{"id":"24fdad18-5016-49ba-b031-67ce97feb913","arxiv_id":"2504.18487","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"For a nucleus of charge Z, a fermionic atom binds fewer than 1.1185 Z + O(Z^{1/3}) electrons, improving the previous best constant 1.22.","lead":"Mathematicians proved a tighter upper limit on how many extra electrons a large atom can bind: fewer than 1.1185 Z plus a lower-order term, improving bounds from 1984 and 2012. The result shows quantum statistics strongly separate atoms from bosonic matter, which can bind about 21 percent excess charge.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"At p=3 the leading constant 1.1185 rests on the exact endpoint of the improved Hardy inequality [EF06, Lemma 2.4]; that lemma is neither stated nor verified, so this external step is the most load-bearing unverified point.","rationale":"The reader's verdict is already CONDITIONAL, and its weakest_assumption points to the same step I find most load-bearing. The paper's main chain is otherwise well-structured: the reduction in Section 3 to alpha_{N,p}, the mean-field comparison in Section 5, and the weighted kinetic-energy bounds in Section 6 are coherent, and the final optimization for p=3 is plausible. The genuine soft spot is not an internal inconsistency but an external endpoint: Lemma 4.3 uses an improved Hardy inequality with constant d^2/4 for functions orthogonal to radial functions, and at p=3 the coefficient d_H - p^2/4 vanishes, so the whole radial-symmetrization argument at the optimal p depends on that constant being exactly 9/4. Since the lemma is only cited, not stated, the proof as written does not allow the reader to verify the decisive inequality. The auxiliary numerical bound in Lemma A.5 is also not reproducible, since it states a computer summation to N=1000 with an error estimate but provides no code or output; this affects the explicit lower-order constants in Proposition 2.5 and should be made reproducible. However, that issue is less structural: the leading coefficient F(3) would survive even if the lower-order constants had to be recomputed. In my read, the central argument is likely correct, but both the external Hardy lemma at its endpoint and the computer-assisted bound need independent verification before the claim can be accepted unconditionally. Hence the reader's CONDITIONAL verdict remains unchanged.","tokens_in":89882,"tokens_out":20631,"duration_ms":204826,"concrete_test":"Retrieve [EF06, Lemma 2.4] and check its exact hypotheses and constant. Then verify the p=3 case of (4.15) from that lemma: for every nu in H^{-1} orthogonal to radial functions, the regularized potential V_{nu,lambda} = (-Delta + lambda^2)^{-1} nu must satisfy the improved Hardy inequality with constant 9/4 in the form integral |grad(phi^{1/2} V_{nu,lambda})|^2 >= (9/4) integral phi |V_{nu,lambda}|^2 / |x|^2, where phi is the regularized weight. A direct way to settle the endpoint is to analyze the ell=1 angular channel for functions orthogonal to radial functions in d=3: the sharp constant is ((2 ell + d - 2)/2)^2 = 9/4 at ell=1. If this reproduces 9/4 and the orthogonality of V_{nu,lambda} passes, the p=3 step is sound; if the constant is only 1/4, Lemma 4.3 fails exactly at p=3 and the proof must fall back to p<3.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Proposition 2.5's coefficient F(3) < 1.1185 depends on the lower bound beta_3 >= F(3)^{-1} from Proposition 4.5, which in turn depends on Theorem 4.2: beta_3 = beta_3^rad. Theorem 4.2 is exactly Lemma 4.3. In the proof of Lemma 4.3, after IMS localization and regularization, the key estimate is (4.15): (1/4pi) C_p^{lambda,epsilon}(nu,nu) >= (d_H - p^2/4) <...>, with d_H = d^2/4 the improved Hardy constant for functions orthogonal to radial functions, cited from [EF06, Lemma 2.4]. At d=3 and p=3, d_H - p^2/4 = 9/4 - 9/4 = 0, so the nonnegativity of the regularized quartic form, and hence C_3(nu) >= 0, rests entirely on the constant in that external lemma being exactly d^2/4 rather than the usual Hardy constant (d-2)^2/4 = 1/4. The manuscript does not quote the lemma or its hypotheses, so the H^1 functions V_{nu,lambda} used here cannot be checked against it from the text. If the cited constant were smaller, or if the orthogonality condition on V_{nu,lambda} were not satisfied, Lemma 4.3 would fail at p=3 and the leading coefficient would not be established by this argument. This is the single most load-bearing place in the chain: unlike the auxiliary computer constant in Lemma A.5, which affects explicit lower-order constants but not the structural bound, this step sets F(3). I do not claim the lemma is false; I claim the proof's leading term is exactly as secure as that unstated external endpoint.