{"id":"f1dad275-44f6-4d1e-b2f7-15af0b9f261d","arxiv_id":"2512.17716","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Short-range antiferromagnetic correlations lower the interaction at which the two-particle charge vertex diverges in CDMFT, and that sign change is the prerequisite for the Mott transition.","lead":"Using cellular dynamical mean-field theory (CDMFT), the authors map where self-consistent perturbation theory breaks down in the two-dimensional Hubbard model and show that short-range magnetic fluctuations shift this breakdown to weaker interactions than the single-site DMFT picture. The same two-particle formalism links that breakdown to the Mott metal-insulator transition and to nearby phase-separation instabilities.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Cluster-size scaling absent: 2x2 CDMFT results may be finite-size artifacts for the lowered divergence line.","rationale":"The reader's weakest assumption was the two-particle periodization approximation (Eq. 14). While this is a genuine caveat, it is not the most load-bearing concern for the central claim. The divergence line in Fig. 5 is a property of the irreducible vertex of the impurity cluster, obtained before any periodization; the necessary-condition proof in Sec. IV B uses the impurity susceptibility and the hopping variance t^2_{q=0}, which is positive semidefinite, so the requirement of a negative impurity eigenvalue does not depend on the periodization scheme. The periodization mainly affects the identification of the specific eigenvalue of the periodized lattice susceptibility (Fig. 6) and the associated eigenvector projection, but the core 'prerequisite' statement survives. By contrast, the entire numerical case is built on a single 2x2 cluster. Finite-size effects are notoriously strong for antiferromagnetic correlations in the 2D Hubbard model, and the paper's own benchmark shows the 2x2 Néel temperature is higher than for 8x8. Therefore the systematic lowering of the divergence line could be quantitatively (or even qualitatively) altered by cluster size, and the absence of any cluster-size scaling leaves the central physical claim under-verified. The proposed 4x4 and 8x8 CDMFT calculations would directly test this. I agree with the reader's CONDITIONAL verdict; the concern reinforces the need for further verification without demonstrating an error, so no change in verdict is warranted. The absence of error bars and data release are secondary issues that also support the conditional assessment.","tokens_in":36762,"tokens_out":11956,"duration_ms":137772,"concrete_test":"Recompute the first charge-vertex divergence line and the leading impurity charge eigenvalue crossing using CDMFT with a 4x4 cluster (and, if feasible, 8x8) at the same temperatures (e.g., T/t = 1/15 and T/t = 0.4). Compare the location of the divergence line relative to DMFT and the position of the eigenvalue sign change relative to U_c2. If the lowering persists and the crossing still occurs before the MIT, the 2x2 result is robust; if the shift shrinks or reverses, the central claim is a finite-size artifact. Also compute the same quantities for an isolated 4x4 cluster to confirm the trend observed with the 2x2 plaquette.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central numerical claims rest entirely on a single 2x2 CDMFT cluster. The systematic lowering of the first charge-vertex divergence line relative to DMFT (Fig. 5) and the eigenvalue sign-change that precedes the Mott transition (Fig. 7) are demonstrated for this one cluster size at selected temperatures. A 2x2 cluster has a particularly strong antiferromagnetic response due to perfect nesting; the paper itself shows in Fig. 4 that the 2x2 Néel temperature is higher than that of 8x8 CDMFT. Thus the short-range AF fluctuations invoked to explain the lowering may be overestimated by the cluster geometry, and the 'systematic' lowering could be a finite-size artifact rather than a robust property of spatial correlations. The periodization caveat raised by the reader is less load-bearing: the vertex divergence line is computed directly from the impurity cluster before periodization, and the necessary-condition argument for a negative impurity eigenvalue relies on the positive-semidefinite hopping variance t^2_{q=0}, not on the periodized lattice susceptibility. Periodization mainly affects which specific eigenvalue is tracked in Sec. IV, not the existence of the prerequisite. Hence the missing cluster-size convergence is the more serious gap.