{"id":"571ae5dd-981e-4b5e-9e0b-e181304a50e7","arxiv_id":"1908.05478","paper_version":6,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Near each nucleus of a heavy atom or molecule, the true ground-state electron density is close, with explicit L^p error bounds, to the density of a single non-interacting hydrogenic atom, in both nonrelativistic and relativistic settings.","lead":"This paper proves a sharp asymptotic statement about electron density in heavy atoms and molecules: near a nucleus, the true many-electron ground-state density is close to the density of an isolated atom with no electron-electron repulsion, with explicit L^p error bounds. The result extends the 1996 strong Scott conjecture to larger distances from the nucleus and to relativistic electrons.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.8 is the least secure load-bearing step: for a ≥ Z^{-2/3-δ/2}, the final estimate (1.12) relies on a two-perturbation trace bound whose n ≥ N tail is asserted, not proved.","rationale":"The reader's weakest assumption was the local Holder-type replacement of the Thomas-Fermi potential by the bare Coulomb potential around (4.10). I agree that passage is delicate, but the bound |W'_m(x)-W'_m(y_m)| ≤ C Z^{3/2}|x-y_m|^{1/2} is plausible from TF regularity (|ΔW'| ≤ C Z^{3/2}|x-y_m|^{-3/2}) and is not the point where I would first expect a breakdown. The reader also listed Propositions 2.3, 3.7, and 3.8 as needing scrutiny; my concern is specifically Proposition 3.8, whose proof delegates the decisive n ≥ N tail to the single-perturbation estimate (2.11). If that tail estimate needs an extra smallness condition or a different power of (ε+ς), the theorem's full range a ≤ Z^{-1/2-κ} is not established, although the core a ≍ Z^{-1} regime may survive. The numerical cross-check above would settle whether the estimate is true; until then, the manuscript is best treated as CONDITIONAL, which is exactly the reader's verdict, so I recommend no change.","tokens_in":24756,"tokens_out":32004,"duration_ms":310126,"concrete_test":"Compute the cross-term trace difference T = Tr[(H_{V0+εΦ})_-] - Tr[(H_{V0})_-] - Tr[(H_{V0+εΦ+ςU})_-] + Tr[(H_{V0+ςU})_-] in the spherically symmetric one-particle model with V0 = |x|^{-1}, Φ = U = 1_{B(0,r)}, at r = Z^{1/2-κ}, ε = Z^{-1/2} a^{1/2}, a = r/Z, and ς chosen to minimize the right side of (4.14), for Z = 10^2, 10^3, 10^4. If T / ((ε+ς)^{1-κ} r (ς r^{3/2} + ε r^{5/2})) is not bounded as Z grows, Proposition 3.8 fails; if it is bounded, the concern reduces to a missing but repairable detail in the proof.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.3 is proved by reducing the many-body density integral to one-particle trace differences. For a ≤ Z^{-2/3-δ/2}, Proposition 3.6 suffices, but the advertised range up to a ≤ Z^{-1/2-κ} is obtained only through Proposition 3.8: after (4.10)-(4.13), the proof replaces the remaining potential error by εΦ and uses (3.31) to justify the final F, G in (1.12)-(1.14). The proof of Proposition 3.8 is a sketch at exactly the tail that matters. For n ≤ N it invokes Proposition 3.7, but for n ≥ N it asserts estimates (3.33)-(3.34), saying they follow from estimate (2.11). This is not immediate: (2.11) is a single-perturbation spectral-shift bound for the operator H_{V0}+ςU, proved using the explicit hydrogenic eigenfunctions and their decay (2.18). The operators in (3.33)-(3.34) contain both εΦ and ςU, and the paper does not display a two-perturbation version of Proposition 2.2 nor check the eigenvalue numbering/cluster separation for H_{V0+εΦ+ςU} in the tail. In the application r = Za can be as large as Z^{1/2-κ}, and (ε+ς)r^{3/2} is small only by the margin Z^{-κ}; an unverified power of (ε+ς) or a failed cluster-separation condition at the lower end of the tail would change the exponents in F and G. This is a load-bearing proof gap, not a cosmetic omission.