{"id":"0666d3dd-eb8b-49f1-8808-46e7ad9ef8f9","arxiv_id":"2411.18096","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For all positive integers n, the limit wave speed c0(h) of periodic traveling waves of the perturbed generalized KdV equation is monotonically increasing in the Hamiltonian parameter h, and at most one such wave exists for each speed.","lead":"This paper proves that the limiting wave speed of periodic traveling waves in a family of perturbed KdV equations changes monotonically with the wave amplitude for every positive nonlinearity power. It settles a conjecture that had been open for all powers above four.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Monotonicity direction of c0(h) is not independently secured: it rests on a sign-corrected external theorem and directly contradicts the prior 'decreasing' results for n=1..4 cited in the introduction.","rationale":"The reader identified the sign correction of Theorem B as the weakest assumption. I agree this is critical, but the concern is wider: even if the sign correction is applied exactly as the authors intend, the paper still has an unresolved contradiction with the cited literature reporting decreasing c0(h) for n=1..4. A direct numerical computation of F_n(h) for small n would settle the mathematical direction and simultaneously force the authors to explain whether prior results concern a different quantity. This makes the paper correctly CONDITIONAL: the internal algebra is coherent and likely correct, but the central direction is not independently anchored and the relationship to prior work is not established. The reader's verdict need not change, so I mark UNCHANGED rather than moving to a harsher category. I did not find a clear internal algebraic error in the main calculation; the typos (e.g., fn(u,w) instead of fn(u,v) in (3.9)) appear cosmetic. The endpoint values of F_n in Lemma 3.4, if correct, would favor the decreasing direction, but they are quoted without derivation, so the numerical check remains the most direct way to remove the ambiguity.","tokens_in":13281,"tokens_out":11859,"duration_ms":103773,"concrete_test":"Fix n=1,2,3,4 and compute F_n(h) numerically over a grid of, say, 20 values of h in the interval (-n(n+1)^(2/n)/(2(n+2)), 0), using high-precision quadrature of An(h)=2∫_{α(h)}^{β(h)} u^n sqrt(u^2 - 2u^{n+2}/((n+1)(n+2)) + 2h) du and A0(h)=2∫_{α(h)}^{β(h)} sqrt(u^2 - 2u^{n+2}/((n+1)(n+2)) + 2h) du. If any sequence F_n(h_i) increases with h_i, then the monotonicity theorem is false and the sign-corrected application of Theorem B is wrong. If all sequences decrease, the direction is confirmed, and the prior 'decreasing c0(h)' statements must be explicitly reconciled as referring to a different normalization, parameter, or definition of the limit wave speed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the claim in Theorem 3.6 that c0(h) is increasing, which follows from F_n(h) being decreasing. This direction is obtained in 'Conjecture 3.2' through Theorem B, a result whose original version in [9] had a sign error, corrected only in [25]. The paper quotes the correction but gives no independent derivation, so a misreading of that correction would flip Theorem 3.6 from increasing to decreasing. The stakes are visible in the introduction: the authors state that for n=1,2,3,4, prior work [1,2,11] proved c0(h) is decreasing. If those results use the same definition of c0(h), they cannot coexist with Theorem 3.6, since a decreasing F_n forces an increasing c0 via c0=1/(F_n-1). The endpoint values in Lemma 3.4 do suggest F_n decreases from n+1 at the center to 2(n+1)(n+2)/(3n+4) at the homoclinic orbit, but those endpoints are quoted from [26] without proof. Thus the entire monotonicity direction, and with it the resolution of the open problem and the uniqueness claim, depends on a sign convention plus a literature reconciliation that the manuscript does not provide.