{"id":"32ae83c4-fc90-4f13-8354-b5756ec27ec0","arxiv_id":"2505.22485","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Averaged spectral measures of a class of random Schrödinger operators coincide with Plancherel measures of convolution operators on wreath products, yielding new spectral criteria and Lifshitz tail estimates.","lead":"This paper proves a spectral bridge between certain random Schrödinger operators and deterministic convolution operators on wreath product groups, generalizing the known lamplighter correspondence from binary disorder to arbitrary potential distributions. The bridge yields a new criterion for absolutely continuous spectra on wreath products, Lifshitz tail estimates for polynomial growth groups, and an exact formula for the second moment of the Green function.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Conditional theorem sound, but potential-class hypothesis is restrictive; Theorem 2.3 lower-bound proof has a sign error that should be corrected.","rationale":"The reader's verdict of CONDITIONAL is appropriate. The main theorems are conditional statements whose proofs appear correct; the identified weakest assumption—the restrictive nature of the disorder distribution—is a genuine scope limitation, acknowledged by the author in Remark 3.3 but not reflected in the abstract or the title of Theorem 2.3. The additional sign error in the lower-bound proof of Theorem 2.3 is a concrete flaw that should be corrected, but it is localized and unlikely to change the mathematical result. Since the stress-test finds no error in the central correspondence itself, the reader's CONDITIONAL verdict stands without modification. The proposed concrete tests would settle (1) whether the sign error is merely typographical and (2) whether the claimed Lifshitz-tail theorem actually covers natural disorder distributions, thereby clarifying the paper's true scope.","tokens_in":12377,"tokens_out":20009,"duration_ms":226763,"concrete_test":"Re-derive the lower-bound step in Theorem 2.3 with the corrected inequality ¯µ_H(I_ε) ≥ k(t) − e^{−εt}, choose εt = 2c t^{d/(d+2)} log^α(t), and verify that the conclusion ¯µ_H(I_ε) ≥ (C/2) exp(−c' ε^{−d/2+o(1)}) still follows; if so, the sign error is typographical and Theorem 2.3 is sound. Separately, test the scope restriction: take the uniform distribution on [-1,1] (even moments 1/(2k+1)) and compare with the moment sequence of µ_B^{eΛ} for any symmetric ℓ¹ measure m on a finitely generated Abelian group, whose moments are return probabilities p_k of a random walk on Λ (e.g. p_{2k} ~ C k^{-d/2} on Z^d). If no such m reproduces the uniform moments, then uniform disorder lies outside the correspondence class, confirming that the abstract and Theorem 2.3 title overstate generality.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central correspondence (Theorems 1.1 and 1.2) is correct as a conditional statement: if the i.i.d. potential has distribution equal to the spectral measure µ_B^{eΛ} of a convolution operator B = L_{m_Λ} with m_Λ ∈ ℓ¹, then the averaged spectral measure equals the Plancherel measure of the wreath-product operator M = L_{m̂_Γ + m̂_Λ}. The proof via moment expansions is solid, and the direct integral construction for Abelian Λ is valid. The load-bearing weakness is the hypothesis itself: requiring the single-site disorder to be exactly the spectral measure of some ℓ¹ convolution operator on a lamp group is a strong modelling constraint, not satisfied by generic bounded i.i.d. potentials. Remark 3.3 explicitly concedes this, noting that the Fourier representation may fail to be ℓ¹. Consequently, the abstract's unqualified 'random Schrödinger operators' and the title 'Lifshitz tail on any polynomial growth group' overstate the scope: Theorem 2.3 applies only to potentials in the correspondence class, not to e.g. uniform or Gaussian disorder. Additionally, the proof of Theorem 2.3's lower bound contains a sign error: the step '¯µ_H(I_ε) ≥ −e^{εt} + C e^{−ct^{d/(d+2)} log^α(t)}' should read '¯µ_H(I_ε) ≥ k(t) − e^{−εt}', i.e. the first term is e^{−εt}, not e^{εt}. As printed, the lower bound is vacuous. This is a concrete, localizable flaw in an advertised application, though it is likely typographical and fixable.