{"id":"08782317-8bdf-425d-84f8-02fa7976edad","arxiv_id":"2607.17159","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every q=αγ>1 and every pair of shifts, the shifted lattice cross on the positive hyperbola branch has an infinite-dimensional Fourier pre-annihilator: uniqueness fails.","lead":"For a hyperbola branch and shifted lattice crosses with density product q>1, this paper proves the Fourier pre-annihilator is always infinite-dimensional, so uniqueness never holds. The result closes an open problem and supplies a spectral normal form that may guide Fourier-interpolation and Klein–Gordon studies.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; the external periodization lemma is standard and does not threaten the proof.","rationale":"The reader correctly identifies the one-sided periodization lemma as a dependency, but not as a live weakness: the lemma is a standard Fourier/periodization fact and the proof sketch in Lemma 2.3 supplies the convergence needed. I independently checked the sign propagation from (2)-(3) through scaling, y=q/x substitution, and the tail elimination leading to (13); the algebra is consistent. The essential spectral radius argument (Theorem 3.3) is careful about the non-uniform weak constants, and Proposition 4.6's cocycle rigidity correctly handles the unimodular phases; the fixed-point/power-limit step is valid because BV right limits exist. Proposition 4.7's untwisted compatibility condition is exactly the mean-zero identity (34), and the Fredholm-plus-kernel argument closes. No step uses the data to choose parameters, and the paper's own limitations (q=1, intrinsic Green-domain description, Remark 5.8 equality) are stated. Hence the ACCEPT verdict stands unchanged.","tokens_in":30666,"tokens_out":36379,"duration_ms":313448,"concrete_test":"Reprove Lemma 2.3 without citing [15]: for F in L1(R+), compute int_0^1 (sum_{k>=0}F(x+k)) e^{-2pi i m x} dx = int_0^infty F(t)e^{-2pi i m t}dt and use completeness of exponentials in L1(0,1) to conclude equivalence; repeat for G2. If this direct derivation matches (7) and yields equation (13) with the stated signs, the load-bearing bridge is verified independently.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central chain from the shifted cross to nonuniqueness is traced: Lemma 2.3 (one-sided periodization) -> graph equation (13) -> Theorem 3.3 (twisted Lasota-Yorke) -> Proposition 4.6 (peripheral rigidity) -> Proposition 4.7 (BV extension). The only external step singled out by the reader, [15, Lemma 2.1], is actually elementary in the form used: for F in L1(R+), vanishing of all Fourier coefficients in m on (0,1) is equivalent to the one-sided periodization F0(x)=sum_{k>=0}F(x+k) vanishing a.e.; the paper's own Tonelli argument proves the required absolute convergence. The other imported result, [4, Theorem C], is invoked with the relevant hypotheses (partial filling, uniform expansion, second-derivative bound) checked explicitly in Lemma 4.5. I found no internal contradiction, no circularity, and no hidden parameter fitting. The strongest claim (i)-(iii) follows from the displayed propositions; Section 5's Green-domain results are extra and not needed for the HUP conclusion. Thus the concern about an unproved bridge does not land as a correctness risk.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Heisenberg uniqueness pairs for the positive hyperbola branch and shifted lattice crosses in the supercritical regime q=αγ>1. Theorem 1.1 states that every BV function on (1,q) is the middle restriction of a global BV pre-annihilator; that this extension is unique when (ζ,η)≠(1,1) and has a one-dimensional ambiguity in the untwisted case; and that the normalized pre-annihilator is infinite-dimensional, so the pair is not a Heisenberg uniqueness pair. The proof reduces the shifted annihilation conditions to a graph equation for a twisted Perron–Frobenius operator, proves a phase-uniform Lasota–Yorke inequality, classifies the unit-circle point spectrum, and derives an exact operator-theoretic normal form for the entire L^1 pre-annihilator via the maximal Green domain. Section 5 contains additional spectral results, including