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the perturbed generalized KdV equation (1.3) and claims three results: (i) the ratio of Abelian integrals F_n(h)=A_n(h)/A_0(h) is strictly decreasing on the periodic annulus (-n(n+1)^(2/n)/(2(n+2)),0); (ii) consequently the limit wave speed c0(h) is monotonically increasing there, answering an open problem of Yan et al. and a conjecture of Ouyang et al.; (iii) equation (1.3) has at most one isolated periodic traveling wave for each wave speed. The monotonicity proof is algebraic, using the involution of the Hamiltonian level curves and a criterion of Liu et al. with a sign correction from Wei et al. The paper also gives an n=5 numerical example of a limit cycle and a plot of c0(h).","tokens_in":13567,"tokens_out":18837,"duration_ms":153925,"significance":"If the PDE-level reduction were correct, the paper would resolve a natural open problem for all n with a parameter-free computation: the proof of Conjecture 3.2 is explicit, the algebraic identities in Lemma 3.1 are checked in detail, and the numerical n=5 example is consistent with the claimed monotonicity of c0(h). The use of the ratio of Abelian integrals, rather than case-by-case arguments, is a methodological strength. However, the paper's central claims are not yet supported as written because the derivation of the reduced system (2.6) from the original PDE appears inconsistent, and the new 'increasing' conclusion directly contradicts the cited earlier 'decreasing' results for n=1..4 without any reconciling discussion.","major_comments":[{"comment":"The derivation of the reduced system is not self-consistent. Substituting U=c^{1/n}u and ξ=η√c (as stated) into (2.3) gives, after multiplying by c^{1-1/n}, -c^2 u + c^2 u^{n+1}/(n+1) + u_{ηη} + ε(c^{1/2}u_η + c^{-1/2}u_{ηηη})=0, not (2.4). If the intended scaling is ξ=η/√c, then (2.4) is recovered, but the third equation of the singular-perturbation system should contain ε/√c y rather than ε√c y. In either case the slow-manifold reduction to (2.6) does not follow from (2.5): from (2.5) one obtains y' = u - u^{n+1}/(n+1) - ε√c y (or, with the corrected scaling, -ε/√c y), whereas (2.6) contains +ε√c(u^n - 1 - 1/c)y. Because the Abelian integral (2.11) is computed from (2.6), the connection between the original PDE (1.3) and the monotonicity theorem is not established as written.","section":"Section 2, Eqs. (2.4)-(2.6)"},{"comment":"The introduction states that for n=1,2,3,4 the earlier works [1,2,11] proved c0(h) is decreasing, whereas Theorem 3.6 proves c0(h) is increasing for all n. Since c0(h)=1/(F_n(h)-1) by (3.18) and F_n(h) is decreasing, these statements cannot both hold with the same definition of c0(h) and the same parameter h. The paper does not reconcile this contradiction. The authors should either identify a different normalization or branch (e.g., u<0) in the earlier papers, or explain why the earlier 'decreasing' results do not apply; otherwise the claimed resolution of the open problem is not credible.","section":"Section 1 vs. Theorem 3.6"},{"comment":"Theorem 3.3 asserts that (1.3) has at most one isolated periodic traveling wave for all positive integers n, but the proof only analyzes the periodic annulus around (\\sqrt[n]{n+1},0), and the remark correctly limits the conclusion to 'when u(x,t)>0 or u(x,t)<0'. For even n there are two centers, (\\pm\\sqrt[n]{n+1},0), and by the symmetry of (2.6) a limit cycle around each center can coexist. Thus the global statement of Theorem 3.3 is stronger than what is proved; the theorem should be restated as 'at most one such wave in each component u>0 and u<0'.","section":"Section 3, Theorem 3.3 and Remark"},{"comment":"The monotonicity direction of F_n(h) is inherited from Theorem B of Liu et al. [9] with a sign correction quoted from [25]. The manuscript gives only a terse remark about the corrected sign and does not state the full corrected theorem or map its variables to the system (2.12), so the reader cannot verify that the direction has been applied correctly. Similarly, the endpoint values used in (3.17)-(3.20) are quoted from Yan et al. [26] without proof. These external results are load-bearing for the c0(h) monotonicity claim, and the paper should state them precisely or provide self-contained derivations.","section":"Section 3, Theorem B Remark and Lemma 3.4"}],"minor_comments":[{"comment":"The notation 'U = n√c u' is ambiguous; it should