This problem records the five open questions posed in McMullen's 2010 paper on dynamics of Blaschke products on the unit disk. For a Blaschke product ff and a cycle CC of the model map pd(z)=zdp_d(z)=z^d on the unit circle, the cycle length L(C,f)=log⁡∣(fq)′(z)∣L(C,f)=\log|(f^q)'(z)| is computed at the periodic orbit of ff corresponding to CC under the canonical marking. The paper proves that short cycles (below log⁡2\log 2) are simple, that every Blaschke product has a simple cycle of length O(d)O(d), and that binding collections of cycles give compact sublevel sets in the moduli space of rational maps. It then asks: (Q1) whether L(C,f)L(C,f) is free of critical points on the Blaschke space for simple CC; (Q2) whether the binding sum ∑iL(Ci,f)\sum_i L(C_i,f) achieves its minimum at a unique point; (Q3) whether a uniform simplicity threshold δd>0\delta_d>0 exists over the larger space QBd\mathrm{QB}_d of quasi-Blaschke products; (Q4) whether the compactness theorem for a binding pair of cycles extends from Blaschke products to QBd\mathrm{QB}_d; and (Q5) whether there is a useful notion of intersection number for pairs of cycles. The hyperbolic-surface analogs of these statements are classical theorems.

Background

A Blaschke product of degree dd is a holomorphic self-map of the unit disk of the form f(z)=z∏i=1d−1z−ai1−ai‾zf(z)=z\prod_{i=1}^{d-1}\frac{z-a_i}{1-\overline{a_i}z} with ∣ai∣<1|a_i|<1; the space of such maps is denoted Bd\mathcal{B}_d and is identified with the polydisk Δd−1\Delta^{d-1} via the zeros aia_i. Each f∈Bdf\in\mathcal{B}_d restricts to a covering map S1→S1S^1\to S^1 of degree dd; f∣S1f|_{S^1} preserves Lebesgue measure on the circle and is expanding there, with ∣f′∣>1|f'|>1, so every periodic orbit (cycle) on S1S^1 has multiplier of modulus greater than 1, where the multiplier of a periodic point zz of period qq is the derivative (fq)′(z)(f^q)'(z), the infinitesimal expansion rate of the qq-th iterate at zz. The model map is the monomial pd(z)=zdp_d(z)=z^d. The canonical marking ϕf:S1→S1\phi_f:S^1\to S^1 is the unique orientation-preserving homeomorphism that varies continuously with ff, conjugates ff to pdp_d on the circle, and equals the identity when f=pdf=p_d; it labels the periodic orbits of ff by the cycles of pdp_d. For a cycle CC of pdp_d of period q=∣C∣q=|C|, the cycle length is L(C,f)=log⁡∣(fq)′(z)∣L(C,f)=\log|(f^q)'(z)| for any z∈S1z\in S^1 with ϕf(z)∈C\phi_f(z)\in C, i.e. for any point of the periodic orbit of ff marked by CC; this is independent of the chosen point on that orbit. For a finite pdp_d-invariant set X⊂S1X\subset S^1, the degree deg⁡(pd∣X)\deg(p_d|_X) is the least e>0e>0 such that pd∣Xp_d|_X extends to a covering map S1→S1S^1\to S^1 of degree ee; in particular deg⁡(pd∣X)=1\deg(p_d|_X)=1 if and only if pd∣Xp_d|_X preserves the cyclic ordering of XX. A cycle CC (viewed as a finite subset of S1S^1) is simple if deg⁡(pd∣C)=1\deg(p_d|_C)=1, i.e. the marking parametrizes it as a single winding; equivalently its points form an orbit whose cyclic order matches that of a rigid rotation. A finite set X⊂S1X\subset S^1 is renormalizable if it admits a nontrivial partition into disjoint unlinked subsets (subsets whose convex hulls in the disk are pairwise disjoint) that are permuted by the map; a collection of cycles C1,…,CmC_1,\dots,C_m of pdp_d is binding if deg⁡(pd∣⋃Ci)=d\deg(p_d|_{\bigcup C_i})=d and pd∣⋃Cip_d|_{\bigcup C_i} is not renormalizable. Binding collections play the role, for circle dynamics, that filling multicurves — systems of closed curves that cut a surface into disks — play for hyperbolic surfaces. The space QBd\mathrm{QB}_d of quasi-Blaschke products consists of rational maps f(z)=z∏i=1d−1z−ai1−bizf(z)=z\prod_{i=1}^{d-1}\frac{z-a_i}{1-b_i z} with ∏i∣ai∣<1\prod_i|a_i|<1 and ∏i∣bi∣<1\prod_i|b_i|<1, whose Julia set (the closed set of points at which the family of iterates of ff fails to be normal, i.e. the locus separating the regular basins from the chaotic dynamics) is a Jordan curve; such a map is obtained by gluing two Blaschke products f1,f2∈Bdf_1,f_2\in\mathcal{B}_d along the circle, so expressions like L(Ci,fi)L(C_i,f_i) refer to the restriction on each side of the Jordan curve. MRatd\mathcal{M}\mathrm{Rat}_d denotes the moduli space of rational maps of degree dd up to conjugation by Möbius transformations. The paper proves: (Theorem 1.1) any cycle with L(C,f)<log⁡2L(C,f)<\log 2 is simple, and all such short cycles share a rotation number; (Theorem 1.2) every f∈Bdf\in\mathcal{B}_d has a simple cycle with L(C,f)=O(d)L(C,f)=O(d); (Theorem 1.3) for a binding collection (Ci)(C_i), the sublevel set {f∈Bd:∑iL(Ci,f)≤M}\{f\in\mathcal{B}_d:\sum_i L(C_i,f)\le M\} has compact closure in MRatd\mathcal{M}\mathrm{Rat}_d; (Theorem 1.4) the closure of the set of simple cycles has Hausdorff dimension zero. These are analogs of classical results for hyperbolic surfaces — surfaces carrying a complete metric of constant curvature −1-1, studied via their Teichmüller space, the space of marked conformal structures on the surface up to isotopy — due to Kerckhoff, Otal, and Thurston: closed geodesics shorter than log⁡(3+22)\log(3+2\sqrt{2}) are simple and disjoint, every closed surface of genus gg has a simple closed geodesic of length O(log⁡g)O(\log g), and a bound on the total length of a filling multicurve confines the surface to a compact part of Teichmüller space.

