{"id":"30cddba2-3b0e-4bc7-928e-22fe3fd9259d","arxiv_id":"1908.02693","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Transverse 3-parameter unfoldings of vector fields with 'ears' or 'glasses' separatrix graphs carry the numerical invariant φ = -ln ρ / ln λ, so every such family is structurally unstable.","lead":"This paper proves that generic three-parameter families of vector fields on the sphere built around the phase portraits 'ears' and 'glasses' are structurally unstable, because their classification carries an invariant number built from the saddles' eigenvalue ratios. It adds two new examples alongside the known 'tears of the heart' case and reworks the classical saddle-loop bifurcation into a reusable tool.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central proof rests on imported estimate (3) from [16, Lemma 4]/[18, Lemma 6]; the paper states it as a black box and does not verify the essential uniform O(1) in β, so the invariant's validity is inherited, not established.","rationale":"I read the paper in good faith. The central claim (Theorem 4) is that for the 'ears' and 'glasses' manifolds, φ = -ln ρ/ln λ is robustly invariant under weak topological equivalence with Sep-tracing. The proof has two quantitative inputs: (i) Theorem 3, which gives the asymptotic relation (1)-(2) between two equivalent unfoldings of a saddle loop, and (ii) Lemma 1, the synchronizing subfamily computation. The reader's weakest_assumption was (i) and Eq. (10) from (ii). My reading agrees that (i) is the single most load-bearing: all limit formulas in Sec. 4.2 and the final equality in Sec. 4.3 are obtained by dividing the limits from Theorem 3, so if (3) fails in the O(1)-uniform sense, the invariant is not forced. The paper is transparent about importing (3) from [16, Lemma 4] and [18, Lemma 6] but provides no statement of the lemma or its proof, and the O(1) uniformity in β is essential. This is a missing support rather than an internal contradiction. I do not think Eq. (10) is equally load-bearing: the synchronizing lemma's conclusion (9) holds for any relation ln δ = K(α) ln ε + O(1) with K bounded away from 0 and ∞, so the exact product λρ is not needed for the proof; the form of φ comes from (7)-(8), which depend only on the growth law (3). Other asserted-but-unproved statements (M embedded codim-3, topologically distinguished) are secondary for the local invariant and are standard perturbations of the given conditions. The verdict CONDITIONAL is appropriate: the result is likely correct, but the paper should either re-derive or fully state the imported estimate, and expand the terse assertions in Sec. 4.1.2. I therefore recommend no change to the reader's verdict.","tokens_in":13869,"tokens_out":21865,"duration_ms":229304,"concrete_test":"Independently verify the imported estimate (3) by (a) extracting the exact statement and hypotheses of [16, Lemma 4] and [18, Lemma 6] and checking that the hypotheses hold in the setting of Theorem 3 (saddle loop with characteristic λ>1, a winding separatrix, transverse unfolding), in particular that the O(1) is uniform for ‖β‖<C; or (b) numerically simulating a model family (e.g., a polynomial vector field with a saddle loop and a winding separatrix), computing ε_n(β) for n up to large N and β in a grid, and testing whether sup_β |ln(-ln ε_n(β)) - n ln λ(0,β)| remains bounded as n grows. If a proof or numerical check confirms the bound, the concern is settled.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative engine is Eq. (3): ln(-ln ε_n(β)) = n ln λ(0,β) + O(1), imported verbatim from [16, Lemma 4] and [18, Lemma 6] (Sec. 3.2). Theorem 3, and hence the limit formulas (7)-(8) in Sec. 4.2 and the equality φ(v0)=φ̃(ṽ0), all reduce to this estimate. The paper gives no proof, no statement of the lemma's hypotheses, and no discussion of the uniformity of the O(1) term in β; yet (4), (7), (8) require that the bound is uniform for ‖β‖<C. If the O(1) constant grows as β→0, the limits (2), (7), (8) and the synchronizing argument in Sec. 4.3 break down. The authors' candid remark that Theorem 3 'reinterprets results of [16]' (Sec. 3.1) makes the dependence explicit but does not discharge it. Overlapping authorship between this paper and [16,18] increases the need for an independent check, not as evidence of error, but because the new examples' invariant inherits the validity of that lemma. Secondary gaps (the one-sentence proof that M is an embedded codim-3 submanifold and topologically distinguished, Sec. 4.1.2; and the terse derivation of (10) in the ears case) are less load-bearing: the synchronizing lemma (9) only needs a power-law relation with logarithmically bounded exponent, not the exact product λρ.