{"id":"6e2bd431-cff8-4302-ae96-daf9f6e5f396","arxiv_id":"2412.00924","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Patch bosonization cannot be used for a two-dimensional metal at a quantum critical point, except in artificial small- or large-flavor limits.","lead":"Patch bosonization, a method used to describe electrons in metals, is argued to fail for two-dimensional metals at a quantum critical point, except in artificial limits with very few or very many particle types. The paper proposes new toy limits where the method is controllable and warns that results at realistic particle numbers are unreliable.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Nonperturbative sum-versus-integral claim in §4.2 relies on smoothness; the actual integrands have poles, so the discrepancy may be algebraic or O(1), not exponentially small.","rationale":"The reader's weakest_assumption matches our stress-test: the nonperturbative crossover from synthetic continuum limits to the physical discrete-patch model rests on the sum-versus-integral property after Eq. (66). We agree this is the least secure link. The author's argument is an assertion plus a Matlab check on generic smooth functions; it does not account for the poles and branch cuts in the actual Green's functions. Since the central claim is a negative no-go statement ('fundamentally inapplicable'), the burden is on this nonperturbative separation. If the discrepancy is polynomial, the strongest form of the claim fails, though the cutoff inconsistency of §4.1 would still stand as a strong critique of standard continuum bosonization. Thus the reader's CONDITIONAL verdict is appropriate: the paper makes a coherent case but its central conclusion depends on an unproved mathematical property. No change to the verdict is needed.","tokens_in":25977,"tokens_out":9846,"duration_ms":96514,"concrete_test":"Compute the discrete-sum version of the one-loop boson self-energy, replacing the χ-integral in Eq. (19) by a sum over N_patches = 2π/Δχ patches: Π_discrete(ω,k) = (g² k_F/4π²) Σ_m Δχ [k cosχ_m / (ω − v_F k cosχ_m + iε)]. Compare to the continuum result Eq. (19) for fixed ω/k and various pairs (Δχ, ε), including ε = Δχ, ε = Δχ², ε = Δχ³, and ε fixed. Because the integrand has a pole, also derive the leading discrepancy analytically via Poisson summation. If |Π_discrete − Π_cont| decays as Δχ^p (p ≥ 1) or is O(1) near the pole, rather than faster than any power, then the §4.2 nonperturbative claim fails. Repeat for the fermion self-energy integrand in Eq. (53) to confirm.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.2's central no-go step is the assertion after Eq. (66) that the difference between a discrete sum over patches and its continuum integral is smaller than any power of the patch spacing δx. This is justified by Euler-Maclaurin with vanishing boundary terms and a Matlab check, but the Euler-Maclaurin remainder estimate assumes the summand is smooth on the integration contour. The actual diagrammatic integrands are not smooth: the boson self-energy Eq. (19) contains a pole at ω = v_F k cosχ (regulated by iε), and the fermion self-energy Eq. (53) inherits the same singular structure. For such functions, the Poisson summation formula shows the discrepancy contains Fourier modes of the pole, which decay algebraically in |m| unless the regulator width ε is much larger than δx. In the critical limit ε→0 and δx→0, the two limits do not commute; the discrepancy can be O(1) or a power of δx. The author's Matlab experiments on generic smooth functions do not probe this regime. If the discrepancy is polynomial, ordinary perturbation theory in the patch spacing could connect the continuum synthetic limits to the discrete N=1 theory, converting the claimed nonperturbative crossover into a controlled expansion and undermining the 'fundamentally inapplicable' conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that patch bosonization of Fermi surfaces cannot describe the two-dimensional metal at a quantum critical point for N=1, except in contrived small-N and large-N limits. The argument has three parts: (i) Section 4.1 presents an internal-consistency check showing that the boson self-energy requires the continuum patch limit with ω^(1/3) much larger than the patch spacing, while the fermion self-energy requires the opposite inequality; (ii) Section 4.2 introduces a small-N bosonizable limit and claims that the crossover from the continuum of patches to the discrete-patch N=1 model is nonperturbative because