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a strong form of the Scott conjecture for heavy atoms and molecules: under the nuclear-separation assumption min_{m<m'} |y_m-y_{m'}| >= Z^{-1/3+sigma} with sigma >= 0, and for test functions U supported in B(y_m,a) with a <= Z^{-1/2-kappa} and |U| <= 1, the ground-state density rho_Psi satisfies the weak-L^1 estimate (1.12) with the explicit error terms F and G in (1.13)-(1.14), where the comparison density rho_{m,beta} is the spectral density of a single non-interacting hydrogenic operator with charge Z_m. The result is claimed in both the non-relativistic and the relativistic case. The proof reduces the many-body density integral to one-particle spectral trace differences via Thomas-Fermi energy bounds, then analyzes the one-particle operator H_{V0} perturbed by ςU and epsilon Phi using cluster decomposition of the hydrogenic spectrum and eigenfunction decay estimates developed in Appendices A and B.","tokens_in":25047,"tokens_out":11583,"duration_ms":112998,"significance":"If the proof is completed as indicated, this is a substantial result: it gives a parameter-free comparison density (no fitting parameters, and the comparison object is defined solely by the single-particle hydrogenic spectral projector), explicit L^p error rates, uniformity in the nuclear charge Z, and coverage of the relativistic case. The paper also gives credit where due to the earlier work of Iantchenko-Lieb-Siedentop and to the author's own Thomas-Fermi estimates. The main caveat is that the proof is a long chain of spectral and functional-analytic estimates, several of which are delegated to the reader at precisely the points that control the advertised range of a; the three load-bearing issues below should be addressed before the theorem can be regarded as fully proven.","major_comments":[{"comment":"Proposition 3.8 is the only ingredient that extends Theorem 1.3 from a <= Z^{-2/3-delta/2} to the advertised range a <= Z^{-1/2-kappa}: after (4.13) the proof invokes (3.31) to obtain (4.14). In the proof of Proposition 3.8, the tail estimates (3.33) and (3.34) are introduced with the remark 'which follow from estimate (2.11)', but (2.11) is stated for a single perturbation ςU of H_{V0} and does not, as stated, cover the operator H_{V0+εΦ+ςU} whose eigenvalues define lambda_{n,k}(ε,ς). The missing step is either a two-perturbation version of Proposition 2.2, including cluster separation and eigenvalue numbering for H_{V0+εΦ+ςU}, or an explicit argument applying (2.11) to the normalized combined perturbation P = εΦ + ςU and then using the triangle inequality. This is not merely a cosmetic omission: the final powers in F and G at (1.13)-(1.14) depend on the n^{-3} tail rate and on the precise way (ε+ς) enters the estimate, and the paper explicitly notes in Remark 1.5(ii) that without Proposition 3.8 only the weaker bounds (1.19)-(1.20) are obtained.","section":"§3, Proposition 3.8"},{"comment":"The estimate |W'_m(x) - W'_m(y_m)| <= C Z^{3/2} |x - y_m|^{1/2} is asserted with only a brief footnote and no proof or reference. This estimate is load-bearing: it is exactly what allows the replacement of the Thomas-Fermi potential W by the bare Coulomb potential V_m^0 plus a constant, and it converts the error term in (4.7) into the C Z r^2 term in (4.11). If this Hölder-type control fails, or holds with a different exponent or with a constant not uniform in Z and in the nuclear separation, then the passage from (4.7) to (4.11), and hence the minimization leading to (1.12)-(1.14), does not follow. The manuscript should either provide a derivation of (4.10) from the Thomas-Fermi equation or cite a precise statement that covers this uniformity.","section":"§4, inequality (4.10)"},{"comment":"The trace bound (3.29) is central to the finite-n part of Proposition 3.8: it controls the sum over n <= N after the decomposition of the trace difference. The proof, however, is delegated with the sentence 'Proof repeats the proof of Proposition 3.3. We leave easy details to the reader.' The two-perturbation setting is not literally the same as Proposition 3.3: the projectors pi_n(ε,ς) are those of H_{V0+εΦ+ςU}, and one needs the analogue of the contour-expansion argument with two independent perturbation parameters ε and ς, plus the trace estimates (2.23) applied to both U and Phi. Since Proposition 3.7 feeds directly into the proof of Proposition 3.8 and hence into Theorem 1.3, the details should be written out or reduced to Proposition 3.3 in a way that is verifiable without reconstructing the argument.","section":"§3, Proposition 3.7"}],"minor_comments":[{"comment":"The phrase 'the mode l with no interactions' should read 'the model with no interactions'.","section":"Abstract"},{"comment":"The sentence 'combining (4.2), (4.2)' should reference (4.2) and (4.3).","section":"§4, proof of Proposition 4.1"},{"comment":"The displayed formula (3.32) appears to contain a stray 'U' before the integral, and the sign convention in the trace identity should be checked; the text says 'without traces or absolute value' but the expression as printed is not clearly an identity.","section":"§3, equation (3.32)"},{"comment":"There are several typos: 'snd' for 'and', 'resztriction' for 