{"id":"c0a1403e-1f5d-4f70-9144-290b39ac4edf","arxiv_id":"2501.18325","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Nearest-neighbor, low-frequency spin fluctuations, not local ones, are identified as the processes that drive the Mott metal-insulator transition in a 2x2 cellular dynamical mean-field calculation.","lead":"This paper introduces a real-space diagnostic that decomposes the electron self-energy of the Hubbard model into contributions from local and non-local magnetic fluctuations. Applied to a 2x2 cluster calculation across the Mott transition, it shows that low-energy nearest-neighbor spin fluctuations are what open the insulating gap.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that nearest-neighbor spin-boson diagrams drive the Mott gap rests on a single 2x2 CDMFT cluster; without a cluster-size check the word 'unambiguously' is unsupported.","rationale":"The Hedin decomposition in App. A is standard and internally consistent; the diagrams in Figs. 6, 9, and 10 sum to the full self-energy within the expected truncation error. The identification of the w10/G00 diagram as the source of the insulating pole in the 2x2 spin-channel decomposition is supported by the frequency-resolved analysis in Fig. 7 and the SBE comparison in App. C3. However, all evidence is for one small cluster. The central claim is about the model, not merely the Nc=2x2 impurity; the abstract's 'unambiguously' therefore requires cluster-size convergence, and this is the most load-bearing gap. The finite Matsubara grid and the Fierz channel ambiguity are secondary: the grid error is quantified and presumably not sign-flipping, and the spin channel is the natural one at half-filling, but neither has been tested across cluster sizes. A 4x4 RFD calculation is the decisive test. Since the paper can be accepted only if the mechanism survives cluster-size checks, the CONDITIONAL verdict is appropriate; my analysis does not move it.","tokens_in":28609,"tokens_out":10003,"duration_ms":100218,"concrete_test":"Perform the same real-space fluctuation diagnostics for a 4x4 CDMFT cluster (or 8x8 if feasible) at half-filling, T=0.067t, for U values around the cluster's MIT (e.g., U=6t and 6.5t), measuring the Hedin vertex and decomposing the on-site self-energy into w00, w10, w20 spin diagrams. If the w10/G00 diagram no longer dominates the negative ΔΣ, or if the local w00 diagram controls the insulating pole, the central claim is falsified for the model. Use the same frequency grids and error propagation as the original paper; the authors' public data repository [41] provides a starting point.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline claim (abstract; Sec. VII C) is that nearest-neighbor, low-frequency dynamic antiferromagnetic spin-boson excitations are responsible for the Mott MIT. The evidence is the RFD decomposition of the on-site self-energy in the spin channel for a single 2x2 CDMFT solution at U=6t (Fig. 6): the w10/G00 diagram has negative ΔΣ ≈ -1.0, the local w00 diagram has positive ΔΣ ≈ +0.9, and the total is -1.2. Nothing shows this decomposition is stable against increasing the cluster size. Sec. II A references [31,69,70] only for single-particle cluster-size dependence; no RFD or two-particle-level cluster-size test is presented. In 2D the 2x2 plaquette has a specific U_c ≈ 6t and emphasizes short-range AF correlations; if the dominance of w10 over w00 is reduced or reversed in a 4x4 or 8x8 cluster, the causal statement in Sec. VII C describes only the Nc=2x2 approximation, not the Hubbard model. The finite Matsubara grid (App. A4, Fig. 13) compounds this: the spin channel systematically overestimates the self-energy, and the w10 contribution carrying the insulating pole is part of that overestimated channel, so a larger grid could reduce its magnitude. Thus 'unambiguously' is not warranted by the presented evidence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a real-space fluctuation diagnostics (RFD) approach based on the Hedin equation, which expresses the self-energy as a sum over real-space bosonic propagators w, fermionic propagators G, and the fermion-boson vertex λ. Using CDMFT on a 2x2 cluster for the half-filled Hubbard model on the square lattice, the authors compute all ingredients and decompose the on-site and non-local self-energies across the Mott metal-insulator transition. They identify the nearest-neighbor bosonic spin propagator