Epstein-Poincaré surfaces for $G-$opers

Given a complex, simple Lie group $G$ of adjoint type, we introduce the notion of an Epstein-Poincaré surface associated to a $G$-oper. These surfaces generalize Epstein’s classical construction for $G=PGL_2 (\mathbb{C})$. As an application, we provide a criterion that ensures that the holonomy of the oper is $Δ-$Anosov. Finally, we discuss how the developing map of the oper interacts with domains of discontinuity of the holonomy (whenever Anosov) and the transversality properties it satisfies. Along the way, we provide a quick review of opers that we hope serves as a self-contained introduction.

💡 Research Summary

The paper introduces a higher‑rank analogue of Epstein’s classical construction of surfaces in hyperbolic three‑space, extending it from the setting of complex projective structures (G = PGL₂(ℂ)) to arbitrary complex simple Lie groups G of adjoint type. The authors define an “Epstein‑Poincaré (EP) surface” associated to any G‑oper on a compact hyperbolic Riemann surface X. A G‑oper is a holomorphic immersion D : ˜X → B (the full flag manifold of G) equivariant with respect to a representation ρ : π₁(X) → G, satisfying a tangency condition to the set O of principal directions determined by a principal embedding φ : PGL₂(ℂ) → G. Starting from a Fuchsian oper (the analogue of a Fuchsian projective structure) and the principal embedding, one obtains a canonical EP‑surface Ep₀. For any other G‑oper (D, ρ) the EP‑surface is defined by composing the oscillating map Osc(P φ∘D_F, D) with Ep₀; this construction is independent of the choice of φ and recovers the classical Epstein surface when G = PGL₂(ℂ).

The central technical achievement is a quantitative criterion guaranteeing that the EP‑surface is τ_Θ‑nearly geodesic for a Weyl‑orbit Θ of simple roots. The τ_Θ‑nearly geodesic condition, introduced by Davalo, is a convexity requirement that prevents the surface from entering any tangent horoball based at points of a distinguished stratum τ_Θ of the visual boundary ∂X. By computing the second fundamental form of the EP‑surface, the authors translate this geometric condition into explicit inequalities involving the L²‑norms of the Hitchin differentials α_i ∈ H⁰(X, K^{m_i+1}) (where m_i are the exponents of g) and their covariant derivatives ∇α_i. The inequalities also involve Lie‑theoretic data: the Killing form κ, a principal sl₂‑triple (e, h, f), and the minimal angle φ_Θ^S between h and the walls defined by Θ. If the inequalities (Theorem 1.5 for general opers, Theorem 1.6 for cyclic opers) hold pointwise on X, then Davalo’s theorem implies that the holonomy ρ is Θ‑Anosov, i.e. Δ‑Anosov. The authors call such opers “quasi‑Hitchin” and denote their moduli by QH(X). The EP‑witnessed opers (those whose EP‑surfaces satisfy the τ_Θ‑nearly geodesic condition) form an open subset E(X) ⊂ Op_X(G) containing the Fuchsian oper; by Davalo’s work, each connected component of E(X) containing the Fuchsian point lies inside QH(X).

Beyond the Anosov criterion, the paper studies the interaction between the developing map D and domains of discontinuity. For a τ_Θ‑nearly geodesic EP‑surface, the projection π : B → P (where P is the partial flag manifold associated to τ_Θ) sends the image of D into a co‑compact domain of discontinuity Ω_ρ ⊂ P constructed via the balanced ideal method of Kapovich‑Leeb‑Porti. Consequently, π ∘ D descends to a smooth section of the fiber bundle Ω_ρ / ρ(π₁(S)) → S. In the special case G = PGL₃(ℂ), the full flag manifold coincides with P, so the developing map itself lands in a cocompact domain of discontinuity and extends continuously to the closed disk, with its boundary agreeing with the Anosov limit map (Theorem 1.11).

The final part of the work investigates transversality properties of G‑opers. A pair of points z, w ∈ ˜X is called transverse if the corresponding flags D(z) and D(w) are transverse in B. The authors prove that for any G‑oper the set of non‑transverse points relative to a fixed z₀ is discrete (Theorem 1.13); in the PGL₃(ℂ) EP‑witnessed case this set is finite (Corollary 1.14). They raise natural questions about whether all quasi‑Hitchin opers are (locally maximally) transverse, pointing to a rich interplay with known results on transverse circles for Anosov representations in the real setting.

The paper is organized as follows: Sections 2–3 provide background on opers, principal embeddings, and the holonomy map (including a self‑contained proof of its local injectivity). Section 4 defines EP‑surfaces for G‑opers. Section 5 develops criteria for immersion and regularity of these surfaces. Section 6 carries out the detailed curvature computations leading to Theorems 1.5 and 1.6. Section 7 discusses domains of discontinuity and the extension of developing maps. Section 8 treats transversality. Appendices A and B contain auxiliary material on equidistant families of EP‑surfaces and analytic estimates for the constants appearing in the main inequalities.

In summary, the authors provide a novel geometric construction (EP‑surfaces) that translates the analytic data of a G‑oper into a concrete surface in the symmetric space of G. By controlling the geometry of this surface they obtain explicit, verifiable conditions for the holonomy to be Anosov, describe the associated domains of discontinuity, and explore transversality phenomena. This work bridges complex projective geometry, higher Teichmüller theory, and the theory of Anosov representations, opening new avenues for studying complex opers via their associated Epstein‑Poincaré geometry.