{"id":"d60269fb-3a72-4c26-a6ca-a20394900055","arxiv_id":"2507.19179","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For every (3,3N+1) spectral curve with h2=0, the function w=-3℘1,1 evaluated on (x,t,0,...,0) solves the Boussinesq equation.","lead":"Researchers constructed an integrable hierarchy for the Boussinesq equation using Lie algebra methods and wrote exact finite-gap solutions in terms of Kleinian sigma functions. A general reader would care because explicit quasi-periodic solutions of this classic water-wave equation are rare and the method may extend to other integrable hierarchies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The exact-solution formula hinges on the non-special divisor claim in Theorem 3, which is delegated to [4] without proof here; if the separation divisor (41) can be special, (61) and Theorem 6 do not follow.","rationale":"I read the central claim as the finite-gap formula and its derivation. I checked the main algebraic steps: elimination in Theorem 2 leading to (38)-(39) is correct; the interpolation sums in Theorem 6 have the right degrees (P(z)=z^{N-n}(z^{2N}+B3) is monic of degree 3N-n, giving δ_{n,1}; the other sums vanish); the ℘-identities (68) do combine to (67). The sign discrepancy between (62) and (64) cancels in the final expressions and is not load-bearing. The genuinely unproven point is the non-specialness of the separation divisor required by Theorem 5 of [5]. This is the same assumption the reader identified. The paper's exclusion of Discr is about the curve, not about the divisor; Theorem 3 is asserted with a proof reference, not a derivation. Since this is a conditional gap rather than a demonstrated error, I do not move the verdict: CONDITIONAL remains appropriate, pending the non-specialness check. The overclaim about being the first explicit quasi-periodic solution is also worth tempering given [30], but it does not affect the mathematical claim.","tokens_in":20224,"tokens_out":28252,"duration_ms":258593,"concrete_test":"Use N=2 (genus 6). Pick a generic curve (51) with h2=0 and h outside Discr, and random initial dynamic variables satisfying the Casimir constraints (31). Compute the six roots (z_k,w_k) of (41), form D, and its Abel image u=A(D)=Σ∫_{∞}^{(z_k,w_k)}du using the first-kind differentials (53). Then evaluate θ(A(D)-K;τ) with the period matrix τ; by the Riemann singularity theorem D is non-special iff this value is nonzero. Repeat for several random curves/orbits. If any generic sample yields zero, Theorem 3's non-special claim fails and formulas (61)/(65) are not established. As a cross-check, verify that the polynomials (60) with coefficients (60c) vanish at all six computed points.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The solution formula w=-3℘1,1(u+C) in Theorem 6 rests on the uniformization step in Section 5.3: the divisor D={(z_k,w_k)} defined by the separation system (41) is identified with the common zero divisor of the two polynomials (60) whose coefficients are Kleinian ℘-functions. This identification is Theorem 5 of [5], and its hypothesis is that D is a non-special positive divisor of degree 3N on the (3,3N+1)-curve V. The paper asserts this in Theorem 3, but the proof is only 'made by the method proposed in [4]' and is not reproduced. The discriminant exclusion h∉Discr in Section 5.1 guarantees the curve has genus 3N, but it does not by itself imply that the particular divisor produced by (41) avoids the theta divisor of special divisors. If for some orbit the divisor is special, the coefficient equalities (61), and hence the expressions β3;1=-℘1,1 and β2;1=..., would be invalid, and the linear flow u=(x,t,0,...) in Theorem 6 would not describe the Boussinesq trajectory. No numerical or symbolic check of non-specialness is supplied for the plotted examples.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs the Boussinesq hierarchy from an sl(3) loop-algebra r-matrix by the orbit method, obtains finite-gap Hamiltonian systems on coadjoint orbits, and identifies their spectral curves as (3,3N+1)-curves. Separation of variables is used to reduce the flows to a Jacobi inversion problem, whose solution is expressed through Kleinian ℘-functions. The central result is an explicit quasi-periodic solution w(x,t) = -3℘_{1,1}((x,t,0,...,0)^T + C) of the Boussinesq equation 3w_tt + 4w w_xx + 4w_x^2 + w_xxxx = 0, together with a companion formula for v, and numerical plots for two genus-3 curves.","tokens_in":20497,"tokens_out":10813,"duration_ms":97816,"significance":"If the derivation