On the Lavrentiev gap for manifold-valued maps

This paper investigates the conditions under which smooth maps are modulary dense on compact manifolds, specifically analyzing the validity and failure of this property in the context of the Lavrentiev gap.

The Gist

Here is an explanation of the paper "On the Lavrentiev Gap for Manifold-Valued Maps," translated into simple language with creative analogies.

The Big Picture: Smoothing Out the Rough Edges

Imagine you are an artist trying to paint a perfect, smooth picture on a curved surface (like a globe or a donut). You have a "rough draft" of the painting that has some jagged, messy lines. Your goal is to see if you can always replace that rough draft with a perfectly smooth version that looks almost exactly the same, without changing the overall shape or meaning of the art.

In mathematics, this is called approximation. The "rough draft" is a function with some mathematical roughness (a Sobolev map), and the "perfect smooth version" is a smooth, infinitely differentiable function.

Usually, if your canvas is simple enough, you can always smooth out the rough draft. But sometimes, the rules of the game change, and you hit a wall. This paper explores when you can smooth things out and when you hit a wall that cannot be crossed.

The Setting: The "Double-Phase" Canvas

The authors are working with a very specific type of canvas. Imagine a material that behaves differently depending on where you are on it.

• In some spots, it's soft and stretchy (like rubber).
• In other spots, it's stiff and rigid (like steel).

This is called a Double-Phase material. The "rules" for how much energy it takes to stretch the material change from point to point. The paper asks: If we have a map (a shape) drawn on this weird, changing material, can we always find a smooth version of it?

The Two Main Rules for Success

The authors found that there are two main ways to guarantee you can smooth out your map. Think of these as two different "keys" to unlock the door to smoothness.

Key 1: The "Super-Stretchy" Rule (The Growth Condition)

Imagine the material gets incredibly stretchy as you pull it harder. If the material is "stretchy enough" (mathematically, if the energy function grows fast enough), then the rough spots in your map are naturally forced to smooth themselves out.

• The Analogy: It's like trying to fold a piece of paper that is so thin and flexible that it naturally settles into a flat, smooth sheet. No matter how crumpled you try to make it, it wants to be smooth.
• The Result: If the material follows this rule, you can always approximate your rough map with a smooth one. There are no topological tricks needed; the physics of the material does the work for you.

Key 2: The "Shape-Shifter" Rule (The Topology Condition)

What if the material isn't super-stretchy? What if it's stiff? Then, the shape of the target object matters.

• The Analogy: Imagine trying to wrap a gift. If the gift is a simple ball (a sphere), you can wrap it smoothly. But if the gift is a donut (a torus) with a hole in the middle, and your wrapping paper has a tear, you might not be able to fix the tear without cutting the paper.
• The Result: If the target shape (the manifold) has "holes" or specific loops that are too complex for the material to handle, you might get stuck. However, if the target shape is "simple enough" (mathematically, if it is k-connected, meaning it has no holes up to a certain size), then you can still smooth it out.
• The Catch: If the material is just on the edge of being too stiff, you might only be able to get "close" to the smooth version (weak density) rather than hitting it perfectly (strong density).

The Villain: The Lavrentiev Gap

Now, let's talk about the problem the paper is famous for solving: The Lavrentiev Gap.

Imagine you are trying to find the path of least resistance (minimum energy) to get from Point A to Point B.

• Scenario A: You are allowed to walk on a bumpy, rough path.
• Scenario B: You are forced to walk on a perfectly smooth path.

Usually, the smooth path is just a slightly polished version of the rough path, and the energy cost is the same.
But the Lavrentiev Gap is a trap. It happens when the smooth path is significantly more expensive (higher energy) than the rough path. It's as if the smooth road requires you to climb a mountain, while the rough road lets you take a secret tunnel through the mountain.

The Paper's Discovery:
The authors proved that if the "Double-Phase" material changes its rules too abruptly (specifically, if the transition between the soft and stiff parts happens too fast), this gap appears.

• The Metaphor: Imagine a road that is paved with asphalt for a mile, then suddenly switches to jagged rocks for a foot, then back to asphalt. If the switch is too sudden, a smooth car (a smooth map) cannot drive over it without breaking its suspension (infinite energy). But a bumpy, off-road vehicle (a rough map) can drive right over it easily.
• The Consequence: Because the smooth car can't drive the road, the "smooth approximation" fails. You cannot get close to the rough map using smooth maps. The gap is real, and the smooth maps are not dense.

