a clear scope limitation, not a defect, but the title and abstract should continue to frame the result as a Cauchy-type inverse problem rather than a pure spectral rigidity theorem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this paper proves a genuinely new conditional inverse result. For a 2D Schrödinger operator with only one reflection symmetry, it shows that the first two layers of the quantum Birkhoff normal form, plus the sign of a30 and the transverse-line data {a_{1,2k}}, determine the full Taylor series of V. That is a real step beyond Hezari and Guillemin–Uribe, both of which needed more symmetry or a restrictive structural form. The explicit identity a30^2 = 2v1 b_{1,0,0} + (v2^2/(v1^2-4v2^2)) a12^2 is a nice concrete hook, and the base case is worked out in detail.\n\nCredit where due: the paper is honest about what it does and does not do. Remark 1.2 explicitly labels the problem as a Cauchy problem with prescribed transverse-line data, not pure spectral rigidity. The algebraic inversion is direct, no fitted parameters. The appendix computes the main diagonal contributions to the recurrence, and the displayed coefficient 64 a30 n(n-1)^2 shows why the backward recursion is invertible.\n\nThe soft spot is the induction step in §3.2. The proof asserts that off-diagonal, cross-derivative contributions to the resonant projection only involve coefficients of U_{2N-1} with larger second index, making the recurrence triangular. That claim is stated in a sentence, and the term Γ_m is never computed. Equation (3.14) also lists \"known terms\" that include ℏ^2-layer Poisson contributions such as -1/2{S^2_{2N-1}, U3}, without explicit bookkeeping. The stress-test check of the index balance suggests the triangularity claim is actually correct — the dangerous p=(2,1) pattern cannot occur with the available z1-derivative count — but the paper as written does not prove it. This is a gap, not a demonstrated error. A referee should ask for the N=3 step written out, or for a bound on Γ_m.\n\nThe transverse-line data requirement is a limitation for anyone hoping for pure spectral rigidity, but it is acknowledged and framed correctly.\n\nWho is this for: specialists in semiclassical inverse spectral theory. It deserves a serious referee. The architecture is sound and the missing verification is likely fixable. Send it to review, with the request that the induction step be expanded.","headline":"Genuinely new conditional inverse result with a fully worked base case; the induction step needs to be repaired before the result is fully reliable.","tokens_in":13351,"tokens_out":2991,"would_cite":false,"duration_ms":31060,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35R30","81Q20","37J40"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a 2D Schrödinger well with one reflection symmetry, the first two quantum normal-form layers determine the entire Taylor series of the potential, provided the sign of $a_{30}$ and the transverse-line data $\\{a_{1,2k}\\}$ are also known.","keywords":["semiclassical Schrödinger operator","quantum Birkhoff normal form","inverse spectral problem","single reflection symmetry","Taylor series recovery","Moyal product","resonant terms","transverse-line data"],"falsifier":"Construct two explicit $C^\\infty$ potentials sharing the same $v_1,v_2$ with $v_1/v_2\\notin\\mathbb{Q}$, the same sign of $a_{30}$, and the same transverse-line data $\\{a_{1,2k}\\}$, but differing in a higher coefficient such as $a_{50}$; compute their quantum Birkhoff normal forms through the layers $b_{0,k,\\ell}$ and $b_{1,k,\\ell}$. The theorem asserts those layers must separate the two potentials, so a single example where they coincide would disprove it.","tokens_in":12364,"feed_emoji":"⚛","tokens_out":13523,"duration_ms":138195,"temperature":0.7,"pith_summary":"This paper proves a formal inverse theorem for two-dimensional semiclassical Schrödinger operators whose potential well is symmetric about a single axis. The author shows that the first two layers of the quantum Birkhoff normal form—the classical coefficients $b_{0,k,\\ell}$ and the $\\hbar^2$ coefficients $b_{1,k,\\ell}$—determine the entire Taylor series of the potential at the bottom of the well, provided one also knows the sign of the cubic coefficient $a_{30}$ and the values of the first transverse derivative along the symmetry axis, $\\partial_{x_1}V(0,x_2)$. If the claim is correct, it widens the class of potentials that can be recovered from the spectrum: full evenness is no longer required, and the $\\hbar^2$ quantum layer compensates for the missing symmetry. The result is formal and local; it concerns Taylor coefficients, not global determination of the potential.","feed_headline":"First two quantum layers