🔢 mathematics

Generic non-uniqueness of minimizing harmonic maps from a ball to a sphere

This paper demonstrates that any boundary map from the 3-ball to the 2-sphere can be slightly perturbed in the $W^{1,p}$ norm (for $p<2$ ) to admit multiple minimizing harmonic maps, utilizing a novel homotopy construction to resolve questions regarding norm control in homotopies.

Original authors: Antoine Detaille, Katarzyna Mazowiecka

Published 2026-02-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Antoine Detaille, Katarzyna Mazowiecka

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

Imagine you are a sculptor trying to stretch a piece of elastic fabric (representing a mathematical map) from a flat, circular table (a 3D ball) onto a perfect sphere (the target shape). Your goal is to stretch it in the most efficient way possible, using the least amount of "energy" (like stretching a rubber band as little as possible). This is the world of Harmonic Maps.

Usually, if you fix the edge of the fabric to a specific shape on the table, you expect there to be one single, perfect way to stretch the rest of the fabric to the sphere. This is called uniqueness.

However, this paper by Antoine Detaille and Katarzyna Mazowiecka reveals a surprising twist: Sometimes, there isn't just one best way; there are multiple equally good ways. Even worse, they show that this "confusion" isn't a rare accident. It's actually everywhere, hiding just beneath the surface of almost any shape you choose.

Here is the breakdown of their discovery using simple analogies:

1. The Setup: The Elastic Sheet and the Sphere

Think of the Ball ( $B^3$ ) as a hollow ball of clay. The Sphere ( $S^2$ ) is the surface of a beach ball.

You paint a specific pattern on the surface of the clay ball (the boundary).
You want to fill the inside of the clay ball with a smooth, continuous sheet that matches that pattern on the outside and lands perfectly on the beach ball.
The Goal: Do this with the least amount of "stretching energy."

2. The Old Belief vs. The New Reality

For a long time, mathematicians knew that if you picked a very specific, "generic" pattern on the edge, you would usually get one unique solution. It was like saying, "If I tie a knot in a specific way, there's only one way to pull the rope tight."

But they also knew that for some weird patterns, you could pull the rope tight in two different ways (maybe one way creates a knot inside, and the other doesn't).

The Big Question: Is this "two ways" thing a rare fluke, or is it common?

3. The Discovery: The "Tiny Nudge"

The authors prove that non-uniqueness is generic. This means:

No matter what pattern you start with, you can make a tiny, almost invisible change to the edge pattern.
This tiny change is so small that if you looked at it with a standard microscope, you wouldn't notice.
But, this tiny change creates a situation where there are at least two different ways to fill the ball with the least energy.

The Analogy: Imagine a tightrope walker. Usually, there is one perfect path to balance. The authors show that if you wiggle the rope just a microscopic amount (a "small $W^{1,p}$ change"), you can create a scenario where the walker has two equally stable, but completely different, paths to take. One path might have a "kink" (a singularity) in the middle, while the other is smooth. Both use the same amount of energy.

4. The Secret Weapon: The "Homotopy" (The Shape-Shifter)

How did they prove this? They used a mathematical tool called a homotopy.

Imagine you have two different patterns on the edge of your clay ball.
A homotopy is like a smooth movie that morphs Pattern A into Pattern B.
The authors needed to show that you can morph these patterns into each other without the fabric stretching wildly in between.

The Challenge: In higher dimensions, morphing shapes usually causes the fabric to stretch a lot (high energy) during the transition.
The Breakthrough: They invented a new way to morph the shapes. They showed that if the two patterns are identical everywhere except for a tiny speck, you can morph them while keeping the "stretching energy" of the transition very low.

Think of it like this: If you want to change the color of a giant flag from red to blue, you usually have to repaint the whole thing. But if you only change a tiny dot in the corner, you can do it without disturbing the rest of the flag. The authors found a mathematical way to do this "dot change" morphing so smoothly that the energy stays low.

5. The "Singularities" (The Kinks)

In this math world, a singularity is like a sharp kink or a tear in the fabric where the smoothness breaks down.

The authors found that by making their tiny change, they could create a boundary condition where one solution has zero kinks (perfectly smooth) and another has one kink (a singularity).
Even though one looks "messier" than the other, they both use the exact same amount of energy. Nature, in this mathematical universe, can't decide which one to pick!

