Higher-order analogues of Fuchs' fixed-branch-point conditions and an effective Fuchsian method for the moduli of a Riemann surface

In 1885 Poincaré proved that first-order algebraic differential equations satisfying Fuchs' conditions — equations all of whose solutions have only fixed branch points — never define new classes of transcendents: by the genus p of the associated algebraic surface they reduce to Riccati equations (p = 0), to a single quadrature (p = 1), or to purely algebraic integration (p > 1). The paper's conclusion leaves two questions explicitly open: whether analogous necessary-and-sufficient fixed-branch-point conditions can be established for higher-order equations, and whether such equations define essentially new transcendents; and whether the theorem yields an effective method for finding the moduli of a Riemann surface. Later literature resolved the second-order, first-degree instance of the first question affirmatively — the Painlevé transcendents $P_I$–$P_{VI}$ and the complete fifty-equation Painlevé–Gambier classification — and produced substantial partial results at third order (Chazy, Garnier, Bureau, Cosgrove, Adjabi et al.), but no complete classification exists for order ≥ 3 or degree > 1 (Levi–Sekera–Winternitz 2018). Poincaré's moduli method was never developed in its original form; its modern descendants — effective uniformization via accessory parameters (solved only for four-punctured spheres, Bogo 2022) and algorithmic computation of curve moduli via invariants (hyperelliptic case) — remain open for genus ≥ 2. The problem therefore remains open in both parts, with narrowed, precisely documented open cores.

Background

An algebraic ordinary differential equation of first order is a relation $F(z, y, y') = 0$ where $F$ is a polynomial in the unknown function $y(z)$, its derivative $y'$, and the independent variable $z$. Its solutions are in general multivalued analytic functions: analytic continuation around a closed path can permute branches at branch points (points critiques). A branch point of a solution is called fixed if its location is determined by the equation alone (for instance at singularities of the coefficients), and movable if its location depends on the integration constants. In 1884 Lazarus Fuchs gave conditions — degree bounds on $F$ in $y$ and $y'$, together with the requirement that the discriminant of $F$ as a polynomial in $y'$ vanish only at values of $y$ that are themselves solutions of the equation, with multiplicity constraints — that are necessary and sufficient for every branch point of the general solution of a first-order algebraic equation to be fixed. (Hill and Berry later exhibited the equation $(y')^m = y^{m+r}$, showing that the conditions as originally stated were incomplete; the corrected framework is what survives.)

Poincaré's 1885 theorem determines what such equations can produce. One attaches to $F(z, y, y') = 0$ its associated algebraic surface and its genus $p$ in the sense of the birational classification of surfaces: two algebraic surfaces are birationally equivalent if they are related by mutually inverse rational maps, and the genus is a birational invariant. Poincaré proved: for $p = 0$ the equation is reducible by a birational change of variables to a Riccati equation, hence to a linear second-order equation; for $p = 1$ it is integrable by a single quadrature (its solutions are expressible through elliptic functions with constant modulus); for $p > 1$ it is integrable by purely algebraic procedures — the moduli of the associated surface family remain constant along the flow, and the finitely many birational self-transformations, obtained through his theory of Fuchsian groups, generate all solutions algebraically from a single one. Fuchsian functions are Poincaré's automorphic functions attached to a Fuchsian group, a discrete group of Möbius transformations preserving a disk or half-plane; by uniformization, quotients of the disk by such groups parametrize (uniformize) algebraic curves and surfaces. Conclusion: first-order equations with fixed branch points never define new transcendents.

The conclusion of the paper leaves two questions explicitly open. First: 'At most we might suppose that [essentially new classes] exist among higher-order equations; but this can only be established by a special discussion, analogous to the one preceding.' Second: 'Moreover, one can base on this theorem a method for finding the moduli of a Riemann surface; but this is a point I cannot develop at the moment.'

