{"id":"28d3ed22-2878-4ca6-83ec-3b58a468f0fe","arxiv_id":"1908.02048","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Modern proofs of Liouville's two theorems, an elementary proof of Rosenlicht's solvability criterion for linear differential equations, and an outline of topological Galois theory.","lead":"This paper modernizes classical 19th century theorems about when integrals and solutions of differential equations can be written with familiar functions like exponentials, logarithms and roots. It also sketches a geometric 'topological Galois theory' that uses branching patterns to prove when equations cannot be solved in finite terms.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproved generalized Newton theorem (Theorem 2.15) is load-bearing for the Puiseux-series proof of Theorem 2.12; without it the central proof is conditional.","rationale":"The reader's weakest_assumption precisely identifies Theorem 2.15 as the linchpin. My review of the proof of Theorem 2.12 confirms that the Puiseux expansions (22) and (24) in §§2.4.5.3 and 2.4.6.3 are essential: the leading-term analyses in Theorems 2.18 and 2.19 would be meaningless without the existence and convergence of these expansions. The paper itself flags the missing proof in §2.4.1 and §2.4.3, stating that the modern proof will appear in a separate paper. This is not a routine application of a standard textbook theorem; it is a nontrivial parameter-dependent version of Newton's theorem, and the paper provides neither a complete proof nor a sufficiently precise reference. The theorem may well be true—it aligns with standard results on families of algebraic curves—but a central result of a paper cannot be considered fully proven while resting on an unproved assertion. The reader's CONDITIONAL verdict is therefore appropriate: the proof should be accepted only if Theorem 2.15 is supplied or a complete reference is given. I see no other concern that would change the verdict: the paper is largely expository in Part 3 and clearly labels the parts without proofs, and the new analytical argument is internally coherent assuming Theorem 2.15. The numerous cross-referencing errors (e.g., references to 'Theorem 35', 'Lemma 44', and an apparent mix-up between Theorems 2.12 and 2.13 in §2.4.7) are distracting but do not alter the mathematical substance. Hence I agree with the reader's assessment and recommend no change to the verdict.","tokens_in":46676,"tokens_out":15447,"duration_ms":160949,"concrete_test":"Supply a complete proof of Theorem 2.15 by carrying out the Newton polygon method over the field of convergent Puiseux series in the parameter: for a polynomial P(z) ∈ K[y][z] with K a field of meromorphic functions on U, show that after a finite base change UP → U\\OP, the roots at y = ∞ admit expansions (18) with coefficients in a single finite extension KP, and that the radius of convergence r(a) can be chosen continuous in a. A concrete computational check would be to take a nontrivial family, e.g., P(z) = z^3 − y z − y − t(x) with t(x) having branch points on U, and verify symbolically that all Puiseux coefficients lie in one finite extension of K and that the expansions converge for |y| > r(a) uniformly on compact subsets of UP.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central new contribution is the proof of Theorem 2.12 via Rosenlicht's theorem, using one-parameter families of solutions in generalized integral and exponential-integral extensions (§§2.4.5–2.4.7). The decisive step is the Puiseux expansion of the algebraic function z over K(y) in powers of θρ or θμ, as in expansions (22) and (24). These expansions are justified solely by the 'generalized Newton theorem', Theorem 2.15, stated in §2.4.3. The paper explicitly says: 'Theorem 2.15 plays a crucial role in section 2.4.3. For the sake of completeness I will present its modern proof in a separate paper.' No proof or precise reference is given for the exact statement, which requires that all m branches expand in series whose coefficients are regular on a single finite covering UP of U\\OP and lie in one finite extension KP of K, with a continuous positive radius function r(a). If Theorem 2.15 fails—for instance, if the coefficients cannot be made regular on a common covering, or a continuous r(a) does not exist—then the leading-coefficient arguments in Theorems 2.18 and 2.19 (which force the exponent k/p to be zero) have no foundation, and the proof of Rosenlicht's theorem and hence of Theorem 2.12 collapses. The theorem is plausible—it is a parameter-dependent version of the classical Newton–Puiseux theorem—but the paper's main result is conditional on an unproved statement, exactly as the author acknowledges. Other weaknesses (numerous cross-referencing errors, deferred proofs in Chapter 3) are secondary and do not affect the central argument as directly.