{"id":"71a56c9d-b466-47a7-b05e-ad65a6deee7f","arxiv_id":"1908.07666","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Liouvillian spectral set is a locally finite union of algebraic varieties of codimension at most n, and quasi-solvable Schrödinger potentials have at most d+1 Liouvillian energy values.","lead":"This paper gives an algebraic description of all second-order linear differential equations with polynomial coefficients that have Liouvillian (finite-form) solutions. It also bounds how many energy levels of a polynomial-potential quantum system can have such solutions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof of Theorem 3.3 depends on the unproven nonvanishing of the AIM sequences ℓ_d and ℓ_{d-1}; the one-line justification is insufficient, although a degree induction appears to close the gap.","rationale":"The paper's central result, Theorem 1.3, is a countable union of algebraic varieties bounding Liouvillian solutions. The proof chain runs through Kovacic's algorithm (Theorem 2.2), the AIM obstruction (Theorems 3.1-3.3), and a rank argument (Proposition 3.6). I examined the nonvanishing issue flagged by the reader; it is genuine as a gap in exposition because Theorem 3.2(ii) explicitly requires ℓ_pℓ_{p−1}≠0. However, the underlying claim is very likely true: with A monic of degree n, the recurrence forces deg ℓ_j to grow by n each step, with no cancellation of leading terms. I also considered other possible defects — the absolute value in the arithmetic condition in Section 4, the sketchiness of the codimension proof, and the typo in Example 3.4 — but none of these threatens the core algebraic description. Thus the appropriate verdict remains CONDITIONAL: the paper needs a short proof of the nonvanishing claim (and correction of typos) but the central argument is not overturned. My read agrees with the reader's weakest_assumption.","tokens_in":15445,"tokens_out":31112,"duration_ms":290714,"concrete_test":"In a revision, replace the one-sentence assertion in Theorem 3.3 with a complete induction: from recurrence (21) prove deg_x ℓ_j^±=(j+1)n and deg_x r_j^±≤(j+1)n−1 for all j≥0, so ℓ_j^± never vanishes. As an independent computational check, use a computer algebra system to evaluate ℓ_0^±,...,ℓ_10^± for n=1,2,3 with generic A,B and also with random specialized coefficients, and verify ℓ_dℓ_{d−1}≠0 whenever b_{n−1}=±(n+2d). If any specialization gives a zero, test whether δ_d=0 holds without a polynomial solution, which would disprove the equivalence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Theorem 3.3 the paper characterizes L′_{2n,d} by the vanishing of Δ_d(A,B)Δ_d(−A,B) together with b_{n−1}^2=(n+2d)^2. The 'if' direction invokes Theorem 3.2(ii), which requires ℓ_pℓ_{p−1}≠0; otherwise δ_p=0 does not imply a polynomial solution. The proof merely says 'This comes easily from the fact that ℓ±0=±2A(x) is of bigger degree than r0=B±A′'. That observation only covers the initial step. One needs an induction showing that the leading term of ℓ_{j+1} comes from ℓ0ℓ_j and is nonzero for every j. While such an induction is straightforward (deg ℓ_j=(j+1)n, deg r_j≤(j+1)n−1), the paper does not supply it. If the nonvanishing were false for some admissible A,B, the algebraic description of L′_{2n,d} would admit false positives, and Theorem 1.3(b)-(d) would be undermined. This is the same load-bearing gap flagged by the reader; it is a proof gap rather than a demonstrated error.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the general second-order linear differential equation with polynomial coefficients, y'' + P(x)y' + Q(x)y = 0, under a non-degeneracy condition. It defines the spectral set L2n of equations admitting a Liouvillian solution and spectral subvarieties L2n,d according to polynomial-hyperexponential solutions of polynomial degree d. The main result, Theorem 1.3, states that L2n is the countable union of pairwise disjoint algebraic varieties L2n,d of codimension at most n in P2n, and that any compact subset of P2n meets only finitely many of them. The proof uses D'Alembert reduction to monic trace-free form, a dichotomy theorem from the authors' earlier work (Theorem 2.2) characterizing Liouvillian solutions via polynomial solutions of two auxiliary equations, and the Asymptotic Iteration Method to express solvability as the vanishing of universal differential polynomials Δ_d. A further result, Theorem 4.2, bounds by d+1 the number of energy parameters for which an algebraically quasi-solvable