Degenerate biquadratic regimes for the string-equation Dirichlet problem: John-mapping periodicity, Poncelet dynamics, and the Pell-Abel correspondence

For a plane domain bounded by a biquadratic algebraic curve (a real curve of bidegree (2,2)), Burskii and Zhedanov proved that the homogeneous Dirichlet problem for the string equation $u_{xy}=0$ is non-unique exactly when the John mapping (the characteristic billiard on the boundary) is periodic, when an associated Poncelet problem for two conics has a periodic trajectory, and when an associated quartic Pell-Abel equation $P(t)^2-R(t)Q(t)^2=L$ is solvable in polynomials. Their 2008 note stated the periodicity criterion through the two elliptic periods and deferred a sharper parametrization, the explicit multiplication-theorem solutions, and the degenerate cases. The deferred nondegenerate items were then completed by the authors themselves (2013): real Jacobi parametrizations of all real canonical cases, one-real-period periodicity criteria, the Cayley-Malyshev Hankel-determinant identity, and explicit solution families. What remains open is the degenerate side: classify the degenerate biquadratic curves (geometric genus 0, reducible, or singular), derive the John-mapping periodicity criteria and the corresponding uniqueness criteria for the Dirichlet and Neumann problems, construct explicit solution families, and formulate the correct Pell-Abel correspondence when the associated quartic is not square-free.

Background

The string equation $u_{xy}=0$ is the hyperbolic second-order partial differential equation whose characteristics are the coordinate axes; its solutions have the separated form $u(x,y)=f(x)+g(y)$. For a bounded plane domain the homogeneous Dirichlet problem ($u=0$ on the boundary) is ill-posed: uniqueness may fail, and whether it fails depends on the global shape of the boundary. Bourgin and Duffin (1939) showed for a rectangle that uniqueness holds exactly when the ratio of side lengths is irrational. Fritz John (1941) reduced the question, for domains convex with respect to the characteristic directions, to the dynamics of the John mapping (also called the characteristic billiard) $T$ on the boundary: $T$ is the composition of two involutions, each sending a boundary point to the second intersection of the boundary with the vertical, respectively horizontal, line through it. Uniqueness is governed by the orbit structure of $T$: absence of periodic points (transitivity) yields uniqueness in the classical settings, while a periodic orbit produces nontrivial solutions.

Burskii and Zhedanov studied this problem when the boundary is a biquadratic curve, i.e. a real algebraic curve of bidegree (2,2), $F(x,y)=\sum_{i,k=0}^{2}a_{ik}x^{i}y^{k}=0$; a generic such curve has genus 1 (its compactification is an elliptic curve, birationally parametrizable by Jacobi elliptic functions sn, cn, dn with modulus k and complete elliptic integral $K=K(k)$), and the symmetric Euler-Baxter normal form $x^{2}y^{2}+1+a(x^{2}+y^{2})+2bxy=0$ plays a distinguished role. In a series of papers (2006-2008), including their 2008 note [1], they established a chain of equivalences for such domains: the homogeneous Dirichlet problem for the string equation has a nontrivial solution if and only if the John mapping on the curve is periodic, if and only if an associated Poncelet problem for two conics is periodic, and if and only if an associated Pell-Abel equation with quartic right-hand side is solvable. The Poncelet problem is the classical porism of projective geometry: given two nested conics, iterating tangents to the inner one from points of the outer one produces a closed polygon either for every starting point or for none; its periodicity is decided by Cayley's Hankel-determinant criterion in the Taylor coefficients of $\sqrt{\det(A-\lambda B)}$ for the matrix pencil of the two conics. The Pell-Abel equation $P(t)^{2}-R(t)Q(t)^{2}=L$, with unknown polynomials $P,Q$ and constant $L$, is the polynomial analogue of the Diophantine Pell equation; its solvability goes back to Abel's study of Abelian integrals expressible by logarithms, is equivalent to periodicity of the continued fraction of $\sqrt{R}$, and underlies Chebyshev-type least-deviation problems (as surveyed by Malyshev, who gave a Hankel-determinant solvability criterion for quartic $R$); for quartic $R$ it is governed by torsion-type conditions on the elliptic curve $y^{2}=R(t)$.