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies global bifurcations in generic 3-parameter families of vector fields on the two-sphere. It introduces two new classes of degenerate vector fields, called 'ears' and 'glasses', and proves (Theorems 1 and 4) that the function φ(v) = -ln ρ(v)/ln λ(v) is a robust invariant for transverse 3-parameter unfoldings with respect to weak topological equivalence with Sep-tracing. Consequently, all such unfoldings are structurally unstable, and an open set of non-local 3-parameter families carries a numerical invariant (Theorem 2). The proof reduces the invariance to asymptotic comparisons of separatrix splitting parameters around the 'ears' and 'glasses' graphs, using the theory of 'sparkling separatrix connections' imported from prior work.","tokens_in":14018,"tokens_out":12327,"duration_ms":121499,"significance":"If correct, the paper provides two new examples of locally generic structurally unstable 3-parameter families, complementing the 'tears of the heart' example and simplifying the construction because the separatrix graph is not a polycycle. The invariant is explicitly computable and the overall reduction from a dynamical equivalence to an algebraic identity is transparent. The paper also contains a clean reusable statement (Theorem 3) for the saddle-loop bifurcation. However, the central quantitative estimate (3) is not proved in the manuscript and is only cited, with no explicit statement of hypotheses or uniformity, and the proof of the synchronizing lemma is compressed. These points need to be addressed.","major_comments":[{"comment":"The paper's central asymptotic estimate ln(-ln ε_n(β)) = n ln λ(0,β) + O(1) is imported from [16, Lemma 4] and [18, Lemma 6] as a black box. The text asserts that the O(1) term is uniformly bounded in a neighborhood {0<ε<C, ‖β‖<C}, but no proof is given and the hypotheses of the cited lemmas are not stated. This uniformity is essential because it is used in the limit formulas (7)-(8) and in the synchronizing argument of Sec. 4.3; if the O(1) constant grew as β→0, those limits would fail. The authors should either prove (3) with the required uniformity or state and prove a precise lemma, giving a full reference to the exact statement in [16] or [18] and explaining how the uniformity follows.","section":"Sec. 3.2, Eq. (3)"},{"comment":"The proof that the synchronizing subfamily E satisfies ln(-ln ε)/ln(-ln δ) → 1 is too compressed. The derivation of Eq. (10), ln δ = λ(α)ρ(α) ln ε + O(1), relies on an assertion about correspondence maps of a chain of saddles that is not proved in detail; the text says 'for a chain of maps one should multiply the exponents' and cites references that do not clearly cover the composition of a loop and a bridge map. In the 'ears' case the argument is reduced to two displayed equalities followed by 'Similarly'. Since Lemma 1 is the step that forces φ(v0) = φ̃(ṽ0), a fuller proof is needed.","section":"Sec. 4.3, Lemma 1"}],"minor_comments":[{"comment":"The word 'arbitrafily' is a typo for 'arbitrarily' in the description of metrically generic 3-parameter families.","section":"Sec. 1, p.2"},{"comment":"The citation 'see e.g. [16, Lemma 5, 10, Lemma 1]' appears to be miscopied, since reference [10] concerns the Krylov–Bogolyubov procedure and does not contain a saddle correspondence map. Please correct the citation or explain which lemma supplies the saddle map estimate.","section":"Sec. 4.3, proof of Lemma 1"},{"comment":"The claim that M is an embedded Banach submanifold of codimension 3 is justified in one sentence by taking ψ = (ε,σ,δ). The full rank of this map is not demonstrated; please add a brief argument showing that the three splitting parameters are independent.","section":"Sec. 4.1.2"},{"comment":"The step 'Taking logarithms of both sides, we get (9)' is