'restriction', and 'definitely is not rigt' for 'definitely is not right'.","section":"Remark 1.5"},{"comment":"The notation ⟨x⟩ is used in (1.12)-(1.14) but is defined only after (1.16); the definition should be moved before its first use.","section":"Notation, after (1.16)"},{"comment":"The proof of Proposition 2.3 is also delegated ('We leave easy details to the reader'); since the proposition is used in the small-n regime, a few lines indicating how (2.30) enters the trace estimates would improve verifiability.","section":"§2, Proposition 2.3"}],"recommendation":"major_revision","confidential_remarks":"The central claim is plausible and the comparison density is genuinely parameter-free, but the proof of Theorem 1.3 relies on several estimates that are either asserted or delegated at load-bearing points. I do not see a reason to doubt the result, but the manuscript as submitted does not yet contain a fully checkable proof of the advertised range a <= Z^{-1/2-kappa}; the authors should supply the missing arguments for Proposition 3.8, (4.10), and Proposition 3.7. If those are provided, I expect the paper to be acceptable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Vitya asked me to look at the Ivrii paper. The short version: this is a genuine strengthening of the strong Scott result, not a repackaging. It moves the approximation distance from ~Z^{-1} out to Z^{-1/2-kappa}, adds Lp error terms, and covers the relativistic operator. The comparison density comes from a single-particle spectral projector, so there is no circularity. The proof strategy is functional-analytic rather than microlocal, and the local Holder control of the Thomas-Fermi potential in (4.10) is a nice touch.\n\nThe soft spot is exactly where the stress-test note puts it. Proposition 3.8 is the step that buys the advertised range when a is larger than Z^{-2/3-delta/2}. Its proof estimates the tail n>=N via (3.33)-(3.34), and claims they follow from (2.11). I do not see how that follows directly. Estimate (2.11) is proved for one perturbation around the pure Coulomb operator, using explicit hydrogenic eigenfunctions and their decay. The operators in (3.33)-(3.34) have two perturbations, epsilon*Phi and sigma*U, around H_{V0}, and the text does not display a two-perturbation analogue of Proposition 2.2 nor check the cluster separation and eigenvalue numbering for H_{V0+epsilon*Phi+sigma*U} in the tail. At the application end r=Za can be of order Z^{1/2-kappa}, and the smallness margin of (epsilon+sigma)r^{3/2} is only Z^{-kappa}. So a lost power of (epsilon+sigma) or a failed gap condition at the lower end of the tail would alter the exponents in F and G. This is not a cosmetic omission.\n\nSmaller concerns: Propositions 2.3 and 3.7 are delegated with 'easy details to the reader'. They look plausible, but the referee should ask for the details. The paper leans on the author's own Thomas-Fermi energy bounds rather than re-deriving them; that is acceptable for a specialist, but the reader should know the lemma is in a 3000-page book.\n\nIf the Prop 3.8 gap is filled, the theorem is a solid advance. As it stands, the result is conditional, not established. Do not desk-reject. Send it to a referee who knows spectral asymptotics for Coulomb Hamiltonians and ask specifically for a full proof of (3.33)-(3.34) or a counterexample. This deserves referee time.","headline":"Real extension of the strong Scott conjecture with a mostly sound functional-analytic proof, but the advertised distance range leans on an asserted two-perturbation trace estimate that needs to be checked before the theorem is trusted.","tokens_in":25636,"tokens_out":2807,"would_cite":true,"duration_ms":26165,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P20","81V70"],"pacs":[],"model":"deepseek-v4-flash","headline":"Near a heavy nucleus, the many-electron ground-state density is approximated in $L^p$ by a non-interacting hydrogenic density, with explicit error bounds, in both nonrelativistic and relativistic settings.","keywords":["electronic density","Scott conjecture","relativistic Schrödinger operator","heavy atoms","molecules","Thomas-Fermi potential","L^p approximation","ground state density"],"falsifier":"Compute the Thomas-Fermi potential $W$ for a diatomic molecule with nuclear separation $d=Z^{-1/3}$ and numerically check the bound $|W'_m(x)-W'_m(y_m)| \\le C Z^{3/2}|x-y_m|^{1/2}$ on $B(y_m,Z^{-1/2})$; a violation would break the error control in (4.11). Alternatively, a direct numerical ground-state calculation of a two-electron ion at moderately large $Z$, testing