Problem Statement

As established in McMullen 2010 (Comment. Math. Helv. 85 (2010), 723-749), the following five questions are open. They are independent open problems: no one of them is implied by another, and settling any one of them is a separate result.

Q1. Let CC be a simple cycle. Is the function L(C,f)L(C,f) free of critical points in Bd\mathcal{B}_d?

Q2. Let (Ci)(C_i) be a binding collection of cycles. Does ∑iL(Ci,f)\sum_i L(C_i,f) achieve its minimum at a unique point f∈Bdf\in\mathcal{B}_d?

Q3. Let QBd\mathrm{QB}_d denote the space of rational maps f(z)=z∏i=1d−1z−ai1−bizf(z)=z\prod_{i=1}^{d-1}\frac{z-a_i}{1-b_i z} with ∏i∣ai∣<1\prod_i|a_i|<1 and ∏i∣bi∣<1\prod_i|b_i|<1, such that the Julia set of ff is a Jordan curve (equivalently, ff is obtained by gluing two Blaschke products f1,f2∈Bdf_1,f_2\in\mathcal{B}_d along the circle). Does there exist δd>0\delta_d>0 such that, for all f∈QBdf\in\mathrm{QB}_d, all cycles CC with L(C,f)<δdL(C,f)<\delta_d are simple? (Here L(C,f)L(C,f) denotes the cycle length computed for the rational map ff itself, on its Jordan Julia curve.)

Q4. Suppose the cycles (C1,C2)(C_1,C_2) are binding. Does the set of f∈QBdf\in\mathrm{QB}_d with L(C1,f1)+L(C2,f2)≤ML(C_1,f_1)+L(C_2,f_2)\le M have compact closure in the moduli space of all rational maps of degree dd? (Here f1,f2f_1,f_2 are the two Blaschke products glued along the circle to form ff, and L(Ci,fi)L(C_i,f_i) is the cycle length computed on each side of the Jordan Julia curve.)

Q5. Is there a useful notion of intersection number for a pair of cycles?