not immediate. Since (10) is an asymptotic relation for the logarithms themselves, please add one line showing that ln(-ln δ) = ln(-ln ε) + ln(λρ) + o(1), which gives the desired ratio.","section":"Sec. 4.3, after Eq. (10)"},{"comment":"The definition of 'topologically distinguished' could be phrased more clearly: it should state explicitly that there exists a neighbourhood U of M such that no vector field in U\\M is orbitally topologically equivalent to a vector field in M.","section":"Sec. 2.2.1, Definition 5"}],"recommendation":"major_revision","confidential_remarks":"The central estimate (3) is imported from [16] and [18], which share authorship with the present paper. This makes it particularly important that the manuscript state the exact result and its uniformity, since a reader cannot easily separate the new contribution from the prior work. I do not see evidence of circularity, but the reliance is heavy and should be made fully explicit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: the central claim is a new result, not a repackaging. Ears and glasses are the first non-polycycle separatrix graphs shown to have generically structurally unstable 3-parameter unfoldings, with -lnρ/lnλ as a numerical invariant. The synchronizing-subfamily trick is a new device, and Theorem 3 is a useful packaging of the sparkling-connection estimate from tears of the heart.\n\nWhat it does well: the logic from Theorem 3 to the equality φ(v0)=φ̃(ṽ0) is transparent. Equations (7) and (8) come from applying the packaged theorem to the two loops, their ratio is the ratio of φ's, and Lemma 1 forces the synchronizing ratios to 1. I could follow this part line by line. The paper is also honest about what it imports: it explicitly says Theorem 3 reinterprets [16].\n\nSoft spots: the main one is real. The engine is (3), imported verbatim from [16, Lemma 4] and [18, Lemma 6], with overlapping authorship and no statement of hypotheses or proof of the uniform O(1) in β. The limit formulas and the synchronizing argument all reduce to that uniformity. This does not make the paper unserious, but it means the new invariant inherits its validity from an external lemma. A referee should check that lemma or ask for a sketch in the paper. Secondary: M being embedded codim-3 and topologically distinguished is asserted in one or two sentences. The codim-3 part is standard; the topological distinction is plausible given the structurally stable complement, but since it underpins Theorem 2, it deserves a paragraph. The ears case of Lemma 1 is compressed relative to glasses; the missing step is routine but should be written out.\n\nThe stress-test note overstates the dependence on the exact product λρ in (10): the synchronizing lemma only needs a power-law with logarithmically bounded exponent. So that particular worry is not the load-bearing one; the load-bearing is (3).\n\nWho this is for: specialists in planar global bifurcation theory. Someone outside that area will likely not need it. It deserves a serious referee: the claim is important within the subfield, the proof architecture is checkable, and the gaps are fillable. I'd send it out.","headline":"New non-polycycle examples with a real numerical invariant and a clean proof architecture; the main risk is the imported sparkling-connection estimate, which a referee should verify.","tokens_in":14772,"tokens_out":2249,"would_cite":true,"duration_ms":23583,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37C15","37C29","37G15"],"pacs":[],"model":"deepseek-v4-flash","headline":"Ears and glasses vector fields are structurally unstable under weak equivalence, with a ratio of saddle characteristic numbers as the invariant.","keywords":["structural instability","weak topological equivalence","Sep-tracing","numerical invariants","saddle loop bifurcation","separatrix graphs","ears and glasses","planar vector fields"],"falsifier":"Compute the splitting parameters $\\varepsilon$ and $\\delta$ along the synchronizing subfamily of an explicit ears or glasses unfolding and compare $\\ln\\delta$ with $\\lambda(\\alpha)\\rho(\\alpha)\\ln\\varepsilon$: a deviation larger than $O(1)$ as $\\varepsilon,\\delta\\to0$ would break Eq. (10) and with it the invariant. The direct decisive test would be to exhibit