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a spectral correspondence between random Schrödinger operators on a countable group Γ and deterministic convolution operators on the wreath product Λ≀Γ. Theorem 1.1 states that if H(ω)=A+V on ℓ²(Γ) has A=L_{mΓ} and i.i.d. single-site potential distributed as the spectral measure of B=L_{mΛ} at the identity, then the averaged spectral measure at eΓ equals the Plancherel measure of M=L_{m̂Γ+m̂Λ} on Λ≀Γ. Theorem 1.2 upgrades this, for finitely generated Abelian Λ, to unitary equivalence between M and the direct integral of H(ω) over the Pontryagin dual of the lamp configurations. Applications include an absolutely-continuous-spectrum criterion via Wegner estimates (Theorem 2.1), Lifshitz tail asymptotics for polynomial-growth base groups (Theorem 2.3), and a Fourier/Parseval formula for the second moment of the Green function (Proposition 2.5). Proofs are based on moment expansions and explicit Fourier transforms.","tokens_in":12643,"tokens_out":46565,"duration_ms":494115,"significance":"The central identity is exact and is proved by a clean moment computation, with the direct-integral version giving a constructive unitary. If correct, it provides a new bridge between random operator theory and group convolution, and the applications—especially the AC criterion and Lifshitz tails—are concrete and nontrivial. The proof of Theorem 1.1 is elementary and self-contained apart from standard spectral theory; Theorem 2.3 correctly uses Erschler's return-probability asymptotics as an external input. The main limitation is the modelling assumption that the disorder distribution coincides with the spectral measure of an ℓ¹ convolution operator on a lamp group, which is not satisfied by a generic bounded i.i.d. potential; this is acknowledged in Remark 3.3 but should be foregrounded. Overall the paper is a solid conditional contribution.","major_comments":[],"minor_comments":[{"comment":"The displayed lower bound '¯µ_H(I_ε) ≥ −e^{εt} + C e^{−ct^{d/(d+2)} log^α(t)}' should read '¯µ_H(I_ε) ≥ C e^{−ct^{d/(d+2)} log^α(t)} − e^{−εt}'. As printed the right-hand side is eventually negative and the bound is vacuous, while the following line already uses the corrected exponent e^{−εt}; this is a typographical error but it must be fixed because the lower bound is needed for the theorem.","section":"§2.2, proof of Theorem 2.3, lower bound"},{"comment":"The abstract and the heading 'Lifshitz tail on any polynomial growth group' overstate the scope: the results apply only when the single-site potential distribution is exactly the spectral measure μ_B^{eΛ} of some convolution operator B=L_{mΛ} with mΛ∈ℓ¹, and Remark 3.3 concedes that not every distribution can be realized in this way. Please state this hypothesis explicitly in the abstract and temper the 'random Schrödinger operators' phrasing accordingly.","section":"Abstract and Theorem 2.3"},{"comment":"The final sentence 'M is purely absolutely continuous operator' appears to require that the support of m_{Λ≀Γ} generates the group so that δ_{e_{Λ≀Γ}} is cyclic; otherwise the bounded-density conclusion applies only to the cyclic subspace. This is satisfied in Example 2.2 because ˆS⊔ˆT is a generating set, but the theorem statement should include an explicit generating or cyclicity assumption.","section":"Theorem 2.1"},{"comment":"In the upper-bound argument for Theorem 2.3, the log^α(t) factor is dropped when passing from k(t) to the bound '≤ 2Ce^{−ct^{d/(d+2)}+εt}'. The inequality is valid for large t since log^α(t)≥1, but the notation should be kept consistent so the reader can follow which constants absorb the logarithmic factor.","section":"§2.2, upper-bound chain"},{"comment":"Please correct minor typos: the section heading 'Lifchitz tail' should be 'Lifshitz tail'; Theorem 1.1 contains 'on of the form'; the abstract repeats 'Lifschitz'; and Theorem 2.3's assumption 'containing a finite generating set' should specify that the measure is symmetric and that the generating set is meant in the group-theoretic sense.","section":"Various"}],"recommendation":"minor_revision","confidential_remarks":"The paper is a solid conditional contribution; the central correspondence Theorems 1.1 and 1.2 appear correct, and the applications are interesting. The potential-class restriction is the main caveat and should be prominently disclosed in the abstract and introduction. The sign typo in the Lifshitz proof is easily fixed. I support publication after a minor revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Core is solid; the packaging is too loud. Theorem 1.1 is a clean generalization of the lamplighter–percolation correspondence, and I believe the proof. The moment expansion on closed walks is the right way to see it, and the direct integral version for Abelian lamps (Theorem 1.2) is a genuine strengthening, not just a reinterpretation. The applications are real: the ac criterion via Wegner estimate works, the Green function second moment/delocalization dictionary is attractive, and using Erschler's return probabilities to get Lifshitz tails is a nice transfer from group theory to random operators.