be written U = \\sqrt[n]{c}\\,u to avoid confusion.","section":"Section 2, scaling line"},{"comment":"The second displayed formula for J0(0) should be J_n(0); as printed, the two formulas for J0(0) are inconsistent.","section":"Lemma 3.4"},{"comment":"In the first line of (3.9), 'fn(u,w)' should read 'fn(u,v)' in both occurrences.","section":"Eq. (3.9)"},{"comment":"There is a duplicated word: 'there is at most one one limit cycle'; the same duplication appears in the conclusion ('when when').","section":"Remark after Theorem 3.3"},{"comment":"The phrase 'closely approximates a closed trajectory' is vague; a quantitative measure, such as the maximum gap after one period, would strengthen the numerical evidence for the limit cycle.","section":"Section 4, numerical simulation"},{"comment":"The statement 'This document does not have any associated manuscript' appears to mean that there is no associated data and should be reworded accordingly.","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The algebraic monotonicity proof has the appearance of a solid, self-contained computation, and the n=5 numerical test is consistent. However, the paper is not ready for publication: the derivation of the reduced system (2.5)-(2.6) contains apparent algebraic/scaling errors that break the link between the PDE and the Abelian integral, and the contradiction with the cited earlier 'decreasing' results for n=1..4 is unresolved. Before resubmission, the authors should repair the reduction, clarify the sign convention in the quoted theorem, and restate Theorem 3.3 with the correct branch qualification."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proves that the ratio F_n(h)=A_n/A_0 is strictly decreasing on the periodic annulus for every positive integer n, and uses that to show the limit wave speed c0(h) is increasing. The algebraic identity in Lemma 3.1 is the genuinely new piece—it yields a compact proof of the monotonicity that previously had been checked only for n=1..4. The authors also give a simpler uniqueness proof via the Christopher–Li criterion. That is a real contribution.\n\nBut the paper has a conspicuous unresolved tension. The introduction cites prior work [1,2,11] showing c0(h) is decreasing for n=1,2,3,4. The present Theorem 3.6 says increasing. Since c0 = 1/(F_n - 1), a decreasing F_n forces an increasing c0, so the two claims cannot both hold under the same definition of c0. The paper never addresses this. Either the previous results used a different convention (which should be said explicitly), or there is a sign/coherence problem in one of the two lines. This is not a minor formatting issue; it is the load-bearing conclusion.\n\nThe proof of the monotonicity itself is algebraically coherent and the n=5 numerical figure supports the increasing direction. The reliance on Theorem B from Liu et al., whose statement had to be corrected in [25] for a sign, is a bit fragile—the authors adopt the correction without independent derivation. That is acceptable if the correction is right, but given the contradiction with the earlier literature, the readers need more assurance.\n\nAlso, Theorem 3.3 as stated claims the equation has at most one isolated periodic traveling wave, but the proof only gives at most one per wave speed c; the quantification should be made explicit. Lemma 3.4 quotes endpoint values from [26] without proof; that's a minor complaint since it is cited.\n\nFor all that, the core derivation is neat and the result, if the direction is sorted out, is a solid advance. The paper deserves a serious referee, but the referee should insist on a reconciliation with the existing decreasing results before it is accepted.