recover a partially symmetric 2D potential","feed_subtitle":"Once the sign of a30 and the data on the symmetry axis are given, the spectrum fixes all Taylor coefficients.","key_machinery":"The central object is the quantum Birkhoff normal form (QBNF), a formal Weyl symbol $B = H_2 + \\sum_{2r+k+\\ell\\ge 2} b_{r,k,\\ell}\\;\\hbar^{2r}\\Omega_1^k\\Omega_2^\\ell$ in the two harmonic-action variables $\\Omega_j = x_j^2+\\xi_j^2$. It is obtained from the Schrödinger symbol by unitary conjugation, and its coefficients are spectral invariants. The proof's engine is the third-order Moyal bracket $\\{\\cdot,\\cdot\\}_3$ together with the resonant projection $\\Pi$ onto the kernel of $L(\\cdot)=\\{\\cdot,H_2\\}$, which is exactly the polynomial algebra spanned by $\\Omega_1,\\Omega_2$ when $v_1/v_2$ is irrational. Two structural facts carry the argument: Lemma 3.1, that every even-degree homogeneous component of the potential is determined by its resonant projection (the relevant binomial coefficients are always nonzero), and a backward recursion at odd degrees in which the $\\hbar^2$ layer of the normal form yields an equation whose coefficient is $64\\,(a_{30}/v_1)\\,2^{-(2N-1)}\\binom{2m}{m}\\binom{2n-1}{n}n(n-1)^2$ times the unknown $a_{2N-1-2m,2m}$. Because $a_{30}\\neq 0$, this coefficient never vanishes along the steps where the recursion actually runs, so each Taylor coefficient is solved uniquely.","core_discovery":"Theorem 1.1 states that, under the assumptions that $V$ has a non-degenerate minimum at the origin, is reflection-symmetric in $x_2$, and has harmonic frequencies $v_1,v_2$ with $v_1/v_2\\notin\\mathbb{Q}$, the coefficients $b_{0,k,\\ell}$ and $b_{1,k,\\ell}$ of the quantum Birkhoff normal form (the classical layer and the $\\hbar^2$ layer) uniquely determine the full Taylor series of $V$, once the sign of $a_{30}$ and the sequence $\\{a_{1,2k}\\}_{k\\ge1}$ are prescribed. The explicit starting relation is $a_{30}^2 = 2v_1 b_{1,0,0} + \\frac{v_2^2}{v_1^2-4v_2^2} a_{12}^2$. The proof then proceeds by induction on the degree: at each odd degree $2N-1$ the $\\hbar^2$ layer of the normal form gives a nonsingular linear equation (whose coefficient is proportional to $a_{30}$) that determines the unknown coefficient $a_{2N-1-2m,2m}$ by a backward recursion starting from the known data $a_{1,2N-2}$, and at each even degree $2N$ the classical layer fixes the resonant projection of $U_{2N}$, from which Lemma 3.1 recovers the whole homogeneous piece $U_{2N}$. As a direct consequence (Corollary 1.3), the semiclassical eigenvalues $E(\\hbar)$ themselves determine the Taylor series once the same auxiliary data are given.","pith_inferences":["Because the transverse-line data are prescribed rather than recovered, a natural test is to search for two potentials with identical first two QBNF layers and the same sign of $a_{30}$ that differ in $a_{12}$ (with $b_{1,0,0}$ adjusted via relation (1.6)) and in higher $a_{1,2k}$; if such a pair exists, the auxiliary data are genuinely necessary.","The recursion coefficient vanishes exactly at $n=1$, the step supplied by the transverse-line data; this suggests the algebra leaves the highest-order odd coefficient unconstrained at every degree without that input, which is why the problem is structurally a Cauchy problem rather than pure spectral rigidity.","The complex-coordinate Diophantine matching used here should generalize to potentials in $d$ dimensions symmetric in $d-1$ coordinates, with the analogous data prescribed on a codimension-one hyperplane; the same combinatorics (products of nonvanishing binomial coefficients) would drive the recursion."],"forward_implications":["Corollary 1.3: the family of eigenvalues $E(\\hbar)$ in the interval $[0,\\delta]$ uniquely determines the full Taylor series of $V$ at the origin, given the sign of $a_{30}$ and the transverse-line data $\\{a_{1,2k}\\}$.","The reconstruction is constructive: order by order, the normal-form coefficients yield explicit linear equations (relation (1.6) and the degree-$2N$ recursion), so each Taylor coefficient is computed without solving nonlinear systems.","The class of recoverable potentials strictly widens Hezari's class: the cross-odd terms $a_{1,2k}$ may be nonzero, and the auxiliary symmetry $V(x_1,0)=V(-x_1,0)$ used in earlier 2D results is replaced by the prescribed transverse-line data.","The non-resonance condition $v_1/v_2\\notin\\mathbb{Q}$ plays a double role: it guarantees the quantum Birkhoff normal form exists and is spectrally determined, and it implies $v_1^2-4v_2^2\\neq 0$, so all formulas are well defined."],"supporting_citations":[{"why":"Supplies the quantum Birkhoff normal form construction through formally conjugate transformations, on which the whole