{"id":"59d80068-73c6-4620-9b00-404ed186b4fe","arxiv_id":"2501.05315","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"high","formal_verification":"none","parameter_count":2,"one_line_summary":"A new Bezout based upper bound 2^n(3n-2)^3 on equilibria of n charges, a record ratio 25/7 construction, and a counterexample to a 2007 conjecture.","lead":"The paper improves the known upper bound on how many equilibrium points n electric charges can create, and it builds configurations with a record-high ratio of equilibria per charge. It also claims to disprove a 2007 conjecture by Gabrielov, Novikov and Shapiro, but the counterexample uses a special symmetric setup the conjecture's generic position clause excludes.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Result 3's counterexample to Conjecture 2 ignores the generic-position hypothesis: the truncated octahedron is symmetric, and no perturbation argument shows #1(V1) > #1(E) for a generic nearby configuration.","rationale":"The reader's verdict is CONDITIONAL and identifies two weaknesses: the unproven additivity formula in Eq. (9) behind Result 2, and the generic-position issue in Result 3. I focus on the latter as the single most load-bearing concern because the abstract explicitly promises a counterexample to a published conjecture; if the generic-position gap cannot be closed, the paper's most headline-worthy claim fails. The Bezout upper bound (Result 1) appears sound: Lemma 2 adequately handles the generic choice of one charge, and the affine Bezout bound is applied to the polynomial system P with degrees (3n−2)^3·2^n. The additivity formula is also unproven, but a failed record ratio would be less damaging than an invalid counterexample. The proposed test would settle whether the truncated octahedron can be perturbed generically while retaining #1(V1) > #1(E). If the test shows the inequality cannot survive generic perturbation, the paper should be revised to present Result 3 as a numerical observation, not a counterexample. Since the current conditional verdict already requires such a revision, my read does not move the verdict.","tokens_in":23841,"tokens_out":10382,"duration_ms":106280,"concrete_test":"Randomly perturb each of the 24 vertices of the truncated octahedron by independent Gaussian displacements of magnitude δ = 10^-k, k = 1,...,8, and for each perturbation (a) solve ∇V1 = 0 numerically and count index-1 and index-2 equilibria with a certified solver (e.g., interval arithmetic); (b) compute the effective cells of the Voronoi diagram of the perturbed points and count #1(E). If there is any δ for which #1(V1) > #1(E), Result 3 can be salvaged; if #1(E) ≥ #1(V1) for all small δ, the truncated octahedron is not a counterexample to Conjecture 2 and the claim must be withdrawn or softened to a numerical observation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Result 3 (Section 4.2) claims that unit charges at the vertices of the truncated octahedron contradict Conjecture 2, but Conjecture 2 is restricted to 'generic position' of the charges. The truncated octahedron is highly symmetric—the center is a degenerate equilibrium, so the configuration is exactly of the kind excluded by the hypothesis. The paper gives no argument that a small generic perturbation of the vertices preserves the asserted inequality #1(V1) > #1(E). The 18 index-1 and 36 index-2 equilibria of V1, if non-degenerate, would persist under perturbation by the implicit function theorem, but the limiting E-count #1 is the number of effective Voronoi cells, which can change (typically increase) under perturbation; a generic 24-point set in R3 has a much richer Voronoi/Delaunay complex than the symmetric one. Without a proof (or certified computation) that some generic perturbation has #1(E) < 18 while #1(V1) ≥ 18, the example does not refute the conjecture. Moreover, the asserted counts 18 and 36 are not rigorously established in the paper—they are justified by symmetry-adapted observations ('we observe', 'can be deduced') and no complete analytic proof or reproducible code is supplied.