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a two-particle Bethe-Salpeter formalism for cellular dynamical mean-field theory (CDMFT) on a real-space 2×2 cluster, including Ward identities in all channels and a detailed particle-particle BSE derivation. It computes the first divergence line of the two-particle irreducible charge vertex for the 2D Hubbard model at half-filling and finds that, relative to DMFT, the divergence occurs at systematically lower interactions, which it attributes to short-range antiferromagnetic spin fluctuations (analyzed via a single-boson-exchange decomposition). The paper then studies the Mott transition region at the two-particle level, showing that a single eigenvalue of the generalized charge susceptibility diverges at the transition, and argues that a sign change of the corresponding cluster-impurity eigenvalue—the same event as the vertex divergence—is a necessary prerequisite for the Mott transition and phase-separation instabilities. The work also connects the strong-coupling approximation of the lattice bubble to an effective hopping amplitude t2_eff,q.","tokens_in":37038,"tokens_out":3044,"duration_ms":36288,"significance":"If the central numerical claims hold, the paper is a significant contribution: it provides a consistent and carefully benchmarked two-particle CDMFT framework, with Ward identities and applied-field checks, and it extends the DMFT vertex-divergence program to include short-range spatial correlations. The formal BSE/Ward-identity sections are carefully derived and the small-cluster results are benchmarked against DiagMC and larger-cluster data, which increases credibility. The paper also makes a falsifiable prediction: in CDMFT, the charge-vertex divergence line is shifted to lower U than in DMFT, and the same eigenvalue sign change precedes the Mott transition. The machine-checkable derivations and reproducible benchmarks are notable strengths. However, the numerical support for the 'systematic' shift and the 'essential prerequisite' claim rests entirely on a single 2×2 cluster, and this is load-bearing for the central assertions.","major_comments":[{"comment":"The central claim that the vertex divergence line occurs 'systematically' at lower interactions in CDMFT than DMFT is supported only for a single cluster size (Nc=2×2). The paper's own Fig. 4 shows that the 2×2 cluster has a substantially higher Néel temperature than 8×8 CDMFT, so the short-range AF fluctuations invoked to explain the lowering may be quantitatively overestimated by the perfect-nesting geometry of the 2×2 plaquette. Without a cluster-size scaling test (e.g., 4×4 CDMFT at selected temperatures, or at least a systematic discussion of the 8×8 benchmark data), the assertion that this is a robust property of short-range correlations rather than a finite-size artifact is not established. The authors do discuss the isolated-cluster comparison, but that is a different limit and does not control the imprint of the embedding bath on the 2×2 geometry.","section":"Fig. 5 and §III.A"},{"comment":"The derivation of the necessary condition for the MIT relies on projecting onto the leading lattice eigenvector V∞ at Uc2. Equation (47) sums over impurity eigenvalues Ei with projections P∞(Xi). The plot in Fig. 7 shows that one eigenvalue E1 dominates at Uc2, but the subsequent tracking of X1U as a function of U (green line) must be protected against eigenvalue crossing or level repulsion. The authors state that they follow the eigenvector with 'largest overlap' with X1^Uc2, but the numerical value of that overlap is not given. If the overlap is not close to 1, the 'necessary prerequisite' conclusion could be weakened. Please provide the overlap values or a more quantitative tracking criterion. In addition, the inversion leading to Eq. (48), where (t2_q)^{-1} is placed on the left-hand side, should be justified more carefully; the strong-coupling approximation t2_eff,q is introduced in","section":"§IV.B, Eqs. (44)–(48)"},{"comment":"The paper notes after Eq. (14) that the reciprocal-lattice mapping 'implicitly assumes an equivalence of each atom, which is strictly speaking an approximation.' The reader's report raised the concern that this periodization could bias the lattice response. In this paper, the central vertex-divergence line (Fig. 5) is computed directly from the impurity cluster before periodization, and the eigenvalue necessary condition in §IV.B also uses the impurity cluster and the superlattice equation without the reciprocal-lattice mapping. Thus the periodization caveat is less