All terms are as defined in the source paper: a cycle is a periodic orbit of the model map pdp_d on S1S^1, its length is L(C,f)=log⁡∣(fq)′(z)∣L(C,f)=\log|(f^q)'(z)| where q=∣C∣q=|C| and zz is any point of the periodic orbit of ff marked by CC (i.e. any zz with ϕf(z)∈C\phi_f(z)\in C, for the canonical marking ϕf\phi_f conjugating ff to pdp_d), a cycle is simple if deg⁡(pd∣C)=1\deg(p_d|_C)=1, where for a finite pdp_d-invariant set X⊂S1X\subset S^1 the degree deg⁡(pd∣X)\deg(p_d|_X) is the least e>0e>0 such that pd∣Xp_d|_X extends to a covering map S1→S1S^1\to S^1 of degree ee (so that deg⁡(pd∣X)=1\deg(p_d|_X)=1 if and only if pd∣Xp_d|_X preserves the cyclic ordering of XX), a finite subset of S1S^1 is renormalizable if it admits a nontrivial partition into disjoint unlinked subsets permuted by the map, and a collection of cycles C1,…,CmC_1,\dots,C_m of pdp_d is binding if deg⁡(pd∣⋃Ci)=d\deg(p_d|_{\bigcup C_i})=d and the union is not renormalizable. The paper notes that the hyperbolic-surface analogs of these questions have positive answers, citing work of Kerckhoff, Otal, and Thurston: on a hyperbolic surface, geodesics shorter than log⁡(3+22)\log(3+2\sqrt{2}) are simple and disjoint, and a bound on the lengths of a filling multicurve confines the surface to a compact part of Teichmüller space; for Q5 the surface counterpart is the geometric intersection number of simple closed curves.

References

  1. C. T. McMullen, Dynamics on the unit disk: Short geodesics and simple cycles, Comment. Math. Helv. 85 (2010), no. 4, 723-749. DOI: 10.4171/CMH/209. URL: https://doi.org/10.4171/CMH/209
  2. Y. M. He, H. Lee, I. Park, Analytic theory on the space of Blaschke products: simultaneous uniformization and pressure metric, arXiv:2507.17077 (2025). URL: https://arxiv.org/abs/2507.17077
  3. Y. M. He, H. Lee, I. Park, Pressure metrics in geometry and dynamics, arXiv:2407.18441 (2024). URL: https://arxiv.org/abs/2407.18441
  4. Y. Luo, On geometrically finite degenerations I: boundaries of main hyperbolic components, J. Eur. Math. Soc. 26 (2024), no. 10, 3793-3861. DOI: 10.4171/jems/1342. URL: https://arxiv.org/abs/2101.07880
  5. Y. Luo, On geometrically finite degenerations II: convergence and divergence, Trans. Amer. Math. Soc. 375 (2022), no. 5, 3469-3527. DOI: 10.1090/tran/8597. URL: https://arxiv.org/abs/2102.00357
  6. Y. Luo, Trees, length spectra for rational maps via barycentric extensions, and Berkovich spaces, Duke Math. J. 171 (2022), no. 14. DOI: 10.1215/00127094-2022-0056. URL: https://arxiv.org/abs/1907.06220
  7. N. Payne, M. Thatte, Hausdorff dimension of closure of cycles in d-maps on the circle, arXiv:2208.11837 (2022). URL: https://arxiv.org/abs/2208.11837
  8. S. Zakeri, Cyclic permutations: degrees and combinatorial types, J. Combin. Theory Ser. A 184 (2021), 105518. DOI: 10.1016/j.jcta.2021.105518. URL: https://arxiv.org/abs/1909.03300
  9. C. L. Petersen, S. Zakeri, On combinatorial types of periodic orbits of the map x↦kx(modZ)x\mapsto kx \pmod{\mathbb{Z}}, Adv. Math. 361 (2020), 106953. DOI: 10.1016/j.aim.2019.106953. URL: https://arxiv.org/abs/1712.04506
  10. K. Drach, V. Kaloshin, Unmarked spectral rigidity of expanding circle maps, arXiv:2511.17452 (2025). URL: https://arxiv.org/abs/2511.17452
  11. C. T. McMullen, The question mark function, welding, and complex dynamics, Invent. Math. 244 (2026), no. 2, 769-814. DOI: 10.1007/s00222-026-01405-9. URL: https://doi.org/10.1007/s00222-026-01405-9