closed-disk L^1 spectrum, nonclosed range of I−Q, and exact norm 2N for the Green partial sums outside a countable exceptional set.","tokens_in":30882,"tokens_out":56100,"duration_ms":430677,"significance":"The result settles the infinite-dimensionality clause of [9, Open Problem 1.4] for arbitrary shifts and all q>1, a substantial extension of [4,7,8,9]. The proof is detailed and internally coherent: the chain from periodization to graph equation, twisted Lasota–Yorke, Fredholm index, peripheral rigidity, and BV extension is carefully executed. Strengths include the phase-uniform variation estimate, the complete peripheral spectral classification, the explicit invariant density for integer q, and the exact norm computation for Green sums. The Section 5 normal form is an interesting bonus, though not required for Theorem 1.1.","major_comments":[],"minor_comments":[{"comment":"D is defined as the open unit disk, but (42) asserts σ(Q;X)=σ_{e,F}(Q;X)=D. A spectrum is closed; the proof actually establishes σ(Q)=σ_{e,F}=overline{D}. Please correct the notation and the inclusion chain near the end of the proof (D⊂σ_{e,F}⊂σ(Q)⊂overline{D}, not ⊂D).","section":"§5.2, Proposition 5.5"},{"comment":"The equality should read \\overline{\\mathrm{Ran}(I−Q)}=\\ker\\Pi (with the displayed spaces). As written, equality without the closure contradicts the immediately following statement that Ran(I−Q) is a proper dense subspace.","section":"§5.2, Eq. (43)"},{"comment":"The load-bearing equivalence with the one-sided periodization is imported from [15, Lemma 2.1]. The paper proves the required absolute convergence by Tonelli, so the application is sound, but including a short proof or explicit statement of the corrected lemma would improve self-containedness.","section":"§2.4, Lemma 2.3"},{"comment":"The phrase 'the new range in Corollary 4.8 is 1<γ≤b' is slightly misleading because the corollary yields nonuniqueness for all γ>1; the novelty is indeed the range up to b, but the sentence could be phrased more precisely.","section":"§4, Corollary 4.8"}],"recommendation":"minor_revision","confidential_remarks":"The central HUP result (Theorem 1.1) is convincing and the paper is suitable for the journal. The Section 5 results are intriguing but contain a couple of notation/statement errors that should be fixed. I do not see a need for additional experiments or numerics."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Zhiqiang Wan's paper is the real thing: it removes the shift restrictions in the supercritical regime q>1 for the positive hyperbola branch and proves the infinite-dimensionality clause of Giri–Manna's open problem. The main theorem is the BV extension dichotomy: every BV middle datum extends, uniquely unless both phases are trivial, with an affine line in the untwisted case. That is a substantive advance over the unshifted and specially shifted cases in [4,7,9]. The proof chain—periodization to a twisted graph equation, Lasota–Yorke with uniform phases, Fredholm index, peripheral spectral rigidity, and the Green-domain normal form—is coherent and detailed. The extra material in Section 5 (closed unit disk spectrum, non-closed range, norm 2N of the partial sums outside a countable algebraic set) is a genuinely nice operator-theoretic payoff.\n\nThe reader's flagged concern about the imported one-sided periodization lemma [15, Lemma 2.1] does not land. The lemma is elementary in the form used here; the paper's Tonelli argument proves the needed absolute convergence, and the equivalence is just Fourier coefficients of the periodized function on (0,1). The other imported result, [4, Theorem C], is invoked after explicit verification of its hypotheses in Lemma 4.5. I found no circularity and no parameter fitting.\n\nSoft spots are mostly expository. The paper is long and terse in places; Lemma 3.1's proof of finiteness of cylinder images is a bit quick, and Lemma 3.2's coefficient inequalities are asserted without proof. Neither is load-bearing. The maximal Green domain is characterized only operator-theoretically, not in a standard function space; the authors say so themselves. The critical case q=1 remains open, and they do not overclaim it. The citation pattern is appropriate: the zero-middle generator is attributed to prior work, and the new contribution is clearly separated.