load-bearing for the main claims than it could appear. However, the paper should state this explicitly, since the abstract and §IV.A refer to 'two-particle CDMFT' without clarifying that the key divergence line is a cluster (not periodized) quantity. The periodization mainly affects which specific eigenvalue is identified in","section":"§II.B, Eq. (14) and §III.A"}],"minor_comments":[{"comment":"The Ward identity check is shown for only one orbital combination and one parameter point. It would be helpful to state that the Ward identity was verified for all orbital combinations and for the parameter range used in the phase diagram, especially in the strong-coupling regime near the MIT.","section":"§II.C, Fig. 4"},{"comment":"The notation 'w.o. sp' in Fig. 5 is defined in the text, but the orange dashed line's exact construction (subtraction of all non-local spin-transverse SBE diagrams) could be re-stated more clearly in the caption. The same applies to the yellow dot-dashed line in the left panel of Fig. 5.","section":"§III.B, Eq. (34)"},{"comment":"The strong-coupling approximation t2_eff,q is written as a 16×16 matrix in App. B2. The main text should mention that the approximation is diagonal in Matsubara frequency and that it is valid when the self-energy is large compared to the hopping, as stated in the text, but the limits of this validity for the intermediate-coupling regime T=1/15t, U≈5.9t should be discussed.","section":"§IV.B, Eq. (43)"},{"comment":"The labels in Fig. 6c) and d) are small and difficult to read. Consider enlarging the eigenvector plots or providing a separate panel for the leading eigenvector structure.","section":"Fig. 6"},{"comment":"There are a few typos and small errors: in App. A 'estaimated' should be 'estimated' and 'plance' should be 'plane'; in Fig. 9 caption 'T=0/15t' should presumably be 'T=1/15t'; in the references, some entries are incomplete (e.g., Ref. [29] has missing page numbers).","section":"General, typos"}],"recommendation":"major_revision","confidential_remarks":"The paper is clearly a strong and careful contribution to the vertex-divergence and cluster-DMFT literature. The main gap is the lack of cluster-size scaling for the central numerical claim. The 2×2 cluster is known to overestimate AF correlations, and the paper itself shows that its Néel temperature is higher than that of 8×8 CDMFT. If the authors can provide at least 4×4 results at selected temperatures, or alternatively restrict the claim to 'in the 2×2 CDMFT approximation', the paper would be significantly strengthened. The eigenvector-tracking issue in §IV.B is secondary but should be addressed quantitatively. The formal BSE/Ward-identity material is publishable as is, and the benchmarks against DiagMC and applied-field data are a strong point."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Co-worker — the thing to know: this is a careful, useful paper that moves the vertex-divergence story from DMFT to CDMFT and delivers the two-particle formalism to do it. The main numerical claim — that the first charge-vertex divergence line shifts to lower U once short-range correlations are included — is persuasive for the 2x2 cluster they study, but the paper does not yet show it survives larger clusters. That's the soft spot to keep in mind.\n\nWhat's new: they derive the real-space BSE set for CDMFT in all channels, including a particle-particle real-space BSE they correctly note was missing, and real-space Ward identities that they verify numerically. The AFM response is benchmarked against applied-field CDMFT and DiagMC, and the Ward identities are satisfied in the parameter regime used. That is real formal work, and the benchmarks are appropriate. The systematic T–U map of the first divergence line, the SBE decomposition separating non-local spin-transverse contributions, and the eigenvalue analysis linking the vertex divergence to the Mott transition and phase-separation instability go beyond the sporadic cluster points in the literature.