\n\nThe main caveat is structural: the correspondence only holds when the single-site disorder is exactly the spectral measure of an ell^1 convolution operator on the lamp group. That is a real modelling constraint, and Remark 3.3 concedes that not every distribution has this form. So the abstract's unqualified 'random Schrödinger operators' and the title of Theorem 2.3 overstate the scope. The result is still interesting for the correspondence class, but it is not a Lifshitz tail theorem for arbitrary bounded i.i.d. potentials on polynomial-growth groups.\n\nThere is also a concrete sign error in the lower bound of the Theorem 2.3 proof: the inequality should subtract e^{-\\varepsilon t}, not e^{\\varepsilon t}. As printed the bound is vacuous. It looks typographical—the next line already uses the correct e^{-\\varepsilon t+···}—but it has to be fixed.\n\nCitations look honest; Erschler, Pittet–Saloff-Coste and the Wegner estimate are used as inputs, not manufactured. No circularity. The paper is written carefully apart from the abstract/title overreach and a few typos.\n\nWho is this for? People working at the intersection of random Schrödinger operators and geometric group theory, especially anyone interested in spectral measures of wreath products. I would send it to a serious referee; with the sign fixed and the scope qualified, it should be publishable.","headline":"Core conditional theorems are sound and cleanly proven; the abstract overstates scope, and a sign error in the Lifshitz tail proof needs fixing.","tokens_in":13237,"tokens_out":3272,"would_cite":true,"duration_ms":33208,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H25","47B80","20E22","60B15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The averaged spectrum of a random Schrödinger operator is the Plancherel measure of a deterministic convolution operator on a wreath product.","keywords":["random Schrödinger operators","wreath products","convolution operators","Plancherel measure","density of states","Lifshitz tails","direct integral","lamplighter groups"],"falsifier":"For $\\Gamma = \\mathbb{Z}$, $\\Lambda = \\mathbb{Z}/2\\mathbb{Z}$, with $m_\\Gamma$ and $m_\\Lambda$ each uniform on $\\{\\pm 1\\}$, compute both sides of the moment identity (7) for $n = 1,2,3,4$: the left side is an explicit finite sum over closed walks on $\\mathbb{Z}$ with Bernoulli weights, and the right side is an explicit finite sum over closed walks on the lamplighter group. A disagreement for any single $n$ would refute Theorem 1.1; agreement for all $n$ is a direct check of the proof's closed-walk bookkeeping.","tokens_in":12099,"feed_emoji":"🎲","tokens_out":11781,"duration_ms":100442,"temperature":0.7,"pith_summary":"This paper establishes a spectral bridge between random Schrödinger operators $H(\\omega) = A + V$ on a countable group $\\Gamma$, with $V$ an i.i.d. potential, and deterministic convolution operators on the wreath product $\\Lambda \\wr \\Gamma$ built by attaching a lamp group $\\Lambda$ to every site of $\\Gamma$. The central identity (Theorem 1.1) says that the averaged spectral measure at the identity, i.e. the density of states of $H(\\omega)$, equals the Plancherel measure of the convolution operator $M = L_{\\hat m_\\Gamma + \\hat m_\\Lambda}$ on $\\Lambda \\wr \\Gamma$, provided the single-site potential is distributed according to the spectral measure of a convolution operator $B$ on $\\Lambda$. When $\\Lambda$ is finitely generated and Abelian, the correspondence is upgraded (Theorem 1.2) to a unitary equivalence between $M$ and the direct integral $\\int_\\Omega^\\oplus H(\\omega)\\,P(d\\omega)$. A sympathetic reader should care because the identity lets deterministic group-harmonic tools bear on random operators and, conversely, lets random-operator regularity tools bear on convolutions on wreath products; from it the paper derives an absolute-continuity criterion, Lifshitz tails with exponent $-d/2$ on any polynomial-growth base group, and an exact formula for the second moment of the Green function.","feed_headline":"Random disorder is the Plancherel measure of a wreath product","feed_subtitle":"Averaged spectra of disordered lattices equal the spectrum of one convolution operator on a wreath product, bridging two fields.","key_machinery":"The central object is the wreath product $\\Lambda \\wr \\Gamma = (\\bigoplus_\\Gamma \\Lambda) \\rtimes \\Gamma$, whose elements $(f,g)$ describe a lamplighter at position $g \\in \\Gamma$ with finitely many lamps $f: \\Gamma \\to \\Lambda$ switched on. The engine of the proof is a moment expansion: powers of $H(\\omega)$ enumerate closed walks on $\\Gamma$ with potential weights, while powers of $M$ enumerate closed walks on $\\Lambda \\wr \\Gamma$ where each visit of a potential is replaced by a lamp switch; identity (7) matches these sums term by term using the fact that the moments of the single-site potential are exactly the return moments of the lamp-group convolution operator. For Abelian lamps, the Fourier transform on $\\bigoplus_\\Gamma \\Lambda$ turns the configuration space into the probability space $\\Omega$ of the i.i.d. potential and yields the direct-integral unitary equivalence of Theorem 1.2.","core_discovery":"On its own terms, the discovery is that randomness and wreath-product geometry are two faces of the same spectral object. Theorem 1.1 identifies the density of states of $H(\\omega) = L_{m_\\Gamma} + V$, where the i.i.d. potential follows $\\mu^{e_\\Lambda}_{L_{m_\\Lambda}}$, with the Plancherel measure $\\mu^{e_{\\Lambda \\wr \\Gamma}}_{L_{\\hat m_\\Gamma + \\hat m_\\Lambda}}$ on the wreath product. Theorem 1.2 upgrades this to a unitary equivalence for finitely generated Abelian lamp groups: after Fourier transform on the lamp configurations, the deterministic operator $M$ on $\\Lambda \\wr \\Gamma$ is exactly the direct integral of $H(\\omega)$ over the probability space of potentials. The paper then uses the correspondence in both directions: a Wegner-estimate argument makes the density of states absolutely continuous with bounded density whenever the lamp-group spectral measure is, giving an absolute-continuity criterion for convolutions on wreath products; return-probability asymptotics on wreath products yield Lifshitz tails for the random operator on any polynomial-growth base group; and a reverse correspondence expresses matrix elements of $f(H(\\omega))$ as Fourier series over lamp configurations, with the Parseval identity giving the second moment of the Green function.","pith_inferences":["A consequence the paper leaves implicit is that any numerical method able to approximate Plancherel measures of convolution operators on finite truncations of wreath products becomes a numerical method for the density of states of the random operator, bypassing ensemble averaging entirely.","The restriction identified in Remark 3.3 suggests an inverse classification problem: characterize the probability measures on the real line that are spectral measures of $\\ell^1$ convolution operators on countable Abelian groups; the paper's results would apply exactly on that class.","The Green-function second-moment formula rephrases the open problem of low-disorder delocalization on amenable groups as a deterministic question about decay of matrix coefficients of resolvents of wreath-product convolution operators along the lamp factor, a potentially more tractable target than direct random-operator analysis.","The correspondence is likely to extend to generalized wreath products with site-dependent lamp groups, as the paper itself notes in Remark 1.3, which would cover inhomogeneous or correlated disorder."],"forward_implications":["The density of states of $H(\\omega)$ becomes a deterministic spectral object, so Wegner-type regularity estimates for random operators translate into absolute-continuity results for convolution operators on wreath products; in particular, if the lamp-group spectral measure is absolutely continuous with bounded density, then the Plancherel measure of $M$ is purely absolutely continuous (Theorem 2.1","Lifshitz tails hold on any infinite polynomial-growth base group $\\Gamma$ with the universal exponent $-d/2$, where $d$ is the growth degree of $\\Gamma$, for every potential in the correspondence class (Theorem 2.3).","The reverse correspondence gives an exact formula for the second moment of the Green function, $E|\\langle \\delta_{e_\\Gamma}, (H(\\omega)-z)^{-1}\\delta_{e_\\Gamma}\\rangle|^2$, as a sum of squared matrix elements of $(M-z)^{-1}$ over lamp configurations, linking delocalization of $H$ to decay of $M$'s eigenvectors along the lamp group (Proposition 2.5 and Remark 2.6).","The known lamplighter-to-Bernoulli correspondence is generalized from binary disorder to arbitrary spectral-measure disorder and from the lamplighter group to arbitrary base and lamp groups."],"supporting_citations":[{"why":"Establishes the earlier correspondence between lamplighter groups and percolation Hamiltonians that Theorem 1.1 generalizes.","marker":"[27]"},{"why":"Provides the first explicit computation of the spectrum of the lamplighter group, the prototype for wreath-product spectral analysis.","marker":"[19]"},{"why":"Supplies the spectral-representation theorem used to define the spectral measures at the core of the