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the number of critical points (equilibria) of the electrostatic potential of n point charges in R^3. It claims three main results: (1) an upper bound of 2^n(3n-2)^3 on the number of isolated critical points for generic position of one charge, obtained by clearing denominators and applying Bezout's theorem; (2) a family of iterated anti-prism configurations with equilibrium-to-charge ratio exceeding 25/7 - epsilon; and (3) a counterexample to Conjecture 1.8(a) of Gabrielov, Novikov and Shapiro, based on 24 unit charges at the vertices of a truncated octahedron. The paper also surveys equilibria for Platonic, Archimedean and Catalan solids, discusses the one-parameter family V_p, and formulates several conjectures about slices of Voronoi tessellations.","tokens_in":24189,"tokens_out":20069,"duration_ms":196225,"significance":"If fully established, the improved upper bound would be a substantial advance over the previous Thom-Milnor bound, and the Bezout-based clearing-of-denominators idea is attractive. The ratio construction would give a new lower-bound record, and a valid counterexample to the GNS conjecture would be an important result. However, in the present version the counterexample is not a counterexample to the conjecture as stated, the record ratio rests on an unproved additivity formula, and the proof of the upper bound has a nontrivial gap. The systematic polyhedral tables and the local-homology observations are useful computational and structural contributions.","major_comments":[{"comment":"The proof that the Jacobian of the polynomial map P has full rank is incomplete. The argument shows only that the upper-left block dR/dx is nonsingular when the displayed sum is nonzero. Because the lower-right block dQ/du is invertible, full rank of the full Jacobian requires invertibility of the Schur complement dR/dx - (dR/du)(dQ/du)^{-1}(dQ/dx), which is never established. In addition, the displayed formula for dR_j/du_q omits the factor (x_j - p_{ij}); with that factor restored, the Schur complement is different and the claimed conclusion is not immediate. Thus the correspondence between non-degenerate equilibria and isolated non-degenerate zeroes of P, and hence the Bezout bound of Result 1, is not proven as written.","section":"Section 2.3, Lemma 2"},{"comment":"The degree computation in Proposition 1 is incorrect. For p = 2r, each factor r_m^{p+2} = r_m^{2r+2} = (||x - A_m||^2)^{r+1} has degree 2(r+1), not r-1. The numerator after clearing denominators therefore has degree 1 + 2(r+1)(n-1), not 1 + (n-1)(r-1). Consequently the stated bound (r(n-1)+2-r)^3, and in particular the n^3 bound for V_2, does not follow from the argument. For n = 2 and p = 2, the cleared numerator already has degree 5, showing the discrepancy directly.","section":"Section 2.5, Proposition 1"},{"comment":"The additivity formula m(n^L - 1)/(n - 1) for the iterated construction is asserted without proof. A rigorous multi-scale argument must show that, at sufficiently small scale, every equilibrium of each layer persists, that the layers do not interact to create additional equilibria, and that the equilibria of the outer configuration survive the replacement of a charge by a tiny cluster. No such perturbation or convergence argument is supplied. Moreover, the base value m = 25 for the square anti-prism is reported from experiments in Section 3.3; Appendix C proves existence of some equilibria but not an exhaustive count. Therefore Result 2 is not established.","section":"Section 3.4, Eq. (9)"},{"comment":"The truncated octahedron configuration is not in generic position: the center is a degenerate equilibrium, which is exactly the case excluded by Conjecture 2. To refute the conjecture one must exhibit a generic configuration, or at least prove that a small generic perturbation of the truncated octahedron preserves the inequality #1(V1) > #1(E). No such argument is given. The asserted counts of 18 index-1 and 36 index-2 equilibria are justified by symmetry-adapted observations and not by a complete analytic proof or certified computation. As presented, the example does not contradict the conjecture as stated.","section":"Section 4.2, Result 3"}],"minor_comments":[{"comment":"The statement of Theorem 1 contains a typo: 'd1, ..., dd' should be 'd1, ..., dm'. The theorem should also state the standard hypotheses under which the affine Bezout bound applies to non-degenerate zeros.","section":"Section 2.2, Theorem 1"},{"comment":"The formula for dQ_m/dx_s contains a typo: 'xr - p_{ms}' should be 'x_s - p_{ms}'. The displayed formula for dR_j/du_q is also missing the factor (x_j - p_{ij}).","section":"Section 2.3, Lemma 2"},{"comment":"The phrase 'Our experiments show' should be replaced by a clear distinction between proven results and numerical evidence; Appendix C proves the existence of some equilibria but not the exhaustive counts used in Eq. (8).","section":"Section 3.3"},{"comment":"There is a typo in the proof: 'if h < sqrt(2)h' should read 'if h < sqrt(2)R', and 'x = 2 = 0 = x3' should read 'x2 = 0 = x3'.","section":"Appendix C, Theorem 5"},{"comment":"The question about exceeding 25n/7 should be phrased conditionally on the validity of Result 2, since the claimed record depends on the