w10 combined with the local fermionic propagator G00 as the dominant diagram producing the insulating pole of the local self-energy, while the local spin diagram w00 remains metallic. The paper further analyzes the frequency structure of the Hedin vertex to explain the emergence of this insulating contribution, and connects the result to momentum-space antiferromagnetic fluctuations. The central claim is that nearest-neighbor, low-frequency dynamic antiferromagnetic spin-boson excitations are responsible for the occurrence of the MIT.","tokens_in":28749,"tokens_out":10755,"duration_ms":95495,"significance":"The RFD method is a useful addition to the fluctuation-diagnostics toolbox: it substitutes the three-point Hedin vertex for the computationally heavier four-point vertex, and it provides a real-space picture complementary to existing momentum-space diagnostics. The paper is careful in deriving the Hedin equation in real space (App. A), in documenting sources of error (App. A4), and in making raw data publicly available. The central conclusion—that nearest-neighbor low-frequency spin fluctuations drive the MIT within the 2x2 CDMFT solution—is plausible and consistent with earlier momentum-space studies, but its validity for the Hubbard model in larger clusters is not yet established. The method itself is a worthwhile contribution regardless of the eventual cluster-size fate of the specific mechanism.","major_comments":[{"comment":"The conclusion that nearest-neighbor spin-boson diagrams are 'responsible for the occurrence of the MIT' (Sec. VII C, Fig. 6) rests entirely on a single Nc=2x2 CDMFT solution. The paper explicitly restricts itself to this cluster size in Sec. IV and points to Refs. [31,69,70] for cluster-size dependence, but those references concern single-particle quantities. No evidence is presented that the RFD decomposition—in particular the dominance of the w10/G00 diagram over the local w00/G00 diagram—is stable when the cluster is enlarged to 4x4 or 8x8. Since the 2x2 plaquette can only capture the shortest non-local distance, the word 'unambiguously' in the abstract is not supported. Please provide a cluster-size check at the two-particle level, or qualify the claim to the 2x2 CDMFT approximation.","section":"Sec. IV and Sec. VII C"},{"comment":"The spin-channel Hedin self-energy systematically overestimates the directly measured self-energy because of the finite Matsubara grid (82 positive fermionic and 81 positive bosonic frequencies). The key insulating contribution in Fig. 6 (w10, sum=-1.0) belongs to this overestimated spin channel, so the reported balance between the metallic local diagram w00 (+0.9) and the insulating nearest-neighbor diagram w10 (-1.0) is potentially affected by grid truncation. The authors should demonstrate that the qualitative decomposition—specifically the sign and relative magnitude of ΔΣ from the w10 and w00 classes—is robust to increasing the number of Matsubara frequencies, or provide an explicit high-frequency tail extrapolation. Figure 13 shows that the overshoot is frequency-dependent, so it does not cancel in the difference ΔΣ by construction.","section":"App. A4 and Fig. 13"}],"minor_comments":[{"comment":"The critical temperature is given as 'T_c^{2x2} ≈ 0.09/t'; the units appear to be wrong and should read 0.09t.","section":"Sec. IV"},{"comment":"The site labeling '0, 1, 1, 2' in the text and Fig. 2 is confusing; the figures use an overbar on one of the '1' sites (e.g., w10 vs w10). Please clarify the notation.","section":"Sec. III B"},{"comment":"In the equation for Σs,e(iνn), the summation index 'd' in 'X d,f' appears to be a typo for the bosonic orbital index 'b' used in the preceding equations.","section":"App. A2, Eq. (A18)"},{"comment":"The word 'unambiguously' is too strong given the limitations discussed above; consider replacing it with a more qualified formulation such as 'for the 2x2 CDMFT solution we identify...'.","section":"Abstract"},{"comment":"The first sentence says the finite Matsubara grid 'results underestimate correlations', but the charge channel underestimates while the spin channel overestimates; please rephrase to describe the opposite signs.","section":"App. A4"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid methods-and-application contribution, and the RFD construction is internally consistent. The main risk is overinterpretation of a single 2x2 cluster without a two-particle-level cluster-size check, and the finite-grid error in the spin channel that directly affects the key diagram decomposition. The authors should be encouraged to either add such checks or soften the causal claims. The data availability and the detailed appendices are strengths that fit the journal's standards."