{"id":"a4214e45-c6bb-42be-b240-70eeb8603c2f","arxiv_id":"1908.06781","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For general regularizations of the planar visible fold, the paper proves existence and local uniqueness of a saddle-node bifurcation of limit cycles at the grazing parameter value, with distance scaling ε^{2k/(2k+1)}.","lead":"This mathematics paper proves that a specific smooth approximation of a piecewise-smooth 'visible fold' produces exactly one saddle-node bifurcation when a limit cycle just touches the switching surface. The proof uses consecutive blow-up transformations and gives a concrete scaling law that matches numerical simulations of a mass-spring friction oscillator.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 3.14's derivative profile is the quantitative core of Theorem 1.3(d)(ii) and hence of Theorem 1.7, but its proof is only sketched; the endpoint asymptotics of X'_C,0 need independent verification.","rationale":"I independently traced the proof of Theorem 1.7. The reduction to the implicit function theorem in Section 4 is clean and the nondegeneracy conditions (4.7)-(4.8) are satisfied provided Lemma 4.3 and Lemma 4.4 hold as stated. Lemma 4.3 in turn is a direct consequence of Lemma 3.17, whose quantitative content is exactly the derivative profile of Lemma 3.14. The proof of Lemma 3.13 itself (the S-shape of the central map) is present and appears internally consistent, but Lemma 3.14's endpoint estimates are asserted rather than proved. Since these estimates are the only source of the uniform negativity of X'' and the precise range of X', they are load-bearing for the uniqueness argument. I did not find other internal inconsistencies: the blowup construction is coherent, the scaling exponent 2k/(2k+1) is consistent across Theorem 1.3(b), Theorem 1.7, and the numerical slope of about 0.8 for k=2, and the limitations stated in Section 5 are compatible with the theorem as stated. The reader's conditionality is therefore appropriate: the result is plausible and well-structured, but the central quantitative lemma needs either a full analytic proof or a rigorous numerical verification before the uniqueness claim can be accepted with confidence.","tokens_in":36749,"tokens_out":19533,"duration_ms":186239,"concrete_test":"Verify Lemma 3.14 directly for a representative case, e.g. k=2, by numerically integrating the Chini-type system (3.34) and its first and second variational equations from Sigma_L to Sigma_R for the u-values corresponding to x1=-1-zeta and x1=-1+zeta as nu decreases (e.g. nu = 10^-3, 10^-4, 10^-5). Compute X'_C,0 and X''_C,0 from (3.37) and check that X'_C,0 stays within (-1+c,-c) and X''_C,0<-c for the chosen c, and that the endpoint values approach -1 and 0^- respectively. If the numerical profile matches, the concern is resolved; if not, Lemma 3.17 and Theorem 1.7 need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim Theorem 1.7 depends on the unique saddle-node construction in Section 4, which uses Lemma 4.3 and Lemma 4.5. Both rely on Theorem 1.3(d)(ii), specifically on the quantitative bounds X'_epsilon in (-1+c,-c) and X''_epsilon < -c^{-1}. These bounds are derived in Section 3.5 through Lemma 3.17, Lemma 3.16, and ultimately Lemma 3.14. Lemma 3.14 asserts that for any c>0 there is nu0 such that for all nu in (0,nu0], X'_C,0(x1) lies in (-1+c,-c) and X''_C,0(x1)<0 for x1 in [-1-zeta,-1+zeta]. The argument given is a sketch: for x1=-1+zeta it invokes 'a simple calculation' plus an odd-function variational equation to conclude X'_C,0 approaches -1, while for x1=-1-zeta it asserts convergence to 0^- by 'following the flow ... up close to the center manifold of pa'. The intermediate monotonicity is supplied by Lemma 3.13, which is proven, but the endpoint limits and the uniform negativity of X''_C,0 are not demonstrated in detail. If either endpoint asymptotics fails, or if X''_C,0 is not bounded away from zero on the interval, then Lemma 3.17's bound X''_epsilon < -c^{-1} can fail, the implicit function theorem in Lemma 4.5 may lose the nondegeneracy condition (4.7), and the uniqueness of the saddle-node bifurcation in Theorem 1.7 would be unsupported. This is a genuine gap in the proof, not merely a matter of presentation, because no alternative argument for these quantitative bounds is provided elsewhere in the manuscript.