{"id":"3a8dc09b-d06a-4153-8285-38433c786afb","arxiv_id":"1908.06390","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Under an outside-segment piercing condition, any n-point general-position set together with its n piercing points is contained in a cubic curve.","lead":"If every line through two points of a general-position n-point set is pierced by an n-point set lying outside the corresponding segment, then the union of both sets must lie on a single cubic curve. The result proves a conjecture of Erdos and Purdy in this special case and ties extremal piercing configurations to algebraic geometry.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.3 is not proved for n=5,6 or for any hull with exactly four extreme vertices; both small-case gaps are real and unaddressed.","rationale":"The reader's weakest assumption correctly identifies the improper use of Theorem 1.1 for small extreme sets, so I agree partially. My stress-test also found a second small-case obstruction that the reader did not explicitly list: the Chasles induction needs seven distinct consecutive P-points, and this fails for n = 5 and n = 6. Both are genuine gaps in the proof as written, but neither is a demonstrated counterexample; for n >= 7 with k != 4 the two-step argument seems coherent, and the counting plus Chasles machinery gives independent support. Because the theorem is asserted without an n restriction, a moderate-confidence accept is too strong unless the authors supply the missing k = 4 and n = 5,6 arguments. I therefore recommend conditional acceptance.","tokens_in":8747,"tokens_out":37942,"duration_ms":383435,"concrete_test":"Perform the missing small-case check analytically: (i) for a 4-point extreme set, prove |R'| >= 4 using the fact that an exterior point can lie on at most two of the six vertex-pair lines and that the two diagonals of a quadrilateral meet inside, so three exterior points cannot cover all six lines; (ii) for n = 6, construct a valid 3x3 Chasles grid on the nine points x5,...,x0,r5,r7,r9 and a second grid forcing r0,r2,r4, and verify all nine intersections are distinct; if either construction is impossible, the theorem as written is unproved for that case.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central theorem is stated for every n, but two places in the proof stop working exactly at small parameters. First, after defining P' (the extreme subset) and R' (the points of R outside the hull), the proof asserts |R'| >= k by Theorem 1.1. Theorem 1.1 is stated only for even n > 4. For odd k the trivial counting bound gives |R'| >= k, and for even k >= 6 Theorem 1.1 applies, but k = 4 is not covered. Thus a configuration whose convex hull has exactly four vertices is not eliminated by the written argument. Second, the Chasles step chooses a cubic through x_{n-1},...,x_{n-7}; for n = 5 and n = 6 these seven indices are not distinct (e.g. n = 6 gives x5,...,x1,x5), so the initial cubic and Claims 2.2 and 2.3 require seven distinct consecutive P-points. The paper gives no separate argument for n = 5,6. These are proof gaps, not known counterexamples; the n >= 7 case appears sound.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the following problem: given a set P of n points in general position in the plane and a disjoint set R of n points such that every line through two points of P contains a point of R outside the segment determined by those two points, show that P∪R lies on a cubic curve. The main theorem (Theorem 1.3) is proved in two steps. The first step is combinatorial: the outside-segment condition forces P to be in convex position, all points of R to lie outside the convex hull of P, and the collinear triples to follow the explicit rule i+j+k=0 modulo n after suitable labeling. The second step is algebraic: using Chasles's theorem in the style of Green and Tao, the authors construct a cubic curve through seven consecutive points of P and two points of R and then extend containment to the whole set P∪R via two claims (Claims 2.2 and 2.3). A bipartite analogue (Theorem 3.1) is also stated and proved along similar lines.","tokens_in":8938,"tokens_out":21941,"duration_ms":190617,"significance":"If correct, this is a strong and elegant result. It proves a special case of Conjecture 1.2, the conjecture attributed to Erdos and Purdy that equal-size piercing sets force a cubic curve, and it generalizes Jamison's characterization of extremal slope configurations. The proof is largely explicit: the combinatorial classification is clear and the Chasles applications are checkable, following the Green-Tao ordinary-lines framework. The outside-segment hypothesis is the key new input and it is used in a principled way to derive a rigid collinearity structure. No circularity or free parameters appear; external theorems are used as black boxes. The main caveat is that the proof as written has genuine small-case gaps that are not acknowledged in the manuscript.","major_comments":[{"comment":"The claim that |R′| ≥ k is justified by an application of Theorem 1.1 to the extreme subset P′. However, Theorem 1.1 is stated only for even cardinalities larger than 4. Thus the argument does not cover the case where P′ has exactly four extreme vertices. For odd k the trivial counting bound |R′| ≥ k is indeed available, but this is not stated, and for k = 4 the trivial bound gives only |R′| ≥ 3, which is insufficient. The equality |R′| = k and all subsequent steps depend on this lower bound, so the case of a convex hull with exactly four extreme vertices is not handled. A separate argument for k = 4 (or a modification of the counting argument) is needed for the theorem to hold for all n.","section":"Section 2, paragraph after the definition of P′"},{"comment":"The algebraic step requires a cubic Γ passing through x_{n−1}, …, x_{n−7} and then repeatedly applies Claim 2.2, whose statement presupposes seven distinct consecutive points of P for the initial configuration and six distinct consecutive points plus three distinct points of R for the Chasles grid. For n = 5 and n = 6 these indices are not distinct modulo n (for example, with n = 6 the list x_5, ..., x_{−1} repeats x_5), so the claims are not applicable. The paper does not provide a separate treatment of n = 5 and n = 6, and the same omission affects the even-n case in Claim 2.3 for n = 6. Thus the main theorem is not proved for n = 5 and n = 6. The same issue affects Theorem 3.1 when 2n is small.","section":"Section 2.1, construction of the cubic Γ and Claims 2.2–2.3"}],"minor_comments":[{"comment":"The statement contains a grammatical typo: “Let P is a set” should read “Let P be a set”.","section":"Theorem 1.1 statement"},{"comment":"A few typographical errors are present, including “colllinear” in the proof of the collinearity structure and “the the line” in the proof of Claim 2.3. These are harmless but should be corrected.","section":"Throughout the proof"},{"comment":"The term “relevant” is introduced informally. A short formal definition (e.g., “r is relevant for x if there exists y ∈ P \\ {x} such that r was chosen for the pair (x,y)”) would improve readability, though the intended meaning is clear from context.","section":"Section 2, notational clarity"},{"comment":"In the illustration for Claim 2.3, the label “m3” appears to be repeated in a way that may confuse the reader; the intended three m-lines should be labeled distinctly.","section":"Figure 4"}],"recommendation":"major_revision","confidential_remarks":"The two gaps identified in the major comments are localized and appear repairable: the k = 4 case may be handled by a separate counting or geometric argument, and the small values n = 5, 6 may be checked by a finite analysis or by a modified initial cubic. The central idea of the proof is sound for all sufficiently large n and the paper is a good fit for the journal. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [colleague],\n\nQuick take on Keller–Pinchasi, arXiv:1908.06390. The headline is a genuinely new structural result: under the outside-segment piercing hypothesis, P∪R must lie on a cubic. The two-step proof—first forcing P into convex position with a rigid i+j+k=0 collinearity rule, then using Chasles to get the cubic—is elegant and mostly checkable. I think the paper deserves a real referee, but not as-is: there are two small-case gaps that need to be closed before the theorem as stated is proved.\n\nWhat is actually new: Theorem 1.3 and the bipartite Theorem 3.1. These generalize Jamison's few-slopes classification to a wider family, and they are the first positive evidence for Conjecture 1.2 under a plausible extra assumption. The Chasles argument in Claims 2.2 and 2.3 is transparent; the reader can verify the nine points are the claimed ones. The combinatorial step—that the outside-segment condition forces convex position and the modular collinearity rule—is the real contribution, and it is not in the cited literature.