polynomial Schrödinger potential admits a Liouvillian solution, where d is an arithmetic condition read from the potential's coefficients.","tokens_in":15686,"tokens_out":25199,"duration_ms":224776,"significance":"If correct, the results provide a genuinely algebraic, degree-independent description of Liouvillian integrability for a large family of second-order equations, together with explicit equations for spectral varieties in low degree (Tables 2-3) and a quantitative bound on quasi-exactly-solvable eigenvalues of polynomial Schrödinger potentials (Theorem 4.2). The paper relies on a published dichotomy theorem [1] and on an external AIM criterion [15], so there is no circular dependence on the new results. The concrete computations, including the Turbiner potential example in Example 4.3, give explicit and falsifiable spectral predictions. However, two proof steps that are load-bearing for Theorem 1.3 are insufficiently detailed: the non-vanishing of the AIM sequences in Theorem 3.3 and the codimension argument in Proposition 3.6. The contribution is potentially valuable, but the central claims are not yet fully demonstrated.","major_comments":[{"comment":"The proof must justify that ℓ±_{d-1}ℓ±_d ≠ 0 for the auxiliary equations (7)-(8) before applying Theorem 3.2(ii), which requires exactly that condition to conclude that δ_p = 0 implies a polynomial solution. The sentence 'This comes easily from the fact that ℓ±_0 = ±2A(x) is of bigger degree than r0 = B ± A′' addresses only the initial step of a recurrence. A complete proof should include an induction showing, for instance, that deg ℓ±_j = (j+1)n and deg r±_j ≤ (j+1)n - 1 for every j, so that ℓ±_j is never the zero polynomial. Without this, the equivalence in Theorem 3.3 is not established, and Theorem 1.3(b)-(d) and Theorem 4.2 inherit the gap.","section":"Section 3.2, proof of Theorem 3.3"},{"comment":"The rank assertion rk(M±_{d-1,n}(A,B)) = d is stated without proof, and the passage from that rank to the conclusion codim(L′_{2n,d}, M2n) ≤ n is only sketched. The text does not identify precisely which n equations are locally independent, nor why the hyperplane condition b_{n-1} = ±(2d+n) together with the determinant conditions gives codimension at most n rather than n+1. Since Theorem 1.3(b) depends on this codimension bound, a rigorous local argument, including a transversality statement around a smooth point of L′_{2n,d}, is required.","section":"Section 3.3, Proposition 3.6 and Remark 3.7"},{"comment":"The displayed identity Δ_d(x,b_0) = Δ_d(-x,b_0) = 2^{d+1} ∏_{k=0}^d (d-k) is incorrect as written: the factor k=d makes the product zero for every d, and the equality of Δ_d(x,b_0) with Δ_d(-x,b_0) is false in general (for d=0, Δ_0(x,b_0)=b_0-1 and Δ_0(-x,b_0)=b_0+1). Since this example is meant to illustrate the general method, the formula should be corrected; the final displayed form of L′_{2,d} is, however, correct.","section":"Section 3.2, Example 3.4"}],"minor_comments":[{"comment":"The phrase 'which is a polynomial in x of λ' is imprecise: Δ_d(A(x),B(x)+λ) is a polynomial in x and λ, and for a fixed λ the Liouvillian condition is that this polynomial vanish identically in x, i.e., all its x-coefficients vanish. The bound on the number of λ-values follows from the degree in λ of the gcd of these coefficient polynomials, which is at most d+1 by Lemma 4.1; this should be stated explicitly.","section":"Section 4, Theorem 4.2 proof"},{"comment":"There is a typo in the sentence 'v,β are solutions of u′ = αu, v′ = ℓ0 v y β′ = u2v'; it should read 'u′ = αu, v′ = ℓ0 v, and β′ = u²v'.","section":"Section 3.1, Theorem 3.1 proof"},{"comment":"The bounds 'small or equal to (d+1)/2' and 'small or equal to (d+2)/2' should use integer floors or ceilings to avoid the impression that non-integer degrees are allowed.","section":"Section 3.2, Lemma 4.1"},{"comment":"The manuscript contains many typographical errors and infelicities, for example 'inpendently' in the Introduction, 'Corolary 2.6', 'Liovullian' in the Section 2 heading, 'posibilities' in Proposition 2.4, and inconsistent punctuation in the display of Tables 2-3. A thorough language and copy-editing pass is needed.