{"id":"c5f7e897-e8f1-4886-b972-d1027f61db55","arxiv_id":"2608.23139","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"An explicit two-step bisection recurrence shows that longest-edge bisection of tetrahedra can degenerate when tied longest edges are chosen arbitrarily, violating standard quality conditions.","lead":"A short mathematical note constructs a family of tetrahedra for which the longest-edge bisection rule, applied with arbitrary tie-breaking, produces shapes that become infinitely flat. The result warns designers of mesh-refinement software to specify and analyze tie-breaking rules explicitly.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The counterexample is correct, but the proof of Eq. (9) contains an arithmetic error in the distance d(B,CD); this needs correction before publication.","rationale":"The manuscript's core claim is a valid counterexample. I re-derived the edge lengths, the recurrence, the volume ratio, and the two dihedral-angle formulas. Eq. (10) checks out: the three projection quantities are exactly as stated, and the cosine follows. The proof of Eq. (9), however, has an erroneous distance: d(B,CD) is sqrt(14a)/4, not sqrt(7a)/8, and the displayed simplification from the printed value to sqrt(8a/7) does not compute. Since Eq. (9) is nonetheless correct, this is a repairable typo, not a counterexample failure. The reader's conditional verdict is therefore retained, with the condition sharpened: correct the distance in §3 and include the short projection derivation for (10). No further objections.","tokens_in":4744,"tokens_out":23041,"duration_ms":213911,"concrete_test":"Recompute the dihedral angle at CD of E(a) from the face normals (or from the corrected distance) and the dihedral angle at MC of O(a) from the three projection identities in §3; if the results reproduce (9) and (10) for a=1 and in the limit a→0, the remaining issue is purely typographical.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Independent recomputation confirms the central construction. The edge-length recurrence (5), the volume ratio (8), and the projection calculation (10) are all consistent; in particular, (10) yields cos Θ(a) = (2a−7)/sqrt((4a+7)(8a+7)). The one substantive flaw is in the proof of (9): the text states d(B,CD) = 2|BCD|/|CD| = sqrt(7a)/8. With |BCD| = sqrt(7)a/4 and |CD| = sqrt(2a), the correct value is d(B,CD) = sqrt(14a)/4. Moreover, even taking the displayed value at face value, a/(sqrt(7a)/8) simplifies to sqrt(64a/7), not sqrt(8a/7). Eq. (9) itself is correct once the distance term is fixed, so this is a local typo rather than a fatal gap. The minimum-angle conclusion is still established by the corrected calculation, and the maximum-angle conclusion is supported by (10), which has been checked.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs an explicit one-parameter family of tetrahedra E(a), 0 < a ≤ 1, and proves an exact two-bisection recurrence: after one longest-edge bisection of E(a) and retaining the appropriate child O(a), a second longest-edge bisection (choosing BD when BD and DC are tied) yields a tetrahedron congruent to E(a/2). Iterating from a = 1 produces an infinite admissible longest-edge bisection sequence E(2^{-k}), O(2^{-k}). The main theorem states three simultaneous failures: the normalized volume |E_k|/h_{E_k}^3 tends to zero; an interior dihedral angle δ(a_k) of E(a_k) at edge CD tends to zero; and an interior dihedral angle Θ(a_k) of O(a_k) at edge MC tends to π. Explicit formulas (8)--(10) are given for the decay rates. The paper also notes a deterministic tie-breaking rule that selects the degenerating branch and sketches an embedding of the construction into conforming global longest-edge partitions.","tokens_in":4918,"tokens_out":15685,"duration_ms":122288,"significance":"If the result is correct, it closes a gap in the literature on three-dimensional longest-edge bisection: the bare geometric rule with arbitrary tie-breaking does not preserve shape regularity or the minimum/maximum-angle conditions, contrary to the reassuring behavior in two dimensions. The construction is self-contained and parameter-free in the sense that only the scaling variable a is varied; the recurrence is proved by exact edge-length sextuples, and the asymptotic limits are explicit. The paper is careful to distinguish the existence of a