{"id":"2266e5ba-ce3a-4924-bb0f-ff238b255bbb","arxiv_id":"1908.00536","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A channel decomposition with equal and opposite charge and spin bare interactions removes the leading irreducible vertex contributions, yielding an effective fermion-boson theory (TRILEX2) that largely avoids the Fierz ambiguity.","lead":"The authors derive an effective fermion-boson theory for the extended Hubbard model and propose a specific channel mixing (Uc=U/2, Us=-U/2) that suppresses the Fierz ambiguity that usually plagues partially bosonized theories. The resulting TRILEX2 method matches ladder dual-fermion results for the nonlocal self-energy at moderate coupling and offers a cheap route to include spin fluctuations beyond DMFT.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"M≈Λ w̄ Λ is exact only when Λ²=1; the residual four-fermion term in Eq. (B13) is O(U(Λ²−1)) and is not shown to be small at low frequencies.","rationale":"Good-faith reading: the paper proposes a concrete scheme and supports it with numerical comparisons. The central claim is not internally inconsistent; the issue is a missing control on a specific algebraic replacement. The reader's weakest assumption already names M≈Λ w̄ Λ, so this is the same concern, sharpened. I do not see a more basic flaw: the local vertex comparison (Fig. 3-4) is honest evidence, the numerical benchmarks are meaningful, and the Uc=-Us=U/2 construction is at least well-defined. The decisive question is whether the residual four-fermion term in Eq. (B13) is actually negligible. It can be answered directly with the authors' own data. Until then, the claim that the Fierz ambiguity is 'solved' is not fully supported, so the CONDITIONAL verdict stands; no verdict change.","tokens_in":21561,"tokens_out":10309,"duration_ms":100332,"concrete_test":"Using the same ALPS CT-HYB impurity data (U=1.0, T=0.1), compute the residual four-fermion vertex R_{νν′ω}=Γ_{νν′ω}−[2Λ_{νω}w̄_ωΛ_{ν′+ω,−ω}−(vertical terms)] with w̄=w−U/2, following Eq. (7)/(A8) and Eq. (B13). Report the normalized weighted norm ‖R‖/‖Γ‖ (e.g., over fermionic ν,ν′ and bosonic ω, weighted by |G_ν G_ν′ G_{ν+ω} G_{ν′+ω}|) and the low-frequency value of (U/2)(Λ_{νω}Λ_{ν′+ω,−ω}−1) at ν=ν′=ω=πT. If ‖R‖/‖Γ‖ is not small (≲0.1) in the low-frequency region, the dropped four-fermion term is not negligible and action (9) is not the theory whose self-energy is compared in Fig. 5. A secondary check: rerun the TRILEX2 self-energy with R kept as a perturbative one-shot correction; if the correction changes Σ̃ by more than the plotted difference from ladder DF, the cancellation claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The effective action (9) is reached by dropping the four-fermion term Γ−4M in Eq. (B13). This requires two separate approximations. First, (A7) replaces Γ by 4(Λ w Λ − U/2), which is the w-reducible approximation. Second, Sec. II D and Appendix A replace M=Λ w Λ − U/2 by Λ w̄ Λ with w̄=w−U/2. The second step is not a consequence of suppressing w-irreducible diagrams; algebraically it requires Λ_{νω}Λ_{ν′+ω,−ω}=1 for every pair of frequencies, because M−Λ w̄ Λ=(U/2)(Λ_{νω}Λ_{ν′+ω,−ω}−1). The only given justification is the high-frequency asymptotics Λ→1 and agreement with more elaborate methods. Figure 1 shows Λ deviates substantially from 1 at low frequencies, so the dropped four-fermion vertex has magnitude ∼2U(Λ_{νω}Λ_{ν′+ω,−ω}−1) and is not small by construction. If this residual is significant, the action (9) is not equivalent to the original problem, and the claimed Fierz-ambiguity-free description is not established. This is the load-bearing soft spot: it is exactly the step that makes the cancellation in (B13) work, but it is an uncontrolled algebraic approximation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript proposes a partially bosonized dual action for the extended Hubbard model that is claimed to be free of the Fierz ambiguity. Starting from the dual-boson representation, the authors rewrite the local Coulomb interaction with channel-dependent bare couplings, argue that the choice Uc=-Us=U/2 removes the leading w-irreducible (ladder-like vertical) contributions to the local fermion-fermion vertex, and approximate that vertex by the boson-reducible form M^ς=Λ^ς w^ς Λ^ς - U^ς/2. They then replace M^ς by Λ^ς \\bar{w}^ς Λ^ς with \\bar{w}^ς=w^ς-U^ς/2, which permits a Hubbard-Stratonovich transformation that eliminates the fermion-fermion vertex from the