{"id":"f49b56e4-658d-47f0-a9ad-54532aacaeef","arxiv_id":"2511.03654","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A rigorous proof shows the momentum distribution of a Fermi-gas trial state is the Fermi step plus the RPA correction, with k_F^{-2+ε} errors for summable interactions.","lead":"This paper proves that a specially built trial state of a Fermi gas has the momentum distribution predicted by Daniel and Vosko: the free Fermi step plus a random-phase-approximation correction, with rigorous error bounds. It is the first rigorous version for Coulomb interactions and for momenta arbitrarily close to the Fermi surface.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bootstrap closing step in §7 (Eq. 7.2) is not justified: C_ε k_F^{-1+ε} Ξ^{1/2} is not o(1)Ξ; the proof as written does not yield Ξ≤C k_F^{-1}.","rationale":"The reader identified the imported energy bounds from [CHN24]/[CHN23a] as the weakest assumption. That is a valid concern but it is a standard reliance on prior work, not an internal flaw. By contrast, the bootstrap step in Section 7 is internal to the proof of Proposition 1.2, the central claim of the paper. The displayed inequality (7.2) contains an unjustified o(1)Ξ absorption: the term k_F^{-1+ε} Ξ^{1/2} is not asymptotically smaller than Ξ under the only available a priori bound Ξ≤1. This is precisely the kind of technical estimate the reader said could not be machine-checked. The gap appears fixable by a standard iteration, and the final result may survive, but as written the proof does not rigorously establish the central error bounds. Therefore the appropriate verdict is CONDITIONAL: accept only after the bootstrap step is corrected and the resulting exponents are verified. I do not see grounds for rejection, because the rest of the argument is detailed, the leading-order identification (Lemma 6.1) is consistent with (1.9) (despite the 'summing' typo), and the error estimates are plausible.","tokens_in":42648,"tokens_out":22745,"duration_ms":165422,"concrete_test":"Re-derive the bootstrap in Section 7 without the step C_ε k_F^{-1+ε} Ξ^{1/2} = o(1)Ξ. Iterate the inequality starting from the trivial bound Ξ≤1: first show Ξ≤C k_F^{-1+ε}, then substitute to show Ξ≤C k_F^{-1} for ε<1/3. If this works, verify that inserting the resulting bound into (7.1) with γ=1/3 still yields the claimed error exponents in (1.25)/(1.26). If the iteration only gives Ξ≤C k_F^{-2+2ε}, check whether the final error bounds still hold; if not, the theorem's error estimates are unsupported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The proof of Proposition 1.2 is closed in Section 7 via the bootstrap inequality\n\nΞ ≤ sup_q n_RPA(q) + sup_q n_ex(q) + C_ε(k_F^{-1} + k_F^{-1+ε} Ξ^{1/2}) ≤ C k_F^{-1} + o(1)Ξ ⇒ Ξ ≤ C k_F^{-1}.\n\nThe second inequality is not justified. From the preceding bound, using only the trivial Ξ≤1, one gets C_ε k_F^{-1+ε} Ξ^{1/2} ≤ C_ε k_F^{-1+ε}, and k_F^{-1+ε} is not o(1)Ξ for Ξ of order k_F^{-1} or smaller. For any ε>0, k_F^{-1+ε} is asymptotically larger than k_F^{-1}, so it cannot be absorbed as an o(1) error relative to Ξ. A rigorous treatment would iterate the quadratic inequality: first Ξ≤C k_F^{-1+ε}, then, for ε<1/3, the term k_F^{-1+ε}Ξ^{1/2} becomes O(k_F^{-3/2+3ε/2}), which is o(k_F^{-1}), yielding Ξ≤C k_F^{-1}. Thus the stated conclusion is likely salvageable, but the displayed step in (7.2) is a genuine gap at the load-bearing point: the bootstrap controls the error terms in (7.1) and is used to derive the per-momentum bound (7.3) that gives the final error estimates (1.25)/(1.26). If the corrected bootstrap only gave Ξ≤C k_F^{-2+2ε}, the final exponents could change, so this needs to be checked rather than assumed.