🔢 mathematics

Characterisation of some multivalued harmonic functions on $\mathbb{R}^{n}$

This paper proves that $\mathbb{Z}_2$ -harmonic functions on $\mathbb{R}^n$ with quadratic growth at infinity and $\mathcal{O}(r^{3/2})$ local growth near a smooth codimension-two branching set are uniquely characterized up to rigid motions, confirming the uniqueness of recent constructions by Donaldson and Yan.

Original authors: Yang Li, Parsa Mashayekhi, Yichen Zhang

Published 2026-08-18

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yang Li, Parsa Mashayekhi, Yichen Zhang

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the vast landscape of modern geometry, mathematicians often study shapes that minimize area, much like a soap film stretching across a wire frame. These minimal surfaces are usually smooth and predictable, but in higher dimensions, they can develop complex singularities where the surface folds or branches. To understand these sharp corners and self-intersections, researchers use special mathematical tools called multivalued harmonic functions. Think of these not as single, simple curves, but as sheets that wrap around a central line, taking on two different values as you circle it, before snapping back together. These objects appear everywhere in advanced geometry, from the study of how space-time might be structured to the behavior of exotic materials. For years, mathematicians have known that these functions can exist, but a fundamental question remained: if you see a specific pattern of growth far away and a specific way the function behaves near its branching point, is that pattern unique? Could there be many different shapes that look the same from a distance and behave the same way up close, or is there only one true form?

A team of researchers has now answered this question with a definitive proof. They demonstrated that for a specific class of these multivalued functions in spaces of three or more dimensions, the behavior at the edges and the behavior near the center completely determine the shape, provided the branching set is a smooth, compact submanifold. If you have a function that grows in a specific quadratic manner as you move infinitely far away, and if it vanishes in a particular way near its branching line, then that function must be the exact same shape as a known model, up to simple movements like rotation or shifting its position. The researchers did not just guess this; they proved it rigorously. They showed that any other shape that tried to fit these descriptions would inevitably lead to a mathematical contradiction. This means that the "Lawlor neck," a specific geometric shape previously constructed by other mathematicians, is the only possible solution for these conditions.

The journey to this proof required the team to bridge two very different worlds of mathematics. On one side, they used analytical tools to understand how the function behaves locally, proving that the function cannot have any unexpected zeros or irregularities away from its main branching line. On the other side, they employed a sophisticated method from symplectic topology, a field that studies the geometry of phase spaces in physics. They treated the graph of the function as a high-dimensional surface and compared it to a known model surface. By analyzing how these surfaces could intersect, they found that if the two surfaces were different, they would have to cross each other in a way that violates the laws of energy and area. The proof relies on a delicate argument showing that any difference between the unknown function and the known model would create a tiny, impossible geometric object that cannot exist.

The researchers also established that if the branching set is a smooth, compact submanifold and the function satisfies the specific growth conditions, then that branching set must be an ellipsoid. They proved that the function's growth at infinity is tied directly to the shape of this branching set. If the function grows like a specific type of quadratic polynomial, the branching set must be an ellipsoid. This connection between the local behavior near the branch and the global behavior at infinity is what makes the result so powerful. It means that the entire structure is locked into place by these two boundary conditions. The team's work confirms that the examples constructed by Simon Donaldson and Dashen Yan are not just isolated curiosities but are the unique representatives of their class.

This result settles a long-standing uncertainty in the field. Before this work, it was possible that there were many different ways to construct such functions, perhaps with different branching shapes or different growth rates that happened to look similar. The new proof rules out all those possibilities. It shows that the geometry is rigid; once you fix the growth at infinity and the vanishing order at the branch, the entire function is fixed. This rigidity is crucial for applications in other areas of geometry, such as understanding how certain high-dimensional spaces can collapse or deform. By knowing that the model is unique, mathematicians can use it as a reliable building block for more complex theories without worrying about hidden variations.

The paper also touches on what happens if the conditions are relaxed. The authors note that if the growth at infinity were allowed to be degenerate, meaning it flattened out in certain directions, the problem would reduce to a simpler, lower-dimensional version. However, under the strict conditions of non-degenerate growth and smooth branching, the uniqueness holds firm. The proof is a tour de force of modern geometric analysis, combining deep insights from the study of minimal surfaces with the powerful machinery of Floer theory, a tool used to count and classify geometric objects. It stands as a clear example of how abstract mathematical structures, when examined with the right tools, reveal a surprising and absolute order. The multivalued harmonic functions, once thought to be a flexible family of shapes, are now understood to be a single, unique entity defined by its boundaries.

Technical Summary: Characterisation of some multivalued harmonic functions on Rn

Problem Statement
The paper addresses the global uniqueness of $Z_2$ -harmonic functions on $\mathbb{R}^n$ ( $n \geqslant 3$ ) subject to specific growth conditions. A $Z_2$ -harmonic function $u$ is defined as a harmonic section of a flat real line bundle over $\mathbb{R}^n \setminus \Sigma$ , where $\Sigma \subset \mathbb{R}^n$ is a smooth, compact, codimension-2 branching set. The function possesses $Z_2$ -monodromy around loops linking $\Sigma$ .

Recent constructions by Donaldson (on $\mathbb{R}^3$ ) and Yan (on $\mathbb{R}^n$ ) provided explicit examples of such functions where:

The branching set $\Sigma$ is an ellipsoid.
The function exhibits quadratic growth at infinity, asymptotic to a non-degenerate, trace-free quadratic form of index 1 or $n-1$ .
The leading coefficient in the local asymptotic expansion near $\Sigma$ vanishes, implying $u = O(\text{dist}(\cdot, \Sigma)^{3/2})$ . This condition treats $\Sigma$ as a nonlinear free boundary rather than a prescribed set.