Summary

This paper is a bit like discovering that almost every door in a house has a hidden second key.

Before: We thought most doors had only one key (one unique solution).
Now: We know that if you wiggle the lock just a tiny bit (a small mathematical perturbation), you can find a second key that opens the door just as easily.
The Method: They built a special "bridge" (the homotopy construction) that proves you can wiggle the lock without breaking the door frame.

Why does this matter?
It tells us that in the complex world of geometry and physics, "perfect" solutions are often unstable. If you are designing something based on these maps (like materials science or fluid dynamics), you can't assume there is only one answer. You have to be prepared for the possibility that the system might "snap" into a completely different, equally efficient shape with just a tiny nudge.

1. Problem Statement

The paper investigates the uniqueness of minimizing harmonic maps from the unit ball $B^3 \subset \mathbb{R}^3$ to the unit sphere $S^2$ .

Context: A harmonic map $u: B^3 \to S^2$ minimizes the Dirichlet energy $E(u) = \int_{B^3} |\nabla u|^2 dx$ subject to a fixed boundary condition $\phi = u|_{\partial B^3}$ .
The Issue: While it is known that for a generic boundary datum in the $W^{1,2}$ topology, the energy minimizer is unique (Theorem 1.1, attributed to Almgren), there exist specific boundary maps admitting multiple minimizers (some singular, some regular).
The Gap: Previous work (specifically [11]) suggested that if one perturbs the boundary datum in the $W^{1,p}$ topology with $p < 2$ , non-uniqueness might be a generic phenomenon. However, a rigorous proof establishing the density of such non-unique boundary data in $W^{1,p}$ was missing.
Goal: To prove that boundary data leading to non-uniqueness are dense in the space $W^{1,p}(S^2, S^2)$ for $p < 2$ . Specifically, any smooth boundary map can be approximated arbitrarily closely in the $W^{1,p}$ norm by a map that admits at least two distinct energy minimizers with different numbers of singularities.

2. Methodology

The authors employ a combination of variational analysis, homotopy theory, and geometric constructions.

A. The Homotopy Construction (Section 2)

The core technical novelty of the paper is a new homotopy construction that controls the $W^{1,p}$ norm ( $p < 2$ ) during the deformation between two maps.

The Challenge: Given two smooth maps $\phi, \psi: S^2 \to S^2$ that are close in $W^{1,p}$ and coincide outside a small ball $B_r(x)$ , can one construct a homotopy $H_t$ connecting them such that the entire path remains close to $\phi$ in the $W^{1,p}$ norm?
The Solution (Proposition 2.1 & Corollary 2.2): The authors construct a homotopy by rescaling the deformation region.
1. They start with an arbitrary homotopy $G$ between $\phi$ and $\psi$ .
2. They define a family of diffeomorphisms $\Phi_t$ that compress the support of the deformation into a smaller ball $B_\tau(x)$ (where $\tau \ll r$ ) while keeping the maps unchanged outside $B_{2r}(x)$ .
3. The new homotopy $H_t$ is defined by composing $G$ with these rescalings.
4. Key Estimate: Because $p < n$ (here $p < 3$ ), the change of variables theorem implies that the $W^{1,p}$ norm of the rescaled map scales with a factor involving $\tau^{(n-p)/p}$ . By choosing $\tau$ sufficiently small, the authors ensure that the $W^{1,p}$ norm of the difference $\|\phi - H_t\|_{W^{1,p}}$ can be made arbitrarily small, regardless of the initial "size" of the homotopy $G$ , provided the maps coincide outside a small ball.