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies singularly perturbed planar systems that limit, as ε→0, to piecewise smooth systems with a visible fold singularity. The main technical result, Theorem 1.3, gives a detailed description of the transition map near the fold under general assumptions (A0)-(A2) on the regularization function, including contraction away from the grazing point, a derivative estimate in an O(ε^{2k/(2k+1)}) neighborhood of the grazing point, and a description of the derivative profile there. Theorem 1.7 then uses this transition map to prove, for a grazing repelling limit cycle satisfying (B1)-(B3), the existence of a locally unique saddle-node bifurcation of limit cycles at α = ε^{2k/(2k+1)} α_2(ε), with exactly two limit cycles below the bifurcation value and one at the value, and with Hausdorff convergence of the saddle-node orbit to the grazing cycle. The proof is built on two consecutive blowups, a partial linearization near the hyperbolic points p_L and p_R, an analysis of a Chini-type equation, and an implicit function argument. Section 2 applies the result to a mass-spring system on a moving belt with a Stribeck friction law and provides AUTO-based numerical evidence, including the predicted exponent 4/5 for k=2.","tokens_in":37117,"tokens_out":11178,"duration_ms":106344,"significance":"If the quantitative estimates in Section 3.5 are supplied in full, the paper would constitute the first rigorous proof of local uniqueness of the saddle-node bifurcation in the regularized grazing bifurcation for the visible fold, extending earlier work by Bonet-Revés and Seara from Sotomayor-Teixeira regularization functions to the class of asymptotic regularization functions satisfying (A1)-(A2). The consecutive-blowup framework is appropriate and general, and the paper gives explicit, falsifiable predictions: the scaling exponent ε^{2k/(2k+1)} in Theorem 1.7 is verified numerically, and the transition-map derivative profile in Theorem 1.3(d) is a concrete quantitative statement. The assumptions (A0)-(A2) and (B1)-(B3) do not appear to encode the conclusion; there is no circular parameter fitting. The application to the Stribeck friction oscillator is a genuine and well-motivated addition. However, the central derivative profile of the middle transition map, Lemma 3.14, is only sketched, and the chain-rule estimate that converts it into the uniform negativity bound in Lemma 3.17 is asserted rather than demonstrated.","major_comments":[{"comment":"The proof of Lemma 3.14 is only a sketch, and this is the quantitative core of Theorem 1.3(d)(ii). For the endpoint x1 = -1+ζ the argument invokes 'a simple calculation' together with an odd variational equation to conclude X'_C,0 approaches -1; for x1 = -1-ζ it asserts convergence to 0^- by 'following the flow ... up close to the center manifold of p_a' without any estimate. The uniform negativity of X''_C,0 on the whole interval is inferred from Lemma 3.13, but Lemma 3.13 only gives sign information, not a positive lower bound for |X''_C,0|. These facts are load-bearing: Lemma 3.17 needs X''_ε < -c^{-1} after the rescaling, and Lemma 4.5's nondegeneracy condition (4.7) plus the uniqueness of the saddle-node in Theorem 1.7 depend on that bound. An independent, detailed verification of both endpoint asymptotics and of a uniform bound on X''_C,0 is required.","section":"Section 3.5, Lemma 3.14 and Lemma 3.17"},{"comment":"The chain-rule estimate in Eq. (3.47) is not justified by the displayed argument. Using the asymptotics (3.44)-(3.45), the contributions to X''_ε coming from the second derivatives of X_R,ε and ~X_L,ε are of order ε^{-2k/(2k+1)} or larger, not O(1); they can be absorbed into the leading term only if those second derivatives are uniformly O(c) and (X_C,ε)'' is bounded away from zero. The proof says 'simple calculation' but does not track these terms. Since Eq. (3.47) is the bridge from Lemma 3.14 to Lemma 3.17 and hence to the nondegeneracy condition (4.7), a complete derivation with all