In these works the periodicity criterion for the John mapping was expressed through both elliptic periods, as $n\varepsilon=m_{1}\omega_{1}+m_{2}\omega_{2}$ with integers $n,m_{1},m_{2}$, where the mapping is conjugate to a shift by $\varepsilon$ on the elliptic uniformization. The 2008 note explicitly deferred: a more exact parametrization of the biquadratic curve that would reduce the criterion to a single real period; the concrete form of the rational functions $R_n$ and $T_n$ through which the solutions are expressed via elliptic-function multiplication theorems; and the treatment of the degenerate cases. The same circle of ideas covers the Neumann problem (conormal derivative data), an indeterminacy criterion for a moment problem on the curve, static solutions of the classical Heisenberg XY spin chain, and elliptic grids in Padé interpolation.

The deferred nondegenerate items were completed by the authors' full-length treatment (2013): real Jacobi parametrizations of all real canonical cases (covering all generic relative dispositions of the two conics), one-real-period criteria of the form $\operatorname{Re}\eta/2K\in\mathbb{Q}$ (with case-dependent variants), the identity of the Cayley and Malyshev Hankel determinants, and explicit solution families $\Phi_{n}(x,y)=R_{2nN}(x)-R_{2nN}(y)$ built from functions $\varphi$ with the multiplication property $\varphi(nz)=R_n(\varphi(z))$; the even periodic functions with such a property were classified by Ritt (1922) into a rational-in-cosine (trigonometric) case and the elliptic case, with two exceptional parameter-restricted cases (the lemniscatic and equianharmonic cases). In that treatment the degenerate curves are explicitly excluded: irreducible curves of geometric genus 0 (for instance the symmetric canonical form whose constant term vanishes, where the elliptic uniformization degenerates to a rational one), reducible curves (where $F$ factors into a product of real polynomial factors, for instance two factors that are linear in each variable separately, and the boundary becomes a union of components), and curves whose discriminant polynomials acquire multiple roots, so that singular points appear at which the John involutions are undefined or multi-valued (John's case III of the boundary dynamical system).

Problem Statement

Complete the periodicity-and-solvability classification for the degenerate regimes of biquadratic boundaries - the 'remaining regimes' left open in the 2008 note [1] and explicitly excluded in the authors' full-length treatment. Concretely, for real biquadratic curves $F(x,y)=\sum_{i,k=0}^{2}a_{ik}x^{i}y^{k}=0$ in the degenerate cases (irreducible curves of geometric genus 0, reducible curves where $F$ factors into a product of real polynomial factors, and curves whose discriminant polynomials acquire multiple roots so that singular points appear):

(1) Classify the degenerate real biquadratic curves into canonical forms under real linear-fractional changes of the coordinates, in the sense of the canonical-form theory of the genus-1 treatment, and, for each class, determine whether the associated pair of conics (and the corresponding Poncelet problem) exists, matching the class to that geometry whenever it does and giving the substitute formulation where it does not.

(2) For each class, derive an explicit criterion for periodicity of the John mapping on the real curve - the analogue of the one-real-period conditions of the genus-1 theory - specifying the correct classical or modified problem setting wherever the hypothesis that each vertical or horizontal line meets the curve in at most two points fails (multi-component or singular boundaries).

(3) Determine, with proof, or refute by an explicit counterexample together with the corrected substitute theorem, whether the equivalence chain - non-uniqueness of the homogeneous Dirichlet problem for $u_{xy}=0$, periodicity of the John mapping, periodicity of the associated Poncelet problem, solvability of the associated Pell-Abel equation - extends to each degenerate class, including the Neumann-problem and moment-problem analogues; and in the non-unique regime construct explicit infinite linearly independent solution families (expected through the rational-in-cosine multiplication functions of Ritt's classification).

(4) Establish the degenerate counterpart of the Cayley-Malyshev Hankel-determinant identity, i.e. the correct formulation of the correspondence with the Pell-Abel equation when the associated quartic is not square-free: a non-square-free quartic has a square-free part of degree less than four, and whenever that degree is at most two polynomial solutions always exist in the classical Chebyshev theory, so bare solvability cannot be the correct counterpart; the equivalence must instead be matched between specified degrees of the solutions and periods of the mapping, and the precise statement is part of the problem.

The single completion boundary is: for every degenerate biquadratic boundary, an explicit computable criterion that decides uniqueness of the Dirichlet problem for the string equation, together with the Neumann-problem and moment-problem analogues, with explicit solution families when uniqueness fails, and the matching statement on the Pell-Abel side; combined with the established nondegenerate criteria of the genus-1 theory, this yields such a criterion for every biquadratic boundary. Items (1)-(4) are all mandatory deliverables, and the canonical-form classification of item (1) is an explicit part of this boundary rather than an optional preliminary. Coverage of only some of the degenerate classes, or a refutation of the equivalence chain for a single class, resolves the corresponding items for those classes and counts as partial progress rather than as completing the task; a refutation satisfies item (3) for its class only when it is accompanied by the corrected substitute theorem.