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the maximal number N_c(Z) of electrons that a nucleus of charge Z can bind, and proves new explicit upper bounds. Theorem 2.2 states that for every p in (1,3] there is C(p)>0 such that N_c(Z) < F(p)Z + C(p)Z^{1/3}, with F(p) = max_{0<=t<=1}(1+t^{p-1})/(1+t^p). The two concrete cases are p=2, giving N_c(Z) < 1.2072Z + 2.96Z^{1/3} for Z>=2, and p=3, giving the stronger bound N_c(Z) < 1.1185Z + 3.90Z^{1/3} + 0.0134 + 0.184Z^{-1/3} + 0.0196Z^{-2/3} for Z>=4. The proof adapts the Benguria-Lieb-Nam weighted-multiplier method to a general weight |x|^p; the main ingredients are a radial-symmetrization theorem for the mean-field functional beta_p (Theorem 4.2 and Lemma 4.3), comparison lemmas between the many-particle constant alpha_{N,p} and beta_p (Lemmas 5.3, 5.5, 5.9), and an upper bound for the weighted kinetic term (Lemma 6.4). The paper claims the first finite-Z separation between fermionic and bosonic atoms, since the fermionic bound has leading coefficient below the bosonic value t_c approximately 1.21.","tokens_in":90287,"tokens_out":14106,"duration_ms":124137,"significance":"If the proof can be completed at the points flagged below, this is a significant result: it improves the best rigorous upper bound on excess charge for fermionic atoms from 1.22Z + 3Z^{1/3} to 1.1185Z + O(Z^{1/3}), proves Nam's radial conjecture for the range 2 <= p <= 3, and gives a quantitative separation from bosonic atoms at finite nuclear charge. The main chain is modular, uses standard inequalities and classical theorems, and contains no fitted target data; the final constants are derived rather than adjusted. The paper also contains useful methodological contributions, including the multipole-expansion estimates in Section 5 and the refined comparison lemmas. However, the significance is conditional on two verifiability issues: the unstated endpoint Hardy lemma used at p=3, and a computer-assisted tail estimate in Appendix A that is not reproducible from the manuscript.","major_comments":[{"comment":"At p=3 the proof of Lemma 4.3 (and hence Theorem 4.2 and Proposition 4.5) relies on the endpoint of the improved Hardy inequality with constant d^2/4 for functions orthogonal to radial functions, cited as [EF06, Lemma 2.4]. Since d_H - p^2/4 = 9/4 - 9/4 = 0 at p=3, the nonnegativity of the regularized quartic form C_3(nu), and therefore the lower bound beta_3 >= F(3)^{-1}, depends on the exact value of that constant. The lemma is not stated in the manuscript, and its hypotheses are not checked for the potentials Phi_{epsilon,lambda} psi_{nu,lambda} used in the IMS localization step. Please state the lemma explicitly and verify the orthogonality and H^1 membership conditions in the present regularization; without this, the leading coefficient 1.1185 in Proposition 2.5 is not established by the written proof.","section":"Section 4, Lemma 4.3 and Eq. (4.15)"},{"comment":"The tail bound in Lemma A.5 is computer-assisted but not reproducible: the proof selects N=1000, states 'computing those terms on a computer explicitly', and reports the aggregate values 0.0242, 0.0203, and 0.0003 without code, pseudocode, or a table of intermediate sums. This estimate is used in the proof of Lemma 5.9 and contributes to the constants E_3 and E_4 in Proposition 2.5, so it is part of the claimed explicit bounds. To make the proof verifiable, include the computer code or a rigorous interval-arithmetic version with a table of the tail computation.","section":"Appendix A, Lemma A.5"},{"comment":"The claim that the supremum in Eq. (7.33) is attained at r = beta_3^{-1} is not correct as stated. For g(r) = 3(3/10)^{1/3} beta_3^{-2/3} r^{1/3} + lambda beta_3^{-1} r^{-2/3} on the interval [beta_3^{-1}, 5/2], the derivative vanishes at approximately r = 1.27, which lies inside the interval when the numerical values in the paper are used; the maximum is then at an endpoint, and comparison of the endpoints shows that r = 5/2 gives the larger value. The final bound E_1 < 3.893 still appears to hold, but the stated reason does not; please replace this by a correct monotonicity or endpoint argument, or by an explicit numerical bound on the full interval.","section":"Section 7.3, after Eq. (7.33)"}],"minor_comments":[{"comment":"The