remainders accounted for is needed.","section":"Section 3.5, Lemma 3.16, Eq. (3.47)"},{"comment":"The proof of the contraction estimate in Theorem 1.3(d)(i) is only sketched. The first part of the passage from Σ_out,L_L to Σ_in,C_R is described as 'standard and left out of this manuscript completely', and the second part is asserted to be contracting because of exponential contraction toward the center manifold of p_a. This estimate is used in Lemma 4.1 to prove that for α<0 there are exactly two fixed points and for α>0 there are none, which is part of the full statement of Theorem 1.7. Please supply the missing estimates or state and prove the needed contraction bound explicitly.","section":"Section 3.5, proof of Theorem 1.3(d)(i), and Section 4, Lemma 4.1"}],"minor_comments":[{"comment":"In the displayed forms of Ψ_L and Ψ_R, the function ~R_L is written as O(ρ^k_1), but the coordinate change should have the form 1 + O(ρ^k_1) to be invertible near the identity; please clarify the intended expansion.","section":"Section 3.2, Eqs. (3.19) and (3.23)"},{"comment":"There is a typo in the line 'XR(0, 1, 0, ) = 0': an extra comma appears after the third argument.","section":"Section 3.2, Eq. (3.26)"},{"comment":"The phrase 'C2 O(ε_1^{1/(2k+1)})-close' is awkward; it should be rephrased as 'C^2 and O(ε_1^{1/(2k+1)})-close' to avoid ambiguity.","section":"Lemma 3.6"},{"comment":"The paper says it will leave to the reader what 'sufficiently large' smoothness is, but the main theorems rely on C^2 estimates and an implicit function theorem argument. Please state a specific finite order of smoothness for the main results, or at least specify that C^3 or C^4 suffices for Theorem 1.7.","section":"Section 1.4 and throughout"},{"comment":"The subfigure references are inconsistent: the text refers to 'Fig. 5(d)' for the log-log plot of α*−α, while the caption labels that plot as '(b)', and the caption's '(d)' is the phase portrait. Please align the in-text references with the caption.","section":"Section 2, Figure 5 and surrounding text"},{"comment":"There is a typo in the phrase '[26, Thoerem 3.3]'; it should read 'Theorem'.","section":"Section 5, paragraph 'Comparison with previous results'"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the topic is appropriate. The main concern is that the quantitative estimates in Section 3.5, which are the backbone of Theorem 1.7, are not fully proved. I see no concern about novelty or attribution; the issue is purely one of missing technical detail in a load-bearing part of the proof. I would be willing to review a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper does something real: it gives the first rigorous proof that a grazing limit cycle in a piecewise smooth system becomes a locally unique saddle-node bifurcation under a broad class of regularizations, with a precise scaling exponent epsilon^{2k/(2k+1)}. That goes beyond Bonet-Reves and Seara's work, which had the same conclusion only for Sotomayor-Teixeira functions and without a proof. The blowup construction is coherent, the transition map estimates in Theorem 1.3(d) are genuinely new, and the numerical exponent 0.8024 matches the theoretical 4/5. The paper is honest about its limitations and does not overclaim. The citations look appropriate; the self-citations are to prior blowup methods that are directly relevant.\n\nThe soft spot is exactly the one you flagged. Lemma 3.14 is the quantitative core: it claims the derivative of the middle transition map X'_C,0 lies in (-1+c,-c) with X''_C,0<0 on the whole interval, and this is what makes the S-shaped graph and the unique saddle-node work. But the proof at the endpoints is only sketched: \"a simple calculation\" for one end, \"following the flow\" for the other, and the uniform negativity of the second derivative is asserted rather than shown. This is not a matter of taste. If that derivative profile fails, the nondegeneracy condition (4.7) can fail and the implicit function theorem argument in Lemma 4.5 collapses. The stress-test note is fair, and I would not soften it. That said, the gap looks fillable; the rest of the proof is detailed and the claim is plausible. A referee should ask for a complete calculation, not reject the paper on suspicion.