\n\nRecommendation: send it to review, with a request to fix the contradiction and the quantification.","headline":"A clean Abelian-integral proof with a real gap: no reconciliation with prior 'decreasing' c0(h) results for n=1..4, so the central claim is not yet settled.","tokens_in":14044,"tokens_out":4834,"would_cite":false,"duration_ms":38010,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34C05","34C07","34C08","37G15"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every positive integer n, the limit wave speed of periodic traveling waves of the perturbed generalized KdV equation is monotonically increasing, and at most one isolated periodic wave exists for each wave speed.","keywords":["perturbed generalized KdV equation","periodic traveling wave","limit wave speed","Abelian integral ratio","monotonicity","geometric singular perturbation theory","limit cycle uniqueness"],"falsifier":"For n=5, compute $F_5(h)$ numerically on a fine grid of $h$-values in $(-5\\cdot 6^{2/5}/14,0)$; the claimed result requires strict decrease at every point, and any local increase would refute it. Equivalently, check the endpoint values: the theorem predicts $c_0(0)=(19)/(50)$ for $n=5$ and $c_0$ tending to $1/5$ at the left endpoint; a deviation from these values would show the monotonicity argument fails.","tokens_in":13115,"feed_emoji":"🌊","tokens_out":7221,"duration_ms":58968,"temperature":0.7,"pith_summary":"The paper studies the perturbed generalized KdV equation $U_t+U^nU_x+U_{xxx}+\\epsilon(U_{xx}+U_{xxxx})=0$ with $\\epsilon>0$ small and $n$ any positive integer. It proves that the ratio of Abelian integrals $F_n(h)=A_n(h)/A_0(h)$ is strictly decreasing on the energy interval $(-n(n+1)^{2/n}/(2(n+2)),0)$, the interval swept out by periodic orbits of the unperturbed Hamiltonian system. This monotonicity is the engine behind two conclusions: the limit wave speed $c_0(h)$ is monotonically increasing on the same interval, and for each wave speed the original equation has at most one isolated periodic traveling wave. The result settles an open problem from [26] and proves a conjecture from [13], extending earlier case-by-case results to all $n$.","feed_headline":"Wave speed increases monotonically for every n in generalized KdV","feed_subtitle":"Strictly decreasing ratio of Abelian integrals settles wave-speed monotonicity and uniqueness of periodic waves.","key_machinery":"The central object is the ratio of Abelian integrals $F_n(h)=A_n(h)/A_0(h)$, where $A_n(h)=\\oint_{\\Gamma_h} u^n y\\,du$ and $A_0(h)=\\oint_{\\Gamma_h} y\\,du$ over the periodic level curves $\\Gamma_h$ of the unperturbed Hamiltonian system. The proof also uses the involution $\\delta(u)$ defined implicitly by $\\Phi(u)=\\Phi(\\delta(u))$, where $\\Phi(u)=-u^2/2+u^{n+2}/((n+1)(n+2))$, and the function $T_n(u)$ built from the ratio of two integrals over the interval between $\\delta(u)$ and $u$. Lemma 3.1 supplies the algebraic identity that makes the derivative of $T_n(u)$ visibly negative; an existing criterion then turns this negativity into monotonicity of $F_n(h)$.","core_discovery":"The central claim is that the monotonicity of the limit wave speed is controlled by a single ratio of Abelian integrals, $F_n(h)=A_n(h)/A_0(h)$, formed from the perturbation terms $u^n y$ and $y$ over the level curves $\\Gamma_h: H(u,y)=h$ of the Hamiltonian $H=y^2/2-u^2/2+u^{n+2}/((n+1)(n+2))$. The paper proves that $F_n(h)$ is strictly decreasing for $h$ in the periodic annulus $(-n(n+1)^{2/n}/(2(n+2)),0)$. The key step is an algebraic identity, Lemma 3.1, that rewrites the numerator of a certain derivative as a sum of positive squares, giving $T_n'(u)<0$; a quoted monotonicity criterion for ratios of Abelian integrals then transfers this to $F_n'(h)<0$. Since a periodic wave exists only when $A_n(h)/A_0(h)=1+1/c_0(h)$, the decreasing $F_n$ makes $c_0(h)$ increasing, with endpoint limits $c_0(0)=(3n+4)/(2n^2+3n)$ and $c_0\\to 1/n$ as $h$ approaches the left endpoint of the annulus.","pith_inferences":["Editorial extension: the same proof strategy should work for other polynomial perturbations of the same Hamiltonian, because only the positivity of the sum-of-squares expression in Lemma 3.1 is used; any perturbation whose analogous numerator stays positive would inherit monotone $F_n$.","Editorial extension: the endpoint bounds $1/n < c_0(h) < (3n+4)/(2n^2+3n)$ are quantitative enough to be tested in numerical simulations of the full PDE for small $\\epsilon$, and a violation would point to a failure of the slow-manifold reduction.","Editorial extension: for even $n$, the reflection symmetry gives one periodic annulus on each side of the saddle; the