effective action, leaving action (9) with a bare boson propagator W=W_EDMFT-U/2 and a local fermion-boson vertex Λ. The method is benchmarked by comparing the approximate local vertex with CT-HYB results (Figs. 3-4) and the nonlocal self-energy with ladder dual-fermion results (Fig. 5) for the 2D half-filled Hubbard model at U=0.5, 1.0 and 1.5; a self-consistent calculation locates the Mott transition around U≈1.7.","tokens_in":21831,"tokens_out":7273,"duration_ms":73923,"significance":"If the central claim is correct, the paper offers a computationally inexpensive TRILEX-like scheme that accounts for charge and spin fluctuations beyond EDMFT without the Fierz ambiguity, and the explicit form of the effective action (9) provides a concrete starting point for realistic GW-like implementations. The numerical benchmarks against exact CT-HYB impurity vertices and against ladder dual-fermion self-energies are independent references and are presented transparently, and the derivation is largely explicit. The main limitation is that the two key steps - the special channel decomposition and the replacement M≈Λ\\bar{w}Λ - are justified only heuristically and by agreement in selected regimes, not by a controlled estimate of the neglected terms; the paper's strongest claim therefore goes beyond what is currently demonstrated.","major_comments":[{"comment":"The replacement M^ς_{νν'ω}≈Λ^ς_{νω} \\bar{w}^ς_ω Λ^ς_{ν'+ω,-ω} is not an identity. From Eq. (8) and the definition \\bar{w}^ς_ω=w^ς_ω-U^ς/2, the difference is (U^ς/2)(Λ^ς_{νω}Λ^ς_{ν'+ω,-ω}-1). The cancellation of the fermion-fermion vertex in Eq. (B13) therefore leaves a residual four-fermion term proportional to U(ΛΛ'-1), and Fig. 1 shows that Λ deviates substantially from 1 at low frequencies. The manuscript justifies this step only by the high-frequency asymptotics and by the final agreement with ladder dual fermion; it should provide a quantitative estimate of the dropped four-fermion vertex (for example, its magnitude over the Matsubara grid or its contribution to the TRILEX2 self-energy) to establish that the action (9) is actually equivalent to the original problem in the tested regime.","section":"Sec. II.D, Appendix A, Eq. (B13)"},{"comment":"The claim that the choice Uc=-Us=U/2 'almost fully suppresses' the missing w-irreducible contributions is not quantified. Fig. 4 shows only two frequency cuts of the vertex at zero bosonic frequency, which is not enough to establish that the irreducible remainder Γ^ς-Γ^ς_approx is small across the full Matsubara-frequency range, especially at U=1.5 where the text reports an increased role of vertical diagrams. Please provide a global measure of the irreducible remainder (e.g., a frequency-summed norm) and, ideally, a parquet-style decomposition of the remainder into particle-particle and other irreducible parts, so that the uniqueness claim is supported by more than selected cuts.","section":"Sec. II.C, Eqs. (7)-(8), Fig. 4"}],"minor_comments":[{"comment":"The notation 'TRILEX2 I' appears in the text but is not defined; please explain that the superscript I denotes the Ising-decoupling version of the method.","section":"Sec. III.A, Fig. 5"},{"comment":"The relation between Uc, Us and U that makes the second line of Eq. (6) follow from the first is not stated explicitly; please write the decoupling relation (e.g., the relevant Fierz identity connecting Uc and Us) so that the decoupling independence of Γ0 can be checked.","section":"Eq. (6)"},{"comment":"The numerical prefactors in Eq. (A5) (the factor 4 in front of ΛUχUΛ and the term 2U) are introduced without derivation; since the text notes that the generalized susceptibility differs by a factor of 2 from Ref. [75], a short explanation of the channel normalization would remove ambiguity.","section":"Appendix A, Eq. (A5)"},{"comment":"There are several minor language issues (e.g., 'We find that in our case, the phase transition occurs', 'Surprisingly, the elimination...'); a careful proofread is recommended.","section":"Sec. IV"}],"recommendation":"major_revision","confidential_remarks":"The central claim is important but currently rests on an uncontrolled approximation. The missing quantitative estimate of the residual four-fermion term and of the irreducible vertex remainder appears feasible with the existing numerical setup, so major revision rather than rejection is appropriate. Please ask the authors to address the specific ΛΛ'-1 estimate and a global measure of the irreducible remainder, not just the selected frequency cuts shown in Fig. 4."