correspondence.","marker":"[33]"},{"why":"Supplies the direct-integral theory needed for the unitary-equivalence formulation in Theorem 1.2.","marker":"[10]"},{"why":"Provides the return-probability asymptotics for wreath products that drive the Lifshitz-tail exponent in Theorem 2.3.","marker":"[13]"},{"why":"Provides the Wegner estimate used to turn the density-of-states identity into the absolute-continuity criterion of Theorem 2.1.","marker":"[3]"},{"why":"Provides the heat-kernel and return-probability comparison used to convert wreath-product return probabilities into spectral-edge decay in the proof of Theorem 2.3.","marker":"[32]"},{"why":"Supplies the polynomial-growth degree and its integrality, which sets the exponent in the Lifshitz-tail statement.","marker":"[21]"}],"fun_headline_variants":["Averaged random spectra equal wreath-product convolution","Wreath-product convolutions reveal random operator spectra","Disorder's spectrum is wreath-product Plancherel measure","Spectral correspondence fuses random operators and wreath products"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results hold only when the single-site disorder distribution is exactly the spectral measure of an $\\ell^1$ convolution operator on some lamp group, a class that does not cover every bounded i.i.d. potential, as the paper itself notes in Remark 3.3.","fun_headline_variants_meta":{"raw":{"variants":["Averaged random spectra equal wreath-product convolution","Wreath-product convolutions reveal random operator spectra","Disorder's spectrum is wreath-product Plancherel measure","Spectral correspondence fuses random operators and wreath products"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000777,"raw_usage":{"total_tokens":3419,"prompt_tokens":910,"completion_tokens":2509,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":526,"completion_tokens_details":{"reasoning_tokens":2443}},"tokens_in":526,"tokens_out":2509,"duration_ms":20927,"temperature":1.0,"reasoning_tokens":2443,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:07:21.940102+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For $\\Gamma = \\mathbb{Z}$, $\\Lambda = \\mathbb{Z}/2\\mathbb{Z}$, with $m_\\Gamma$ and $m_\\Lambda$ each uniform on $\\{\\pm 1\\}$, compute both sides of the moment identity (7) for $n = 1,2,3,4$: the left side is an explicit finite sum over closed walks on $\\mathbb{Z}$ with Bernoulli weights, and the right side is an explicit finite sum over closed walks on the lamplighter group. A disagreement for any single $n$ would refute Theorem 1.1; agreement for all $n$ is a direct check of the proof's closed-walk bookkeeping.","supporting_citations":[{"cited_title":"On th e spectrum of lamp- lighter groups and percolation clusters","cited_arxiv_id":null,"evidence_quote":"Establishes the earlier correspondence between lamplighter groups and percolation Hamiltonians that Theorem 1.1 generalizes."},{"cited_title":"The lamplighter group as a group generated by a 2-state automaton, and its spectrum","cited_arxiv_id":null,"evidence_quote":"Provides the first explicit computation of the spectrum of the lamplighter group, the prototype for wreath-product spectral analysis."},{"cited_title":"Methods of modern mathematical physics, volume 1","cited_arxiv_id":null,"evidence_quote":"Supplies the spectral-representation theorem used to define the spectral measures at the core of the correspondence."},{"cited_title":"Les c [*]-alg´ ebres et leurs repr´ esentations","cited_arxiv_id":null,"evidence_quote":"Supplies the direct-integral theory needed for the unitary-equivalence formulation in Theorem 1.2."},{"cited_title":"Isoperimetry for wreath products of markov chains and multiplicity of selfintersections of random walks","cited_arxiv_id":null,"evidence_quote":"Provides the return-probability asymptotics for wreath products that drive the Lifshitz-tail exponent in Theorem 2.3."},{"cited_title":"Random operators, volume 168","cited_arxiv_id":null,"evidence_quote":"Provides the Wegner estimate used to turn the density-of-states identity into the absolute-continuity criterion of Theorem 2.1."},{"cited_title":"On the stability of the beha vior of random walks on groups","cited_arxiv_id":null,"evidence_quote":"Provides the heat-kernel and return-probability comparison used to convert wreath-product return probabilities into spectral-edge decay in the proof of Theorem 2.3."},{"cited_title":"Groups of polynomial growth and expanding ma ps (with an ap- pendix by jacques tits)","cited_arxiv_id":null,"evidence_quote":"Supplies the polynomial-growth degree and its integrality, which sets the exponent in the Lifshitz-tail statement."}],"review_version":1}