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers N spinless fermions on T^3 in the mean-field scaling H_N = -ΣΔ + k_F^{-1}ΣV(x_i-x_j). For the trial state Ψ_N = Re^{-S}Ω introduced in [CHN23a], it proves that the momentum distribution satisfies n(q) = n_RPA(q)+n_ex(q)+E(q) for |q|≥k_F, and the complementary formula inside the Fermi ball, with explicit pointwise error bounds. Under the Coulomb-class condition (1.14), the error is C_ε k_F^{-1-1/6+ε} e(q)^{-1}; under the stronger summability condition (1.18), it is C_ε k_F^{-2+ε} e(q)^{-1}. The proof uses a Duhamel expansion of e^S a_q^* a_q e^{-S}, an exact evaluation of the leading term via the Sherman–Morrison formula, normal ordering of the many-body errors, and a bootstrap on the global quantity Ξ = sup_{q,λ} ⟨e^{-λS}Ω, a_q^*a_q e^{-λS}Ω⟩. The energy closeness of the trial state is imported from [CHN24] and [CHN23a].","tokens_in":43084,"tokens_out":16505,"duration_ms":138483,"significance":"If the proof is completed, this is a substantial advance: it gives the first rigorous pointwise control of the random-phase-approximation correction to the momentum distribution for Coulomb-type potentials in the mean-field regime, including momenta arbitrarily close to the Fermi surface, thereby refining [BL25]. The trial state is explicit, the leading term n_RPA(q) is parameter-free and computed exactly, and the error analysis is highly detailed and checkable. No fitted constants or ad hoc assumptions are introduced beyond the cited energy bounds. The proof is long, but the organization is clear and the main technical estimates are stated separately. The result is likely correct, but the manuscript contains a load-bearing bootstrap step that needs repair before the theorem can be accepted as proven.","major_comments":[{"comment":"The displayed bootstrap implication is not justified. From the preceding bound and the trivial bound Ξ≤1, the term C_ε k_F^{-1+ε} Ξ^{1/2} is only bounded by C_ε k_F^{-1+ε}, and k_F^{-1+ε} is not o(1)Ξ when Ξ is of order k_F^{-1}. Thus the displayed step 'Ξ ≤ Ck_F^{-1} + o(1)Ξ' does not follow. This step is load-bearing: it is used to obtain the pointwise bound (7.3) and hence the final exponents in (1.25)/(1.26). The gap is likely fixable by a two-step iteration: first use Ξ≤1 to get Ξ≤C_ε k_F^{-1+ε}; then, for fixed ε<1, k_F^{-1+ε}Ξ^{1/2}≤C_ε k_F^{-3/2+3ε/2}=o(k_F^{-1}), yielding Ξ≤Ck_F^{-1}. Please write out this iteration (or an equivalent argument) explicitly. As written, the proof of Proposition 1.2 is incomplete at this point.","section":"§7, Eq. (7.2)"},{"comment":"Lemma 6.2 is stated for potentials with \\g\\V∈ℓ^1(Z^3), but it is used in the proof of Proposition 1.2 for all potentials satisfying (1.23), and Theorem 1.1 covers potentials with |\\g\\V(ℓ)|≤C|ℓ|^{-2}, which are not in ℓ^1 in three dimensions. The proof of (6.9) appears to need only boundedness of \\g\\V together with \\sum_ℓ \\g\\V(ℓ)^2<∞, or a similar condition implied by (1.23). Please restate and prove Lemma 6.2 under the hypotheses actually used, or supply a reduction. As stated, the lemma does not cover the Coulomb case, and since the bootstrap (7.2) invokes this bound, this is a missing justification at a load-bearing point.","section":"§6, Lemma 6.2 and proof of Prop. 1.2"}],"minor_comments":[{"comment":"The condition \\g\\V(ℓ)≤C|ℓ|^{-2} is stated for all ℓ∈Z^3, but |ℓ|^{-2} is undefined at ℓ=0. For the periodic Coulomb potential the zero mode is usually regularized or vanishes; please state the condition for ℓ≠0 or assume \\g\\V(0)=0.","section":"Hypothesis (1.14)"},{"comment":"The term 'optimal error bounds' is used without a matching lower bound. Since no lower bound is proved, 'sharp for the present method' or 'of the expected optimal order' would be more