unproved additivity formula in Eq. (9).","section":"Section 6"}],"recommendation":"reject","confidential_remarks":"The core idea behind the upper bound is promising and deserves further development, but as it stands two of the three headline results are not proven and the proposed counterexample appears to use a configuration outside the hypothesis of the conjecture it claims to refute. I would encourage the authors to repair the upper-bound proof, provide a rigorous multi-scale argument or withdraw the record claim, and either produce a genuine generic-position counterexample or present the truncated-octahedron observation as computational evidence rather than as a theorem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper has one genuinely solid result—a new upper bound on the number of equilibria of the electrostatic potential—wrapped around two headline claims that are not yet proven at the level the text suggests. The upper bound is worth a serious referee; the other two need substantial revision.\n\nThe Bezout argument in Section 2 is the best part. The authors reduce the critical point equations to a polynomial system by introducing the distances as variables, then use the affine Bezout bound after proving that for generic position of one charge the critical points correspond to non-degenerate, hence isolated, zeros. The bound 2^n(3n-2)^3 is a huge improvement over the previous 5*9^{3+n} and is, as far as I can tell, correct. This alone is a worthwhile contribution, and the extension to even p potentials is a nice bonus.\n\nThe problems start with Result 2. The iterative construction that claims a 25/7 ratio depends on Eq. (9), which says the total number of equilibria after replacing each charge by a small copy of the original solid is m(n^L-1)/(n-1). That additivity is asserted, not proven. You need a rigorous multi-scale perturbation argument to show outer equilibria persist and inner clusters contribute independently. The text doesn't supply it. Also, the per-anti-prism counts (25 equilibria for the square anti-prism) are partly based on numerical observation; Appendix C proves some of them, but not the full count for all k>=4. So Result 2 is a plausible conjecture with strong numerical support, not a theorem.\n\nResult 3 has a more serious problem. It claims the truncated octahedron configuration refutes Conjecture 2, but Conjecture 2 is explicitly restricted to generic position. The truncated octahedron is symmetric, with a degenerate equilibrium at the center—exactly the kind of configuration excluded by the hypothesis. The paper gives no argument that a small generic perturbation preserves the inequality #1(V1) > #1(E). Moreover, the asserted counts 18 and 36 are not rigorously established; they come from symmetry-based observations, and no reproducible code or complete analytic proof is provided. As stated, the counterexample doesn't refute the conjecture.\n\nThe numerical explorations and the local homology analysis are interesting, and the paper is clearly written by people who know the area. But the two marquee results need more work.\n\nMy recommendation: send it to peer review—the upper bound deserves scrutiny and the other claims, if fixed, would be significant. A good referee will catch these gaps, and the authors are capable of addressing them.","headline":"The Bezout upper bound is real; the record ratio and counterexample are not yet fully proven, so read carefully before trusting the headline claims.","tokens_in":24640,"tokens_out":3609,"would_cite":true,"duration_ms":34335,"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":"This paper proves a new upper bound on electrostatic equilibria, builds configurations whose equilibrium-per-charge ratio surpasses 25/7, and refutes a 2007 conjecture via the truncated octahedron.","keywords":["electrostatic potential","electric field zeroes","equilibria","Morse theory","Bezout theorem","Voronoi tessellation","truncated octahedron","Maxwell conjecture"],"falsifier":"Track all equilibria numerically to high precision in the two-layer iterated square anti-prism: Eq. (9) predicts $m(n^2-1)/(n-1)=25\\cdot 9=225$ equilibria, so any count different from 225 would falsify the additivity premise behind Result 2. Independently, perturb the 24 truncated-octahedron charges generically and recount index-1 equilibria of $V_1$: a drop from 18 to 14 would show the counterexample to Conjecture 2 rests on the non-generic symmetry.","tokens_in":23666,"feed_emoji":"⚡","tokens_out":10788,"duration_ms":94635,"temperature":0.7,"pith_summary":"This paper studies the number of equilibria—points where the electric field vanishes—of a potential generated by n positive point