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new thing here is the channel choice Uc = -Us = U/2, with the -U/2 absorbed into the boson propagator, which lets the authors cancel the fermion-fermion vertex from the dual action and leaves a simple fermion-boson theory. That is not present in the earlier literature, and the paper is honest that a concurrent work [81] has a similar vertex decomposition but not this Fierz-ambiguity-free action. The benchmarks against the exact CT-HYB vertex and against ladder dual fermion are the right kind of evidence, and the agreement is decent across U = 0.5–1.5. The method is cheap and could be genuinely useful for realistic GW+DMFT-style calculations that want spin fluctuations. The paper is also candid about what is heuristic.\n\nThe main soft spot is exactly where the stress-test note lands. The step M ≈ Λ w̄ Λ with w̄ = w - U/2 is not a consequence of suppressing w-irreducible diagrams; algebraically it requires Λ² = 1. At low frequencies Λ deviates visibly from 1 (Fig. 1), so the dropped four-fermion term in Eq. (B13) is O(U(Λ²−1)) and is not small by construction. The paper justifies the step by high-frequency asymptotics and by agreement with more elaborate methods, which is reasonable but not a proof. The same applies to the broader claim that irreducible diagrams are suppressed: the argument is asymptotic and empirical, not controlled. These are real limitations, but they are stated in the paper, and the empirical benchmarks do support the method in the tested regime.\n\nMinor issues: no Monte Carlo error bars in Figs. 4–5, and the tests are limited to the single-band half-filled square lattice at T=0.1. The Mott-transition result is qualitative. The abstract and conclusions overstate the resolution of the Fierz ambiguity, but the core idea is plausible and worth taking seriously.\n\nWho is this for? Anyone doing diagrammatic extensions of DMFT, dual boson, TRILEX, or GW+DMFT who wants a cheap way to include collective spin/charge fluctuations. It deserves a serious referee: the idea is new, the benchmarks are proper, and the soft spot is identifiable and could be addressed with a more controlled test (e.g., a case where Λ²−1 is large, or a multiband system). I would send it to peer review rather than desk reject.","headline":"A new partial-bosonization scheme with a genuinely clever channel choice that likely sidesteps the Fierz ambiguity, but one uncontrolled algebraic step keeps it from being fully established.","tokens_in":22405,"tokens_out":1756,"would_cite":true,"duration_ms":19661,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A single interaction choice removes the Fierz ambiguity from bosonized Hubbard models.","keywords":["Fierz ambiguity","partial bosonization","extended Hubbard model","dynamical mean-field theory","dual boson","fermion-boson vertex","Mott transition","strongly correlated electrons"],"falsifier":"A decisive test is to compute, from an exact impurity solver, the difference between the full local vertex $\\Gamma^\\varsigma_{\\nu\\nu'\\omega}$ and the proposed boson-exchange approximation at low Matsubara frequencies and stronger coupling (for example, $U = 2$ at lower temperature), or to compare the resulting nonlocal self-energy with a numerically exact lattice calculation such as diagrammatic Monte Carlo; if the difference is not small, the cancellation that eliminates the fermion-fermion vertex fails and the channel dependence should reappear.","tokens_in":21291,"feed_emoji":"🧲","tokens_out":10473,"duration_ms":100080,"temperature":0.7,"pith_summary":"This paper argues that the famous Fierz ambiguity of partially bosonized theories — the dependence of approximate results on the arbitrary choice of how a local interaction is split into charge and spin channels — can be removed by one specific split: $U/2$ in the charge channel and $-U/2$ in each spin channel. With that split, the effective fermion-fermion interaction generated by exchanging a single boson almost reproduces the full local two-particle vertex, so no separate fermion-fermion vertex is needed in the action. The resulting fermion-boson action contains only the local fermion-boson vertex and a boson propagator shifted by $U/2$. Tested on the half-filled two-dimensional Hubbard model, the nonlocal self-energy from this cheap scheme closely tracks the result of a much more elaborate ladder calculation, and a self-consistent