precise.","section":"Abstract and Remark 1"},{"comment":"The first displayed formula in the proof involves pulling number operators through a_q; the identity is correct after using [N,a_q^*a_q]=0 and a_q f(N+1)=f(N+2)a_q, but this should be made explicit to avoid confusion.","section":"§5.2, Lemma 5.6"},{"comment":"The formal continuum limit on the right-hand side is heuristic and contains powers of 2π that are not derived in detail; please state explicitly that this is only a heuristic comparison and not part of the rigorous theorem.","section":"Eq. (1.21)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a serious and detailed contribution, and the central result is likely correct after a local repair. The main issue is the bootstrap step in §7, Eq. (7.2), which is not justified as written; the authors should supply the standard two-step iteration. The mismatch between Lemma 6.2's stated hypothesis and the assumptions used in the proof also needs to be fixed, though it appears to be a straightforward adaptation. I would not reject: the proof is long but well structured, and the gap is localized and fixable within the scope of the paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a serious paper, correct in its main idea, but the bootstrap argument in Section 7 has a genuine gap that needs repair before the proof closes.\n\nThe genuinely new content is the removal of the patch cutoffs from the earlier BL25 analysis, the handling of the Coulomb-class singularity |ell|^-2, and the pointwise bound e(q)^-1 near the Fermi surface. The trial state is the CHN23a state, and the proof works through a very detailed Duhamel expansion with all error terms organized and estimated. The leading-order calculation in Lemma 6.1 via Sherman-Morrison is clean, and the extraction of the exchange term is done carefully. This is honest, reproducible mathematics: no fitted constants, no hidden assumptions beyond those stated.\n\nThe soft spot is exactly the one in the stress-test note. In (7.2), the inequality Xi <= C k_F^-1 + C_eps k_F^(-1+eps) Xi^(1/2) is turned into Xi <= C k_F^-1 + o(1) Xi. That \"o(1) Xi\" is not justified from the previous line, because k_F^(-1+eps) Xi^(1/2) is not little-o of Xi unless you already know Xi >= k_F^(-2+2eps). The fix is straightforward: iterate the quadratic inequality, first getting Xi <= C_eps k_F^(-1+eps), then substituting back to get Xi <= C k_F^-1 for sufficiently small eps. The stress-test note is right that this is likely salvageable, but (7.2) as written is a real gap at the load-bearing point. The final error exponents could change if the iteration only gave a weaker bound, so this needs to be checked.\n\nThe other caveats are minor. The energy closeness (1.15) is imported from CHN24/CHN23a, not proven here; that's a legitimate black box. And the whole thing is about a trial state, not the true ground state; the authors are explicit about that, and the heuristic extension is clearly labeled.\n\nWho should read it: researchers working on rigorous many-body perturbation theory for Fermi gases, bosonization, and RPA. It's a technical paper, but the proof is written in full and the bootstrap gap is fixable. I'd send it to a serious referee. With a corrected (7.2) it should be publishable.","headline":"Solid RPA momentum-distribution paper for a trial state; Section 7 bootstrap has a real but likely fixable gap.","tokens_in":43580,"tokens_out":5909,"would_cite":true,"duration_ms":44492,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81V74","82D20","81Q05"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a Coulomb-interacting Fermi gas in the mean-field scaling limit, the momentum distribution is a step profile corrected by the random phase approximation, with rigorous