the sum-integral difference vanishes faster than any power of the spacing; (iii) Sections 4.3-4.4 outline a random-coupling large-N bosonizable theory and an expansion in intra-patch curvature, with the caveat that the expansion has not been completed. The paper concludes that bosonization is fundamentally inapplicable except in synthetic limits.","tokens_in":26203,"tokens_out":7696,"duration_ms":69236,"significance":"If correct, the paper would be an important and timely critique of a widely used formalism; it also makes a constructive proposal for a new large-N bosonizable model. The paper is commendably explicit about its own open issues (e.g., Section 4.4) and about the distinction between internal inconsistency and mere disagreement with conventional wisdom. However, the central no-go claim depends on a nonperturbative sum-versus-integral assertion that is not proved for the singular propagators of the theory; the paper's strength is therefore more in the clear formulation of the cutoff problem than in a definitive demonstration of the nonperturbative conclusion.","major_comments":[{"comment":"The assertion that the difference between the discrete sum over patches and the continuum integral is smaller than any power of δx is essential for the claim that the small-N continuum theory is not perturbatively connected to the N=1 model, but it is not established for the actual integrands. The Euler-Maclaurin argument assumes the summand is smooth on the integration domain, while the boson self-energy in Eq. (19) has a pole at ω = v_F k cosχ regulated only by iε, and the fermion self-energy in Eq. (53) inherits the same singular structure. The Matlab check on generic smooth functions does not probe this regime. For singular integrands, the discrepancy typically decays as a power of δx (or is O(1) if the pole and the patch spacing scale incompatibly), which would turn the claimed nonperturbative crossover into a perturbatively controlled expansion in the patch spacing and invalidate the conclusion that the continuum limits are fundamentally disconnected from the discrete model. Please provide a proof or careful analytic/numerical treatment of the sum-integral discrepancy for the actual propagators, or weaken the conclusion accordingly.","section":"Section 4.2, after Eq. (66)"},{"comment":"The abstract states that it is 'at least formally possible to construct perturbative expansions around these synthetic limits,' but Section 4.4 (final paragraph) reports that one of the two order-γ² patch-local diagrams was not calculated and concludes that 'it is clear that more work is required to determine whether the perturbation theory around linearized bosonization is sensible.' This is an internal inconsistency between the level of confidence in the abstract and the evidence presented. Either provide the missing calculation and a demonstration that the C(γ) counterterm renders the expansion well-defined, or explicitly reformulate the abstract as a conjecture rather than a shown result.","section":"Abstract and Section 4.4"}],"minor_comments":[{"comment":"The word 'one-dimensinoal' should read 'one-dimensional'; also, the phrase 'the latter of which' is used ambiguously in the following sentence.","section":"Section 2.2, Eq. (18)"},{"comment":"The name 'Euler-Macluarin' is a typo for 'Euler-Maclaurin'.","section":"Section 4.2, Eq. (66)"},{"comment":"The sentence using 'dχdy = 1/N' would benefit from specifying that this holds in the rescaled units with the cutoffs in Eq. (56); as written, the measure appears dimensionally inconsistent.","section":"Section 4.2, after Eq. (62)"},{"comment":"The summation index n in the geometric series is overloaded with the number of fermion flavors n; consider renaming one of them to avoid confusion.","section":"Section 4.3, Eq. (73)"},{"comment":"The notation 'iη∂t' is introduced without defining η; a sentence explaining that η is the coefficient of the irrelevant kinetic term would help.","section":"Section 3.4, Eq. (49)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a provocative single-author perspective. The editor may wish to ensure that the review process includes a referee familiar with both patch bosonization and nonperturbative methods; the main concern is whether the nonperturbative sum-versus-integral assertion is sufficiently supported for the strong negative conclusion."