is correct, the paper gives the first explicit family of quasi-periodic finite-gap solutions of the Boussinesq equation in terms of Kleinian ℘-functions for arbitrary genus 3N, connecting the orbit-method hierarchy to the algebro-geometric uniformization of trigonal curves. The paper also contains useful concrete material: the explicit r-matrix and shift element, the reduction of the h5,h6 flows to Boussinesq in Theorem 2, and the verification that the ℘-identity (67) implies that w = -3℘_{1,1} satisfies (39). The N=1 examples compute period matrices and produce graphical output, which is a valuable check on the formalism. At the same time, the manuscript relies heavily on the authors' previous papers [4,5,6,7] for the decisive separation, non-specialness, Jacobi inversion, and ℘-identity steps; these citations are published and are not circular in themselves, but the present paper does not make the required hypotheses fully explicit.","major_comments":[{"comment":"The claim that the divisor defined by the system (41) is non-special is load-bearing for the entire uniformization argument, but the proof is only delegated: 'A proof is made by the method proposed in [4]'. Theorem 5 of [5], imported in Section 5.3, applies only to a non-special positive divisor of degree 3N, and the coefficient identifications (61) and hence Theorem 6 collapse if the divisor is special for some orbit. The paper should state the precise theorem or proposition in [4] that guarantees non-specialness, verify its hypotheses for the divisor produced by (41), and indicate why the parameter values used in the Section 7 examples lie in the allowed regime; currently no check of non-specialness is supplied for the plotted cases.","section":"Section 4, Theorem 3"},{"comment":"The formula for v is internally inconsistent. From (61), β_{2;1} = (1/2)(℘_{1,1,1} - ℘_{1,2}) and β_{3;1} = -℘_{1,1}; hence 3β_{2;1} - (3/2)∂_x β_{3;1} = 3℘_{1,1,1} - (3/2)℘_{1,2}. The printed equality in Section 5.3 gives instead (3/2)(℘_{1,1,1} - ℘_{1,2}) + (1/2)∂_x℘_{1,1} = 2℘_{1,1,1} - (3/2)℘_{1,2}, which is what Theorem 6 states. More seriously, the Theorem 6 expression v = 2℘_{1,1,1} - (3/2)℘_{1,2} does not satisfy the first equation of (38), w_t = 2v_x, because w_t = -3℘_{1,1,2} whereas v_x = 2℘_{1,1,1,1} - (3/2)℘_{1,1,2}. The correct companion variable should be derived from w_t = 2v_x, which gives v = -(3/2)℘_{1,2} up to a function of t. This affects only the auxiliary field v and not the w-solution, but the inconsistency must be corrected.","section":"Section 5.3 and Theorem 6 (65)"},{"comment":"The identification of the separation divisor (41) with the common zero divisor of the uniformizing polynomials (60) is asserted by comparing coefficients, but this identification is exactly the content of Theorem 5 of [5] and requires the divisor to be non-special. The paper does not prove that the particular divisor produced by the separation system (41) satisfies the hypotheses, nor does it discuss what happens when the curve parameters approach the discriminant locus Discr. The assertion 'The obtained equalities are solvable for the dynamic variables' is also not justified; a short argument or an explicit reference to the relevant theorem in [4] or [5] should be supplied.","section":"Section 5.3, equations (60)-(61)"}],"minor_comments":[{"comment":"There are numerous encoding artifacts from the LaTeX source, such as 'hierarch y', 'Korteweg/emdash.cyrde Vries', '/guillemotleft.cyr', '/greaterorequalslant', and 'hamiltoninans'; these should be cleaned before publication.","section":"Throughout"},{"comment":"The abstract uses 'N ∈ \\Natural' with an undefined symbol \\Natural; this should be written as N ∈ ℕ.","section":"Abstract and Section 7"},{"comment":"In the list of branch points for V2R6C, e8 is listed before e6 and e7; the ordering should be made consistent so that the sheet-joining pattern (70) and the homology basis are unambiguous.","section":"Section 7.2"},{"comment":"The axes of the plots are not labeled, and the captions do not state the values of the Hamiltonians h5, h6, h9, h12 used; adding this information is important for reproducibility.","section":"Section 7 and Figures 1-2"},{"comment":"The conclusion states that bounded real-valued solutions remain an open problem, and Section 6 only presents conjectures on reality conditions; the abstract's phrase 'exact finite-gap solution' should be qualified to indicate that the obtained solutions may be complex-valued or singular in the real case.","section":"Section 8"}],"recommendation":"major_revision","confidential_remarks":"The paper's novelty is the application of the already developed Kleinian ℘-function machinery to the Boussinesq hierarchy, and the main algebro-geometric steps are drawn from the authors' own prior publications. The editorial decision may focus on whether the local errors, especially the v formula and the unproved non-specialness, are fixed and whether the reliance on [4,5,6] is made precise enough for the claimed theorems to be verifiable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Julia, here's my read of Bernatska–Skrypnyk.