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript is an extended commentary on Ritt's book. Part 1 (Section 2) gives a modern proof of Liouville's First Theorem using Lie group actions and locally invariant 1-forms, and develops a generalization of Liouville's Second Theorem: Theorem 2.12 states that a homogeneous linear ODE (12) is solvable by generalized quadratures over K exactly when it has a solution y1 = exp z with z' algebraic over K and the reduced equation is solvable over K(y1). The proof follows Rosenlicht's theorem (Theorem 2.13) and avoids valuation theory by using one-parameter families of solutions and Puiseux expansions based on a generalized Newton theorem (Theorem 2.15). Part 2 (Section 3) outlines topological Galois theory: Section 3.2 is self-contained on representability by radicals and the 13th Hilbert problem, while Sections 3.3 and 3.4 state results on S-functions, monodromy pairs, Fuchsian equations, SC-germs, and holonomic systems essentially without proofs. The paper explicitly announces that most of Section 3 is an outline and that Theorem 2.15 is deferred to a separate paper.","tokens_in":47002,"tokens_out":7305,"duration_ms":68025,"significance":"If Theorem 2.15 is supplied, the proof of Theorem 2.12 gives a genuinely elementary, valuation-free route to Rosenlicht's theorem and to the Liouville-Picard-Vessiot criterion for homogeneous linear equations of arbitrary order. The Lie-group proof of the First Liouville Theorem is a conceptually attractive modernization of the classical arguments. The paper is honest about its limitations: the abstract and Section 2.4.1 explicitly say that Theorem 2.15 is deferred and that most of Section 3 is statement-only. The topological Galois theory portion is a useful survey and includes a complete proof in Section 3.2 of the topological criterion for representability by radicals; the repeated references to [Kho14] are appropriate for a survey. These strengths do not remove the need to make the main new proof self-contained or explicitly conditional on an established theorem.","major_comments":[{"comment":"Theorem 2.15 is the sole justification for the Puiseux expansions (22) and (24) used in Sections 2.4.5 through 2.4.7. The theorem is stated without proof, and the text says its modern proof will appear in a separate paper. The leading-term arguments in Theorems 2.18 and 2.19, which force the exponent k/p to be zero and identify the constant term as a solution, depend on those expansions holding with coefficients regular on a common covering UP and with a continuous positive radius function r(a). As written, the proof of Theorem 2.12 is therefore conditional on an unproved statement. Please include a proof of Theorem 2.15, or give a precise published reference containing exactly this parameter-dependent statement, or state explicitly that Theorem 2.12 is conditional on it.","section":"§2.4.3, Theorem 2.15"},{"comment":"The proof of the First Liouville Theorem invokes 'Lemma 27', 'Lemma 22', 'Theorem 31', and 'Theorem 21', none of which appears in the manuscript; Section 2.3.4.1 similarly refers to 'Lemma 8' and 'Theorem 9' instead of Lemma 2.5 and Theorem 2.7. Because this proof is one of the paper's two advertised new proofs, the broken cross-references make the argument unverifiable as written. Please renumber all internal references and check every such citation.","section":"§2.3.4.3, proof of Theorem 2.10"},{"comment":"The logical relation between Theorems 2.11, 2.12, and 2.13 is mislabeled. The generalized Riccati equation (15) satisfies the hypothesis of Rosenlicht's theorem (Theorem 2.13), not of the criterion Theorem 2.12; the sentence at the end of Section 2.4.2 saying that 'The generalized Riccati equation (15) satisfies the conditions of Theorem 2.12' should refer to Theorem 2.13. Similarly, the proof in Section 2.4.7 is titled 'Proof of Rosenlicht's theorem' and proves exactly the statement of Theorem 2.13 for equation (13), but the text calls it 'Proof of Theorem 2.12' and even 'Proof of Theorem 35'. Correct the numbering and spell out the implication chain: Rosenlicht's theorem for (15) gives Corollary 2.3, which yields Theorem 2.11; Theorem 2.12 then follows from Lemma 2.8 by induction on the order.","section":"§2.4.2 and §2.4.7"},{"comment":"In the proof of Theorem 2.19 the citation 'Lemma 2.20' should be 'Lemma 2.21' in both places: the exponential-integral analogue of Lemma 2.18 is Lemma 2.21, not Lemma 2.20. As printed, the reader cannot tell which statement about leading terms is being used for the case k/p = 0. This is a local error, but it occurs in the proof of a central claim and should be corrected.","section":"§2.4.6 and §2.4.7, proof of Theorem 2.19"}],"minor_comments":[{"comment":"Equation (2) has a stray label 'abeleqn2 (3)' that should be removed.","section":"§2.3.3.2, equation (2)"},{"comment":"The reference 'Theorem 19' should be Theorem 2.5, and in equation (5) the denominator 'Mj(z1)' should be 'Mi(z1)'.","section":"§2.3.3.2, proof of Abel's Theorem"},{"comment":"Theorem 2.4 says it 'follows from Lemma 8', but the intended reference is Lemma 2.3.","section":"§2.2.6, Theorem 2.4"},{"comment":"The statement of Lemma 2.21 has minor typos: 'Piueux' should be 'Puiseux', and the condition for the leading term should be written uniformly with the notation of Lemma 2.18.","section":"§2.4.6, Lemma 2.21"},{"comment":"There is a global renumbering problem in Section 3: the text refers to 'Theorem 10', 'Theorem 16', 'Lemma 32', 'Corollary 38', and 'Theorem 40' where the corresponding numbered statements do not appear under those numbers. A systematic pass to fix all cross-references is needed.","section":"§3"},{"comment":"The chain in Definition 2.1 is written 'K = F0 ⊂ ... ⊆ Fn ⊃ F'; this should presumably be 'F ⊂ Fn' to match the intended definition of an extension.","section":"§2.2.2.2, Definition 2.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is an extended comment and survey, and its most substantial new contribution (the valuation-free proof of Rosenlicht's theorem and Theorem 2.12) is made conditional on Theorem 2.15, which is not proved and is deferred to a separate paper. This is a load-bearing gap rather than a presentation issue. If the journal is willing to accept a conditional advance with a companion-paper arrangement, major revision is appropriate; otherwise the editor may wish to consider whether the unproved main lemma falls within the manuscript's scope. The heavy reliance on [Kho14] in the survey sections is legitimate but should be clearly separated from the claimed new proofs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nWhat you should know: this is Khovanskii's comments on Ritt's book, but you can read it as two working papers. Section 2 contains the substance. Section 2.3 gives a modern proof of Liouville's First Theorem by Lie group actions and invariant 1-forms. Section 2.4 gives a new-looking proof of Rosenlicht's criterion (Theorem 2.12) that avoids valuation theory, replacing it with Puiseux expansions of a one-parameter family of solutions, plus a nonlinear generalization (Theorem 2.14). Section 3 is a survey of topological Galois theory, with most proofs deferred to his book [Kho14].\n\nThe new material is genuinely useful. The Lie-group proof of the First Theorem is clear and really does illuminate the old argument. The Puiseux route to Rosenlicht is a real change of method, and Theorem 2.14 is a modest but real extension. The paper is also honest about its gaps, which counts for something.\n\nThe soft spots, in proportion. The main one is Theorem 2.15, the generalized Newton theorem. It is stated without proof, and the author says a modern proof will appear in a separate paper. That theorem is load-bearing: the Puiseux expansions (22) and (24), and therefore the leading-coefficient arguments in Theorems 2.18 and 2.19, rest on it. If the exact statement—coefficients regular on a common covering, with a continuous radius function—doesn't hold, the proof of Theorem 2.12 collapses. The result is plausible and should be standard, but the paper doesn't provide a proof or a precise reference. So the central new proof is conditional. That's exactly the author's own description, but it is still a serious incompleteness for a paper that advertises a proof of Rosenlicht's theorem.\n\nSecond, the cross-references are a mess. The text cites Lemma 22, Lemma 27, Theorem 19, Theorem 31, and others that don't match the numbering. That doesn't hurt the mathematics, but it makes refereeing harder and suggests the manuscript hasn't been cleaned up.\n\nThird, Section 3 is not really new. It is an outline of [Kho14], and most theorems are stated with \"see [Kho14].\" That's fine as a service to the reader, but it shouldn't be judged as an independent contribution.\n\nThe citation pattern is self-heavy, but mostly legitimate: the surveyed results are genuinely from his own book and papers.\n\nWho is this for? Someone who wants a modern map of Ritt's book, or wants to see the Rosenlicht criterion approached with power series instead of valuations. It would be a good reading-group starting point.