pointwise error bounds down to the Fermi surface.","keywords":["momentum distribution","Fermi gas","random phase approximation","Coulomb interaction","mean-field scaling","bosonization","Bogoliubov transformation","Fermi surface"],"falsifier":"Evaluate the momentum distribution of the explicit trial state (2.8)-(2.14) for a Coulomb potential on the torus at momenta q with e(q)=O(1) and compare it with n_RPA(q). The theorem predicts |n(q)-n_RPA(q)| ≤ C_ε k_F^{-1-1/6+ε}; observing a decay slower than k_F^{-1-1/6} for any fixed distance to the Fermi surface would disprove it. Alternatively, a counterexample to the imported energy lower bound E_gs ≥ E_FS + E_corr - C k_F^{1-1/6+ε} for Coulomb would falsify Theorem 1.1 even if the momentum formula itself remains true.","tokens_in":42539,"feed_emoji":"⚛️","tokens_out":7387,"duration_ms":62060,"temperature":0.7,"pith_summary":"This paper proves that, in the mean-field scaling limit, a three-dimensional Fermi gas with Coulomb interaction has a trial state whose momentum distribution is the free Fermi-ball step profile corrected by the random phase approximation (RPA), with rigorous pointwise error bounds that remain valid arbitrarily close to the Fermi surface. The RPA correction, predicted by Daniel and Vosko on physical grounds, is shown to scale as k_F^{-1} e(q)^{-1}, where e(q) measures the distance of q to the Fermi surface; the deviation from this profile is controlled by k_F^{-1-1/6+ε} e(q)^{-1} for Coulomb-type potentials, improving to k_F^{-2+ε} e(q)^{-1} when the potential has summable Fourier transform. The result matters because the momentum distribution is the basic one-body observable of a Fermi liquid, and this analysis extends rigorous control of the interaction correction to the physically most relevant Coulomb case and to momenta closer to the Fermi surface than previous work. The proof uses a trial state built from approximate bosonic particle-hole pairs and a double bootstrap on the maximum of the momentum distribution.","feed_headline":"Coulomb Fermi gas momentum distribution proven to match RPA","feed_subtitle":"Rigorous pointwise error bounds now reach momenta arbitrarily close to the Fermi surface.","key_machinery":"The trial state is Ψ_N = R e^{-S} Ω, where R is the particle-hole transformation and S is a Bogoliubov generator written as a sum over momentum transfers ℓ of kernels K(ℓ) acting on approximate pair creation operators b_p^*(ℓ) = a_p^* a_{p-ℓ}^*, with p in the 'lens' L_ℓ = B_F^c ∩ (B_F + ℓ). These pair operators obey approximate canonical commutation relations, so the Hamiltonian can be treated as a quasi-bosonic Bogoliubov Hamiltonian. The RPA momentum distribution emerges from a Duhamel expansion of e^S a_q^* a_q e^{-S}: the leading term is 1/2 Σ_ℓ 1_{L_ℓ}(q) (cosh(2K(ℓ))-1)_{q,q}, which is evaluated in closed form using the Sherman-Morrison formula; all remaining terms are controlled throu","core_discovery":"The central claim is Theorem 1.1: for a radial, decreasing, nonnegative potential with Fourier transform bounded by C|ℓ|^{-2} — which includes the Coulomb potential — there exists a sequence of trial states Ψ_N with energy within k_F^{1-1/6+ε} of the ground state, and such that for every momentum q, n(q)=n_RPA(q)+E(q) for |q|≥k_F and n(q)=1−n_RPA(q)+E(q) for |q|<k_F, with |E(q)| ≤ C_ε k_F^{-1-1/6+ε} e(q)^{-1}. If the Fourier transform is summable, the error improves to C_ε k_F^{-2+ε} e(q)^{-1}. Here n_RPA(q) is the explicit random phase approximation integral (1.9), and e(q) is the excitation energy measuring the distance of q to the Fermi surface. The paper also extracts a smaller exchange","pith_inferences":["If a matching lower-bound energy estimate for the Coulomb gas becomes available, the same bootstrap