\n\nThe paper is a solid piece of algebro-geometric integration. What's actually new: for the whole family of (3,3N+1) spectral curves, the authors write down an explicit finite-gap solution w = -3℘_{1,1}((x,t,0,...)+C) of the Boussinesq equation, derived uniformly from an sl(3) r-matrix of Holod–Flashka–Newell–Ratiu type. The reduction in Theorem 2 from the h5,h6 flows to the Boussinesq equation checks out, and the ℘-identity (67) does imply (39). Earlier trigonal examples (Matveev–Smirnov) were genus three with split Jacobian; this covers all N, so it is a genuine step forward.\n\nThe soft spots are all in the middle section. The biggest: Theorem 3 asserts the separation divisor (41) is non-special, and the proof is delegated to [4]. That is load-bearing, because Theorem 5 from [5]—which turns the divisor into the zero divisor of the two polynomials (60) and yields (61)—assumes non-specialness. The discriminant exclusion h ∉ Discr only guarantees the curve has genus 3N, not that this particular divisor avoids the theta divisor. No check of non-specialness is given for the plotted examples. If the divisor were special, (61) and the linear flow in Theorem 6 would fail. This is probably fixable, but the paper needs to either prove it or state it as an assumption with a genericity argument.\n\nSecond, the formula for v is inconsistent: Section 5.3 gives v = 3℘_{1,1,1} - (3/2)℘_{1,2} (up to sign conventions), while Theorem 6 gives 2℘_{1,1,1} - (3/2)℘_{1,2}. One of these is a typo, but it matters because v appears in the reduction to Boussinesq.\n\nThird, the reality section is openly conjectural. The conclusion admits bounded real solutions are still an open problem, and the plotted solutions are singular. That's an honest limitation, but it means the phrase \"first explicit quasi-periodic solution\" should be tempered: these are complex/singular quasi-periodic solutions.\n\nMinor: no code for the figures, which makes the period computations hard to reproduce.\n\nThe citation pattern is self-heavy but the cited results are real and relevant. The paper deserves a serious referee; I'd send it to review, with the non-special divisor question as the main request, plus reconciliation of v and ideally verification code.","headline":"A genuine new family of explicit finite-gap Boussinesq solutions, but the non-special divisor step is delegated and load-bearing; worth a serious referee.","tokens_in":21089,"tokens_out":3479,"would_cite":true,"duration_ms":31179,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14H70","37K10","35Q53","14H42"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that every $(3,3N+1)$ spectral curve with $h_2=0$ yields an exact quasi-periodic Boussinesq solution $w=-3\\wp_{1,1}((x,t,0,\\dots)^\\top+C)$.","keywords":["Boussinesq equation","integrable hierarchy","orbit method","coadjoint orbits","loop algebra sl(3)","trigonal spectral curves","Kleinian sigma functions","Jacobi inversion problem"],"falsifier":"Take an explicit Lax matrix for $N=1$ whose spectral curve has a double branch point, lying on the discriminant locus, compute the divisor from (41), and check whether $w(x,t)=-3\\wp_{1,1}((x,t,0)^\\top+C)$ still satisfies $3w_{tt}+4ww_{xx}+4w_x^2+w_{xxxx}=0$; a nonzero residual, or a divisor whose Abel image does not move linearly with $(x,t)$, would falsify the claim.","tokens_in":19989,"feed_emoji":"🌊","tokens_out":8681,"duration_ms":82265,"temperature":0.7,"pith_summary":"The paper constructs an integrable Boussinesq hierarchy on coadjoint orbits in the loop algebra of $\\mathfrak{sl}(3)$ and proves that its finite-gap systems are solved by $w(x,t)=-3\\wp_{1,1}((x,t,0,\\dots,0)^\\top+C)$, where $\\wp_{1,1}$ is a Kleinian multiply periodic function on a trigonal spectral curve. Separation of variables turns the Boussinesq flow into a straight line on the Jacobian variety, and the Jacobi inversion problem is solved uniformly for every genus $3N$. This yields exact quasi-periodic