The Counter-Example: The "Fractal" Trap

To prove this isn't just a theory, the authors built a specific mathematical "monster."

• They created a map on a cube that looks smooth everywhere except for a very strange, fractal-like set of points (a Cantor set).
• On this set, the material rules break down.
• They showed that while a rough map can exist with low energy, no sequence of smooth maps can ever get close to it. The smooth maps would require infinite energy to navigate the "fractal cliffs."

Summary: What Does This Mean for Us?

1. Smoothness is usually possible: If the material rules are well-behaved (not changing too wildly) or if the target shape is simple, you can always smooth out your rough mathematical maps.
2. The "Gap" is real: If the material rules change too abruptly (violating a specific mathematical condition), a "Lavrentiev Gap" opens up. This means the smoothest possible version of your map is fundamentally different from the rough version. You cannot approximate the rough one with the smooth one.
3. Why it matters: This helps engineers and physicists understand when their models will work. If they are designing a material with rapidly changing properties (like a composite material), they need to know if their mathematical models will break down or if they need to account for these "gaps" where smooth solutions don't exist.

In a nutshell: The paper tells us exactly when we can smooth out a crumpled piece of paper on a weird, changing surface, and when the paper is so crumpled and the surface so tricky that no amount of smoothing will ever make it flat.

Technical Summary

Here is a detailed technical summary of the paper "On the Lavrentiev Gap for Manifold-Valued Maps" by Carlo Alberto Antonini, Filomena De Filippis, and Cintia Pacchiano Camacho.

1. Problem Statement

The paper addresses the fundamental problem of approximating Sobolev maps between Riemannian manifolds in non-homogeneous function spaces. Specifically, it investigates the density of smooth maps ($C^\infty(M, N)$) within the Musielak-Orlicz Sobolev space $W^{1,\varphi}(M, N)$, where $M$ and $N$ are compact Riemannian manifolds.

The core issue is the Lavrentiev phenomenon: a situation where the infimum of a variational integral over smooth functions is strictly greater than the infimum over the natural energy space ($W^{1,\varphi}$). If this gap exists, smooth maps are not dense in the Sobolev space, preventing standard approximation techniques.

The authors focus on Musielak-Orlicz spaces, where the energy density $\varphi(x, t)$ depends explicitly on both the spatial variable $x \in M$ and the magnitude $t$. The prototypical example is the double-phase integrand:
$$\varphi(x, t) = t^p + a(x)t^q$$
where $a(x)$ is a non-negative Hölder continuous coefficient that vanishes on a set of measure zero, creating a transition between $p$-growth and $q$-growth regimes. This setting models strongly anisotropic materials and non-uniformly elliptic systems.

2. Methodology

The authors employ a combination of geometric analysis, functional analysis, and fractal constructions to establish their results.

• Geometric Extension (The Double): To handle manifolds with boundaries, the authors utilize the "double" of the manifold $D(M)$, allowing them to reduce boundary problems to the case of closed manifolds ($\partial M = \emptyset$).
• Vanishing Web Oscillations: A central technique is the concept of "vanishing web oscillations." A map has this property if, for any $\epsilon$, the manifold can be decomposed into a "web" (a set of measure zero) and open components such that the map's oscillation on the boundaries of these components is small. Proving that maps in $W^{1,\varphi}$ possess this property is crucial for density.
• Integral Conditions and Growth: The authors derive sharp integral conditions on the function $\varphi$ (specifically its infimum $\varphi^-_M$) that guarantee the existence of vanishing web oscillations. These conditions relate to the integrability of the inverse of the Young function.
• Topological Obstructions: When integral conditions fail, the authors investigate topological obstructions. They utilize the $k$-connectedness of the target manifold $N$ (triviality of homotopy groups $\pi_j(N)$ for $j \le k$) to construct approximations via retractions and triangulations, adapting methods from Bethuel, Hang, and Lin.
• Counterexample Construction: To demonstrate the sharpness of their assumptions, they construct a counterexample using fractal Cantor sets. They adapt the construction of Balci, Diening, and Surnachev to the manifold-valued setting, creating a map that cannot be approximated by smooth maps when the growth condition on the exponents is violated.