run places the Mott transition near $U \\approx 1.7$. This matters because it makes simultaneous charge and spin fluctuations practical in GW-like calculations for correlated materials.","feed_headline":"A single interaction choice removes the Fierz ambiguity","feed_subtitle":"A compact fermion-boson action matches costlier ladder calculations and opens realistic magnetic GW studies","key_machinery":"The carrying machinery is a decomposition of the local fermion-fermion vertex into horizontal (bosonic frequency $\\omega$) and vertical (transfer frequency $\\nu'-\\nu$) boson-exchange pieces, combined with a unique choice of the bare channel interaction $U_c = -U_s = U/2$. The key identity is the reducible approximation $M^\\varsigma_{\\nu\\nu'\\omega} = \\Lambda^\\varsigma_{\\nu\\omega} \\bar{w}^\\varsigma_\\omega \\Lambda^\\varsigma_{\\nu'+\\omega,-\\omega}$ with $\\bar{w}^\\varsigma_\\omega = w^\\varsigma_\\omega - U^\\varsigma/2$, which packages the effect of a bosonic line dressed by two fermion-boson vertices. This $M$ is then generated by a second Hubbard-Stratonovich transformation, so that the full four-fermion vertex disappears from the action and is replaced by a shifted bosonic propagator $W^\\varsigma = W^\\varsigma_{\\mathrm{EDMFT}} - U^\\varsigma/2$. What this machinery does is to turn a problem with an expensive four-fermion vertex into a fermion-boson problem whose simplest self-energy diagram already contains the leading ladder physics, without any dependence on the decoupling recipe.","core_discovery":"The central claim is that, for the extended Hubbard model, the Fierz ambiguity can be avoided entirely: the local Coulomb interaction $U$ is decoupled separately in every channel with the unique assignment $U_c = -U_s = U/2$ for all three spin components. Under this assignment the ladder-like irreducible contributions to the local fermion-fermion vertex are almost completely suppressed, and the reducible boson-exchange part $M^\\varsigma_{\\nu\\nu'\\omega} = \\Lambda^\\varsigma_{\\nu\\omega} \\bar{w}^\\varsigma_\\omega \\Lambda^\\varsigma_{\\nu'+\\omega,-\\omega}$, with $\\bar{w}^\\varsigma_\\omega = w^\\varsigma_\\omega - U^\\varsigma/2$, approximates the full vertex $\\Gamma^\\varsigma$ well enough that $\\Gamma^\\varsigma$ can be dropped from the dual action. The result is the effective action (9) for fermions interacting with bosons through the local vertex $\\Lambda^\\varsigma$, with bare boson propagator $W^\\varsigma_{q\\omega} = W^\\varsigma_{\\mathrm{EDMFT},q\\omega} - U^\\varsigma/2$. In the tested regime — half-filled two-dimensional Hubbard model, temperature $0.1$, $U = 0.5$, $1.0$, and $1.5$ — the nonlocal self-energy obtained from the simplest double-triangular diagrams agrees closely with ladder dual-fermion results, and a fully self-consistent solution gives a metal-to-Mott transition near $U \\approx 1.7$, lower than the plain DMFT value.","pith_inferences":["Editorial extension: The same cancellation could be tested for a particle-particle (superconducting) decoupling, where the paper notes the construction is possible in principle; a suppression check on the pairing vertex would show whether the method survives competing superconducting fluctuations.","Editorial extension: The quality of the scheme should degrade where the fermion-boson vertex deviates strongly from unity; a systematic scan of $U$ and temperature would map the boundary of the regime where the Fierz-free action remains accurate.","Editorial extension: A natural multiorbital generalization would assign channel-dependent bare interactions satisfying the same cancellation conditions; if it works, magnetic fluctuations and charge screening would enter realistic materials calculations on equal footing."],"forward_implications":["The nonlocal self-energy and screened interaction can be obtained from a single simple diagram set, avoiding a Bethe-Salpeter inversion and making simultaneous charge and spin fluctuations numerically cheap.","The screened interaction $W$ is improved in both charge and spin channels, which is precisely what a magnetic, realistic GW-style calculation needs.","Because the fermion-fermion vertex is removed from the action, the effective fermion-boson model can be solved by fRG, parquet, or diagrammatic Monte Carlo without the usual Fierz-induced channel dependence.","For the half-filled two-dimensional Hubbard model, the metal-to-Mott transition is found near $U \\approx 1.7$, below the DMFT value, in