proof would likely transfer the RPA momentum-distribution formula to the actual ground state; the only external input needed is the energy bound.","The bootstrap quantity Ξ and the e(q)^{-1} structure suggest that the shape of the momentum distribution near the Fermi surface is universal, determined by the pair-excitation spectrum rather than by the details of the interaction, so similar profiles should appear for other singular potentials with summable square.","The method's reliance on the summability lemma Σ_{r∈S} e(r)^{-1} ≤ C k_F^{1+ε} indicates the proof would break, or need modification, for potentials with slower decay than |ℓ|^{-2-α}, offering a concrete test of where the RPA description fails.","The explicit form of n_RPA(q) permits numerical checks of the bootstrap prediction for moderate k_F, and the sharp error exponents provide a benchmark for future many-body simulations of the interacting Fermi gas."],"forward_implications":["For Coulomb potentials, the Daniel-Vosko RPA prediction for the momentum distribution is proven rigorously for a state energetically within k_F^{1-1/6+ε} of the true ground state.","The error bounds are pointwise in q and hold even when q is within a distance ~k_F^{-1} of the Fermi surface, where the excitation energy e(q) is of order one.","For potentials with summable Fourier transform, the error is k_F^{-2+ε} e(q)^{-1}, which is optimal in view of the leading RPA term scaling as k_F^{-1} e(q)^{-1}.","The same trial state reproduces the Gell-Mann-Brueckner correlation energy upper bound, so the energy and momentum distribution are derived from one consistent state.","The authors argue the same RPA momentum distribution should be expected for the true ground state, since two independent bosonization constructions now give consistent formulas."],"fun_headline_variants":["Coulomb Fermi gas momentum distribution matches RPA","Fermi gas momentum: RPA match proven for Coulomb","Coulomb Fermi gas: RPA momentum distribution proven","Rigorous RPA error bounds for Fermi gas momentum","Fermi gas with Coulomb: momentum near RPA proven"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The theorem relies on previously proven bounds for the ground-state energy of the Coulomb gas, E_gs = E_FS + E_corr + O(k_F^{1-1/6+ε}) and ⟨Ψ_N,H_NΨ_N⟩ ≤ E_FS + E_corr + O(k_F^{1/2}); if either imported bound fails, the energy-closeness part of the theorem collapses.","fun_headline_variants_meta":{"raw":{"variants":["Coulomb Fermi gas momentum distribution matches RPA","Fermi gas momentum: RPA match proven for Coulomb","Coulomb Fermi gas: RPA momentum distribution proven","Rigorous RPA error bounds for Fermi gas momentum","Fermi gas with Coulomb: momentum near RPA proven"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0007,"raw_usage":{"total_tokens":2992,"prompt_tokens":731,"completion_tokens":2261,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":475,"completion_tokens_details":{"reasoning_tokens":2181}},"tokens_in":475,"tokens_out":2261,"duration_ms":15185,"temperature":1.0,"reasoning_tokens":2181,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T23:51:28.969445+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the momentum distribution of the explicit trial state (2.8)-(2.14) for a Coulomb potential on the torus at momenta q with e(q)=O(1) and compare it with n_RPA(q). The theorem predicts |n(q)-n_RPA(q)| ≤ C_ε k_F^{-1-1/6+ε}; observing a decay slower than k_F^{-1-1/6} for any fixed distance to the Fermi surface would disprove it. Alternatively, a counterexample to the imported energy lower bound E_gs ≥ E_FS + E_corr - C k_F^{1-1/6+ε} for Coulomb would falsify Theorem 1.1 even if the momentum formula itself remains true.","supporting_citations":[],"review_version":1}