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: solid methods paper with a genuinely new tool—real-space fluctuation diagnostics based on the Hedin three-leg vertex—and a clear, internally consistent application to the 2x2 CDMFT Mott transition. The headline claim, that nearest-neighbor low-frequency spin fluctuations produce the insulating self-energy, holds up within the model. The abstract's 'unambiguously' does not.\n\nWhat is new: prior fluctuation diagnostics used four-point functions in momentum space; here they use the three-point Hedin vertex, which is cheaper and works directly in real space. That is a real step forward. The decomposition is bookkeeping of an exact identity, not a fit, and the diagram sums in Figs. 6, 9, and 10 reproduce the full self-energy to the expected accuracy. Raw data is deposited on Edmond, so the central numbers can be checked. The cross-check against the single-boson-exchange decomposition in App. C3 is a good-faith robustness test, and the explanation of why the local w00 diagram stays metallic while w10/G00 produces the insulating pole is careful and physically plausible.\n\nSoft spots, in proportion. First, 'unambiguously' overstates the evidence. The whole mechanism claim rests on one cluster size (2x2) and essentially one interaction value (U=6t). The cluster-size dependence is cited only at the single-particle level [31,69,70]; there is no RFD-level or two-particle-level cluster-size check. So the causal statement in Sec. VII C is a statement about 2x2 CDMFT, not yet about the Hubbard model. That is a wording fix, but an important one.\n\nSecond, the finite Matsubara grid is a real caveat. App. A4 and Fig. 13 show the spin channel systematically overestimates the directly measured self-energy. The w10 contribution carrying the insulating pole lives in that channel, so a larger grid could reduce its magnitude. The authors argue the error is not qualitative; I would believe them more with a grid-convergence check. Third, the main decomposition plots have no error bars, though the appendix reports statistical errors from the Hedin vertices. Adding error bars would cost little.\n\nNone of this sinks the central claim within the model. The conclusion that dynamic short-range spin fluctuations drive the Mott transition in 2x2 CDMFT is legitimate and consistent with the broader spin-fluctuation literature. The paper is honest about its limitations in the appendices; the overclaim is confined mostly to the abstract.\n\nWho this is for: people working on cluster DMFT, fluctuation diagnostics, and the Mott transition. It deserves a serious referee. My recommendation: send it to review, with the referee asked to push for a grid-convergence check and a tempered abstract. I would not desk-reject.","headline":"New real-space fluctuation diagnostics with an honest within-model result, slightly oversold by the abstract's 'unambiguously'.","tokens_in":29441,"tokens_out":2818,"would_cite":true,"duration_ms":24515,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.30.+h","71.10.Fd"],"model":"deepseek-v4-flash","headline":"The Mott gap in the square-lattice Hubbard model is opened by short-range, nearest-neighbor spin excitations, not by local ones.","keywords":["Mott metal-insulator transition","real-space fluctuation diagnostics","Hedin vertex","cellular dynamical mean-field theory","Hubbard model","spin-boson excitations","antiferromagnetic fluctuations","self-energy decomposition"],"falsifier":"Repeat the real-space fluctuation diagnostics on a 4×4 (or larger) CDMFT cluster for the half-filled Hubbard model at the same temperature: if the on-site self-energy's insulating pole no longer comes from the w10 + G00 spin-boson diagram, or if the local w00 diagram becomes the dominant negative contribution, the central claim is falsified. A less costly check: extend the Matsubara grid for the Hedin vertex (using more frequencies than the 82 fermionic and 81 bosonic points used here) and verify that the spin-channel overestimate of the self-energy (App. A4, Fig. 13) does not flip the sign of the w10 versus w00 