charges in $R^{3}$. It establishes three things: the electric field has at most 2^n(3n−2)^3 isolated critical points when one charge is placed generically; iterating a square anti-prism construction gives configurations whose equilibrium-to-charge ratio exceeds 25/7−ε for any ε>0; and the 24 unit charges at the vertices of a truncated octahedron violate a 2007 conjecture of Gabrielov, Novikov and Shapiro, since V1 has 18 index-1 equilibria while the limiting distance function has only 14. The new upper bound is the best known to date, and the counterexample shows that the electrostatic potential can have more equilibria than the distance function defined by the same point charges. The results leave Maxwell's 1873 quadratic upper bound open while suggesting it is far from tight.","feed_headline":"24 charges defeat an equilibrium conjecture","feed_subtitle":"A truncated octahedron gives 18 saddle points where the limit has 14, and a new bound improves on older ones.","key_machinery":"The upper bound is carried by a polynomial reformulation: clearing denominators in $\\nabla V=0$ and introducing variables $u_m$ with $u_m^2=\\|x-A_m\\|^2$ turns the critical-point equations into a system $P(x,u)=0$ whose degrees are $3n-2$ for the three gradient components and $2$ for the n radius equations. Affine Bezout then bounds isolated zeroes by the product $2^n(3n-2)^3$, and Lemma 2 shows the non-degenerate critical points are isolated zeroes when one charge is generic. The record ratio is built by iterated substitution: replacing each charge by a small copy of the same solid and assuming the equilibria simply add gives $m(n^\\ell-1)/(n-1)$ equilibria for $\\ell$ layers, and the square anti-prism attains $m/(n-1)=25/7$ in the limit. The counterexample is the truncated octahedron, whose symmetry lets the authors locate equilibria along intersections of reflection planes and associate them with facets and edges; the count is 18 index-1 saddles for $V_1$ versus 14 for the distance-function limit.","core_discovery":"The central assertions are: (1) for any n and any charges, after moving a single chosen charge to a generic position, the isolated critical points of $V$ number at most $2^n (3n-2)^3$; (2) for every $\\varepsilon>0$ there are unit-charge configurations with $k/n > 25/7 - \\varepsilon$, obtained by layering iterated copies of a square anti-prism; and (3) the potential generated by unit charges at the vertices of the truncated octahedron has $18$ index-1 equilibria, whereas the limiting distance function $E(x)=\\min_i \\|x-A_i\\|$ has only $14$ index-1 equilibria and the same $36$ index-2 equilibria, so the total for $V_1$ exceeds the total for $E$, contradicting Conjecture 1.8(a) of Gabrielov, Novikov and Shapiro. The paper also proves that, for even $p=2r$, the modified potential $V_p$ has at most $(r(n-1)+2-r)^3$ isolated critical points, and it reports numerical and analytic equilibrium counts for charges placed at vertices of Platonic, Archimedean, Catalan, prism and anti-prism solids.","pith_inferences":["A rigorous multi-scale perturbation proof for the additivity in Eq. (9) would promote the 25/7 record from construction to theorem; the paper states the formula without that proof.","The truncated octahedron is not generic, so whether the failure of Conjecture 2 persists under the generic-position hypothesis remains open; a small generic perturbation test would settle it.","One testable extension is to iterate other Archimedean solids and compare limiting ratios; the paper's census shows several candidates with per-vertex ratios above 3, but only the square anti-prism is iterated.","The layered construction suggests that high ratios can be amplified by substitution; if a similar additivity held for other solids, the 25/7 record might be a starting point rather than a ceiling."],"forward_implications":["The previously known upper bound—roughly $5\\cdot 9^{3+n}$—drops to $2^n(3n-2)^3$, though this is still far above Maxwell's conjectured $(n-1)^2$.","For the even-power potentials $V_{2r}$, the bound $(r(n-1)+2-r)^3$ is polynomial in $n$ for fixed $r$, with $V_2$ at most $n^3$ isolated critical points.","Conjecture 1.8(a) of Gabrielov, Novikov and Shapiro is false as stated: the electrostatic potential of equal unit charges can have more equilibria than the distance function of the same points.","The iterative anti-prism construction produces equilibrium-to-charge ratios approaching $25/7\\approx 3.57$, the highest ratio found so far, from a base ratio of $25/8$ for the square anti-prism.","For $p>1$, the potentials $V_p$ have no maxima when all charges are positive, and the octahedron example shows that minima, forbidden for