\n\nSoft spots: first, the cluster-size issue. Everything numerical is 2x2. They even show the 2x2 Néel temperature sits above the 8x8 one, so the short-range AF fluctuations they invoke are plausibly enhanced by the cluster geometry. The qualitative direction — nonlocal correlations lower the divergence — is consistent with earlier DCA/CDMFT points, but \"systematic\" is a strong word until there is a 4x4 or larger point. Second, there are no error bars on the divergence line and no released code/data, so the numerics are not independently reproducible yet. Third, the \"essential prerequisite\" language overstates a bit: via Eq. (11) the eigenvalue sign change and the vertex divergence are the same event, so calling the eigenvalue crossing a prerequisite is partly a restatement. The paper earns independence by showing the crossing occurs before the MIT and by isolating the spin-fluctuation contribution, so this is a wording issue more than a logic flaw. The periodization caveat they admit is real but is not the load-bearing part for the divergence-line claim, which is computed from the impurity cluster itself.\n\nBottom line: the formal machinery and the physical interpretation are credible and worth engaging. The central quantitative claim needs cluster-size scaling before I'd bet on it. Send it to referees — a serious one will ask for exactly that check.","headline":"Solid formal extension of the vertex-divergence program to CDMFT, with a credible but not yet cluster-size-converged picture of how short-range AF fluctuations lower the first divergence line.","tokens_in":37511,"tokens_out":2650,"would_cite":true,"duration_ms":29468,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.30.+h","71.27.+a","71.10.Fd"],"model":"deepseek-v4-flash","headline":"Short-range spin fluctuations shift the breakdown of perturbation theory to lower interaction strengths in the two-dimensional Hubbard model.","keywords":["cellular dynamical mean-field theory","Bethe-Salpeter equation","vertex divergences","Hubbard model","Mott transition","short-range correlations","spin fluctuations","charge susceptibility"],"falsifier":"Compute the first vertex-divergence line with a larger cluster (e.g., 4x4 or 8x8 CDMFT) at the same temperatures and interaction range, or with an alternative periodization scheme that enforces momentum conservation in the bubble. If the divergence line shifts back toward the DMFT line, or if the eigenvalue zero-crossing no longer precedes the Mott transition, the central claim would be a finite-cluster artifact.","tokens_in":36665,"feed_emoji":"🧲","tokens_out":2505,"duration_ms":30194,"temperature":0.7,"pith_summary":"This paper aims to show that, once short-range spatial correlations are included beyond a purely local description, the breakdown of self-consistent perturbation theory—signalled by a divergence of the two-particle irreducible charge vertex—occurs at smaller Coulomb repulsion U than in dynamical mean-field theory. Using cellular dynamical mean-field theory on a 2x2 cluster, the authors derive the full Bethe-Salpeter formalism in all channels, verify it with Ward identities, and compute the divergence line across the temperature-interaction phase diagram. They further identify the change of sign of an eigenvalue of the generalized charge susceptibility as the essential prerequisite for the Mott metal-insulator transition and for phase-separation instabilities at larger U. If correct, the result reframes the physics of the 2D Hubbard model: non-local antiferromagnetic fluctuations actively prepare the charge sector for the Mott transition, not merely accompany it.","feed_headline":"Spin fluctuations push vertex-divergence line to lower U","feed_subtitle":"Short-range antiferromagnetic correlations make perturbation theory break down earlier, reshaping the route to the Mott transition in 2D.","key_machinery":"The central object is the two-particle irreducible charge vertex Γ_ch of the cluster impurity, extracted from the cluster generalized susceptibility via the inverse Bethe-Salpeter equation. The authors derive explicit real-space BSE expressions for the charge, spin, and particle-particle channels within CDMFT, together with Ward identities that benchmark the calculation. The divergence of Γ_ch corresponds to a vanishing eigenvalue of the generalized charge susceptibility of the impurity, and the lattice response is built from the same vertex through a cluster-local BSE with a Fourier periodization. A strong-coupling approximation of the lattice bubble difference t²_eff,q is used to turn the","core_discovery":"The central claim is that the first divergence of the two-particle irreducible charge vertex Γ_ch in CDMFT sits systematically below the DMFT divergence line at intermediate and low temperatures, and that this downward shift is caused by short-range transverse spin fluctuations. The authors show that removing the non-local spin-transverse diagrams from the vertex moves the divergence line back toward higher U, flattening its distinctive 'belly' shape. They also find that the eigenvalue of the generalized