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this one. The paper's real contribution is the cutoff-inconsistency argument in Section 4.1: the boson self-energy needs the continuum patch limit with omega^(1/3) much larger than the cutoff, while the fermion self-energy needs the opposite limit, so one cutoff cannot serve both. That is a simple, internally coherent criterion drawn from the theory's own equations, not from fitted parameters. It gives a concrete reason to doubt the bosonized action (18) and its non-Fermi liquid solution (36) as descriptions of the physical N=1 critical metal.\n\nThe paper is also a fair, readable review of patch bosonization, and it adds two synthetic limits. The small-N limit is attributed to earlier work, but the random-coupling double-large-N model in Section 4.3 is new and genuinely exactly bosonizable after disorder averaging. The treatment of Klein factors and the proposed C-term is honest: the author explicitly says the perturbation theory is incomplete and that more work is needed before it can be assessed.\n\nThe soft spots are where the reader and stress-test point. The strongest form of the no-go claim rests on the assertion after Eq. (66) that the difference between a discrete patch sum and its continuum integral vanishes faster than any power of the patch spacing. That is not proved. The Euler-Maclaurin argument assumes smoothness, and the actual integrands have poles (Eqs. 19 and 53). The stress-test concern is legitimate: the epsilon-to-zero and delta-x-to-zero limits may not commute, and the discrepancy could be algebraic or even O(1). If so, a perturbative expansion in patch spacing could connect the continuum synthetic limits to the discrete N=1 theory, which would soften \"fundamentally inapplicable\" to \"not obviously controlled.\" That is a real gap, but it does not undo the Section 4.1 cutoff inconsistency, which stands independently.\n\nThe small-N and large-N expansions are also admittedly incomplete, and the critique of coadjoint orbit bosonization is more schematic than the rest. Those are proportionate reservations, not fatal flaws.\n\nWho should read this: theorists working on non-Fermi liquids, bosonization, or quantum critical metals. It is a negative result, but a useful one, and the proposed large-N model gives it independent value. I would send it to a serious referee with both bosonization and Eliashberg expertise, asking them to press on the sum-versus-integral step and the regulator non-commutativity. Recommendation: accept for peer review, not desk reject, with revision expected on the nonperturbative crossover claim.","headline":"A genuinely useful negative result about patch bosonization at quantum criticality, with a clean cutoff-inconsistency argument and an honest admission that the strongest no-go claim rests on an unproved sum-versus-integral step.","tokens_in":26747,"tokens_out":2334,"would_cite":true,"duration_ms":24031,"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":"Patch bosonization cannot describe the 2D critical metal at N ~ 1; only synthetic limits are solvable.","keywords":["patch bosonization","non-Fermi liquid","quantum critical metal","Fermi surface","small-N limit","large-N limit","cutoff consistency","boson-fermion model"],"falsifier":"Compute the exact difference between the discrete sum and continuum integral in the one-loop boson and fermion self-energy integrals at finite $\\Delta\\chi$; if the difference decays as a power of $\\Delta\\chi$ rather than faster than every power, the claimed nonperturbative barrier to reaching $N=1$ is absent. A complementary check is to compare the bosonized zero-frequency fermion propagator at nonzero momentum with a numerically exact calculation of the physical critical metal: the paper predicts the two disagree already at leading order.","tokens_in":1891,"feed_emoji":"⚛️","tokens_out":2223,"duration_ms":75375,"temperature":0.7,"pith_summary":"Two-dimensional metals at a quantum critical point are among the simplest models with no Landau quasiparticles, and patch bosonization has often been used to solve them exactly. The paper argues that this method is fundamentally inapplicable to the physical case: the critical regime is not semiclassical, so the patched theory only works by treating its momentum cutoff inconsistently. Bosonization becomes exact only in synthetic limits, a small-$N$ limit already known in the literature and a new random double-large-$N$ limit, and neither is perturbatively connected to $N=1$. If the paper is right, the widely used bosonized non-Fermi liquid solution is an artifact, and controlled results for the critical metal must come from