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper's main novelty is the AIM-based algebraic description of the spectral varieties. The proof gaps in Theorem 3.3 and Proposition 3.6 are real but appear fixable: the non-vanishing of ℓ_j can be shown by a simple degree induction, and the codimension argument can be made rigorous with a local rank/transversality analysis. I would not reject on these grounds, but the manuscript should not be accepted until these points are fully written out. The reliance on Theorem 2.2 from the authors' own [1] is legitimate since that result is published and independent of the present AIM framework. The authors should also correct the erroneous product formula in Example 3.4 and improve the overall presentation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [colleague],\n\nThe short version: this paper is in better shape than the reader's conditional verdict might suggest. The advertised result—the Liouvillian spectral set for second-order polynomial equations is a countable union of algebraic varieties of codimension at most n, plus a bound of d+1 Liouvillian eigenvalues for quasi-solvable potentials—holds up. The weak spot flagged in the review is real but minor: it is a proof-writing problem in Theorem 3.3, not a genuine gap.\n\nWhat is new: the universal differential polynomials Δ_d give a degree-independent algebraic description of the spectral varieties, and Theorem 4.2's bound is a new tool for quasi-solvable Schrödinger equations. The paper leans on the authors' own published dichotomy (Theorem 2.2) and on external AIM results; the former is a parameter-free derivation via Kovacic's algorithm, so citing it is legitimate, not circular.\n\nOn the main gap: the one-line justification \"this comes easily from the fact that ℓ_0 = ±2A(x) is of bigger degree than r_0\" is indeed insufficient as written. But an induction closes it: deg ℓ_j = (j+1)n and deg r_j ≤ (j+1)n−1, because the leading term of ℓ_{j+1} comes from ℓ_0 ℓ_j, and the leading coefficient of ℓ_0 is ±2, which never cancels. So the non-vanishing ℓ_d ℓ_{d−1} needed in Theorem 3.2(ii) is true. The equivalence in Theorem 3.3 is safe.\n\nOther soft spots are cosmetic. Example 3.4's product formula is garbled—as written the product ∏_{k=0}^d (d−k) is zero for every d, so the displayed formula cannot be right. The codimension argument in Proposition 3.6 is sketched via a matrix rank statement that is plausible but not fully justified; a referee should ask for details. Theorem 4.2's proof would be cleaner if it explicitly noted that for quasi-solvable potentials the relevant polynomial in λ is not identically zero; otherwise the degree-bound argument is vacuous. None of this affects the central claims.\n\nThis paper deserves a serious referee. It gives the QES community a concrete bound that is not in the literature, and the algebraic geometry framework is a genuine extension of earlier work. I would send it to review and expect minor revision, mainly to expand the induction in 3.3 and the rank argument in 3.6.","headline":"The spectral-set description and the d+1 eigenvalue bound are sound; the flagged gap in Theorem 3.3 is a terse proof, not a flaw.","tokens_in":16229,"tokens_out":7245,"would_cite":true,"duration_ms":637832,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34M15","81Q35"],"pacs":[],"model":"deepseek-v4-flash","headline":"For each fixed degree, the equations admitting Liouvillian solutions form a countable union of disjoint algebraic varieties of codimension at most n, and quasi-solvable Schrödinger potentials have at most d+1 Liouvillian energy levels.","keywords":["Anharmonic oscillators","Asymptotic iteration method","Kovacic algorithm","Liouvillian solutions","parameter space","quasi-solvable models","Schrödinger equation","spectral varieties"],"falsifier":"Take an explicit point on the variety $V^+_{4,2}$ from Table 2, say $b_1=6$ and $b_0^3+16a_0b_0-16=0$ with a generic $a_0$, and compute the AIM sequences $\\ell_0^+$, $\\ell_1^+$, $\\ell_2^+$ and the obstruction $\\delta_2^+$ from the recurrence (21) for the auxiliary equation (7). The theorem asserts $\\delta_2^+=0$ exactly when a degree-2 polynomial solution exists; if any point on this variety has $\\ell_2^+\\ell_1^+\\equiv 0$, or has $\\delta_2^+=0$ without a polynomial solution, the algebraic criterion of Theorem 3.3 is false.","tokens_in":15246,"feed_emoji":"📐","tokens_out":13386,"duration_ms":116661,"temperature":0.7,"pith_summary":"This paper studies the family of all second-order linear differential equations $u''+P(x)u'+Q(x)u=0$ with polynomial coefficients, parameterized by the coefficients of $P$ and $Q$. It characterizes, within that parameter space, the subset of equations that admit a Liouvillian solution—one built by exponentials, integrals, and