\n\nThis paper deserves a serious referee. I would send it out. Minor suggestions: make the paper self-contained for the periodization lemma (it is two lines), and expand the two short proofs above. The result is important for the HUP subfield and the proof looks sound.","headline":"A solid, genuinely new proof of arbitrary-shift supercritical nonuniqueness for a hyperbola branch; the external-lemma worry is a non-issue and the paper merits refereeing.","tokens_in":31393,"tokens_out":6206,"would_cite":true,"duration_ms":55940,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B10","37C30","37A46","47A10","47A53"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that on the positive hyperbola branch, every shifted lattice cross with q>1 fails to be a Heisenberg uniqueness pair, because every bounded-variation density on the middle interval extends to a global pre-annihilator.","keywords":["Heisenberg uniqueness pair","shifted lattice cross","hyperbola branch","pre-annihilator","Perron-Frobenius operator","Lasota-Yorke inequality","Klein-Gordon equation","Green domain"],"falsifier":"Take a compactly supported G∈L¹(R₊) and compute the one-sided sum S(x)=Σ_{k≥0} G(x+k); if the Fourier integrals over all integer frequencies vanish but S is nonzero on a positive-measure set, the periodization lemma is false and the bridge breaks. Alternatively, for q=2 and trivial shifts, test the constant function v=1 on (1,2) numerically: if it does not admit a global BV pre-annihilator satisfying both periodization identities, Theorem 1.1(i) is contradicted.","tokens_in":30512,"feed_emoji":"🌀","tokens_out":6083,"duration_ms":59103,"temperature":0.7,"pith_summary":"The paper studies Heisenberg uniqueness pairs: a curve and a frequency set such that any measure on the curve whose Fourier transform vanishes on the set must be zero. For the positive branch of the hyperbola and a lattice cross (two perpendicular lattices of frequencies, each possibly shifted), the question is whether the pair is unique when the product of the two lattice densities q=αγ exceeds 1. The paper proves that for every q>1 and for arbitrary shifts of both arms, the answer is no: the pre-annihilator is infinite-dimensional, so the pair is never a Heisenberg uniqueness pair. In fact, every bounded-variation function on the middle interval (1,q) extends to a global pre-annihilator; the extension is unique unless both twist phases are trivial, in which case the ambiguity is exactly one dimension. The proof reduces the annihilation conditions to a spectral problem for a twisted Perron–Frobenius operator, yielding an exact L^1 normal form in terms of a maximal Green convergence domain.","feed_headline":"Nonuniqueness holds for every shifted lattice cross with q>1","feed_subtitle":"Every BV density on the middle interval extends, settling the arbitrary-shift open problem.","key_machinery":"The central object is the twisted Perron–Frobenius operator Q_{q,ζ,η} = L_{ζ,q} L_{η,q} associated with the Gauss-type map T_q(x)={q/x}, together with the forcing operator A_{q,ζ,η}v = −S_ζ v + L_{ζ,q}(S_η J_q v). The annihilation conditions on the shifted lattice cross are equivalent to the graph equation (I−Q)F₀ = A v on L¹((0,1)); the middle datum v is free, and the tail is then determined by the weighted inversion/periodization map B_{η,q}. The proof makes Q quasi-compact on BV via a phase-uniform Lasota–Yorke inequality (essential spectral radius ≤ q^{−2}) and rules out all unit-circle eigenvalues except the trivial phase case by iterating a full-branch phase equation. This yields the e","core_discovery":"Theorem 1.1: for q=αγ>1 and arbitrary real shifts θ₁,θ₂, the restriction map from the BV core A^{BV}_{q,ζ,η} to BV((1,q)) is surjective; the extension is unique unless both twisting phases ζ,η equal 1, in which case the extensions form a one-dimensional affine line. Consequently the normalized pre-annihilator is infinite-dimensional and the pair (Γ₊, Λ_{α,γ,θ₁,θ₂}) is not a Heisenberg uniqueness