inequality (1.12) with $a=Z^{-1/2-\\kappa}$, could settle whether the stated bound holds.","tokens_in":24489,"feed_emoji":"⚛️","tokens_out":13476,"duration_ms":113859,"temperature":0.7,"pith_summary":"The paper proves a quantitative version of the strong Scott conjecture: in a heavy atom or molecule, on distances $a$ up to $Z^{-1/2-\\kappa}$ from a nucleus of charge $Z_m$, the true many-electron ground-state density $\\rho_\\Psi$ is close in $L^p$ to the density of a single isolated hydrogenic atom with no electron-electron interactions. The closeness is controlled by inequality (1.12), with error terms $F$ and $G$ that depend on $Z$, $a$, and the separation between nuclei, and the estimate is uniform for molecules whose nuclei are separated by at least $Z^{-1/3+\\sigma}$. The theorem covers both nonrelativistic kinematics and relativistic kinematics, subject to the stability condition $Z_m\\beta \\le 2/\\pi$. The claim matters because it means the complicated many-body structure of the atomic core can be computed from a one-particle Coulomb model with controlled error.","feed_headline":"Core density matches a single non-interacting atom","feed_subtitle":"Heavy-atom ground states reduce locally to hydrogenic density, even relativistically.","key_machinery":"The load-bearing identity is the trace-energy comparison (4.1)--(4.6): for a test perturbation $U$ supported near $y_m$, $\\varsigma\\int U\\rho_\\Psi dx$ is bounded by $\\operatorname{Tr}[H^-_{W+\\nu}]-\\operatorname{Tr}[H^-_{W+\\varsigma U+\\nu}]$ plus a Thomas-Fermi error $Z^{5/3-\\delta}$. The argument then replaces the self-consistent Thomas-Fermi potential $W$ by the bare one-nucleus Coulomb potential $V_m^0=Z_m|x-y_m|^{-1}$ up to a constant chemical potential, an operation controlled by (4.10), the bound $|W'_m(x)-W'_m(y_m)| \\le C Z^{3/2}|x-y_m|^{1/2}$. After rescaling $x\\mapsto Z_m(x-y_m)$, the comparison is made on the hydrogenic operator $T_\\beta-|x|^{-1}$, whose negative spectrum splits into clusters separated by gaps of order $n^{-3}$; Sections 2 and 3 estimate how eigenvalues and spectral projectors shift under the perturbation $U$, and these estimates are summed over principal quantum number $n$ to produce the final error terms $F$ and $G$. The reference object $\\rho_{m,\\beta}$ is the diagonal of the spectral projector $\\theta(\\tau-H_{\\beta,V^0})$ of this toy hydrogenic Hamiltonian.","core_discovery":"The central statement is Theorem 1.3: under the separation assumption $\\min_{m\\neq m'}|y_m-y_{m'}| \\ge Z^{-1/3+\\sigma}$, for any test function $U$ supported in $B(y_m,a)$ with $a \\le Z^{-1/2-\\kappa}$ and $|U|\\le 1$, $$|\\int U(\\rho_\\Psi-\\rho_{m,\\$\\beta$})dx| \\le C $F^{{1/2}}$\\big((Za)^{3/2}\\|\\langle Z(x-y_m)\\$rangle^{{-3/2}}$U\\|_{$L^{1}$}\\big)^{1/2}+CG,$$ where $F$ and $G$ are given by (1.13)--(1.14). Here $\\rho_{m,\\beta}(x)=qZ_m^3\\bar\\rho_{Z_m\\beta}(Z_m(x-y_m))$ is the diagonal of the spectral projector of the hydrogenic operator $T_\\beta-|x|^{-1}$, with $T_\\beta$ the relativistic or nonrelativistic kinetic energy; that is, the reference density is that of a single atom with the same nuclear charge and no electron-electron repulsion. The proof reaches this by comparing the true density with the one-particle Coulomb density through trace-energy inequalities, controlling the replacement of the self-consistent Thomas-Fermi potential by the bare Coulomb potential via the H\\\"older-type bound (4.10). The approximation holds in $L^1$ on balls or shells around a nucleus and in $L^p$ for $p=2,3,\\dots$ whenever a pointwise upper bound on $\\rho_\\Psi$ is known.","pith_inferences":["Editorial inference: if the H\\\"older bound (4.10) is not sharp, the same trace comparison might hold for radii beyond $Z^{-1/2-\\kappa}$; Remark 1.5(ii) already shows a weaker estimate is available without Proposition 3.8, so a numerical check of (4.10) on Thomas-Fermi potentials would reveal whether the main restriction is an artefact.","Editorial inference: the same machinery should control the full one-particle density matrix, not only its diagonal $\\rho_\\Psi$, because Sections 2 and 3 estimate spectral projectors; inserting rank-one kernels would yield spatially resolved off-diagonal decay estimates.","Editorial inference: the paper leaves open whether the relativistic core density $\\rho_\\beta$ differs from the nonrelativistic density $\\rho_0$; a direct numerical evaluation of the series (1.21) for $\\beta$ near the stability bound $2/\\pi$ would test whether relativistic effects are detectable in the core and could support or