\n\nMy recommendation: send it to peer review, but ask the author to include a proof of Theorem 2.15 or cite a precise published source, and to fix the internal references. The classical theorems are true and the new proof route is plausible; the paper should not be desk-rejected. But as it stands, the advertised contribution is conditional, so acceptance should wait for the missing theorem.","headline":"A useful modern companion to Ritt with a genuinely new-looking Puiseux proof of Rosenlicht's criterion, but that proof is conditional on an unproved generalized Newton theorem and the internal references need cleanup.","tokens_in":47518,"tokens_out":3656,"would_cite":false,"duration_ms":35179,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["12H05","34A30","34M15","14H05","32S40"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper modernizes Liouville's theorems and proves that an n-th order homogeneous linear differential equation is solvable by generalized quadratures exactly when it possesses an algebraic-exponential solution and the reduced equation…","keywords":["Liouville's theorems","generalized quadratures","differential fields","Puiseux series","topological Galois theory","monodromy","solvability in finite terms","linear differential equations"],"falsifier":"Exhibit an algebraic function defined by a polynomial over a functional differential field $K$ whose Puiseux expansion at infinity requires coefficients that generate an infinite extension of $K$, or exhibit a homogeneous linear equation over $K$ that is solvable by generalized quadratures but violates condition (1) of Theorem 2.12 by having no solution $\\exp z$ with $z'$ algebraic over $K$.","tokens_in":46474,"feed_emoji":"∫","tokens_out":6119,"duration_ms":110878,"temperature":0.7,"pith_summary":"This paper gives modern proofs of the two classical Liouville theorems and extends the second one to homogeneous linear differential equations of arbitrary order. The central result is an if-and-only-if criterion: an n-th order equation over a functional differential field is solvable by generalized quadratures exactly when it has a solution of the form $\\exp z$ with $z'$ algebraic over the coefficient field and the reduced $(n-1)$-st order equation is then solvable over the enlarged field. The proof is elementary in the sense that it avoids valuation theory and instead uses convergent Puiseux series expansions furnished by a generalized Newton theorem. The same criterion is shown to hold for nonlinear homogeneous equations satisfying a weighted-degree condition. The second half of the paper outlines a topological Galois theory in which branching and monodromy, rather than differential Galois groups, obstruct solvability in finite terms.","feed_headline":"Liouville's second theorem grows to every order","feed_subtitle":"Puiseux series replace valuation theory, and solvability becomes a recursive algebraic check.","key_machinery":"The mechanism that carries the proof is the generalized Newton theorem, stated as Theorem 2.15: an algebraic function over a field of meromorphic functions on a Riemann surface has convergent Puiseux expansions at infinity with coefficients in a single finite extension of that field. From it, the paper builds a one-parameter family of solutions in a generalized extension by an integral or an exponential of an integral, expands the family in Puiseux series in the varying transcendental element, and plugs the expansion into the differential equation. The leading-term analysis then forces the leading coefficient of the expansion to satisfy the original equation, which is precisely the algebraic solution required by the criterion. This machinery lets the argument avoid valuation theory and remain inside functional differential fields of meromorphic functions.","core_discovery":"On the paper's own terms, the discovery is that Liouville's Second Theorem is not a second-order fact. For any order $n$, a homogeneous linear differential equation $y^{(n)}+a_1y^{(n-1)}+\\cdots+a_ny=0$ with coefficients in a functional differential field $K$ is solvable by generalized quadratures over $K$ if and only if it has a solution $y_1=\\exp z$ with $z'$ algebraic over $K$ and the reduced equation of order $n-1$ over $K(y_1)$ is solvable by generalized quadratures over $K(y_1)$. The proof replaces the standard Picard-Vessiot and valuation-theoretic route with a Puiseux-series argument: a solution living in an extension by an integral or by an exponential of an integral generates a one-parameter family of solutions, and the leading term of the family's Puiseux expansion must already supply an algebraic solution of the associated generalized Riccati equation. The same leading-term argument, together with a weighted-degree condition, extends the criterion to nonlinear homogeneous equations. For second-order equations the statement reduces to the classical form of Liouville's Second Theorem.","pith_inferences":["Although the paper does not spell this out, the recursive form of the criterion looks like a decision procedure: for each candidate algebraic extension, the Puiseux leading-term computation could be turned into an algorithm that either