The first question became the Painlevé program. An ODE has the Painlevé property if its general solution has no movable critical points (movable branch points or essential singularities). Painlevé and Gambier classified all second-order, first-degree algebraic ODEs with this property into fifty canonical equations, all integrable by classical means except six that define new transcendents, the Painlevé transcendents $P_I$–$P_{VI}$. Chazy attacked third and higher order — and equations of degree greater than one in the highest derivative — obtaining coefficient-degree limitation theorems and new classes (the Chazy equations), and Garnier found new higher-order classes with fixed critical points (the Garnier systems). The program is unfinished: for order $n \geq 3$, and for degree $> 1$ even at second order, only partial classifications exist (Bureau's polynomial class, Cosgrove, Adjabi et al.), and the standard Painlevé test (leading-order analysis with resonance compatibility conditions) supplies only necessary conditions for the property.

The second question connects to uniformization and moduli. A compact Riemann surface of genus $p > 1$ depends on $3p - 3$ complex moduli, and two such surfaces are isomorphic exactly when these invariants coincide; 'finding the moduli' of a concretely given surface means computing effective invariants of its biholomorphic (equivalently, for algebraic curves, birational) equivalence class. By the uniformization theorem every surface of genus $p > 1$ is a quotient of the disk by a Fuchsian group, but effective computation of the uniformizing data runs into the accessory parameter problem: the second-order linear differential equation that uniformizes the surface contains parameters not determined by the local (singularity) data, and selecting the values for which the monodromy group is Fuchsian — the quotient of two solutions single-valued — is a global, currently unsolved problem for genus $\geq 2$; only the four-punctured sphere case has an effective solution (Bogo 2022). A second modern route is algebraic: computing curve isomorphism and moduli through explicit invariant rings (effective for hyperelliptic curves, Lercier–Ritzenthaler). Independently, the isomonodromic deformations — deformations of a family of Fuchsian equations under which the monodromy data remain constant — studied by R. Fuchs, Garnier, Schlesinger and later Jimbo, Miwa and Ueno, of which $P_{VI}$ is the simplest case, realize the dictionary between Fuchsian differential equations and moduli spaces of punctured Riemann spheres, and are the deep modern form of Poincaré's equation-to-surface correspondence.

Problem Statement

The problem consists of the two questions Poincaré left open at the end of « Sur un théorème de M. Fuchs » (1885), modernized but retained at the source's generality.

Part A (higher-order Fuchs conditions and new transcendents). Consider algebraic ODEs $F(z, y, y', \ldots, y^{(n)}) = 0$ of order $n \geq 2$, with $F$ polynomial in all arguments. Establish conditions on $F$ that are necessary and sufficient for all critical points of the general solution to be fixed, in the spirit of Fuchs' first-order conditions and of the Painlevé property; and classify all such equations up to the natural equivalence (changes of dependent and independent variables, and birational transformations), determining for each equivalence class whether it is integrable in terms of previously known functions (linear equations, elliptic and abelian functions, the previously classified Painlevé, Garnier and Chazy equations) or defines essentially new transcendents. A complete answer is known only for second order and first degree (Painlevé–Gambier, fifty canonical equations, six new transcendents); the open target is order $n \geq 3$, and degree $> 1$ in the highest derivative at any order, for which only partial classifications exist.

Part B (the moduli method). Poincaré asserted that one can base on his theorem 'a method for finding the moduli of a Riemann surface'. Develop this into an effective procedure: given a concretely presented compact Riemann surface of genus $p \geq 2$ (for example the algebraic surface associated with an equation as in Part A, or a plane algebraic curve), compute its moduli — concretely, either determine the uniformizing Fuchsian group, that is, the accessory parameters of the uniformizing second-order linear differential equation, with proven uniqueness and convergence, or compute an equivalent complete set of algebraic invariants of its birational class. The genus-0 case with four punctures now has an effective solution (Bogo 2022); the open target is genus $\geq 2$ in general, or a rigorously characterized class going substantially beyond current partial results (for example beyond the hyperelliptic case of Lercier–Ritzenthaler).