algebraic operations. The main theorem states that for each fixed polynomial degree this 'spectral set' is a countable union of pairwise disjoint algebraic varieties $L_{2n,d}$, each of codimension at most $n$, and any compact part of the parameter space meets only finitely many of them. A reader should care because this turns a transcendental integrability question into finite algebraic equations and yields a concrete bound on the number of energy values for which a polynomial Schrödinger potential has Liouvillian eigenfunctions.","feed_headline":"Liouvillian-solvable ODEs: a countable union of algebraic varieties","feed_subtitle":"Integrable equations form disjoint algebraic varieties, bounding Liouvillian energy levels by d+1.","key_machinery":"The load-bearing object is the sequence of universal differential polynomials $\\Delta_d\\in\\mathbb{Q}\\{a,b\\}$ generated by the asymptotic iteration method. Applied to the auxiliary equations (7) and (8) of Theorem 2.2, the recurrence (21) with initial values $\\ell_0^\\pm=\\mp 2A$, $r_0^\\pm=B\\mp A'$ produces obstructions $\\delta_d^\\pm = r_d^\\pm\\ell_{d-1}^\\pm-\\ell_d^\\pm r_{d-1}^\\pm$, and $\\Delta_d(A,B)$ is the specialization giving $\\delta_d^+$ while $\\Delta_d(-A,B)$ gives $\\delta_d^-$. The vanishing of their product, together with the arithmetic condition $b_{n-1}^2=(n+2d)^2$, exactly detects the degree-$d$ polynomial factor $P_d$ in a Liouvillian solution $P_d e^{\\pm\\int A\\,dx}$; the same polynomials work for every $n$, which is what makes the description degree-independent.","core_discovery":"On the paper's own terms, the central discovery is that Liouvillian integrability of (1) is exactly membership in one of the varieties $L_{2n,d}$ defined by the existence of a polynomial-hyperexponential solution of polynomial degree $d$, i.e. a solution $P_d(x)e^{\\int A_k(x)\\,dx}$ with $P_d$ of degree $d$ and $A_k$ a polynomial of degree $k$. Theorem 1.3 assembles these varieties into a countable disjoint union $L_{2n}=\\bigcup_{d=0}^\\infty L_{2n,d}$, with the codimension bound and the local finiteness property. The computational engine is Theorem 3.3: after writing the coefficient in the trace-free equation as $M(x)=A(x)^2+B(x)$, the equation $y''=M(x)y$ has a polynomial-hyperexponential solution of degree $d$ if and only if $b_{n-1}^2=(n+2d)^2$ and $\\Delta_d(A,B)\\,\\Delta_d(-A,B)=0$, where $\\Delta_d$ is a universal differential polynomial independent of $n$. This algebraic description supports Theorem 4.2: for a quasi-solvable polynomial Schrödinger potential with arithmetic condition $d$, at most $d+1$ values of the energy parameter admit a Liouvillian solution.","pith_inferences":["The $d+1$ bound is likely sharp in the computed examples (e.g. $d=3$ gives two $\\lambda$-values), so one could test numerically on other quasi-solvable potentials whether the bound is generically attained; a counterexample would point to an unstated assumption in the asymptotic-iteration criterion.","Because $\\Delta_d$ is independent of $n$, the method suggests a purely algebraic elimination procedure for quasi-solvability: compute $\\Delta_d(A(x),B(x)+\\lambda)$ once, eliminate $x$, and obtain polynomial equations in the potential coefficients and $\\lambda$.","The D'Alembert reduction preserves the polynomial degree of polynomial-hyperexponential solutions, so the same variety description should extend to equations with rational coefficients after clearing denominators; the paper does not develop this extension."],"forward_implications":["For fixed $n$, the spectral set $L_{2n}$ is a singular analytic submanifold of the parameter space $P_{2n}$, a countable disjoint union of algebraic components of codimension at most $n$.","Any bounded region of parameter space involves only finitely many possible solution degrees $d$, so a search for Liouvillian solutions can be truncated after checking finitely many $\\Delta_d$ conditions.","For a quasi-solvable polynomial Schrödinger potential with integer arithmetic condition $d$, the number of energy values with a Liouvillian eigenfunction is at most $d+1$; explicit spectral systems for $x^6-(4J+1)x^2$ are computed through $d=13$.","The spectral varieties are nonempty for infinitely many degrees: $L_{2n,d}\\neq\\varnothing$ whenever $d\\equiv 0$ or $1\\pmod{n+1}$.","The varieties $L_{2n,d}$ and $L_{2n,k}$ are disjoint for $d\\neq k$, so a given equation cannot admit polynomial-hyperexponential solutions of two different polynomial