Algebraicity of exterior Cauchy transforms of algebraic ovals: a homological formulation

Technical Summary: Algebraicity of Exterior Cauchy Transforms of Algebraic Ovals

Problem Statement
The paper investigates the conditions under which the exterior Cauchy transform of a bounded domain $\Omega \subset \mathbb{C}$, whose boundary $\Gamma = \partial \Omega$ is an oval of a real algebraic curve, is an algebraic function. The transform is defined as:
$$C_\Omega(z) = \frac{1}{\pi} \int_\Omega \frac{dA(\zeta)}{z - \zeta}, \quad z \in \mathbb{C} \setminus \Omega.$$
This question is distinct from, and weaker than, the classical problem of identifying quadrature domains, where the transform is required to be rational. While rationality implies algebraicity, the converse is not true (e.g., the ellipse has an algebraic but non-rational transform). The authors aim to characterize when $C_\Omega(z)$ is algebraic, particularly in cases where the boundary curve has positive genus or multiple components.

Methodology and Homological Framework
The authors employ a homological approach rooted in the theory of Schwarz correspondences and Riemann surfaces.

1. Schwarz Correspondence and Normalization: If the boundary $\Gamma$ lies on a real algebraic curve $P(z, \bar{z}) = 0$, the authors consider the complex curve $P(z, w) = 0$. They let $X$ be the normalization of the relevant irreducible component of this curve. The boundary $\Gamma$ lifts to a contour $\gamma \subset X$. The Cauchy transform is expressed as a boundary integral on $X$:
   $$C_\Omega(z) = \frac{1}{2\pi i} \int_\gamma \frac{\eta d\pi}{z - \pi},$$
   where $\pi$ and $\eta$ are meromorphic functions on $X$ induced by $z$ and $w$ (with $\eta = \bar{\pi}$ on the real locus).
2. Fixed Normalization vs. Moving Poles: A central methodological insight is the distinction between the variation of the surface and the variation of the integrand. The surface $X$ is fixed; only the pole divisor $F_z = \pi^{-1}(z)$ moves as $z$ varies.
3. Monodromy Analysis: The authors analyze the monodromy of the integral as $z$ moves. They argue that the natural monodromy is "point-pushing" (braid monodromy) on a fixed punctured surface. Crucially, they prove (Lemma 3.1) that after filling the moving punctures, this monodromy acts trivially on the absolute homology $H_1(X, \mathbb{Z})$.
   • Implication: The usual Picard–Lefschetz transvection mechanism, which typically generates non-algebraic behavior via absolute cycles in varying families, does not apply here. Non-algebraicity cannot be inferred merely from the existence of non-trivial absolute homology; it must arise from genuine logarithmic terms or specific Abelian period obstructions.

Key Contributions and Results

1. The Residue Criterion for Algebraicity
The principal positive result is a residue-based criterion (Theorem 4.1). The exterior Cauchy transform is algebraic if the lifted boundary $\gamma$ is separating on the normalization $X$.

• Definition: A lifted oval $\gamma$ is separating if it bounds an integral two-chain $\Sigma$ on $X$ (i.e., $[\gamma] = 0$ in $H_1(X, \mathbb{Z})$).
• Mechanism: If $\gamma = \partial \Sigma$, the integral can be evaluated via the residue theorem on the chain $\Sigma$. The result is a finite sum of residues at the fixed poles of the differential and the moving poles $\pi^{-1}(z)$. Since the residues are algebraic functions of $z$, the sum is algebraic.
• Consequences:
  • Rational Curves: Every smooth compact oval on a rational real algebraic curve has an algebraic exterior Cauchy transform (Corollary 4.4).
  • Dividing Curves: If a real curve is of Type I (dividing) and the domain is bounded by the complete real locus, the transform is algebraic.
  • Positive Genus: Positive genus does not inherently obstruct algebraicity; separating ovals in positive genus yield algebraic transforms.

2. Non-Separating Ovals and Period Obstructions
For non-separating ovals (where $[\gamma] \neq 0$), the residue theorem does not apply directly. The paper posits that algebraicity is not guaranteed and likely fails generically.

• The Obstruction: In the non-separating case, the integral may contain a non-vanishing Abelian period component that cannot be reduced to residues.
• Period-Rank Test: The authors propose a "period-rank test" (Section 6) to detect potential non-algebraicity. This involves checking if the homology class of the oval is invisible to the subspace of holomorphic differentials generated by the family $\eta \pi^n d\pi$. If the class is not invisible, a non-algebraic Abelian contribution is expected.
• Conjectures:
  • Conjecture 6.2: Separating ovals yield algebraic transforms; non-separating ovals are generically non-algebraic.
  • Conjecture 6.4: For a nonsingular real plane curve of even degree $2k \ge 4$ with exactly one real oval, the exterior Cauchy transform is not algebraic. This is because such curves are necessarily non-dividing (Proposition 5.4), making the unique oval non-separating.

3. Distinction from Quadrature Domains
The paper clarifies the relationship between algebraicity and quadrature domains.

• Quadrature domains have rational Cauchy transforms.
• The class of domains with algebraic transforms is strictly larger.
• The authors construct conditional examples (Proposition 5.6, Corollary 5.8) of simply connected domains with positive-genus boundaries that have algebraic (but non-rational) Cauchy transforms. These are separating examples that lie strictly outside the class of quadrature domains.

4. Specific Examples and Counter-Examples

• Ellipse: Demonstrates an algebraic (degree 2) but non-rational transform.
• Nodal Cubic: Shows that even on a rational normalization, if the lifted contour is not closed (i.e., relative endpoints at a singularity), logarithmic terms survive, rendering the transform non-algebraic.
• Smooth Weierstrass Cubic: Serves as a test case for the elliptic period obstruction. The transform reduces to an elliptic integral involving quasi-periods ($\zeta$-functions). The authors conjecture that for generic smooth cubics, these terms do not cancel algebraically, resulting in a non-algebraic transform.

Significance and Claims
The paper claims to provide a "homological formulation" that corrects previous misconceptions regarding the role of Picard–Lefschetz monodromy in this specific problem. By establishing that the normalization is fixed, the authors shift the focus from topological monodromy to the specific homological status of the lifted boundary (separating vs. non-separating).

• Positive Result: Separating boundaries (including all rational ovals and complete real loci of dividing curves) guarantee algebraicity via a residue sum.
• Negative Result (Conjectural): Non-separating boundaries generally lead to non-algebraic transforms due to surviving Abelian periods or logarithmic terms, though a rigorous proof for all cases (specifically smooth plane curves of degree $\ge 4$) remains an open problem.
• Nuance: The paper emphasizes that algebraicity is a property of the specific lifted contour's homology class relative to the differential, not merely a property of the curve's genus or the existence of a rational parametrization.

The authors conclude by framing the non-separating case as a research program, offering the period-rank test as a tool to isolate potential Abelian contributions, while acknowledging that proving non-algebraicity requires demonstrating the failure of finite monodromy or the persistence of logarithmic/quasi-periodic terms.