line with cluster and second-order dual-fermion estimates."],"supporting_citations":[{"why":"Provides the dual fermion ladder method whose nonlocal self-energy the new scheme is benchmarked against.","marker":"[58]"},{"why":"The dual boson formalism that supplies the starting dual action with bosonic fields and local vertex functions.","marker":"[59, 60]"},{"why":"Derives the nonlocal EDMFT propagators and vertex structure of the dual action used to build the effective fermion-boson model.","marker":"[61]"},{"why":"The previous cluster-based attempt to reduce the Fierz ambiguity, which the present single-site construction avoids.","marker":"[44]"},{"why":"Benchmark studies establishing the accuracy of dual theories with two-particle interactions, the standard the new self-energy is compared with.","marker":"[62–64]"},{"why":"Earlier vertex parametrizations with fermion-boson vertex corrections in a single channel, which the multi-channel parametrization extends.","marker":"[76, 77]"},{"why":"Supplies a weak-coupling parametrization of the vertex and supports the claim that particle-particle contributions affect the self-energy only weakly.","marker":"[75]"},{"why":"Provides the single-site TRILEX Mott-transition result with which the lower transition point is contrasted.","marker":"[43]"}],"fun_headline_variants":["Channel-wise bosonization resolves Fierz ambiguity","One interaction split removes Fierz ambiguity","Bosonize each channel, no more Fierz ambiguity","Consistent partial bosonization avoids Fierz problem","Fierz ambiguity cured by channel-separated bosons"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the only important two-particle correlations are the ladder-like ones built from a single boson line, so the part of the local vertex that is not captured by exchanging one boson stays small.","fun_headline_variants_meta":{"raw":{"variants":["Channel-wise bosonization resolves Fierz ambiguity","One interaction split removes Fierz ambiguity","Bosonize each channel, no more Fierz ambiguity","Consistent partial bosonization avoids Fierz problem","Fierz ambiguity cured by channel-separated bosons"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000458,"raw_usage":{"total_tokens":2311,"prompt_tokens":973,"completion_tokens":1338,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":589,"completion_tokens_details":{"reasoning_tokens":1264}},"tokens_in":589,"tokens_out":1338,"duration_ms":12861,"temperature":1.0,"reasoning_tokens":1264,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:48:40.975732+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test is to compute, from an exact impurity solver, the difference between the full local vertex $\\Gamma^\\varsigma_{\\nu\\nu'\\omega}$ and the proposed boson-exchange approximation at low Matsubara frequencies and stronger coupling (for example, $U = 2$ at lower temperature), or to compare the resulting nonlocal self-energy with a numerically exact lattice calculation such as diagrammatic Monte Carlo; if the difference is not small, the cancellation that eliminates the fermion-fermion vertex fails and the channel dependence should reappear.","supporting_citations":[{"cited_title":"Dual fermion approach to nonlocal correlations in the hubbard model,","cited_arxiv_id":null,"evidence_quote":"Provides the dual fermion ladder method whose nonlocal self-energy the new scheme is benchmarked against."},{"cited_title":"From local to nonlocal correla- tions: The Dual Boson perspective,","cited_arxiv_id":null,"evidence_quote":"Derives the nonlocal EDMFT propagators and vertex structure of the dual action used to build the effective fermion-boson model."},{"cited_title":"Fierz convergence criterion: A controlled approach to strongly inter- acting systems with small embedded clusters,","cited_arxiv_id":null,"evidence_quote":"The previous cluster-based attempt to reduce the Fierz ambiguity, which the present single-site construction avoids."},{"cited_title":"Fluctuation Di- agnostics of the Electron Self-Energy: Origin of the Pseudogap Physics,","cited_arxiv_id":null,"evidence_quote":"Supplies a weak-coupling parametrization of the vertex and supports the claim that particle-particle contributions affect the self-energy only weakly."},{"cited_title":"Mott physics and spin fluctuations: A functional viewpoint,","cited_arxiv_id":null,"evidence_quote":"Provides the single-site TRILEX Mott-transition result with which the lower transition point is contrasted."}],"review_version":1}