decomposition.","tokens_in":28285,"feed_emoji":"🧲","tokens_out":6628,"duration_ms":54298,"temperature":0.7,"pith_summary":"This paper asks which physical fluctuations in real space destroy metallicity at the Mott metal-insulator transition. To answer it, the authors introduce a real-space fluctuation diagnostics: they decompose the Hedin equation for the self-energy into contributions classified by the distance traveled by the bosonic (spin or charge) propagator and the fermionic propagator on a 2×2 cluster, computed by cellular dynamical mean-field theory. Applied to the half-filled square-lattice Hubbard model, the decomposition shows that the insulating pole of the on-site self-energy comes from a nearest-neighbor spin-boson diagram with a local fermionic propagator, while the fully local spin-boson diagram stays metallic. The authors conclude that the MIT in this cluster solution is driven by low-frequency dynamic antiferromagnetic spin fluctuations over the nearest-neighbor distance. If true, this changes the picture from a purely local Mott mechanism to one dominated by short-range magnetic fluctuations.","feed_headline":"Mott gap traced to nearest-neighbor spin-boson exchange","feed_subtitle":"A real-space decomposition of the Hedin self-energy pins the insulating pole on nearest-neighbor spin-boson exchange, not local ones.","key_machinery":"The central object is the real-space Hedin equation, Σ = ± U T Σ_{ω} Σ_{b,f} G_{s,f} w_{b,s} λ_{b,f,e} + const., which expresses the self-energy as a convolution of the fermionic propagator G, the renormalized bosonic (spin/charge) propagator w, and the three-point fermion-boson Hedin vertex λ. The new 'real-space fluctuation diagnostics' classifies each diagram by the distance covered by w (local w00, nearest-neighbor w10, second-neighbor w20) and by G (G00, G01, ...), and inspects the bosonic-frequency-resolved contribution ΔΣ̃(iω_m) to the self-energy difference between the first two fermionic Matsubara frequencies. The load-bearing combination is λ100 with G00 and w10, which produces the insulating pole.","core_discovery":"Using the Hedin equation with the three-point fermion-boson vertex λ computed in 2×2 CDMFT, the paper decomposes the on-site self-energy Σ00 into real-space classes. In the spin channel, the local bosonic diagram w00 with the local fermionic propagator G00 produces a metallic contribution (positive ΔΣ), whereas the nearest-neighbor bosonic diagram w10 with G00 produces the negative, pole-like contribution that drives the insulating self-energy at U=6t just above the transition. The pivotal mechanism is the frequency structure of the Hedin vertex λ100: in the insulator it develops a node along iω + iν = 0 that cancels the nodal structure of G00 w10, so the bosonic Matsubara sum no longer cancels, generating a large static self-energy. This behavior is opposite to single-site DMFT, where the local vertex changes sign at the transition.","pith_inferences":["If the 2×2 result survives larger clusters, the same decomposition could serve as a diagnostic to test whether the pseudogap in the doped Hubbard model has the same nearest-neighbor spin-boson origin as the half-filled Mott gap.","The finding that the local Hedin vertex changes sign only for non-local distances suggests that cluster-size extrapolations of the MIT mechanism will need to track the frequency structure of λ at the Fermi node, not just the magnitude of the spin propagator.","The method could be applied within the single-boson exchange decomposition to test whether the multi-boson diagrams in the spin channel, which the paper notes cancel part of the SBE contribution, remain subdominant at larger U.","A direct extension would be to vary the Fierz parameter r to check whether the classification of 'local' versus 'nearest-neighbor' mechanisms is robust to the channel decomposition; the paper fixes to spin and charge channels."],"forward_implications":["The Mott critical interaction in CDMFT (U_c ~ 6t versus ~9.3t in DMFT) is explained by the onset of the nearest-neighbor spin-boson contribution, connecting the reduced critical scale to a specific real-space process.","The nodal/antinodal dichotomy of the momentum-resolved self-energy (periodized) is dominated by the w10 spin-boson contribution, which is insulating at the antinode and metallic at the node.","The local spin-boson diagram alone would keep the system metallic; the Mott gap is misattributed if