References

1. V. P. Burskii and A. S. Zhedanov, "Boundary-value problems for the string equation, the Poncelet problem, and the Pell-Abel equation: Links and relations", Journal of Mathematical Sciences 149 (2008), no. 5, 1483-1487 (English translation of an article in Sovremennaya Matematika. Fundamental'nye Napravleniya 16 (2006)). DOI: 10.1007/s10958-008-0077-1. URL: https://doi.org/10.1007/s10958-008-0077-1
2. V. P. Burskii and A. S. Zhedanov, "Dirichlet and Neumann Problems for String Equation, Poncelet Problem and Pell-Abel Equation", Symmetry, Integrability and Geometry: Methods and Applications (SIGMA) 2 (2006), 041. DOI: 10.3842/SIGMA.2006.041. URL: https://arxiv.org/abs/math/0604278 (also arXiv:math/0604278)
3. V. P. Burskii and A. S. Zhedanov, "On Dirichlet, Poncelet and Abel problems", Communications on Pure and Applied Analysis 12 (2013), no. 4, 1587-1633. DOI: 10.3934/CPAA.2013.12.1587. URL: https://arxiv.org/abs/0903.2531 (also arXiv:0903.2531)
4. V. P. Burskii and A. S. Zhedanov, "On Dirichlet problem for string equation, Poncelet problem, Pell-Abel equation, and some other related problems", Ukrainian Mathematical Journal 58 (2006), no. 4, 487-504. DOI: 10.1007/s11253-006-0081-x. URL: https://doi.org/10.1007/s11253-006-0081-x
5. V. P. Burskii, "On Poncelet problem, Abel equation and a Goursat-type boundary value problem", International Journal of Modern Physics A 37 (2022), no. 20-21, 2243007. DOI: 10.1142/S0217751X22430072. URL: https://doi.org/10.1142/S0217751X22430072
6. J. F. Ritt, "Periodic functions with a multiplication theorem", Transactions of the American Mathematical Society 23 (1922), no. 1, 16-25. DOI: 10.1090/S0002-9947-1922-1501186-3. URL: https://doi.org/10.1090/S0002-9947-1922-1501186-3
7. V. A. Malyshev, "Abel equation", Algebra i Analiz 13 (2001), no. 6, 1-55; English translation in St. Petersburg Mathematical Journal 13 (2002), no. 6, 893-938. URL: https://www.mathnet.ru/eng/aa982
8. F. John, "The Dirichlet problem for a hyperbolic equation", American Journal of Mathematics 63 (1941), 141-154. DOI: 10.2307/2371285. URL: https://doi.org/10.2307/2371285
9. D. G. Bourgin and R. Duffin, "The Dirichlet problem for the vibrating string equation", Bulletin of the American Mathematical Society 45 (1939), 851-858. DOI: 10.1090/S0002-9904-1939-07103-6. URL: https://doi.org/10.1090/S0002-9904-1939-07103-6
10. A. Iatrou and J. A. G. Roberts, "Integrable mappings of the plane preserving biquadratic invariant curves II", Nonlinearity 15 (2002), 459-489. DOI: 10.1088/0951-7715/15/2/313. URL: https://doi.org/10.1088/0951-7715/15/2/313
11. Q. Gendron, "Équation de Pell–Abel et applications", Comptes Rendus. Mathématique 360 (2022), no. G9, 975-992. DOI: 10.5802/crmath.346. URL: https://doi.org/10.5802/crmath.346 (preprint arXiv:2010.09915)
12. F. Barroero, L. Capuano, and U. Zannier, "Betti maps, Pell equations in polynomials and almost Belyi maps", Forum of Mathematics, Sigma 10 (2022), e84. DOI: 10.1017/fms.2022.77. URL: https://arxiv.org/abs/2109.13552 (also arXiv:2109.13552)
13. A. Bogatyrëv and Q. Gendron, "The space of solvable Pell-Abel equations", Compositio Mathematica 161 (2025), no. 7, 1483-1511. DOI: 10.1112/S0010437X25007158. URL: https://arxiv.org/abs/2306.00884 (also arXiv:2306.00884)
14. F. Ballini and F. Veneziano, "Hyperelliptic continued fractions in the singular case of genus zero", Rendiconti Lincei, Matematica e Applicazioni 33 (2022), no. 4, 795-832. DOI: 10.4171/rlm/989. URL: https://doi.org/10.4171/rlm/989 (preprint arXiv:2108.06560)