solutions for every $N$, expressed through the same kind of function that solves KdV on hyperelliptic curves. The paper reports these as the first explicit quasi-periodic solutions of the Boussinesq equation with graphical illustrations, while noting that only singular real solutions have been found so far.","feed_headline":"Every trigonal curve yields exact Boussinesq waves","feed_subtitle":"Separation of variables turns Boussinesq flows into straight Jacobian lines; plots show the new solutions.","key_machinery":"The load-bearing object is the spectral curve $V:\\ -w^3+wI_{2N-1}(z)+I_{3N+1}(z)=0$, a $(3,3N+1)$-trigonal curve of genus $g=3N$, together with the Kleinian $\\wp$-functions built from its $\\sigma$ function. The mechanism is separation of variables: the zeros of the pair of polynomials (41) are the separation variables, their quasi-canonical brackets are proven through the auxiliary pair $(A,B)$, and the Abel map sends their divisor to a linear flow on the Jacobian. Uniformization by the imported Theorem 5 rewrites the polynomial coefficients as expressions in $\\wp$-functions, converting the separation system directly into the solution formula. The identity (67), derived from the $\\wp$-function identities of the abelian function field, then shows that this function solves the Boussinesq equation.","core_discovery":"The central claim is Theorem 6: for the hierarchy built from the rational r-matrix and shift element of Section 3, the divisor of separation variables has Abel image $u=(x,t,0,\\dots,0)^\\top+C$, and the field $w(x,t)=-3\\wp_{1,1}(u+C)$ satisfies the Boussinesq equation $3w_{tt}+4ww_{xx}+4w_x^2+w_{xxxx}=0$. The companion field is $v(x,t)=2\\wp_{1,1,1}(u+C)-\\tfrac32\\wp_{1,2}(u+C)$. The equation itself becomes the dynamical identity (67) for $\\wp_{1,1}$ on the Jacobian of any $(3,3N+1)$-curve with $h_2=0$, so each such curve supplies an explicit finite-gap solution.","pith_inferences":["Beyond the paper, the identity (69) suggests that the case $h_2\\neq0$ should also be integrable by the same scheme, with the full canonical Boussinesq equation arising as the dynamical equation for $\\wp_{1,1}$ on $(3,3N+1)$-curves without the $h_2=0$ constraint.","Because the hierarchy is built from a rational r-matrix with an explicit shift element, the same orbit construction can likely generate higher Boussinesq flows whose spectral curves are other $(n,s)$-curves, with solutions given by the corresponding Kleinian functions.","The restriction to singular real solutions may reflect the choice $C=u[K]$; exploring other characteristics or half-period subspaces could produce bounded real solutions, a testable variant of the paper's Conjecture 3.","The linearization proof identifies the first two Abel coordinates with $x$ and $t$ directly, so the same divisor-flow computation should extend to the next hierarchy flows, yielding higher-order Boussinesq-type equations solved by the same $\\wp$-functions."],"forward_implications":["For every $N$, the hierarchy supplies a $6N$-dimensional integrable Hamiltonian system whose two distinguished flows generate the Boussinesq equation, giving finite-gap solutions at every genus $3N$.","The same function $\\wp_{1,1}$ that solves KdV on hyperelliptic curves solves Boussinesq on trigonal curves, unifying the algebro-geometric integration of the two equations.","The zero-curvature representation with explicit matrices $\\nabla h_5$ and $\\nabla h_6$ provides a Lax pair for the whole hierarchy, so all flows of the hierarchy can in principle be integrated by the same Jacobi-inversion scheme.","If the paper's reality conjectures hold, the solution formula specializes to real-valued quasi-periodic waves on two families of spectral curves, with plots given for all-real and mixed-complex branch point configurations.","The paper's conclusion that only singular real solutions have been obtained marks bounded real solutions as the next open target within the same construction."],"supporting_citations":[{"why":"Supplies Theorem 5, the uniformization statement that every non-special degree-$3N$ divisor on a $(3,3N+1)$-curve is the common zero divisor of the two polynomials (60), which the solution formula (61) imports directly.","marker":"[5]"},{"why":"Provides the basis of abelian functions and the identities used to derive (67), the dynamical equation for $\\wp_{1,1}$ that is the Boussinesq equation.","marker":"[6]"},{"why":"Gives the orbit-method proof of separation of variables for $\\mathfrak{sl}(3)$-related systems, which Theorem 3 and the