other methods. The paper also shows where bosonization does remain exact, namely in the high-frequency transport regime, where it agrees with anomaly-based exact results.","feed_headline":"Patch bosonization cannot describe the 2D critical metal","feed_subtitle":"Only synthetic small-N and large-N limits are solvable; at N ~ 1, discrete-patch effects take over.","key_machinery":"Patch bosonization rewrites the Fermi surface as a set of discrete patches labelled by angle $\\chi$ and tangential position $x_\\parallel$, with cutoffs linked by $\\Lambda_\\parallel = \\frac{1}{2}k_F\\Delta\\chi$; the resulting continuum of chiral bosons $\\zeta(t,x,\\chi)$ describes electron-hole pairs. The paper's argument rests on that cutoff structure: scale invariance forces dynamical uncertainty in $\\chi$ and $x_\\parallel$ to scale with the same power of energy, so the semiclassical separation of scales that justifies the continuum limit cannot hold in the critical regime. To make the failure quantitative, the paper uses the Euler-Maclaurin formula to show that a discrete sum over patches and its continuum integral differ by terms smaller than any power of the patch spacing, which is what blocks any perturbative expansion from the continuum bosonized solution back to the physical model.","core_discovery":"The paper's central claim is that Fermi-surface patch bosonization is fundamentally inapplicable to a two-dimensional metal at a quantum critical point, except in synthetic limits that are not adiabatically connected to the physical model. The bosonized action (18) is quadratic, exactly solvable, and yields a scale-invariant non-Fermi liquid with fermion propagator (36); the paper argues that this solution is an artifact of treating the cutoff inconsistently. In the critical regime, the boson self-energy requires a continuum of patches with $\\omega^{1/3}\\gg \\Lambda$, while the fermion self-energy requires $\\Lambda \\gg \\omega^{1/3}$; no honest cutoff can satisfy both. The formalism is exact only in a small-$N$ limit ($N\\to0$) or in a random double-large-$N$ limit, and perturbative expansions around either cannot cross over to $N=1$ because the discrete sum over patches differs from its continuum integral by terms smaller than any power of the patch spacing.","pith_inferences":["If this argument is right, any transport or spectral calculation for strange metals that starts from a bosonized critical Fermi surface inherits the same cutoff inconsistency, even if it dresses the result with higher loops; the problem is in the starting semiclassical description, not in the loop order.","The Euler-Maclaurin argument suggests a precise mathematical test: if a rigorous comparison of patch sums and integrals shows only exponential, not polynomial, suppression, then the crossover from continuum bosonization to the discrete patch model is genuinely nonperturbative, which would support the paper's main claim.","One could try to use anomaly-derived exact constraints at nonzero frequency and momentum; any discrepancy with the bosonized solution outside the transport regime would give an independent falsifier of the bosonized critical theory.","The proposed large-$N$ random model, though complicated, is the only solvable bosonizable non-Fermi liquid found; it may serve as a benchmark for testing whether non-bosonizable corrections at $N\\sim1$ change critical exponents, which the paper leaves open."],"forward_implications":["The bosonized non-Fermi liquid solution, including the anomalous fermion propagator (36), is not the physics of the physical $N=1$ critical metal; it is the exact solution of a different, synthetic model.","Only synthetic limits are controllable: the small-$N$ limit ($N\\to0$), and the proposed random double-large-$N$ limit with many more bosonic than fermionic flavors; in both, bosonization is exact or effectively solvable.","Perturbation theory around these limits (the intra-patch curvature as a cubic $\\zeta$ vertex, with Klein factors) is formally constructible but cannot converge to the correct $N=1$ answer once nonperturbative discrete-patch effects set in.","In the transport regime $\\omega \\gg v_F k$, bosonization is exact and reproduces anomaly-based results for the boson mass and the free Drude optical conductivity; the failure is specific to the critical regime $\\omega \\ll v_F k$.","The renormalized electron-hole continuum found in the bosonized theory is a consequence of its quadraticity and should not be assumed to be the true excitation spectrum of