The central question is whether these specific growth properties (quadratic asymptote and vanishing leading coefficient at the branch locus) uniquely characterize the solutions up to rigid motions of $\mathbb{R}^n$ , without assuming the geometry of $\Sigma$ a priori.

Methodology
The proof of the main uniqueness theorem (Theorem 1.4) is divided into two distinct phases: analytical preparations and a Floer-theoretic argument.

Analytical Preparations:
- Frequency Function: The authors adapt Almgren's frequency function, originally used for $Q$ -valued harmonic functions, to the $Z_2$ -harmonic setting. They establish a monotonicity formula for the $Z_2$ -harmonic 1-form $du$.
- Non-degeneracy: By relating the vanishing order of $du$ near the singular set to the polynomial growth rate at infinity, they prove that $du$ is non-degenerate (the subleading coefficient $B$ in the expansion is nowhere vanishing) and that $|du|^{-1}(0) = \Sigma$ . This ensures the branching set is exactly the zero set of the gradient.
- Approximate Special Lagrangians: For a small parameter $\delta > 0$ , the graph $L_\delta = \text{graph}(\delta du)$ is shown to be a connected, exact, graded, $C^{1,\alpha}$ -embedded Lagrangian in $\mathbb{C}^n \cong T^*\mathbb{R}^n$ . It is asymptotic to two transverse planes $\Pi_\delta^\pm$ and satisfies an approximate special Lagrangian condition with a Lagrangian angle error of order $O(\delta^{1+\alpha})$ .
Floer-theoretic Part:
- Partial Compactification: Adapting the framework of Imagi–Joyce–Oliveira dos Santos (IJOdS), the authors construct a partial compactification $M$ of $\mathbb{C}^n$ by attaching cotangent bundles of spheres at infinity. This allows the non-compact Lagrangians to be treated as compact objects within a symplectic Calabi–Yau manifold.
- Classification in $DbFuk$: Using the derived Fukaya category, they classify the compactified Lagrangian $\bar{L}_\delta$ . The classification relies on the asymptotic angles, which are determined by the Hessian of the asymptotic quadratic polynomial $V$ . They prove that the index of $V$ must be 1 or $n-1$ .
- Hamiltonian Perturbation: To handle the non-transversality of the approximate solutions, the authors introduce a family of Hamiltonian perturbations. A key technical achievement is removing intersection points of Floer degree 0 and $n$ in regions away from the branching locus and infinity. This is achieved via a real analyticity argument (perturbing the Hessian difference) and careful control of the Lagrangian potential near the singular set.
- Holomorphic Strip Argument: Assuming two distinct solutions $u$ $u$ and $v$ $v$ exist, the authors construct a holomorphic strip (or broken strip) connecting their perturbed Lagrangians.
  - Upper Bound: The area of this strip is bounded from above by the difference in Lagrangian potentials, which is shown to be $O(\delta \Lambda^{-1/4})$ (small) due to the matching asymptotic conditions.
  - Lower Bound: Using a local monotonicity formula for holomorphic curves near a point where the gradients differ, they derive a lower bound for the area of order $\eta \delta$ .
  - Contradiction: The discrepancy between the upper bound (vanishing as parameters are tuned) and the lower bound (linear in $\delta$ ) yields a contradiction, implying the solutions must coincide.

Key Contributions and Results

Theorem 1.4 (Global Uniqueness): The paper proves that any $Z_2$ -harmonic function on $\mathbb{R}^n$ ( $n \geqslant 3$ ) with a non-empty smooth compact codimension-2 branch locus, vanishing leading coefficient at the branch locus, and asymptotic to a non-degenerate homogeneous degree 2 polynomial (with index 1 or $n-1$ ) is unique up to rigid motions. It coincides with the model solutions constructed by Donaldson and Yan.
Characterization of the Branching Set: Unlike previous existence results where the branching set was often prescribed or constructed alongside the function, this result demonstrates that the quadratic growth at infinity and the $O(r^{3/2})$ local growth condition uniquely determine the branching set as an ellipsoid.
Extension of IJOdS Techniques: The work successfully adapts the classification of special Lagrangians (specifically the Lawlor necks) to the setting of $Z_2$ -harmonic functions, which serve as linearized models for branched deformations of special Lagrangians.
Analytical Rigor: The paper provides a rigorous proof of the non-degeneracy of the $Z_2$ -harmonic 1-form using Almgren's frequency function in a context where the Dirichlet energy minimization property is not assumed a priori.

Significance
The paper establishes a rigidity result for a class of multivalued harmonic functions that appear ubiquitously in modern differential geometry, including Almgren's regularity theory, gauge-theoretic blow-up limits (Taubes, Haydys–Walpuski), and Donaldson's program on $G_2$ -manifolds.

The authors note that while previous works (Sun, Haydys–Mazzeo–Takahashi) established uniqueness for a fixed branching set $\Sigma$ , this result removes that assumption, showing that the asymptotic behavior at infinity forces the branching set to be the specific ellipsoidal geometry found in the model examples. The paper avoids solving the nonlinear special Lagrangian equation directly, instead utilizing the linear harmonic equation and Floer-theoretic tools to deduce uniqueness, offering a novel pathway for studying branched minimal surfaces and their deformations.

The authors conclude by noting two related open questions: the potential generalization of these results to self-similar solutions of the " $Z_2$ -heat equation" (self-expanders of Lagrangian mean curvature flow) and the behavior of solutions when the asymptotic polynomial is degenerate (which reduces the problem to lower dimensions).

More like this