pair. At the L¹ level, the full pre-annihilator is characterized exactly by the graph equation (I−Q)F₀ = A v, parameterized by the maximal Green domain; outside a countable set of algebraic q>1, the Green partial sums have norm 2N for every N≥1.","pith_inferences":["The corrected one-sided periodization lemma is imported without proof; if that lemma were false at any level of generality, the operator reduction and the theorem would lose their foundation. A self-contained proof of the lemma would make the result more robust.","The norm identity for the Green sums indicates a natural divergence boundary: in generic supercritical cases the maximal Green domain is meagre, so the L¹ pre-annihilator is typically not all of L¹((1,q)).","The explicit zero-middle generator Ψ_q in the untwisted case (two unit-mass lobes for integer q) provides a concrete function that could seed constructions of boundary HUP pairs at q=1 or in neighboring settings.","The phase-uniform Lasota–Yorke estimate should transfer to other partially filling countable-branch maps with a single truncated edge branch, giving analogous arbitrary-shift nonuniqueness results for other curves."],"forward_implications":["For every q>1 and arbitrary real shifts of both arms, the positive hyperbola branch and the shifted lattice cross do not form a Heisenberg uniqueness pair; the infinite-dimensionality clause of the open problem is settled.","Every BV datum on the middle interval (1,q) admits a global BV pre-annihilator; when at least one twist phase is nontrivial the extension is unique, and in the untwisted case the ambiguity is exactly a one-dimensional affine line.","The full L¹ pre-annihilator is described exactly by the maximal Green domain, and outside a countable set of algebraic q the normalized Green sums have norm 2N for every N.","The operator Q has the closed unit disk as its spectrum on L¹((0,1)) and Ran(I−Q) is a proper dense subspace, explaining why the BV inverse does not extend boundedly to all of L¹.","The special one-arm rational case (α=1, θ₁=2/p, θ₂=0) is infinite-dimensional for every integer p≥2 and γ>1, including the previously unresolved range 1<γ≤p."],"fun_headline_variants":["Arbitrary-shift Heisenberg uniqueness fails for hyperbola crosses","Infinite-dimensional pre-annihilator settles arbitrary-shift problem","Supercritical nonuniqueness for all shifted lattice crosses","Every BV density extends, solving the arbitrary-shift open problem","Shifted lattice crosses: nonuniqueness holds for every q>1"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing bridge is the corrected one-sided periodization lemma, imported from the literature without proof, that converts the two Fourier-vanishing families into almost-everywhere identities; if this lemma fails, the operator reduction and Theorem 1.1 collapse.","fun_headline_variants_meta":{"raw":{"variants":["Arbitrary-shift Heisenberg uniqueness fails for hyperbola crosses","Infinite-dimensional pre-annihilator settles arbitrary-shift problem","Supercritical nonuniqueness for all shifted lattice crosses","Every BV density extends, solving the arbitrary-shift open problem","Shifted lattice crosses: nonuniqueness holds for every q>1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000224,"raw_usage":{"total_tokens":1349,"prompt_tokens":849,"completion_tokens":500,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":593,"completion_tokens_details":{"reasoning_tokens":410}},"tokens_in":593,"tokens_out":500,"duration_ms":4958,"temperature":1.0,"reasoning_tokens":410,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T18:49:08.781939+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a compactly supported G∈L¹(R₊) and compute the one-sided sum S(x)=Σ_{k≥0} G(x+k); if the Fourier integrals over all integer frequencies vanish but S is nonzero on a positive-measure set, the periodization lemma is false and the bridge breaks. Alternatively, for q=2 and trivial shifts, test the constant function v=1 on (1,2) numerically: if it does not admit a global BV pre-annihilator satisfying both periodization identities, Theorem 1.1(i) is contradicted.","supporting_citations":[],"review_version":1}