the critical metal."],"supporting_citations":[{"why":"Defines the one-loop $\\omega^{2/3}$ self-energy and the single-patch fixed-point scaling that the bosonized solution purports to reproduce; the paper argues the two approaches cannot agree.","marker":"[2]"},{"why":"Supplies the classic large-$N$ analysis of fermions coupled to a gapless boson and the low-energy non-Fermi liquid target; the paper contrasts its synthetic limits with this standard treatment.","marker":"[3]"},{"why":"Provides the random-coupling large-$N$ self-consistent one-loop theory, the main existing controlled model that the paper uses as a benchmark and contrast for bosonization.","marker":"[5]"},{"why":"Gives the original derivation of the bosonized fermion propagator (36), the object whose physical validity the paper rejects at $N=1$.","marker":"[9]"},{"why":"Provides the equal-time bosonized result used to check the fermion propagator; the paper notes an inconsistency with the equal-position result.","marker":"[10]"},{"why":"Is the coadjoint-orbit bosonization perturbative program that the paper argues remains singular and does not escape the cutoff problem.","marker":"[14]"},{"why":"Contains the patch $U(1)$ symmetry and anomaly that motivates bosonization and anchors the exact results in the transport regime.","marker":"[18]"},{"why":"Gives the anomaly-based exact boson propagator and optical conductivity, used to delimit the regime where bosonization is exact.","marker":"[24]"}],"fun_headline_variants":["Patch bosonization fails for real 2D critical metals","Patch bosonization only valid in synthetic N limits","No patch bosonization for physical 2D metals","2D critical metal escapes patch bosonization","Patch bosonization dead end for 2D quantum criticality"],"cache_read_input_tokens":28800,"weakest_assumption_plain":"The conclusion rests on the assertion that the difference between a discrete sum over patches and its continuum integral vanishes faster than every power of the patch spacing; if the mismatch were polynomial, a perturbative expansion around the continuum solution could in principle reach the realistic $N=1$ model.","fun_headline_variants_meta":{"raw":{"variants":["Patch bosonization fails for real 2D critical metals","Patch bosonization only valid in synthetic N limits","No patch bosonization for physical 2D metals","2D critical metal escapes patch bosonization","Patch bosonization dead end for 2D quantum criticality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00069,"raw_usage":{"total_tokens":3074,"prompt_tokens":842,"completion_tokens":2232,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":458,"completion_tokens_details":{"reasoning_tokens":2151}},"tokens_in":458,"tokens_out":2232,"duration_ms":14792,"temperature":1.0,"reasoning_tokens":2151,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:51:59.014949+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the exact difference between the discrete sum and continuum integral in the one-loop boson and fermion self-energy integrals at finite $\\Delta\\chi$; if the difference decays as a power of $\\Delta\\chi$ rather than faster than every power, the claimed nonperturbative barrier to reaching $N=1$ is absent. A complementary check is to compare the bosonized zero-frequency fermion propagator at nonzero momentum with a numerically exact calculation of the physical critical metal: the paper predicts the two disagree already at leading order.","supporting_citations":[{"cited_title":"Large-N Theory of Critical Fermi Surfaces,","cited_arxiv_id":null,"evidence_quote":"Provides the random-coupling large-$N$ self-consistent one-loop theory, the main existing controlled model that the paper uses as a benchmark and contrast for bosonization."},{"cited_title":"Postmodern Fermi Liquids,","cited_arxiv_id":null,"evidence_quote":"Is the coadjoint-orbit bosonization perturbative program that the paper argues remains singular and does not escape the cutoff problem."},{"cited_title":"Non-Fermi Liquids as Ersatz Fermi Liquids: General Constraints on Compressible Metals,","cited_arxiv_id":null,"evidence_quote":"Contains the patch $U(1)$ symmetry and anomaly that motivates bosonization and anchors the exact results in the transport regime."},{"cited_title":"Gifts from Anomalies: Exact Results for Landau Phase Transitions in Metals,","cited_arxiv_id":null,"evidence_quote":"Gives the anomaly-based exact boson propagator and optical conductivity, used to delimit the regime where bosonization is exact."}],"review_version":1}