sentence 'In particular Proposition 2.4 shows N < 1.12Z + 4Z^{1/3}, Z >= 4' appears to refer to Proposition 2.5, since Proposition 2.4 has leading coefficient 1.2072; please correct the cross-reference.","section":"Section 2, discussion after Proposition 2.4"},{"comment":"The caption contains the typo 'fat' for 'fact', and it would be helpful to state explicitly that the apparent non-monotonicity of alpha_{N,1} in the plot is a numerical artifact and not in contradiction with Lemma 3.2.","section":"Figure 1 caption"},{"comment":"The phrase 'quantum mechanic virial theorem' should read 'quantum mechanical virial theorem'.","section":"Section 6.3"},{"comment":"The statement that the bounds show the fundamental difference between fermionic and bosonic atoms 'for finite Z' is imprecise: the separation holds for sufficiently large finite nuclear charge under the stated explicit bounds, not for every finite Z. Please qualify this wording.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The central strategy is sound and the final numerical constants are plausible, but the two verifiability issues (the unstated EF06 endpoint lemma and the non-reproducible computer-assisted tail bound) should be resolved before publication. The incorrect claim about the supremum in Section 7.3 is a proof gap that is likely fixable without changing the final constants."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short take: this is a real step forward in a problem that has been stuck since Nam's 2012 paper. The leading constant for fermionic excess charge drops from 1.22 to 1.1185, and it is the first bound that strictly separates fermionic from bosonic atoms at finite Z, since the bosonic ratio is about 1.21. The authors also prove Nam's conjecture that the mean-field functional β_p is minimized by radial measures, for p ≤ 3, and the new multipole/Legendre machinery in the comparison lemmas is substantial. The proof reads as coherent: equation (3.10) starts the chain, Lemmas 5.9, 6.4 and Section 7 produce explicit constants with no fitted target data. The radial-symmetry theorem is the real novelty, and the p-family of weights is a natural framework that should be useful.\n\nThere are two soft spots, but neither sinks the paper. The first is Lemma A.5: a tail bound on Legendre moments is justified by \"computing those terms on a computer explicitly\" with no code, no reproducible table, and no exact interval. This is an auxiliary estimate that affects the explicit lower-order constants in Proposition 2.5, not the structural bound, so it is a reproducibility request rather than a correctness objection. The second is the endpoint p = 3 in Lemma 4.3. At p = 3, the nonnegativity of the regularized quartic form rests on the improved Hardy constant d^2/4 for functions orthogonal to radial functions, cited from [EF06, Lemma 2.4]. The stress-test note is right that the margin d_H - p^2/4 is exactly zero at d = 3, p = 3. But I checked: the manuscript does verify the orthogonality condition on the potential, and the cited constant is a standard result. So I am less worried than the stress-test note. Still, because the leading coefficient F(3) sits directly on that equality, the authors should state the lemma and its hypotheses explicitly instead of just referencing it.\n\nThe citation pattern looks appropriate: Lieb, Nam, Benguria-Lieb, Solovej, LSST, and the EF06 Hardy lemma are all standard and used correctly. I found no circularity: the target bound N_c is never used to set constants.\n\nWho is this for? Anyone working on the ionization conjecture, quantitative bounds in many-body quantum mechanics, or symmetry of mean-field functionals. It deserves a serious referee. I would send it to peer review, with a request for a reproducible appendix for Lemma A.5 and an explicit statement of the Hardy lemma.","headline":"Genuine improvement of the Lieb/Nam excess-charge bound with a clean, mostly self-contained argument; two loose ends—an unquoted Hardy endpoint at p=3 and a non-reproducible computer bound in the appendix—should be fixed, but neither looks fatal.","tokens_in":90837,"tokens_out":5652,"would_cite":true,"duration_ms":53096,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81Q10","81V45"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that every fermionic atom of nuclear charge Z≥4 binds fewer than 1.1185Z electrons, with explicit lower-order terms, improving the previous best bound and giving the first finite-Z separation between fermionic and…","keywords":["excess