degenerating admissible orbit from degeneration for every tie-breaking convention, and it offers a simple deterministic tie-breaking rule that produces the bad orbit. I verified the central computations independently, including the projection calculation leading to Eq. (10); the only defect found is an arithmetic slip in the proof of Eq. (9), which does not affect the final formula.","major_comments":[{"comment":"The proof of Eq. (9) contains two arithmetic errors that should be corrected. The distance d(B,CD) is stated as sqrt(7a)/8; from |BCD| = sqrt(7)a/4 and |CD| = sqrt(2a), the area formula gives d(B,CD) = 2|BCD|/|CD| = sqrt(14a)/4. The subsequent simplification a/(sqrt(7a)/8) = sqrt(8a/7) is also arithmetically incorrect. With the corrected distance, a/(sqrt(14a)/4) = sqrt(8a/7), so the final formula (9) is correct; the errors are local, but the written proof must be fixed before publication.","section":"Section 3, proof of Eq. (9)"}],"minor_comments":[{"comment":"The calculation leading to |p_B|^2, |p_D|^2, and p_B dot p_D is presented as a direct calculation without derivation. I verified the formulas independently; nevertheless, providing the projection coordinates or at least the key simplification steps would strengthen the presentation, since this is the only step establishing the maximum-angle failure.","section":"Section 3, Eq. (10)"},{"comment":"The bound 'every new midpoint-to-opposite-vertex edge has length at most (sqrt(3)/2)c' is stated without proof. It follows immediately from the median formula in a triangle with sides at most c; a one-sentence justification would be helpful for completeness.","section":"Section 4"},{"comment":"The abstract's phrase 'arbitrary tie-breaking among longest edges does not guarantee nondegeneration' could be misread as a universal statement about all tie-breaking rules. The introduction correctly clarifies that the paper proves the existence of a degenerating admissible orbit; consider aligning the abstract's wording with that clarification.","section":"Abstract and Introduction"}],"recommendation":"minor_revision","confidential_remarks":"This is a solid short counterexample. The main theorem is correct; the only technical issue is an arithmetic slip in the proof of Eq. (9), which the authors should fix. The paper is within scope for math.NA and should be publishable after minor revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the quick take: this is a real negative result, and it's worth paying attention to. Korotov builds a one-parameter family E(a) where one longest-edge bisection produces O(a), a second bisection (choosing one of two tied edges) produces a tetrahedron congruent to E(a/2). Iterating from E(1) gives an admissible orbit under the rule 'choose any longest edge,' and the normalized volume, a dihedral angle, and the complement of another dihedral angle all decay like 2^{-k/2}. That kills shape regularity and both minimum- and maximum-angle conditions simultaneously. The construction is self-contained and exact; no fitting, no hidden assumptions.\n\nWhat's genuinely new is the mechanism: a persistent tie between BD and DC in every O(a). The existing 3-D literature mostly analyzes marked or conforming longest-edge variants and similarity-class orbits; it doesn't contain a degenerating orbit for the unrestricted rule with arbitrary tie-breaking. So the paper fills a real gap. It also gives a simple deterministic tie rule (choose the tied edge whose opposite edge is longer) that lands on the bad branch, which strengthens the point.\n\nThe computations are mostly clean. The edge-length sextuples are explicit; the recurrence lemma is a direct congruence check; the volume ratio (8) is immediate. Formula (10) for the dihedral angle at MC is asserted as 'direct calculation' without derivation. The stress-test recomputation says it's correct, and the reader's main doubt is that unshown step. It's minor but should be expanded or cited. The proof of (9), however, contains a local arithmetic error: the text states d(B,CD)=sqrt(7a)/8, but the correct value is sqrt(14a)/4 (which equals sqrt(7a/8)). The simplification shown in the text is also wrong. The final formula δ(a)=arctan(sqrt(8a/7)) is correct once the distance is fixed, so the minimum-angle failure stands. This needs a one-line correction, not a rethink.