\n\nSoft spots, in order of seriousness:\n\n1. Small hulls. The proof of |R'|≥k invokes Theorem 1.1, which is stated for even k>4. For k=4 the trivial counting bound only gives |R'|≥3, so the conclusion |R'|=k is not justified. This gap affects any configuration whose convex hull has exactly four vertices, regardless of n.\n\n2. Small n in the Chasles step. The cubic Γ is chosen through x_{n-1},...,x_{n-7}. For n=5 or 6 these indices are not distinct, so Claim 2.2 cannot be applied. No separate argument is given for n=5,6. Thus the theorem as stated is not proved for those n.\n\nNeither gap looks fatal—small cases often get fixed by direct arguments—but the paper as written overclaims. The n≥7, k≥5 (or k≥6) case appears sound to me. The citation pattern is honest; Theorem 1.1 is cited to multiple sources including two manuscripts, which is a bit soft but not unusual for a recent-result-dependent proof. No circularity.\n\nWho this is for: combinatorial geometers working on Erdős–Purdy type problems and on algebraic structure forced by incidences. They will want to read it despite the gaps. Recommendation: send to peer review with a note asking for small-case handling; do not desk reject. If the authors patch the gaps, this is a solid contribution.","headline":"A genuinely new structural result with an elegant two-step proof, but two real small-case gaps (k=4 hulls and n=5,6) need patching before the theorem as stated is proved.","tokens_in":9469,"tokens_out":5273,"would_cite":true,"duration_ms":50494,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52C10","14H50"],"pacs":[],"model":"deepseek-v4-flash","headline":"Under the outside-segment piercing condition, any configuration P∪R of 2n points must lie on a single cubic curve.","keywords":["Erdős–Purdy problem","line piercing","outside-segment condition","general position","cubic curve","Chasles theorem","convex position","slopes problem"],"falsifier":"A direct way to test the theorem is to search oriented matroids for small $n$, especially hull size $k=4$: if any configuration satisfying the outside-segment condition has $x_i,x_j,r_k$ collinear with $i+j+k\\not\\equiv 0\\pmod n$, or has all $2n$ points not lying on a cubic, the claim is false.","tokens_in":8527,"feed_emoji":"📐","tokens_out":15398,"duration_ms":138849,"temperature":0.7,"pith_summary":"This paper proves a structural version of the classical line-piercing problem. If $P$ is a set of $n$ points with no three collinear and $R$ is a second set of $n$ points such that every line through two points of $P$ contains a point of $R$ outside the segment between them, then all $2n$ points lie on one cubic curve. The outside-the-segment condition is exactly what makes the configuration rigid: $P$ must be in convex position, every point of $R$ lies outside the hull, and the only collinear triples are those obeying the cyclic rule $i+j+k\\equiv 0\\pmod n$. Once that structure is known, a Chasles-theorem argument forces the whole set onto a cubic. This confirms the paper's Conjecture 1.2 under a stronger hypothesis and generalizes the earlier extremal characterization of sets with exactly $n$ directions.","feed_headline":"Outside-the-segment piercing forces all 2n points onto one cubic","feed_subtitle":"The outside-segment version of the classical piercing conjecture is proved: all points must lie on a single cubic curve.","key_machinery":"The machinery is two-step. First, the outside-segment hypothesis becomes a rigid combinatorial structure: the points of $P$ are in convex position, the points of $R$ lie outside their convex hull, and collinearities are exactly the modular rule $i+j+k\\equiv 0\\pmod n$. Second, this rule feeds a Chasles-theorem grid: whenever three lines $\\ell_1,\\ell_2,\\ell_3$ meet three lines $m_1,m_2,m_3$ in nine distinct points, a cubic through any eight of those points passes through the ninth. The paper applies this repeatedly to the nine-point grids formed by the cyclic labels, so that one cubic containing seven $P$-points and two $R$-points must contain every point of $P\\cup R$.","core_discovery":"The central claim, Theorem 1.3, is that the outside-segment piercing condition is strong enough to determine the entire combinatorial geometry of $P\\cup R$ and then force it onto an algebraic curve of degree at most three. More precisely, for every pair $x,y\\in P$ the required piercing point $r\\in R$ lies outside the interval $xy$; the paper shows $P$ is in convex position, $R$ lies outside $\\operatorname{conv}(P)$, and after labelling $P$ cyclically as $x_0,\\dots,x_{n-1}$ and $R$ as $r_0,\\dots,r_{n-1}$, the collinearity condition is $x_i,x_j,r_k$ collinear iff $i+j+k\\equiv 0\\pmod n$. It then proves, using Chasles' theorem in the form that a cubic through eight of the nine intersections of two triples of lines passes through the ninth, that any cubic through seven consecutive $P$-points and two $R$-points must contain all of $P\\cup R$. Hence $P\\cup R$ is contained in a cubic curve.","pith_inferences":["A natural next test is whether the outside-segment condition can be weakened to allow at most one piercing point inside the segment per line; if the cubic conclusion persists, the full Conjecture 1.2 would be much closer.","The cyclic rule $i+j+k\\equiv 0\\pmod n$ suggests that equality configurations are essentially residues on a cubic curve, so a plausible classification is that they are affine images of regular polygons on a cubic.","A small-case search should target hull sizes $k=3$ and $k=4$, since the counting step invokes a lower bound stated only for even $n>4$; if a counterexample exists, it is most likely there."],"forward_implications":["Under the outside-segment hypothesis the extremal piercing configuration is completely rigid: $P$ is in convex position, each point of $R$ lies outside the convex hull, and the only collinear triples are those with $i+j+k\\equiv 0\\pmod n$.","Consequently every such configuration is algebraic of degree at most three: there is a single cubic curve containing all $2n$ points, so the combinatorial extremum cannot be realized by generic point sets.","The theorem recovers the known characterization of general-position sets with exactly $n$ directions, because the required $R$ can be placed on the line at infinity, outside every finite segment.","The same Chasles propagation works in the bipartite setting: if blue and green points alternate on the convex hull and every blue-green line is pierced by a red point outside the segment, then the union lies on a cubic."],"supporting_citations":[{"why":"It introduces the original piercing problem and the conjecture that the theorem settles under the stronger outside-segment assumption.","marker":"[EP78]"},{"why":"These are the cited proofs of Theorem 1.1, the lower bound $|R|\\ge n$ for even $n>4$, which Step 1 uses to count $R$-points outside the hull of the extreme subset.","marker":"[ABK+08, Mil18, Pin18, PP19]"},{"why":"It supplies the Chasles-theorem propagation scheme that the second step uses to force a cubic through all of $P\\cup R$.","marker":"[GT13]"},{"why":"It gives the prior extremal characterization, that general-position sets with exactly $n$ directions lie on a conic, which Theorem 1.3 recovers and extends.","marker":"[Jam86]"}],"fun_headline_variants":["Outside-segment piercing forces all 2n points onto one cubic","External piercing condition forces entire set onto a cubic curve","Outside-pierced lines imply all points lie on a single cubic","Cubic inevitability: outside piercing of lines constrains all points"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that for every pair $x,y\\in P$ the required piercing point in $R$ lies strictly outside the open segment between $x$ and $y$; without this, the convex-position and cyclic-collinearity rigidity that feeds the Chasles argument has no basis.","fun_headline_variants_meta":{"raw":{"variants":["Outside-segment piercing forces all 2n points onto one cubic","External piercing condition forces entire set onto a cubic curve","Outside-pierced lines imply all points lie on a single cubic","Cubic inevitability: outside piercing of lines constrains all points"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001123,"raw_usage":{"total_tokens":4630,"prompt_tokens":859,"completion_tokens":3771,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":475,"completion_tokens_details":{"reasoning_tokens":3699}},"tokens_in":475,"tokens_out":3771,"duration_ms":23053,"temperature":1.0,"reasoning_tokens":3699,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:49:40.288565+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct way to test the theorem is to search oriented matroids for small $n$, especially hull size $k=4$: if any configuration satisfying the outside-segment condition has $x_i,x_j,r_k$ collinear with $i+j+k\\not\\equiv 0\\pmod n$, or has all $2n$ points not lying on a cubic, the claim is false.","supporting_citations":[],"review_version":1}