analysis rests.","marker":"[9]"},{"why":"Provides the form and spectral interpretation of the quantum Birkhoff normal form as a formal series in $\\hbar$ and the harmonic actions.","marker":"[10]"},{"why":"Establishes that the eigenvalues near the bottom of the well determine the QBNF coefficients, converting the spectral problem into an algebraic one.","marker":"[5]"},{"why":"Gives the same spectral-invariant reduction to the quantum Birkhoff normal form that Theorem 1.1 exploits.","marker":"[1]"},{"why":"Defines an earlier class of potentials with no symmetry hypothesis that the present result extends by allowing nonzero $a_{1,2k}$.","marker":"[8]"},{"why":"Provides the prior two-dimensional partial-symmetry result that still required symmetry in both coordinates, which the present theorem weakens.","marker":"[7]"},{"why":"Shows in one dimension that the quantum ($\\hbar^2$) part of the normal form can compensate for missing evenness, the template for the induction used here.","marker":"[2]"}],"fun_headline_variants":["Two spectral layers recover a 2D potential from Birkhoff data","First two Birkhoff layers uniquely fix a symmetric 2D potential","Quantum normal form layers determine potential's Taylor series","Partially symmetric 2D potential recovered from two quantum layers","Cubic term unlocks inverse recovery for symmetric Schrödinger wells"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The transverse-line data $\\{a_{1,2k}\\}$, equivalently the function $\\partial_{x_1}V(0,x_2)$, must be supplied from outside the spectrum; without it the backward induction cannot start, because the highest-order unknown in each odd potential layer is only fixed by that condition.","fun_headline_variants_meta":{"raw":{"variants":["Two spectral layers recover a 2D potential from Birkhoff data","First two Birkhoff layers uniquely fix a symmetric 2D potential","Quantum normal form layers determine potential's Taylor series","Partially symmetric 2D potential recovered from two quantum layers","Cubic term unlocks inverse recovery for symmetric Schrödinger wells"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000929,"raw_usage":{"total_tokens":4088,"prompt_tokens":1163,"completion_tokens":2925,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":779,"completion_tokens_details":{"reasoning_tokens":2841}},"tokens_in":779,"tokens_out":2925,"duration_ms":22711,"temperature":1.0,"reasoning_tokens":2841,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:22:37.897720+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct two explicit $C^\\infty$ potentials sharing the same $v_1,v_2$ with $v_1/v_2\\notin\\mathbb{Q}$, the same sign of $a_{30}$, and the same transverse-line data $\\{a_{1,2k}\\}$, but differing in a higher coefficient such as $a_{50}$; compute their quantum Birkhoff normal forms through the layers $b_{0,k,\\ell}$ and $b_{1,k,\\ell}$. The theorem asserts those layers must separate the two potentials, so a single example where they coincide would disprove it.","supporting_citations":[{"cited_title":"Iantchenko, J","cited_arxiv_id":null,"evidence_quote":"Supplies the quantum Birkhoff normal form construction through formally conjugate transformations, on which the whole analysis rests."},{"cited_title":"Vu Ngoc,The quantum Birkhoff normal form and spectral asymptotics, Journ´ ees ´Equations aux d´ eriv´ ees partielles (2006), 1–12","cited_arxiv_id":null,"evidence_quote":"Provides the form and spectral interpretation of the quantum Birkhoff normal form as a formal series in $\\hbar$ and the harmonic actions."},{"cited_title":"Bottom of the well","cited_arxiv_id":null,"evidence_quote":"Establishes that the eigenvalues near the bottom of the well determine the QBNF coefficients, converting the spectral problem into an algebraic one."},{"cited_title":"The semi-classical spectrum and the Birkhoff normal form","cited_arxiv_id":"0902.2470","evidence_quote":"Gives the same spectral-invariant reduction to the quantum Birkhoff normal form that Theorem 1.1 exploits."},{"cited_title":"Hezari,Inverse spectral problems for Schr¨ odinger operators, Comm","cited_arxiv_id":null,"evidence_quote":"Defines an earlier class of potentials with no symmetry hypothesis that the present result extends by allowing nonzero $a_{1,2k}$."},{"cited_title":"Guillemin and A","cited_arxiv_id":null,"evidence_quote":"Provides the prior two-dimensional partial-symmetry result that still required symmetry in both coordinates, which the present theorem weakens."},{"cited_title":"Colin de Verdi` ere and V","cited_arxiv_id":null,"evidence_quote":"Shows in one dimension that the quantum ($\\hbar^2$) part of the normal form can compensate for missing evenness, the template for the induction used here."}],"review_version":1}