The two parts form one problem because the equation-to-surface dictionary is the shared mechanism: Poincaré's own proof that first-order equations with fixed branch points define no new transcendents runs through the constancy of the moduli of the associated surfaces (the primary text states: « les modules de la surface S sont indépendants de z »), so Part B is simultaneously the missing computational arm of Part A's geometric mechanism. A complete higher-order classification resolves Part A, and an effective moduli/uniformization procedure for genus $\geq 2$ resolves Part B; a resolution of either part is a resolution of the problem.

References

1. H. Poincaré, « Sur un théorème de M. Fuchs », Acta Mathematica 7 (1885) 1–32. DOI: 10.1007/bf02402192. https://doi.org/10.1007/bf02402192 (primary source; full text also at Zenodo record 1966875, https://zenodo.org/records/1966875)
2. L. Fuchs, « Ueber Differentialgleichungen, deren Integrale feste Verzweigungspunkte besitzen », Sitzungsberichte der Königlich Preussischen Akademie der Wissenschaften zu Berlin (1884) 699–710; also in Gesammelte Werke II, 355–368. zbMATH Open, JFM 16.0248.01, https://api.zbmath.org/v1/document/_search?search_string=an%3A16.0248.01
3. J. J. Gray, « Fuchs and the theory of differential equations », Bulletin of the American Mathematical Society 10 (1984) 1–26. DOI: 10.1090/S0273-0979-1984-15186-3. https://doi.org/10.1090/S0273-0979-1984-15186-3
4. M. J. M. Hill and A. Berry, « On Differential Equations with Fixed Branch Points », Proceedings of the London Mathematical Society (2) 9 (1911) 231–234. DOI: 10.1112/plms/s2-9.1.231. https://doi.org/10.1112/plms/s2-9.1.231
5. P. Painlevé, « Sur les équations différentielles du second ordre et d'ordre supérieur dont l'intégrale générale est uniforme », Acta Mathematica 25 (1902) 1–85. DOI: 10.1007/bf02419020. https://doi.org/10.1007/bf02419020
6. B. Gambier, « Sur les équations différentielles du second ordre et du premier degré dont l'intégrale générale est à points critiques fixes », Acta Mathematica 33 (1910) 1–55. DOI: 10.1007/bf02393211. https://doi.org/10.1007/bf02393211
7. R. Garnier, « Sur des équations différentielles du troisième ordre dont l'intégrale générale est uniforme et sur une classe d'équations nouvelles d'ordre supérieur dont l'intégrale générale a ses points critiques fixes », Annales scientifiques de l'École normale supérieure (3) 29 (1912) 1–126. DOI: 10.24033/asens.644. https://doi.org/10.24033/asens.644
8. L. Schlesinger, « Über eine Klasse von Differentialsystemen beliebiger Ordnung mit festen kritischen Punkten », Journal für die reine und angewandte Mathematik 141 (1912) 96–145. DOI: 10.1515/crll.1912.141.96. https://doi.org/10.1515/crll.1912.141.96
9. J. Chazy, « Sur les équations différentielles du troisième ordre et d'ordre supérieur dont l'intégrale générale a ses points critiques fixes », Acta Mathematica 34 (1911) 317–385. DOI: 10.1007/bf02393131. https://doi.org/10.1007/bf02393131
10. J. Chazy, « Sur la limitation du degré des coëfficients des équations différentielles algébriques à points critiques fixes », Acta Mathematica 41 (1916) 29–69. DOI: 10.1007/bf02422939. https://doi.org/10.1007/bf02422939
