on degree of mapping and Lipschitz constant

Overview

This post discusses the fundamental relationship between the topological complexity of a map (its degree) and its geometric “stretch” (the Lipschitz constant), a cornerstone of Quantitative Topology popularized by Mikhail Gromov [@gromov1999metric].

🏷️ The Setup: Lipschitz Mappings

Quantitative topology asks how topological invariants scale with geometric parameters. The most basic of these is the mapping degree. We consider a map $f:M^n\to N^n$ between oriented Riemannian manifolds.

Definitions

• Lipschitz Constant ($L$): $L={\sup }_{x\ne y}\frac{d_N(f(x),f(y))}{d_M(x,y)}$. It measures the maximum speed of the map.
• Topological Degree ($\mathrm{deg}(f)$): The integer representing the “wrapping number” of the map. It can be viewed homologically ($f_{\ast }[M]=d[N]$) or analytically (${\int }_M{f}^{\ast }\omega =d{\int }_N\omega$).

A crucial observation is that if $L<(\text{Vol}(N)/\text{Vol}(M){)}^{1/n}$, then the map cannot have degree $\ge 1$ simply because it cannot stretch the volume of $M$ enough to cover $N$. This intuition leads to the following global bound.

The Degree-Lipschitz Inequality

The absolute value of the topological degree $\mathrm{deg}(f)$ is bounded by the $n$-th power of the Lipschitz constant:

$$|\mathrm{deg}(f)|\le \frac{\text{Vol}(M)}{\text{Vol}(N)}{L}^n$$

🔍 Detailed Analytic Derivation

To bridge the gap between the metric definition of $L$ and the topological definition of $\mathrm{deg}(f)$, we use the language of differential forms.

Proof Sketch (Differential Forms)

Let ${\omega }_N$ be the normalized Riemannian volume form on $N$.

1. Integral Identity: By definition, $|\mathrm{deg}(f)|\cdot \text{Vol}(N)=|{\int }_M{f}^{\ast }{\omega }_N|$.
2. Jacobian Bound: Locally, the pullback of the volume form is $f^{\ast }{\omega }_N(x)=\det (d{f}_x){\omega }_M(x)$.
3. Linear Algebra: By Hadamard’s Inequality, $|\det (d{f}_x)|\le \Vert d{f}_x{\Vert }^n$. Since $f$ is $L$-Lipschitz, the operator norm of its derivative $\Vert d{f}_x\Vert$ is bounded by $L$ almost everywhere by Rademacher’s theorem.
4. Integration:

$$|{\int }_M{f}^{\ast }{\omega }_N|\le {\int }_M|\det (d{f}_x)|{\omega }_M\le {\int }_M{L}^n{\omega }_M=L^n\text{Vol}(M)$$

The bound is remarkably general; it does not depend on the specific geometry of $M$ or $N$ beyond their volumes and dimensions. It implies that to achieve a high degree, the map must be locally very fast.

🌊 Sharpness and the Packing Construction

A natural question is whether this bound is “best possible”. For the simplest manifolds, the answer is yes.

The Sphere Case ( $S^n\to S^n$)

For spheres, the bound simplifies to $|\mathrm{deg}(f)|\le L^n$. This is sharp up to a constant.

• Construction: Partition $S^n$ into $d\approx L^n$ disjoint geodesic balls $B_i$ of radius $\sim 1/L$.
• Mapping: Map each “bubble” $B_i$ to cover the entire target sphere. This is a common theme in discretization, similar to on convergence of graph Laplacian to manifold’s Laplacian.

This “bubble” construction suggests that $L^n$ is the “available volume” for wrapping. If we can pack many small regions that each cover the target, we can maximize the degree.

📉 Topological Obstructions: Scalability

However, for more complex manifolds, the topology can “get in the way” of the packing. This leads to the concept of Scalability.

Obstructions from Rational Homotopy

A significant discovery is that for certain manifolds, the $L^n$ growth is unreachable.

• Scalable Manifolds: These are manifolds like $S^n$ or $\mathbb{C}{P}^n$ where the topology doesn’t prevent local stretching. Here, we can achieve $\mathrm{deg}(f)\sim L^n$.
• Non-Scalable Manifolds: These manifolds (often identified via their Rational Homotopy Type) have “higher-order” topological constraints. For instance, if a manifold is non-formal, there are Massey products in its cohomology that must be preserved, which can force the map to “waste” its Lipschitz budget. In these cases, $|\mathrm{deg}(f)|$ might grow only as $L^{\alpha }$ for some $\alpha <n$.

Extension to Infinite Dimensions

The concept of degree is not limited to finite dimensions. In the study of non-linear PDEs, one often uses the Leray-Schauder degree. While the Lipschitz-constant bound doesn’t have a direct infinite-dimensional analog (as volume forms don’t behave nicely), the principle of using topological invariants to bound analytical complexity remains a guiding light.

Technical Note: Rademacher's Theorem

Although $f$ is only Lipschitz, Rademacher’s theorem ensures it is differentiable almost everywhere.

• Implication: The Jacobian $\det (d{f}_x)$ is a well-defined $L^{\infty }$ function, and the set of “kinked” points (measure zero) does not contribute to the volume integral. This matches the treatment of singularities in on boundary integral with stationary phase.

📝 Notes

• Quantitative Topology: This field (pioneered by Gromov and Guth [@guth2010metaphors]) investigates the “geometric cost” of topological features. Similar scaling laws appear in the analysis of on eigenvalue estimate of kernel.
• Hopf Invariant: For maps $f:S^3\to S^2$, the Lipschitz constant $L$ bounds the Hopf invariant by $L^4$, reflecting the 4-dimensional nature of the invariant’s definition via integral geometry.

🔗 See Also

• on convergence of graph Laplacian to manifold’s Laplacian --- Both rely on the local-to-global volume stretching properties of manifolds; the “bubble” construction for degree sharpness is the discrete analog of the tube partitions used here.

📚 References