two unfoldings that are weakly topologically equivalent with Sep-tracing but whose central fields have different values of $-\\ln\\rho/\\ln\\lambda$; the theorem asserts such a pair cannot exist.","tokens_in":13473,"feed_emoji":"🌀","tokens_out":9995,"duration_ms":99894,"temperature":0.7,"pith_summary":"This paper proves that generic three-parameter unfoldings of vector fields on the two-sphere that pass through one of two degenerate configurations—the separatrix graphs called “ears” and “glasses”—are structurally unstable even under the very flexible notion of weak topological equivalence with Sep-tracing. The obstruction is a numerical invariant: for a degenerate field the ratio $\\varphi(v)=-\\ln\\rho(v)/\\ln\\lambda(v)$, where $\\lambda$ and $\\rho$ are the characteristic numbers of its two saddles, must be preserved by any such equivalence. Because $\\varphi$ takes every positive value and has nonzero derivative on the codimension-three manifold $M=E\\sqcup G$, the equivalence classes form a continuum and no transverse three-parameter unfolding can be stable. This adds two new configurations to the known phenomenon of locally generic structurally unstable three-parameter families, previously seen only for the “tears of the heart” polycycle.","feed_headline":"Ears and glasses vector fields are structurally unstable","feed_subtitle":"Their 3-parameter unfoldings carry a ratio invariant that no weak equivalence can erase.","key_machinery":"The engine is the asymptotic law for sparkling separatrix connections: when a saddle loop is broken with splitting parameter $\\varepsilon>0$, the winding separatrix forms a connection with the stable separatrix after $n$ turns exactly for parameters satisfying $\\ln(-\\ln\\varepsilon_n(\\beta))=n\\ln\\lambda(0,\\beta)+O(1)$, with the error uniform in the transverse parameter $\\beta$. This turns the discrete winding count into the logarithmic coordinate $\\ln(-\\ln\\varepsilon)/\\ln\\lambda$, and a homeomorphism of parameter spaces preserves the ordered sequence of these connections, forcing the difference of these coordinates for two equivalent families to stay bounded. The second estimate, $\\ln\\delta=\\lambda(\\alpha)\\rho(\\alpha)\\ln\\varepsilon+O(1)$ on the synchronizing subfamily, is where the product $\\lambda\\rho$ enters and fixes the exact form of the invariant as $-\\ln\\rho/\\ln\\lambda$.","core_discovery":"The central claim is Theorem 4: on the union $M=E\\sqcup G$ of vector fields with “ears” or “glasses” separatrix graphs, the function $\\varphi(v)=-\\ln\\rho(v)/\\ln\\lambda(v)$ is a robustly invariant function for weak topological equivalence with Sep-tracing. The manifold has codimension three and is topologically distinguished, so every transverse three-parameter unfolding of such a field is structurally unstable, and the last two conclusions of the theorem—$\\varphi(M)=\\mathbb{R}_+$ and $d\\varphi\\neq0$—show the invariant is genuinely non-degenerate. The proof compares two equivalent unfoldings, applies the growth law for sparkling separatrix connections to each of the two broken loops to obtain logarithmic splitting asymptotics, and then restricts to a synchronizing subfamily in which the two splitting parameters are linked by a correspondence map; the limit of the ratio of log-log splitting parameters gives $\\lambda\\rho$, which forces $\\varphi(v_0)=\\tilde\\varphi(\\tilde v_0)$.","pith_inferences":["The same two-loop-plus-bridge architecture could plausibly be stacked: each additional loop that feeds a bridge would force another ratio of log-characteristic numbers, so configurations of this shape may give families with many independent invariants at finite codimension.","The invariance argument never invokes Hölder regularity of the parameter homeomorphism, unlike the classical saddle-loop modulus; this suggests ears and glasses are unstable under a strictly weaker equivalence than the saddle-loop example, a consequence the authors do not draw explicitly.","The two asymptotic estimates (3) and (10) could be checked numerically on a concrete unfolding; since they are imported rather than proved here, a direct measurement of their $O(1)$ terms would test the mechanism without waiting for a full proof."],"forward_implications":["Every transverse three-parameter