charge","ionization conjecture","fermionic atoms","bosonic atoms","Benguria–Lieb–Nam argument","mean-field functional","radial symmetrization","Hardy inequality"],"falsifier":"Evaluate the mean-field constant β₃ by numerical optimization over non-radial atomic measures; if a measure is found with the quotient below 1/F(3) ≈ 0.8943, the radial reduction theorem would be false. Alternatively, test the improved Hardy inequality directly: find a smooth compactly supported function orthogonal to all radial functions whose kinetic energy is less than (9/4 − ε) ∫ |u|²/|x|² for ε > 0.","tokens_in":89710,"feed_emoji":"⚛️","tokens_out":6398,"duration_ms":62853,"temperature":0.7,"pith_summary":"The paper establishes a new rigorous upper bound on the maximum number of electrons $N_c(Z)$ a single atom of nuclear charge $Z$ can bind: for $Z \\ge 4$, $N_c(Z) < 1.1185Z + 3.90Z^{1/3} + 0.0134 + 0.184Z^{-1/3} + 0.0196Z^{-2/3}$. This improves on the long-standing bounds of Lieb ($2Z+1$) and Nam ($1.22Z+3Z^{1/3}$), and is the first finite-$Z$ quantitative result showing that fermionic atoms behave differently from bosonic atoms, since bosonic atoms can bind about 1.21 particles per nuclear charge in the large-$Z$ limit. The proof extends the Benguria–Lieb–Nam weighting argument to a general power weight $|x|^p$, proves a radial-symmetrization result for the resulting mean-field functional, and controls the weighted kinetic energy explicitly. If correct, the result shows that the excess charge of real atoms is quantitatively distinct from the bosonic case already at moderate nuclear charge.","feed_headline":"Fermionic atoms bind fewer than 1.1185 electrons per nuclear charge","feed_subtitle":"First rigorous bound separating finite-Z atoms from bosonic systems, which allow about 1.21 particles per charge.","key_machinery":"The central object is the mean-field functional $\\beta_p(\\mu) = \\frac{\\iint \\frac{|x|^p+|y|^p}{2|x-y|}\\,d\\mu(x)d\\mu(y)}{\\int |x|^{p-1}\\,d\\mu(x)}$, together with its $N$-particle analogue $\\alpha_{N,p}$. The key identity is the reduction $\\beta_p = \\beta_p^{\\mathrm{rad}}$ for $2\\le p\\le 3$: the paper proves that the infimum over probability measures is already attained among radially symmetric measures, by showing that the non-radial part contributes non-negative energy. That reduction relies on an improved Hardy inequality with constant $d^2/4$ for functions orthogonal to radial functions, which fixes the allowed range $p\\le 3$. Once $\\beta_p$ is reduced to a one-dimensional optimization problem, explicit constants $F(p)^{-1}$ are obtained, and refined comparison lemmas connect $\\alpha_{N,p}$ to $\\beta_p$, while a separate lemma bounds the weighted kinetic-energy term by an explicit power of $Z N^{-2/3}$.","core_discovery":"The paper's central claim is that for $p \\in (1,3]$, the critical electron number satisfies $N_c(Z) < F(p)Z + C(p)Z^{1/3}$, where $F(p) = \\max_{0\\le t\\le 1}\\frac{1+t^p}{1+t^{p-1}}$. For the two explicitly worked exponents, the authors prove: for $p=2$, $N_c(Z) < \\frac12(\\sqrt{2}+1)Z + 2.96Z^{1/3}$ for all $Z\\ge 2$, with $\\frac12(\\sqrt{2}+1) \\approx 1.2071$; and for $p=3$, $N_c(Z) < F(3)Z + 3.90Z^{1/3} + 0.0134 + 0.184Z^{-1/3} + 0.0196Z^{-2/3}$ for all $Z\\ge 4$, with $1.1184 < F(3) < 1.1185$. The argument resolves Nam's conjecture that the mean-field constant $\\beta_2$ is attained by radially symmetric probability measures, and extends this radial reduction to all $2\\le p\\le 3$. The proof then transfers the mean-field lower bound to the $N$-particle quantity $\\alpha_{N,p}$ through sharp comparison inequalities, and bounds the remaining weighted kinetic-energy term using Lieb's moment inequality, the fermionic kinetic-energy estimate, and the virial theorem.","pith_inferences":["The same weighting scheme could be applied to higher spatial dimensions $d\\ge 4$, where the improved Hardy constant is $d^2/4$ and the method might allow exponents $p>3$, yielding still smaller leading coefficients.","The explicit constants $C(p)$ are not claimed optimal; numerical experiments in the paper suggest $F(3)$ is within a few percent of the true value, so a more refined optimization could lower the rigorous leading coefficient.","Because Lieb's original bound holds for molecules, the generalized weighting argument may transfer to multi-nuclear systems, giving excess-charge bounds for molecules as well.","The radial-symmetrization lemma might extend to $3<p<4$ with a different Hardy-type estimate, which would