paper proves at most one limit cycle per side, but it does not rule out two coexisting periodic waves, one positive and one negative, for the same wave speed."],"forward_implications":["At most one isolated periodic traveling wave exists for each wave speed when $u(x,t)>0$ or $u(x,t)<0$, for every positive integer $n$.","The limit wave speed obeys $1/n < c_0(h) < (3n+4)/(2n^2+3n)$ throughout the periodic annulus.","At the endpoints, $c_0(h)$ tends to $1/n$ near the center and to $(3n+4)/(2n^2+3n)$ near the homoclinic orbit.","Numerical continuation for $n=5$ shows a stable limit cycle and a monotonically increasing $c_0(h)$ curve, matching the theoretical interval."],"supporting_citations":[{"why":"supplies the slow-manifold reduction, the smoothness of the limit wave speed $c_0(h)$, and the endpoint values used in Lemma 3.4 and Lemma 3.5.","marker":"[26]"},{"why":"provides the monotonicity criterion (Theorem B) that turns a negative derivative of $T_n(u)$ into a negative derivative of the ratio $F_n(h)$.","marker":"[9]"},{"why":"corrects the sign error in the statement of Theorem 2.1 of [9]; the corrected sign is what makes the ratio decreasing rather than increasing.","marker":"[25]"},{"why":"states the conjecture that $F_n(h)$ is monotonically decreasing on the periodic annulus, which the paper proves.","marker":"[13]"},{"why":"gives the Poincaré–Pontryagin theorem used to conclude at most one limit cycle from a single simple zero of the Abelian integral.","marker":"[3]"}],"fun_headline_variants":["Abelian integral ratio proves wave speed monotonicity in KdV","Strictly decreasing Abelian integral ratio settles wave speed for KdV","Wave speed monotonicity via Abelian integral ratio in perturbed KdV","Monotone wave speed and uniqueness from Abelian integral ratio"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes the corrected sign convention in the quoted monotonicity theorem is the right one; if the sign is flipped, the limit wave speed would be decreasing rather than increasing.","fun_headline_variants_meta":{"raw":{"variants":["Abelian integral ratio proves wave speed monotonicity in KdV","Strictly decreasing Abelian integral ratio settles wave speed for KdV","Wave speed monotonicity via Abelian integral ratio in perturbed KdV","Monotone wave speed and uniqueness from Abelian integral ratio"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00087,"raw_usage":{"total_tokens":3769,"prompt_tokens":947,"completion_tokens":2822,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":563,"completion_tokens_details":{"reasoning_tokens":2745}},"tokens_in":563,"tokens_out":2822,"duration_ms":18229,"temperature":1.0,"reasoning_tokens":2745,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:33:19.251608+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For n=5, compute $F_5(h)$ numerically on a fine grid of $h$-values in $(-5\\cdot 6^{2/5}/14,0)$; the claimed result requires strict decrease at every point, and any local increase would refute it. Equivalently, check the endpoint values: the theorem predicts $c_0(0)=(19)/(50)$ for $n=5$ and $c_0$ tending to $1/5$ at the left endpoint; a deviation from these values would show the monotonicity argument fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the slow-manifold reduction, the smoothness of the limit wave speed $c_0(h)$, and the endpoint values used in Lemma 3.4 and Lemma 3.5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the monotonicity criterion (Theorem B) that turns a negative derivative of $T_n(u)$ into a negative derivative of the ratio $F_n(h)$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"corrects the sign error in the statement of Theorem 2.1 of [9]; the corrected sign is what makes the ratio decreasing rather than increasing."},{"cited_title":"Ouyang, W","cited_arxiv_id":null,"evidence_quote":"states the conjecture that $F_n(h)$ is monotonically decreasing on the periodic annulus, which the paper proves."},{"cited_title":"Christopher, C","cited_arxiv_id":null,"evidence_quote":"gives the Poincaré–Pontryagin theorem used to conclude at most one limit cycle from a single simple zero of the Abelian integral."}],"review_version":1}