\n\nWho this is for: anyone implementing or analyzing longest-edge bisection in 3-D, and anyone stating regularity theorems without specifying tie-breaking. The paper is short, readable, and the referee can verify every equation in a few hours.\n\nMy recommendation: send it out. It deserves a serious referee. After the typo fix and a derivation for (10), it should be accepted as a note. I'd cite it.","headline":"A correct, exact counterexample: unrestricted tetrahedral longest-edge bisection can degenerate, and the paper's proof is sound apart from one local arithmetic typo.","tokens_in":5382,"tokens_out":3076,"would_cite":true,"duration_ms":26207,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":["65M50","65N50","65N30"],"pacs":[],"model":"deepseek-v4-flash","headline":"An explicit one-parameter family of tetrahedra shows that the unrestricted longest-edge bisection rule can produce degenerating meshes in three dimensions.","keywords":["longest-edge bisection","tetrahedron","mesh degeneration","shape regularity","minimum-angle condition","maximum-angle condition","tie-breaking"],"falsifier":"Take a concrete value such as $a=1/64$, write down the coordinates of $O(a)$ from the definition, and compute the dihedral angle at edge $MC$ by standard linear algebra on the two face normals; compare with the value predicted by formula (10). If the numbers disagree, the maximum-angle part of Theorem 3.1 is not established.","tokens_in":4552,"feed_emoji":"📐","tokens_out":8918,"duration_ms":82509,"temperature":0.7,"pith_summary":"The paper's aim is to show that the three-dimensional longest-edge bisection rule, read as “bisect any currently longest edge, breaking ties arbitrarily,” does not by itself prevent mesh degeneration. The paper constructs a one-parameter family of tetrahedra in which two valid bisections return a scaled copy of the starting tetrahedron with the shape parameter halved, so the process can be iterated indefinitely. Along the resulting admissible sequence the normalized volume $|T|/h_T^3$ tends to zero, one interior dihedral angle tends to zero, and another tends to $\\pi$. This matters because the two-dimensional theory is well behaved and practical tetrahedral schemes usually add marking or tie-breaking conventions; the paper isolates exactly what the unrestricted rule fails to guarantee.","feed_headline":"Longest-edge bisection can shrink tetrahedra to degeneracy","feed_subtitle":"Two valid bisections reproduce the shape at half scale, forcing one dihedral angle to zero and another to pi.","key_machinery":"The load-bearing object is the exact two-step recurrence $E(a) \\xrightarrow{\\text{bisect }AD} O(a) \\xrightarrow{\\text{bisect }BD} E(a/2)$. The first child has a unique longest edge, and the second child has two exactly tied longest edges, one of which leads back to the same family at half the parameter. This recurrence is what converts a local angular deterioration into a permanent geometric progression: every two steps halve the shape parameter, so the normalized volume and the angular gaps all decay like $2^{-k/2}$ without ever leaving the set of admissible longest-edge bisections.","core_discovery":"For $0<a\\le 1$, the paper defines a tetrahedron $E(a)$ whose squared edge lengths are $(a^2,a+a^2,2a+a^2,a,2a,2a)$, so the edge $AD$ is uniquely longest. Bisecting $AD$ and retaining the child $O(a)$ produces a tetrahedron in which the two edges $BD$ and $DC$ are exactly tied for longest; choosing $BD$ and bisecting it yields, after relabeling, a tetrahedron congruent to $E(a/2)$. Iterating from $E(1)$ therefore gives an admissible sequence $E(a_k)\\to O(a_k)\\to E(a_{k+1})$ with $a_k=2^{-k}$, and along this sequence $|E_k|/h_{E_k}^3\\to 0$, one dihedral angle $\\delta(a_k)\\to 0$, and another dihedral angle $\\Theta(a_k)\\to\\pi$. The paper presents this as a counterexample to the claim that the unrestricted rule “choose any longest edge” guarantees regularity in three dimensions.","pith_inferences":["A natural next step, not taken in the paper, would be to test whether simple fixed tie-breaking rules such as “always choose the tied edge with the shorter opposite edge” avoid this orbit, and whether any such rule can be proven to yield only finitely many similarity classes in three dimensions.","The recurrence $a\\mapsto a/2$ suggests a dynamical-systems reading of longest-edge bisection: classifying all two-step