\n\nMinor points: the paper says \"smooth\" means C^l with l sufficiently large without being precise, and the numerical artifacts behind Figure 5 are not documented. Neither is serious for the main argument.\n\nThe paper is for people working in geometric singular perturbation theory and regularized piecewise smooth systems, especially those studying grazing and boundary bifurcations. It is a solid piece of work that deserves a serious referee. I would send it out with a request for a rigorous proof of Lemma 3.14, and I would not be surprised if the paper came back accepted after that.","headline":"A serious blowup analysis that proves the first rigorous version of the grazing saddle-node bifurcation for general regularizations, but the quantitative derivative estimate at the heart of the uniqueness argument is underproved and needs referee scrutiny.","tokens_in":37682,"tokens_out":1583,"would_cite":true,"duration_ms":19896,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34E15","34C23","37G15"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every sufficiently small $\\epsilon>0$, a limit cycle grazing the discontinuity set in the singular limit unfolds into a locally unique saddle-node bifurcation of limit cycles at $\\alpha=\\epsilon^{2k/(2k+1)}\\alpha_2(\\epsilon)$, with…","keywords":["visible fold","piecewise smooth systems","regularization","singular perturbation","blowup method","saddle-node bifurcation","grazing bifurcation","friction oscillator"],"falsifier":"A direct numerical evaluation of the derivative $X'_{C,0}(x_1)$ of the middle transition map, defined through the Chini equation $v'=2u+v^{-k}$, for small $\\nu$ and $x_1\\in[-1-\\zeta,-1+\\zeta]$, would settle the key estimate: it must lie strictly between $-1$ and $0$ with negative second derivative; observing any value outside that interval for some small $\\nu$ would contradict Lemma 3.14 and with it the uniqueness part of Theorem 1.7.","tokens_in":36528,"feed_emoji":"🔄","tokens_out":9835,"duration_ms":92328,"temperature":0.7,"pith_summary":"This paper establishes a precise statement about what happens when a smooth ordinary differential equation approaches a piecewise smooth system with a visible fold. The main result is that if a repelling limit cycle of the upper vector field grazes the discontinuity line in the singular limit, then for every sufficiently small regularization parameter $\\epsilon>0$ the regularized system undergoes a locally unique saddle-node bifurcation of limit cycles, at a parameter value that differs from the grazing value by order $\\epsilon^{2k/(2k+1)}$, where $k$ is the decay rate of the regularization function above the fold. The proof relies on two consecutive blowup transformations that resolve the fold into hyperbolic and center-manifold building blocks, yielding the first detailed description of the transition map in a full neighborhood of the fold. As a consequence, the paper settles a claim left open in earlier treatments and applies the result to a mass-spring system with Stribeck friction, where numerical continuation confirms the predicted scaling.","feed_headline":"A grazing limit cycle becomes a unique saddle-node bifurcation","feed_subtitle":"For every small epsilon, two limit cycles meet and vanish at parameter offset epsilon^(2k/(2k+1))—the first rigorous proof.","key_machinery":"The machinery is the double blowup: first blow up the discontinuity line $y=\\epsilon=0$ to a cylinder, then blow up the nonhyperbolic point $T$, the imprint of the visible fold, to a sphere with weights $(2k,k,1)$. In the charts $(\\bar r=1)_1$ and $(\\bar\\epsilon=1)_2$, this desingularizes the fold: the critical manifold ends at a nonhyperbolic saddle $p_a$, while the grazing orbit enters and leaves through hyperbolic points $p_L$ and $p_R$. Partial linearizations near $p_L$ and $p_R$ are combined with a global analysis of the Chini equation $v'=2u+v^{-k}$, the reduced equation for the middle transition map, to obtain the S-shaped derivative profile in Lemmas 3.14 and 3.17. The saddle-node then follows by writing the Poincar\\'e map as $Q\\circ R^{-1}$ and applying the implicit