11. R. Fuchs, « Über lineare homogene Differentialgleichungen zweiter Ordnung mit drei im Endlichen gelegenen wesentlich singulären Stellen », Mathematische Annalen 63 (1907) 301–321. DOI: 10.1007/bf01449199. https://doi.org/10.1007/bf01449199
12. R. Chalkley, « New contributions to the related work of Paul Appell, Lazarus Fuchs, Georg Hamel, and Paul Painlevé on nonlinear differential equations whose solutions are free of movable branch points », Journal of Differential Equations 68 (1987) 72–117. DOI: 10.1016/0022-0396(87)90187-2. https://doi.org/10.1016/0022-0396(87)90187-2
13. M. Jimbo, T. Miwa and K. Ueno, « Monodromy preserving deformation of linear ordinary differential equations with rational coefficients. I », Physica D: Nonlinear Phenomena 2 (1981), no. 2, 306–352. DOI: 10.1016/0167-2789(81)90013-0. https://doi.org/10.1016/0167-2789(81)90013-0
14. F. J. Bureau, « Équations différentielles du second ordre en y et du second degré en y'' dont l'intégrale générale est à points critiques fixes », Annali di Matematica Pura ed Applicata (4) 91 (1971) 163–281. DOI: 10.1007/bf02428819. https://doi.org/10.1007/bf02428819
15. C. M. Cosgrove, « Higher-Order Painlevé Equations in the Polynomial Class II: Bureau Symbol P1 », Studies in Applied Mathematics 116 (2006) 321–413. DOI: 10.1111/j.1467-9590.2006.00346.x. https://doi.org/10.1111/j.1467-9590.2006.00346.x
16. Y. Adjabi, F. Jrad, A. Kessi and U. Muğan, « Third order differential equations with fixed critical points », Applied Mathematics and Computation 208 (2009) 238–248. DOI: 10.1016/j.amc.2008.11.044. https://doi.org/10.1016/j.amc.2008.11.044
17. D. Levi, D. Sekera and P. Winternitz, « Lie point symmetries and ODEs passing the Painlevé test », Journal of Nonlinear Mathematical Physics 25 (2018), no. 4, 604. DOI: 10.1080/14029251.2018.1503435. arXiv:1712.09811, https://arxiv.org/abs/1712.09811
18. F. Winkler, « The Algebro-Geometric Method for Solving Algebraic Differential Equations — A Survey », Journal of Systems Science and Complexity 32 (2019) 256–270. DOI: 10.1007/s11424-019-8348-0. https://doi.org/10.1007/s11424-019-8348-0
19. T. N. Vo and Y. Zhang, « Rational Solutions of High-Order Algebraic Ordinary Differential Equations », Journal of Systems Science and Complexity 33 (2019) 821–835. DOI: 10.1007/s11424-019-8133-0. https://doi.org/10.1007/s11424-019-8133-0
20. V. Dragović and I. Goryuchkina, « Polygons of Petrović and Fine, algebraic ODEs, and contemporary mathematics », Archive for History of Exact Sciences 74 (2020) 523–564. DOI: 10.1007/s00407-020-00250-3. https://doi.org/10.1007/s00407-020-00250-3
21. R. Lercier and C. Ritzenthaler, « Hyperelliptic curves and their invariants: geometric, arithmetic and algorithmic aspects », arXiv:1111.4152 (2011). https://arxiv.org/abs/1111.4152
22. Y. V. Brezhnev, « On the uniformization of algebraic curves », Moscow Mathematical Journal 8 (2008), no. 2, 233–271. arXiv:math/0205261, https://arxiv.org/abs/math/0205261
23. G. Bogo, « Accessory Parameters for Four-Punctured Spheres », SIGMA 18 (2022) 024. DOI: 10.3842/SIGMA.2022.024. arXiv:2004.02971, https://arxiv.org/abs/2004.02971
24. P. Menotti, « The continuation method and the real analyticity of the accessory parameters: the parabolic case », Journal of Physics A: Mathematical and Theoretical 55 (2022), no. 47, 475203. DOI: 10.1088/1751-8121/ac9ff8. arXiv:2205.06994, https://arxiv.org/abs/2205.06994