degrees."],"supporting_citations":[{"why":"Contains the lemma and differential-Galois theorem for $y''=M(x)y$ used as the starting point (Lemma 2.1 and Theorem 2.2).","marker":"[1]"},{"why":"Defines algebraic spectrum and quasi-solvable potential notions that the paper's spectral varieties extend.","marker":"[4]"},{"why":"Gives the sextic anharmonic oscillator family and orthogonal polynomials used as the main worked example (Example 4.3).","marker":"[5]"},{"why":"Introduces the asymptotic iteration method and the recurrence (21) that generates the universal polynomials $\\Delta_d$.","marker":"[7]"},{"why":"Provides the classification of the differential Galois group of $y''=M(x)y$ and the conditions for Liouvillian solutions used as Theorem 2.2.","marker":"[11]"},{"why":"Supplies the differential-Galois classification of the Whittaker equation used to pin down the integrable parameter values in the canonical equation.","marker":"[12]"},{"why":"States the criterion for a second-order equation to have a polynomial solution, the basis of the $\\delta_p=0$ condition in Theorem 3.2.","marker":"[15]"}],"fun_headline_variants":["Liouvillian ODEs: countable algebraic variety union","Algebraic varieties capture Liouvillian integrability","Liouvillian spectra: countable union of algebraic sets","Countable varieties bound Liouvillian energy levels"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that, for every admissible coefficient pair $(A,B)$, the asymptotic-iteration entries $\\ell_d^\\pm$ and $\\ell_{d-1}^\\pm$ are nonzero polynomials; the paper asserts this 'comes easily' from a leading-degree comparison but gives no detailed argument, so if those entries can vanish identically for some admissible pair, the equivalence between $\\Delta_d=0$ and existence of a degree-$d$ polynomial solution could fail.","fun_headline_variants_meta":{"raw":{"variants":["Liouvillian ODEs: countable algebraic variety union","Algebraic varieties capture Liouvillian integrability","Liouvillian spectra: countable union of algebraic sets","Countable varieties bound Liouvillian energy levels"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000643,"raw_usage":{"total_tokens":2944,"prompt_tokens":916,"completion_tokens":2028,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":532,"completion_tokens_details":{"reasoning_tokens":1966}},"tokens_in":532,"tokens_out":2028,"duration_ms":14697,"temperature":1.0,"reasoning_tokens":1966,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:01:44.103368+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an explicit point on the variety $V^+_{4,2}$ from Table 2, say $b_1=6$ and $b_0^3+16a_0b_0-16=0$ with a generic $a_0$, and compute the AIM sequences $\\ell_0^+$, $\\ell_1^+$, $\\ell_2^+$ and the obstruction $\\delta_2^+$ from the recurrence (21) for the auxiliary equation (7). The theorem asserts $\\delta_2^+=0$ exactly when a degree-2 polynomial solution exists; if any point on this variety has $\\ell_2^+\\ell_1^+\\equiv 0$, or has $\\delta_2^+=0$ without a polynomial solution, the algebraic criterion of Theorem 3.3 is false.","supporting_citations":[{"cited_title":"Acosta-Hum´ anez and D","cited_arxiv_id":null,"evidence_quote":"Contains the lemma and differential-Galois theorem for $y''=M(x)y$ used as the starting point (Lemma 2.1 and Theorem 2.2)."},{"cited_title":"Acosta-Hum´ anez, J.J","cited_arxiv_id":null,"evidence_quote":"Defines algebraic spectrum and quasi-solvable potential notions that the paper's spectral varieties extend."},{"cited_title":"Bender and G","cited_arxiv_id":null,"evidence_quote":"Gives the sextic anharmonic oscillator family and orthogonal polynomials used as the main worked example (Example 4.3)."},{"cited_title":"Ciftci, R","cited_arxiv_id":null,"evidence_quote":"Introduces the asymptotic iteration method and the recurrence (21) that generates the universal polynomials $\\Delta_d$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the classification of the differential Galois group of $y''=M(x)y$ and the conditions for Liouvillian solutions used as Theorem 2.2."},{"cited_title":"Martinet and J.-P","cited_arxiv_id":null,"evidence_quote":"Supplies the differential-Galois classification of the Whittaker equation used to pin down the integrable parameter values in the canonical equation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States the criterion for a second-order equation to have a polynomial solution, the basis of the $\\delta_p=0$ condition in Theorem 3.2."}],"review_version":1}