one analyzes only local fluctuations.","The charge channel is essentially local and plays no role in the MIT, consistent with the bosonic charge propagator being local.","The method is cheaper than full four-point fluctuation diagnostics because it uses the three-point Hedin vertex, enabling real-space analyses on clusters."],"supporting_citations":[{"why":"Provides the Hedin equations that connect the self-energy to the fermion-boson vertex, the formal foundation of the decomposition.","marker":"[53]"},{"why":"Generalizes Hedin's equations to spin-dependent interactions, used in the derivation of the spin and charge channel formulation.","marker":"[54]"},{"why":"Introduces fluctuation diagnostics of the self-energy, the conceptual basis that the paper extends to real space with the Hedin vertex.","marker":"[33]"},{"why":"Defines cellular dynamical mean-field theory, the method that produces the real-space G, w, and λ on the cluster.","marker":"[25]"},{"why":"Supplies the 2×2 CDMFT phase diagram with reduced critical interaction U_c ≈ 6t that the paper analyzes and reproduces.","marker":"[60]"},{"why":"The previous momentum-space DCA calculation of the Hedin vertex, the benchmark that the real-space CDMFT calculation extends.","marker":"[40]"},{"why":"Documents the sign change of the local fermion-boson vertex in DMFT at the MIT, contrasted with the non-local vertex behavior found here.","marker":"[74]"},{"why":"Provides the cluster-size dependence study that supports the caveat that only the 2×2 cluster is considered.","marker":"[31]"},{"why":"Supplies the numerically efficient conventions for the three-point function and spin/charge channels used to compute χ3 and λ.","marker":"[52]"}],"fun_headline_variants":["Mott gap’s true source: nearest-neighbor spin-boson","Nonlocal spin-boson exchange, not local, makes Mott insulator","Nearest-neighbor vertex drives the Mott self-energy","Mott transition: it’s the nonlocal spin boson","Real-space Hedin vertex: neighbor spin-boson kills metal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the 2×2 cluster captures the mechanism of the Mott transition, so that larger clusters would not dethrone the nearest-neighbor spin-boson diagram, and that the finite Matsubara grid used for the Hedin vertex leaves the qualitative decomposition intact.","fun_headline_variants_meta":{"raw":{"variants":["Mott gap’s true source: nearest-neighbor spin-boson","Nonlocal spin-boson exchange, not local, makes Mott insulator","Nearest-neighbor vertex drives the Mott self-energy","Mott transition: it’s the nonlocal spin boson","Real-space Hedin vertex: neighbor spin-boson kills metal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000845,"raw_usage":{"total_tokens":3643,"prompt_tokens":872,"completion_tokens":2771,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":488,"completion_tokens_details":{"reasoning_tokens":2680}},"tokens_in":488,"tokens_out":2771,"duration_ms":20372,"temperature":1.0,"reasoning_tokens":2680,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T23:53:46.910209+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the real-space fluctuation diagnostics on a 4×4 (or larger) CDMFT cluster for the half-filled Hubbard model at the same temperature: if the on-site self-energy's insulating pole no longer comes from the w10 + G00 spin-boson diagram, or if the local w00 diagram becomes the dominant negative contribution, the central claim is falsified. A less costly check: extend the Matsubara grid for the Hedin vertex (using more frequencies than the 82 fermionic and 81 bosonic points used here) and verify that the spin-channel overestimate of the self-energy (App. A4, Fig. 13) does not flip the sign of the w10 versus w00 decomposition.","supporting_citations":[{"cited_title":"Hubbard, Electron Correlations in Narrow Energy Bands","cited_arxiv_id":null,"evidence_quote":"Provides the Hedin equations that connect the self-energy to the fermion-boson vertex, the formal foundation of the decomposition."},{"cited_title":"Sch¨ afer, N","cited_arxiv_id":null,"evidence_quote":"The previous momentum-space DCA calculation of the Hedin vertex, the benchmark that the real-space CDMFT calculation extends."},{"cited_title":"Parcollet, G","cited_arxiv_id":null,"evidence_quote":"Documents the sign change of the local fermion-boson vertex in DMFT at the MIT, contrasted with the non-local vertex behavior found here."}],"review_version":1}