function theorem in the scaled variables $x=\\epsilon^{2k/(2k+1)}x_2$, $\\alpha=\\epsilon^{2k/(2k+1)}\\alpha_2$.","core_discovery":"The paper's central claim is Theorem 1.7. Under assumptions (A0)-(A2) on the system and regularization, and (B1)-(B3) on a repelling limit cycle with a single quadratic tangency to the switching manifold, there exists, for all sufficiently small $\\epsilon>0$, a locally unique saddle-node bifurcation of limit cycles at $\\alpha=\\epsilon^{2k/(2k+1)}\\alpha_2(\\epsilon)$ with $\\alpha_2$ continuous. Locally, limit cycles exist precisely for $\\alpha\\leq\\epsilon^{2k/(2k+1)}\\alpha_2(\\epsilon)$: two for smaller $\\alpha$, one at the bifurcation value, and the saddle-node periodic orbit converges in Hausdorff distance to the grazing limit cycle $\\Gamma_0$ as $\\epsilon\\to 0$. This is obtained not from the discontinuous return map but from the regularized transition map itself: Theorem 1.3(d) shows that on an $\\epsilon^{2k/(2k+1)}$-neighborhood of the grazing point the map's derivative is monotone, crosses every value in $(-1,0)$ exactly once, and has second derivative uniformly negative, which is exactly the profile needed to separate two fixed points before the bifurcation and none after.","pith_inferences":["The transition-map argument is essentially local in the fold, so a natural extension is to higher-dimensional systems whose grazing is a planar quadratic tangency plus strong contraction; one expects the same $\\epsilon^{2k/(2k+1)}$ scaling for the saddle-node in that setting.","If the regularization function is non-monotone and produces a fold in the critical manifold, the same consecutive-blowup framework suggests that the unstable limit cycle can be continued past the tangency to a canard-like bifurcation, a case the paper explicitly leaves open.","The Chini equation $v'=2u+v^{-k}$ is the scalar model for the middle transition; a computer-assisted or analytic bound on its variational equation could replace the sketched proof of Lemma 3.14 and make the uniqueness proof fully checkable.","The power law $\\alpha_{SN}(\\epsilon)-\\alpha_* \\sim \\epsilon^{2k/(2k+1)}$ is a fingerprint of the regularization's decay rate, so measuring the saddle-node shift for different smoothing functions with known $k$ would directly test the theorem's quantitative claim."],"forward_implications":["For any regularization function satisfying (A1)-(A2), the grazing bifurcation in the regularized system is a genuine saddle-node of limit cycles at the $\\epsilon^{2k/(2k+1)}$-scale, where $k$ is the decay rate of the regularization above the fold.","Below the bifurcation value there are locally exactly two periodic orbits: one attracting whose $\\epsilon\\to 0$ limit has a sliding segment, and one repelling that converges to the repelling grazing cycle; above the value there are none.","The saddle-node periodic orbit converges in Hausdorff distance to the grazing limit cycle $\\Gamma_0$ as $\\epsilon\\to 0$, so the bifurcation is a true regularized unfolding of the piecewise-smooth grazing bifurcation.","In the mass-spring-on-belt example with a Stribeck friction law and a subcritical Hopf bifurcation, the theorem predicts a locally unique saddle-node of limit cycles near the grazing parameter $\\alpha_*$ for every small $\\epsilon$; the paper's AUTO computation gives a numerically observed slope of about 0.8024, matching the predicted exponent $4/5$ for $k=2$.","The transition-map description in Theorem 1.3(d) is sharper than previous treatments: in the middle region the map has a single point with any prescribed slope in $(-1,0)$, and the uniform curvature bound rules out additional saddle-nodes."],"supporting_citations":[{"why":"Supplies the local normal form (1.11) for the visible fold and the earlier analysis of the regularized grazing bifurcation whose existence claim Theorem 1.7 completes.","marker":"[3]"},{"why":"Establishes the blowup treatment of regularized piecewise smooth systems, including the directional charts and the division by common factors used in the proof of Theorem 1.3.","marker":"[32]"},{"why":"Provides the blowup method for nonhyperbolic fold and canard points and the center-manifold and