charge susceptibility that drives the vertex divergence must cross zero and become negative before the Mott transition can occur at larger U; this same eigenvalue then controls the divergenc","pith_inferences":["If short-range spin fluctuations lower the vertex-divergence line, then in the thermodynamic limit the first divergence line may begin at U=T=0 with an exponential onset, as the authors speculate; this could be tested with larger clusters or diagrammatic Monte Carlo in the paramagnetic phase.","The finding strengthens the analogy between vertex divergences and local-moment formation: in 2D, the relevant 'local' moment is spatially extended across nearest neighbors, so cluster size should systematically control the divergence scale—a prediction one could verify by comparing 2x2 and 4x4 clusters at fixed temperature.","The same eigenvalue crossing that precedes the Mott transition may also enhance electron-phonon coupling in 2D, as the authors hint; a concrete test would be to compute the phonon self-energy using the CDMFT charge vertex and look for a divergent tendency near the vertex-divergence line.","The periodization ambiguity suggests that a translation-invariant alternative (e.g., a DCA-like BSE with enforced momentum conservation) should produce a different divergence line; comparing the two periodizations would isolate the true non-local effect from the cluster-geometry artifact."],"forward_implications":["If the divergence line is genuinely lower in CDMFT, then perturbation-theory-based methods become unreliable at weaker interactions in two dimensions than previously expected, and the regime of 'strong correlations' is broader than the local picture suggests.","The identification of the eigenvalue sign change as a necessary prerequisite for the Mott transition implies that any successful non-perturbative theory of the 2D Hubbard model must reproduce this zero-crossing before the transition, not merely the transition itself.","The anti-symmetric frequency structure of the critical eigenvector explains why the uniform charge response stays finite at the half-filled Mott transition even though the generalized susceptibility diverges, resolving a long-standing puzzle.","The strong-coupling threshold derived from t²_eff,q provides a practical way to predict the location of the Mott transition from cluster impurity data alone, without full lattice diagonalization.","Because the same mechanism drives phase-separation instabilities at finite doping, the result constrains the shape of the coexistence region in cluster DMFT phase diagrams."],"fun_headline_variants":["Nonlocal spin fluctuations lower the vertex-divergence line","Antiferromagnetic correlations make perturbation theory break earlier","Vertex divergence precedes Mott transition after susceptibility sign change","Short-range spin correlations shift perturbative breakdown to lower U"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The mapping of cluster two-particle quantities onto the lattice assumes that all sites inside the cluster are equivalent, even though the cluster Green's function is not translationally invariant; if this periodization distorts the lattice response, the downward shift of the divergence line could be an artifact of the 2x2 cluster rather than a real physical effect.","fun_headline_variants_meta":{"raw":{"variants":["Nonlocal spin fluctuations lower the vertex-divergence line","Antiferromagnetic correlations make perturbation theory break earlier","Vertex divergence precedes Mott transition after susceptibility sign change","Short-range spin correlations shift perturbative breakdown to lower U"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000275,"raw_usage":{"total_tokens":1461,"prompt_tokens":710,"completion_tokens":751,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":454,"completion_tokens_details":{"reasoning_tokens":688}},"tokens_in":454,"tokens_out":751,"duration_ms":7900,"temperature":1.0,"reasoning_tokens":688,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T15:10:16.578586+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the first vertex-divergence line with a larger cluster (e.g., 4x4 or 8x8 CDMFT) at the same temperatures and interaction range, or with an alternative periodization scheme that enforces momentum conservation in the bubble. If the divergence line shifts back toward the DMFT line, or if the eigenvalue zero-crossing no longer precedes the Mott transition, the central claim would be a finite-cluster artifact.","supporting_citations":[],"review_version":1}