unfolding of an ears or glasses vector field is structurally unstable with respect to weak topological equivalence with Sep-tracing.","The function $\\varphi(v)=-\\ln\\rho(v)/\\ln\\lambda(v)$ becomes a numerical invariant on an open set of non-local three-parameter families, so the instability reaches beyond germs at the degenerate field.","Since $\\varphi$ takes all positive values and $d\\varphi\\neq0$, the classification of these unfoldings has a continuum of distinct classes.","The ears and glasses configurations join the “tears of the heart” polycycle as known codimension-three degeneracies whose generic unfoldings escape structural stability under this equivalence."],"supporting_citations":[{"why":"Supplies the growth law for sparkling separatrix connections and the tears-of-the-heart construction that this paper adapts to ears and glasses.","marker":"[16]"},{"why":"Gives the more precise form of the growth estimate used as Eq. (3).","marker":"[18]"},{"why":"Supplies the lemma that a hyperbolic saddle's correspondence map is close to $x\\mapsto Cx^{\\mu}$, used to derive the synchronizing estimate Eq. (10).","marker":"[10]"},{"why":"Provides the classical saddle loop bifurcation analysis that Section 3 revisits and that motivates the splitting-parameter setup.","marker":"[9]"}],"fun_headline_variants":["Ears and glasses: new structurally unstable vector fields","3-parameter unfoldings of ears/glasses are structurally unstable","Tears of the heart now joined by ears and glasses","Structural instability spreads to ears and glasses"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the imported growth law $\\ln(-\\ln\\varepsilon_n(\\beta))=n\\ln\\lambda(0,\\beta)+O(1)$ for sparkling separatrix connections, with the $O(1)$ term uniform in the parameters; if that estimate, or the analogous synchronizing estimate $\\ln\\delta=\\lambda\\rho\\ln\\varepsilon+O(1)$, fails, the forced equality of the invariants no longer follows.","fun_headline_variants_meta":{"raw":{"variants":["Ears and glasses: new structurally unstable vector fields","3-parameter unfoldings of ears/glasses are structurally unstable","Tears of the heart now joined by ears and glasses","Structural instability spreads to ears and glasses"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000132,"raw_usage":{"total_tokens":1083,"prompt_tokens":849,"completion_tokens":234,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":465,"completion_tokens_details":{"reasoning_tokens":170}},"tokens_in":465,"tokens_out":234,"duration_ms":2881,"temperature":1.0,"reasoning_tokens":170,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:41:00.410009+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the splitting parameters $\\varepsilon$ and $\\delta$ along the synchronizing subfamily of an explicit ears or glasses unfolding and compare $\\ln\\delta$ with $\\lambda(\\alpha)\\rho(\\alpha)\\ln\\varepsilon$: a deviation larger than $O(1)$ as $\\varepsilon,\\delta\\to0$ would break Eq. (10) and with it the invariant. The direct decisive test would be to exhibit two unfoldings that are weakly topologically equivalent with Sep-tracing but whose central fields have different values of $-\\ln\\rho/\\ln\\lambda$; the theorem asserts such a pair cannot exist.","supporting_citations":[{"cited_title":"Global bifurcations in the two-sphere: a new perspective","cited_arxiv_id":null,"evidence_quote":"Supplies the growth law for sparkling separatrix connections and the tears-of-the-heart construction that this paper adapts to ears and glasses."},{"cited_title":"Bifurcations of the polycycle \"tears of the heart\": multiple numerical invariants","cited_arxiv_id":"1808.07459","evidence_quote":"Gives the more precise form of the growth estimate used as Eq. (3)."},{"cited_title":"Convergence of the Krylov–Bogolyubov Procedure in Bowan’s Example","cited_arxiv_id":null,"evidence_quote":"Supplies the lemma that a hyperbolic saddle's correspondence map is close to $x\\mapsto Cx^{\\mu}$, used to derive the synchronizing estimate Eq. (10)."},{"cited_title":"On the saddle loop bifurcation","cited_arxiv_id":null,"evidence_quote":"Provides the classical saddle loop bifurcation analysis that Section 3 revisits and that motivates the splitting-parameter setup."}],"review_version":1}