transition-map techniques used at the nonhyperbolic point T.","marker":"[33]"},{"why":"Fenichel's persistence and foliation theory gives the invariant slow manifold S_epsilon and its stable foliation in Theorem 1.3(a).","marker":"[13]"},{"why":"Gives the invariant manifold near the switching manifold for regularized discontinuous systems, used in the statement of Theorem 1.3(a).","marker":"[6]"},{"why":"Identifies the regularized system as a singular perturbation problem and describes the critical manifold and reduced dynamics near the discontinuity set.","marker":"[38, 39]"},{"why":"Identifies the Chini equation v' = 2u + v^{-k}, the scalar reduced equation governing the middle transition map whose derivative profile is the quantitative core of the proof.","marker":"[42]"},{"why":"Supplies the saddle-node criterion for fixed points of the Poincaré map used to convert the fixed-point equation into a statement about limit cycles.","marker":"[45]"}],"fun_headline_variants":["Grazing limit cycles yield unique saddle-node bifurcation","Blowup analysis proves unique saddle-node from grazing cycles","Saddle-node bifurcation from regularized visible fold","Grazing cycle triggers unique saddle-node bifurcation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole proof hinges on the claimed derivative profile of the middle transition map near the fold—slope strictly between -1 and 0, with decreasing slope—which the paper supports only by a calculation sketch rather than a full proof.","fun_headline_variants_meta":{"raw":{"variants":["Grazing limit cycles yield unique saddle-node bifurcation","Blowup analysis proves unique saddle-node from grazing cycles","Saddle-node bifurcation from regularized visible fold","Grazing cycle triggers unique saddle-node bifurcation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001685,"raw_usage":{"total_tokens":6681,"prompt_tokens":952,"completion_tokens":5729,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":5673}},"tokens_in":568,"tokens_out":5729,"duration_ms":38448,"temperature":1.0,"reasoning_tokens":5673,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:35:46.854287+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct numerical evaluation of the derivative $X'_{C,0}(x_1)$ of the middle transition map, defined through the Chini equation $v'=2u+v^{-k}$, for small $\\nu$ and $x_1\\in[-1-\\zeta,-1+\\zeta]$, would settle the key estimate: it must lie strictly between $-1$ and $0$ with negative second derivative; observing any value outside that interval for some small $\\nu$ would contradict Lemma 3.14 and with it the uniqueness part of Theorem 1.7.","supporting_citations":[{"cited_title":"Bonet-Rev´ es and T","cited_arxiv_id":null,"evidence_quote":"Supplies the local normal form (1.11) for the visible fold and the earlier analysis of the regularized grazing bifurcation whose existence claim Theorem 1.7 completes."},{"cited_title":"Uldall Kristiansen and S","cited_arxiv_id":null,"evidence_quote":"Establishes the blowup treatment of regularized piecewise smooth systems, including the directional charts and the division by common factors used in the proof of Theorem 1.3."},{"cited_title":"Krupa and P","cited_arxiv_id":null,"evidence_quote":"Provides the blowup method for nonhyperbolic fold and canard points and the center-manifold and transition-map techniques used at the nonhyperbolic point T."},{"cited_title":"Fenichel","cited_arxiv_id":null,"evidence_quote":"Fenichel's persistence and foliation theory gives the invariant slow manifold S_epsilon and its stable foliation in Theorem 1.3(a)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the invariant manifold near the switching manifold for regularized discontinuous systems, used in the statement of Theorem 1.3(a)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies the Chini equation v' = 2u + v^{-k}, the scalar reduced equation governing the middle transition map whose derivative profile is the quantitative core of the proof."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the saddle-node criterion for fixed points of the Poincaré map used to convert the fixed-point equation into a statement about limit cycles."}],"review_version":1}