B. Proof of Generic Non-Uniqueness (Section 3)

The proof of the main theorem (Theorem 1.2) follows a "sandwich" strategy:

Start with a Unique Case: Take an arbitrary smooth boundary map $\phi$ . By Theorem 1.1, there exists a nearby map $\phi_0$ (close in $W^{1,p}$ ) that has a unique minimizer $u_0$ with $M$ singularities.
Create a Non-Unique Target: Using a construction from [9] (based on Hardt-Lin), construct a map $\phi_1$ that is homotopic to $\phi_0$ , coincides with $\phi_0$ outside a tiny ball, and admits a unique minimizer $u_1$ with at least $M+1$ singularities.
Connect via Controlled Homotopy: Use the homotopy construction from Section 2 to connect $\phi_0$ and $\phi_1$ via a path $H_t$ that stays within $\epsilon$ of $\phi_0$ in the $W^{1,p}$ norm.
The Intermediate Value Argument: Define $\tau$ $τ$ as the supremum of times $t$ $t$ where the minimizers of $H_t$ $H_{t}$ have at most $M$ $M$ singularities.
- By stability results, $\tau \in (0, 1)$ .
- Consider sequences $s_i \nearrow \tau$ and $t_i \searrow \tau$ . The minimizers $u_{s_i}$ (with $\le M$ singularities) and $v_{t_i}$ (with $> M$ singularities) converge strongly in $W^{1,2}$ to minimizers $u$ and $v$ of the boundary datum $H_\tau$ .
- Singularity Counting:
  - $u$ must have $\le M$ singularities (otherwise, singularities would have appeared "out of nowhere" from the $s_i$ sequence, violating stability).
  - $v$ must have $> M$ singularities (otherwise, the $M+1$ singularities of $v_{t_i}$ would have to collapse into fewer points, which is impossible due to the uniform separation of singularities).
Conclusion: The boundary map $H_\tau$ admits at least two distinct minimizers ( $u$ and $v$ ) with different numbers of singularities. Since $H_\tau$ is arbitrarily close to the original $\phi$ , non-uniqueness is dense.

3. Key Contributions

Resolution of a Conjecture: The paper confirms the suspicion raised in [11] that non-uniqueness is generic in the $W^{1,p}$ topology for $p < 2$ .
Novel Homotopy Technique: The primary mathematical contribution is the norm-controlled homotopy construction (Theorem 2.1). This answers a specific question regarding whether homotopies between $W^{1,p}$ $W^{1, p}$ maps can be kept small in norm when the maps differ only on a small set.
- Note: The authors clarify that they solve a "weaker" version of the general open problem (Theorem 2.3) where the radius of the perturbation ball depends on the desired accuracy $\epsilon$ , rather than being fixed a priori. However, this is sufficient for their density result.
Strengthening Previous Results: Previous examples (e.g., [7]) showed the existence of a single boundary map with non-uniqueness. This paper proves that such maps are dense, meaning non-uniqueness is the rule rather than the exception in the $W^{1,p}$ topology.

4. Main Results

Theorem 1.2: Let $\phi \in C^\infty(S^2, S^2)$ . For every $\epsilon > 0$ , there exists a smooth $\psi$ such that $\|\phi - \psi\|_{W^{1,p}(S^2)} < \epsilon$ (for $p < 2$ ), and $\psi$ serves as a boundary datum for at least two energy-minimizing harmonic maps from $B^3$ to $S^2$ having different numbers of singularities.
Corollary: The set of boundary data admitting non-unique minimizers is dense in $W^{1,p}(S^2, S^2)$ .

5. Significance

Stability vs. Instability: The results highlight a sharp contrast between the $W^{1,2}$ and $W^{1,p}$ ( $p<2$ ) topologies. In $W^{1,2}$ , uniqueness is generic and stable under small perturbations. In $W^{1,p}$ ( $p<2$ ), the topology is "looser," allowing for the creation of singularities through small perturbations, destroying uniqueness generically.
Implications for Variational Problems: This work deepens the understanding of the Lavrentiev phenomenon and the structure of the space of harmonic maps. It suggests that in lower regularity spaces, the energy landscape is highly complex, with multiple local minima (minimizers) accessible via small deformations.
Methodological Impact: The homotopy construction technique developed here is likely applicable to other problems involving Sobolev maps into manifolds where controlling the norm of a deformation path is crucial, particularly in dimensions where $p < n$ .

In summary, Detaille and Mazowiecka demonstrate that the "generic uniqueness" observed in the standard energy space ( $W^{1,2}$ ) is an artifact of the specific topology; in the slightly weaker $W^{1,p}$ topology ( $p<2$ ), non-uniqueness is the dominant feature, and this is proven via a sophisticated rescaling argument that controls the homotopy norm.