recurrent shape maps of this kind might reveal whether degenerate orbits are rare or unavoidable under generic tie rules.","Because the degeneration rate is geometric, adaptive algorithms that refine until an error indicator drops may keep refining these tetrahedra without improving the discretization, so the example is relevant beyond pure regularity theory."],"forward_implications":["In three dimensions, “bisect a longest edge” is not enough to guarantee nondegeneration: the paper exhibits an infinite admissible orbit on which the normalized volume tends to zero.","Any regularity or convergence theorem for tetrahedral longest-edge bisection must restrict tie-breaking or marking, not merely require that a longest edge be chosen.","The failure is simultaneous in all three standard measures: volume ratio, minimum dihedral angle, and maximum dihedral angle all deteriorate at the same geometric rate.","The same local orbit can be scheduled inside a global face-to-face conforming refinement by inserting finitely many waiting bisections of longer edges, so the degeneration can appear in conforming partitions as well."],"supporting_citations":[{"why":"Supplies the equivalence between the normalized-volume bound and other regularity criteria, which is how the paper measures degeneration.","marker":"[3]"},{"why":"Defines the tetrahedral minimum-angle condition whose violation is established by the vanishing dihedral angle.","marker":"[13]"},{"why":"Defines the maximum-angle condition for linear tetrahedral elements that the sequence violates as one angle approaches $\\pi$.","marker":"[15]"},{"why":"Documents the two-dimensional longest-edge bisection behavior that motivates the paper's question and contrasts with the three-dimensional failure.","marker":"[26]"}],"fun_headline_variants":["Longest-edge bisection leads to degenerate tetrahedra","Arbitrary tie-breaking in longest-edge bisection yields degeneration","Bisecting longest edges can shrink tetrahedra to flatness","A sequence of tetrahedra collapses under longest-edge bisection","Longest-edge rule fails: tetrahedra degenerate despite tie-breaking"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that formula (10), $\\cos\\Theta(a)=(2a-7)/\\sqrt{(4a+7)(8a+7)}$ for the dihedral angle of $O(a)$ at the edge $MC$, is correct; it is asserted as a direct calculation, and if it were wrong, the maximum-angle failure would not be established, though the volume and minimum-angle failures would remain.","fun_headline_variants_meta":{"raw":{"variants":["Longest-edge bisection leads to degenerate tetrahedra","Arbitrary tie-breaking in longest-edge bisection yields degeneration","Bisecting longest edges can shrink tetrahedra to flatness","A sequence of tetrahedra collapses under longest-edge bisection","Longest-edge rule fails: tetrahedra degenerate despite tie-breaking"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000674,"raw_usage":{"total_tokens":2991,"prompt_tokens":793,"completion_tokens":2198,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":409,"completion_tokens_details":{"reasoning_tokens":2111}},"tokens_in":409,"tokens_out":2198,"duration_ms":15944,"temperature":1.0,"reasoning_tokens":2111,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-28T00:22:04.970165+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a concrete value such as $a=1/64$, write down the coordinates of $O(a)$ from the definition, and compute the dihedral angle at edge $MC$ by standard linear algebra on the two face normals; compare with the value predicted by formula (10). If the numbers disagree, the maximum-angle part of Theorem 3.1 is not established.","supporting_citations":[{"cited_title":"Brandts, S","cited_arxiv_id":null,"evidence_quote":"Supplies the equivalence between the normalized-volume bound and other regularity criteria, which is how the paper measures degeneration."},{"cited_title":"Korotov and J","cited_arxiv_id":null,"evidence_quote":"Defines the tetrahedral minimum-angle condition whose violation is established by the vanishing dihedral angle."},{"cited_title":"Stynes, On faster convergence of the bisection method for all triangles,Math","cited_arxiv_id":null,"evidence_quote":"Documents the two-dimensional longest-edge bisection behavior that motivates the paper's question and contrasts with the three-dimensional failure."}],"review_version":1}