refute the conjectured lower bound.","Editorial inference: at the matching radius $a\\approx Z^{-1/2-\\kappa}$, the hydrogenic approximation of this paper and the Thomas-Fermi approximation of the larger-distance theory should agree, so combining the two would produce a piecewise global density approximation with a controlled transition layer."],"forward_implications":["Inside any ball of radius $a \\le Z^{-1/2-\\kappa}$ around a nucleus, the many-electron density can be replaced by the hydrogenic density with an explicit $L^p$ error that grows as $a$ increases and degrades sharply beyond $Z^{-1/3}$, marking the end of the core regime.","The earlier $O(Z^{-1})$ core result is extended to distances up to $Z^{-1/2-\\kappa}$, and the same statement is proved with relativistic kinetic energy.","Corollary 1.4 converts the weighted estimate into practical $L^1$ and $L^p$ bounds on balls or annuli around a nucleus, using only an upper bound on the true density.","The error terms $F$ and $G$ expose two regimes: for very small $a$ the leading errors are $Z^{13/6-\\delta}a^{-1/2}$ and $Z^{7/6-\\delta}a^{3/2}$; for $a$ closer to $Z^{-1/2}$ they become $Z^4a^3$ and $Z^2a^3$.","For distances $a\\gg Z^{-1}$, the paper records that the Thomas-Fermi density rather than the hydrogenic density is the appropriate approximation, so the present theorem supplies the inner-core half of a two-scale description of the electronic density."],"supporting_citations":[{"why":"Supplies the trace-comparison argument behind Proposition 4.1: the weighted density integral is bounded by a difference of one-particle energy traces.","marker":"[IaLS]"},{"why":"Provides the Thomas-Fermi upper and lower energy bounds and the estimates used to replace the Thomas-Fermi potential by the single-nucleus Coulomb potential.","marker":"[Ivr1]"},{"why":"Gives the ground-state energy asymptotics in the relativistic setting, the basis for carrying the argument through the relativistic case.","marker":"[Ivr2]"},{"why":"Supplies the pointwise upper bounds on the true density used in Corollary 1.4(ii) to convert $L^1$ estimates into $L^p$ estimates.","marker":"[Ivr3]"},{"why":"Establishes the semiboundedness condition $Z_m\\beta \\le 2/\\pi$ for the one-particle relativistic operator, assumed throughout the relativistic statement.","marker":"[IH]"},{"why":"Provides stability of relativistic matter, which underpins the existence and energy bounds for the many-electron ground state in the relativistic case.","marker":"[LY]"}],"fun_headline_variants":["Heavy atoms locally mimic single non-interacting atoms","Relativistic heavy-atom density reduces to hydrogenic locally","Local density equals hydrogenic for heavy atoms","Heavy-atom core density: hydrogenic, even relativistic","Strong Scott: local density is single-atom density"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof's load-bearing premise is the local bound (4.10): inside a ball of radius $a$ around the chosen nucleus, the difference between the self-consistent Thomas-Fermi potential and the bare Coulomb potential of that nucleus changes by at most $C Z^{3/2}|x-y_m|^{1/2}$; if this H\\\"older-type control fails on the scale $a$, the energy-trace comparisons (4.7) and (4.11) carry uncontrolled errors and the hydrogenic approximation does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Heavy atoms locally mimic single non-interacting atoms","Relativistic heavy-atom density reduces to hydrogenic locally","Local density equals hydrogenic for heavy atoms","Heavy-atom core density: hydrogenic, even relativistic","Strong Scott: local density is single-atom density"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000197,"raw_usage":{"total_tokens":1354,"prompt_tokens":921,"completion_tokens":433,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":537,"completion_tokens_details":{"reasoning_tokens":355}},"tokens_in":537,"tokens_out":433,"duration_ms":4035,"temperature":1.0,"reasoning_tokens":355,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:12:31.699979+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Thomas-Fermi potential $W$ for a diatomic molecule with nuclear separation $d=Z^{-1/3}$ and numerically check the bound $|W'_m(x)-W'_m(y_m)| \\le C Z^{3/2}|x-y_m|^{1/2}$ on $B(y_m,Z^{-1/2})$; a violation would break the error control in (4.11). Alternatively, a direct numerical ground-state calculation of a two-electron ion at moderately large $Z$, testing inequality (1.12) with $a=Z^{-1/2-\\kappa}$, could settle whether the stated bound holds.","supporting_citations":[],"review_version":1}