further separate the fermionic bound from the bosonic value 1.21."],"forward_implications":["Every atom with nuclear charge at least 4 has at most about 1.1185 electrons per unit nuclear charge, a rigorous improvement over the previous 1.22 coefficient.","For large but finite $Z$, fermionic atoms are quantitatively distinct from bosonic atoms, which can bind about 1.21 particles per nuclear charge.","The bound for $p=2$ improves Nam's result for all $Z\\ge 2$ and Lieb's $2Z+1$ bound for $Z>5.3$.","Nam's conjecture that the mean-field constant $\\beta_2$ is attained by radial measures is confirmed, and the radial reduction is extended to the full range $2\\le p\\le 3$."],"supporting_citations":[{"why":"Supplies the weighted Benguria–Lieb argument with weight |x|^2 and the previous record bound 1.22Z+3Z^{1/3} that this paper generalizes and improves.","marker":"Nam (2012)"},{"why":"Proves the 2Z+1 bound that is the benchmark for all Z≥1 and whose positivity argument the authors extend to general powers p.","marker":"Lieb (1984)"},{"why":"Establishes that bosonic atoms have lim inf N_c(Z)/Z ≥ t_c ≈ 1.21, the contrast value the fermionic result is measured against.","marker":"Benguria & Lieb (1983)"},{"why":"Proves asymptotic neutrality of fermionic atoms, the limit statement this paper makes quantitative for finite Z.","marker":"LSST (1988)"},{"why":"Provides the improved Hardy inequality with constant d^2/4 for functions orthogonal to radial functions, on which the radial-symmetrization lemma rests.","marker":"Ekholm & Frank (2006)"},{"why":"Gives the sharp inequality relating ∫ ρ^{5/3} to weighted moments, used to bound the weighted kinetic-energy term in Lemma 6.3.","marker":"Lieb (1976)"},{"why":"Supplies the fermionic kinetic-energy inequality used in Lemma 6.3 to estimate the L^{5/3} norm of the one-particle density.","marker":"Frank, Nam & van den Bosch (2018)"},{"why":"Complements the bosonic lower bound by proving lim sup N_c(Z)/Z = t_c, fixing the bosonic asymptotic value near 1.21.","marker":"Solovej (1990)"}],"fun_headline_variants":["New bound: fermionic atoms hold fewer electrons than bosonic limit","Fermionic atom charge bound improved to 1.1185 per proton","Excess charge problem: tighter electron cap for atoms","Fermionic atoms beat bosonic electron-binding ratio","Improved upper bound on electrons bound by nuclear charge"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof relies on an improved Hardy inequality that holds only for functions orthogonal to radial functions; if that inequality fails at or near p=3, the leading coefficient 1.1185 would not be established by this argument.","fun_headline_variants_meta":{"raw":{"variants":["New bound: fermionic atoms hold fewer electrons than bosonic limit","Fermionic atom charge bound improved to 1.1185 per proton","Excess charge problem: tighter electron cap for atoms","Fermionic atoms beat bosonic electron-binding ratio","Improved upper bound on electrons bound by nuclear charge"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00101,"raw_usage":{"total_tokens":4294,"prompt_tokens":999,"completion_tokens":3295,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":615,"completion_tokens_details":{"reasoning_tokens":3212}},"tokens_in":615,"tokens_out":3295,"duration_ms":23984,"temperature":1.0,"reasoning_tokens":3212,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:16:33.226017+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the mean-field constant β₃ by numerical optimization over non-radial atomic measures; if a measure is found with the quotient below 1/F(3) ≈ 0.8943, the radial reduction theorem would be false. Alternatively, test the improved Hardy inequality directly: find a smooth compactly supported function orthogonal to all radial functions whose kinetic energy is less than (9/4 − ε) ∫ |u|²/|x|² for ε > 0.","supporting_citations":[{"cited_title":"New bounds on the maximum ionization of atoms","cited_arxiv_id":null,"evidence_quote":"Supplies the weighted Benguria–Lieb argument with weight |x|^2 and the previous record bound 1.22Z+3Z^{1/3} that this paper generalizes and improves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes that bosonic atoms have lim inf N_c(Z)/Z ≥ t_c ≈ 1.21, the contrast value the fermionic result is measured against."},{"cited_title":"Asymptotics for bosonic atoms","cited_arxiv_id":null,"evidence_quote":"Complements the bosonic lower bound by proving lim sup N_c(Z)/Z = t_c, fixing the bosonic asymptotic value near 1.21."}],"review_version":1}