the harmonic case $p=1$, can appear for larger $p$."],"supporting_citations":[{"why":"Supplies the 2007 upper bound and poses Conjectures 1.8(a) and 1.9, the first of which the truncated octahedron is built to refute.","marker":"[6]"},{"why":"Maxwell's 1873 treatise states the $(n-1)^2$ conjecture that motivates the whole counting problem.","marker":"[9]"},{"why":"Morse and Cairns provide the perturbation theorem used to unfold degenerate equilibria and the Morse-theoretic lower bound $n-1$.","marker":"[12]"},{"why":"Gives the affine Bezout theorem that bounds the number of isolated zeroes of the polynomial system in Result 1.","marker":"[16]"},{"why":"States the particular case of Bezout's theorem that the paper uses to bound isolated zeroes after clearing denominators.","marker":"[8]"},{"why":"Defines the Voronoi tessellation whose effective cells correspond to equilibria of the limiting distance function, the comparison object for the counterexample.","marker":"[15]"},{"why":"Constructs 2k+2 points in R^3 with quadratically many effective cells, used to contrast Voronoi bounds with the quest for quadratic equilibria of V.","marker":"[5]"},{"why":"Gives the prior upper bound from the Thom-Milnor theorem that Result 1 replaces as the best known.","marker":"[17]"}],"fun_headline_variants":["Truncated octahedron breaks equilibrium conjecture","18 equilibria beat a 14-equilibrium guess","New upper bound and counterexample for charge equilibria","Charge configurations reach 25/7 zeroes per charge"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The record-ratio construction depends on the unproven additivity formula in Eq. (9): after replacing each charge by a tiny copy of the original solid, the total number of equilibria is assumed to be exactly $m(n^\\ell-1)/(n-1)$, and no multi-scale perturbation argument is supplied.","fun_headline_variants_meta":{"raw":{"variants":["Truncated octahedron breaks equilibrium conjecture","18 equilibria beat a 14-equilibrium guess","New upper bound and counterexample for charge equilibria","Charge configurations reach 25/7 zeroes per charge"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000216,"raw_usage":{"total_tokens":1442,"prompt_tokens":966,"completion_tokens":476,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":582,"completion_tokens_details":{"reasoning_tokens":413}},"tokens_in":582,"tokens_out":476,"duration_ms":5190,"temperature":1.0,"reasoning_tokens":413,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:15:13.844973+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Track all equilibria numerically to high precision in the two-layer iterated square anti-prism: Eq. (9) predicts $m(n^2-1)/(n-1)=25\\cdot 9=225$ equilibria, so any count different from 225 would falsify the additivity premise behind Result 2. Independently, perturb the 24 truncated-octahedron charges generically and recount index-1 equilibria of $V_1$: a drop from 18 to 14 would show the counterexample to Conjecture 2 rests on the non-generic symmetry.","supporting_citations":[{"cited_title":"Gabrielov, D","cited_arxiv_id":null,"evidence_quote":"Supplies the 2007 upper bound and poses Conjectures 1.8(a) and 1.9, the first of which the truncated octahedron is built to refute."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Maxwell's 1873 treatise states the $(n-1)^2$ conjecture that motivates the whole counting problem."},{"cited_title":"Morse and S","cited_arxiv_id":null,"evidence_quote":"Morse and Cairns provide the perturbation theorem used to unfold degenerate equilibria and the Morse-theoretic lower bound $n-1$."},{"cited_title":"Polynomial automorphisms of Cn","cited_arxiv_id":null,"evidence_quote":"Gives the affine Bezout theorem that bounds the number of isolated zeroes of the polynomial system in Result 1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States the particular case of Bezout's theorem that the paper uses to bound isolated zeroes after clearing denominators."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Voronoi tessellation whose effective cells correspond to equilibria of the limiting distance function, the comparison object for the counterexample."},{"cited_title":"Proc. 40th Internat. Sympos. Comput. Geom., 2024","cited_arxiv_id":null,"evidence_quote":"Constructs 2k+2 points in R^3 with quadratically many effective cells, used to contrast Voronoi bounds with the quest for quadratic equilibria of V."},{"cited_title":"Upper bounds for the number of isolated critical points via the Thom–Milnor theorem","cited_arxiv_id":null,"evidence_quote":"Gives the prior upper bound from the Thom-Milnor theorem that Result 1 replaces as the best known."}],"review_version":1}