produces the required solution or terminates with a certificate of non-solvability.","The same family-of-solutions and leading-term argument might extend to equations with irregular singular behavior, where the uniform Puiseux expansion of the generalized Newton theorem does not apply and no analogous elementary proof is currently available.","A testable consequence of the paper's viewpoint is that any solution representable by generalized quadratures but not by algebraic exponentials would force the equation to become solvable after adjoining such an algebraic exponential; this can be checked on explicit transcendental equations whose solutions have no obvious algebraic-exponential entry point.","The topological Galois theory outline suggests that, for Fuchsian equations, the only obstructions to solvability in finite terms are branching obstructions; completing the unpublished proofs in the later sections would give a unified 'only branching matters' explanation for many non-solvability results."],"forward_implications":["For any $n$, solvability by generalized quadratures forces an exponential solution $\\exp z$ whose logarithmic derivative $z'$ is algebraic over the coefficient field; transcendental exponential solutions can only appear after an algebraic adjunction.","Solvability therefore becomes a recursive finite-step check: find the algebraic exponential solution, reduce order, and repeat on the new coefficient field.","For $n=2$ the criterion automatically returns Liouville's Second Theorem, since first-order equations are solvable by quadratures.","Nonlinear homogeneous equations with a weighted-degree-dominant monomial obey the same dichotomy, so the criterion is not limited to linear equations.","The paper's proof makes the theorem available over functional fields of meromorphic functions without a side trip through abstract differential algebra and valuation theory."],"supporting_citations":[{"why":"Supplies the classical book being commented on, including the original proofs and methods of Liouville and Ritt that the paper modernizes.","marker":"[Rit48]"},{"why":"Supplies Rosenlicht's theorem, stated as Theorem 2.13, which the paper proves by an elementary Puiseux-series argument and from which Theorem 2.11 follows.","marker":"[Ros73]"},{"why":"Is the presentation followed in Section 2.4, providing the connection between solvability by quadratures and Newton's theorem.","marker":"[Kho18b]"},{"why":"Is the source followed for the proof of Liouville's First Theorem via Lie group actions.","marker":"[Kho18a]"},{"why":"Is the book-length source for the topological Galois theory outlined in Section 3, including monodromy obstructions and the SC-germ machinery.","marker":"[Kho14]"},{"why":"Provides the standard differential Galois theory background used for comparison with the elementary proof and for the Fuchsian solvability theorems.","marker":"[dPS03]"},{"why":"Supplies the original topological obstruction argument for representability of algebraic functions by radicals, cited in Section 3.2.","marker":"[Kho71]"}],"fun_headline_variants":["Liouville's second theorem extends to every order","Puiseux series prove generalized Liouville criterion","Solvability by quadratures for all linear orders","Generalized Liouville theorem via Puiseux argument","Second Liouville theorem: now for any order n"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the generalized Newton theorem, stated without proof in Section 2.4.3: an algebraic function over a field of meromorphic functions has convergent Puiseux expansions at infinity whose coefficients lie in one finite extension of that field, uniformly over the punctured Riemann surface.","fun_headline_variants_meta":{"raw":{"variants":["Liouville's second theorem extends to every order","Puiseux series prove generalized Liouville criterion","Solvability by quadratures for all linear orders","Generalized Liouville theorem via Puiseux argument","Second Liouville theorem: now for any order n"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1344,"prompt_tokens":952,"completion_tokens":392,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":321}},"tokens_in":568,"tokens_out":392,"duration_ms":4360,"temperature":1.0,"reasoning_tokens":321,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:55:10.660991+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit an algebraic function defined by a polynomial over a functional differential field $K$ whose Puiseux expansion at infinity requires coefficients that generate an infinite extension of $K$, or exhibit a homogeneous linear equation over $K$ that is solvable by generalized quadratures but violates condition (1) of Theorem 2.12 by having no solution $\\exp z$ with $z'$ algebraic over $K$.","supporting_citations":[],"review_version":1}