quasi-canonical bracket proof follow.","marker":"[4]"},{"why":"Establishes the loop-algebra orbit-method framework in which soliton hierarchies arise from coadjoint orbits, the construction on which the Boussinesq hierarchy is built.","marker":"[18]"},{"why":"Introduces the standard separation variables as coordinates of a non-special divisor of the spectral curve, the starting point for the separation system (41).","marker":"[33]"},{"why":"Gives the r-matrix used for the Boussinesq equation in this context, from which the Lax matrix of the hierarchy is derived.","marker":"[43]"},{"why":"Supplies the conditions under which the pair $(A,B)$ generates quasi-canonical separation variables, used in the proof of Theorem 4.","marker":"[16]"},{"why":"Provides general Abel-type equations for separated variables and $\\mathfrak{gl}(n)$-valued r-matrix separation of variables, backing Lemmas 1-3.","marker":"[39]"}],"fun_headline_variants":["Exact Boussinesq waves from every trigonal curve","Jacobi inversion unlocks Boussinesq solutions","New finite-gap Boussinesq flows from trigonal curves","Separation of variables solves Boussinesq hierarchy","Trigonal spectral curves give explicit Boussinesq waves"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the divisor of zeros of the separation system (41) is always non-special and that the spectral curve avoids the discriminant locus; on that imported premise the $\\wp$-function expressions (61) and the linear flow $u=(x,t,0,\\dots)$ in Theorem 6 depend.","fun_headline_variants_meta":{"raw":{"variants":["Exact Boussinesq waves from every trigonal curve","Jacobi inversion unlocks Boussinesq solutions","New finite-gap Boussinesq flows from trigonal curves","Separation of variables solves Boussinesq hierarchy","Trigonal spectral curves give explicit Boussinesq waves"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000588,"raw_usage":{"total_tokens":2718,"prompt_tokens":859,"completion_tokens":1859,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":475,"completion_tokens_details":{"reasoning_tokens":1778}},"tokens_in":475,"tokens_out":1859,"duration_ms":18181,"temperature":1.0,"reasoning_tokens":1778,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:59:49.170438+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an explicit Lax matrix for $N=1$ whose spectral curve has a double branch point, lying on the discriminant locus, compute the divisor from (41), and check whether $w(x,t)=-3\\wp_{1,1}((x,t,0)^\\top+C)$ still satisfies $3w_{tt}+4ww_{xx}+4w_x^2+w_{xxxx}=0$; a nonzero residual, or a divisor whose Abel image does not move linearly with $(x,t)$, would falsify the claim.","supporting_citations":[{"cited_title":"Solution of the Jacobi inversi on problem on non-hyperelliptic curves","cited_arxiv_id":null,"evidence_quote":"Supplies Theorem 5, the uniformization statement that every non-special degree-$3N$ divisor on a $(3,3N+1)$-curve is the common zero divisor of the two polynomials (60), which the solution formula (61) imports directly."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the basis of abelian functions and the identities used to derive (67), the dynamical equation for $\\wp_{1,1}$ that is the Boussinesq equation."},{"cited_title":"Orbit approach to separation of va riables in sl(3)-related integrable systems, Commun","cited_arxiv_id":null,"evidence_quote":"Gives the orbit-method proof of separation of variables for $\\mathfrak{sl}(3)$-related systems, which Theorem 3 and the quasi-canonical bracket proof follow."},{"cited_title":"II-III, Physica D , 9:3 (1983), pp","cited_arxiv_id":null,"evidence_quote":"Establishes the loop-algebra orbit-method framework in which soliton hierarchies arise from coadjoint orbits, the construction on which the Boussinesq hierarchy is built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the standard separation variables as coordinates of a non-special divisor of the spectral curve, the starting point for the separation system (41)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the r-matrix used for the Boussinesq equation in this context, from which the Lax matrix of the hierarchy is derived."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the conditions under which the pair $(A,B)$ generates quasi-canonical separation variables, used in the proof of Theorem 4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides general Abel-type equations for separated variables and $\\mathfrak{gl}(n)$-valued r-matrix separation of variables, backing Lemmas 1-3."}],"review_version":2}