3. Key Contributions and Results

A. Density without Topological Restrictions (Theorem 1.1)

The authors establish that if the function $\varphi$ satisfies specific integral conditions (allowing the space to be slightly larger than or contained in $W^{1,m}$), then smooth maps are dense in $W^{1,\varphi}(M, N)$ regardless of the topology of $M$ or $N$.

• Condition: The integral conditions (1.15) for $m=2$ and (1.16) for $m \ge 3$ control the growth of $\varphi$ at infinity.
• Result: Under these conditions, maps in $W^{1,\varphi}$ possess vanishing web oscillations, implying $C^\infty(M, N)$ is dense.
• Significance: This extends previous results from Orlicz spaces (where $\varphi$ depends only on $t$) to the more general Musielak-Orlicz setting (where $\varphi$ depends on $x$).

B. Density with Topological Restrictions (Theorem 1.2)

When the integral conditions of Theorem 1.1 are not met, density can still be achieved if the target manifold $N$ satisfies a topological condition.

• Condition: $N$ must be $k$-connected (i.e., $\pi_j(N) = 0$ for $j \le k$).
• Interplay: The exponent $\gamma$ (from the almost-decreasing condition of $\varphi$) must satisfy $\gamma \le k+1$.
  • If $\gamma < k+1$, strong density holds.
  • If $\gamma = k+1$, weak density holds.
• Significance: This complements Theorem 1.1 by showing that topological constraints can compensate for weaker growth conditions on the energy functional.

C. Absence of the Lavrentiev Phenomenon (Corollary 1.3)

A direct consequence of the density results is the absence of the Lavrentiev gap.
$$\inf_{u \in W^{1,\varphi}} \int_M \varphi(x, |Du|) = \inf_{u \in C^\infty} \int_M \varphi(x, |Du|)$$
This holds in both settings (integral conditions or topological conditions), ensuring that variational problems in these spaces can be approximated by smooth functions.

D. Sharpness and Counterexample (Theorem 1.4)

The paper provides the first vectorial counterexample of the Lavrentiev phenomenon for manifold-valued maps in the double-phase setting.

• Failure Condition: If the condition relating the exponents $p, q$ and the Hölder exponent $\alpha$ of the coefficient $a(x)$ is violated:
  $$q > p + \alpha \max\left\{1, \frac{p-1}{m-1}\right\}$$
  (Note: This is a sharpened version of the standard condition $q \le p + \alpha \max\{1, p/m\}$).
• Construction: The authors construct a map $u$ into a sphere $S^{N-1}_\Lambda$ (which is $(N-2)$-connected) on a cube $Q_M$ that belongs to $W^{1,\varphi}$ but cannot be approximated by smooth maps.
• Significance: This proves that the local assumption (1.8) regarding the interaction between the spatial variable and the growth rates is necessary. Without it, the Lavrentiev phenomenon occurs even if the target manifold has trivial topology.

4. Significance

• Generalization: The work significantly generalizes the theory of smooth approximation from classical Sobolev spaces ($W^{1,p}$) and autonomous Orlicz spaces to non-autonomous Musielak-Orlicz spaces.
• Double-Phase Regularity: It provides a rigorous foundation for the regularity theory of double-phase functionals, which are critical in modeling materials with strongly anisotropic properties (e.g., composite materials).
• Topological vs. Analytic: The paper clarifies the delicate balance between analytic conditions (growth of the integrand) and topological conditions (homotopy groups of the target) required for density.
• Sharpness: By providing a counterexample that violates the growth condition, the authors establish the precise threshold for the Lavrentiev phenomenon in the vectorial, manifold-valued context, resolving a long-standing question about the necessity of the condition $q \le p + \alpha \max\{1, p/m\}$ (or its refined version).

In summary, this paper characterizes the necessary and sufficient conditions for the density of smooth maps in manifold-valued Musielak-Orlicz spaces, proving that the Lavrentiev gap is absent under specific integral or topological conditions, and demonstrating that these conditions are sharp through explicit counterexamples.