Bi-Lipschitz Smoothing under Ricci and Injectivity Bounds

This paper resolves a specific open problem by proving that any complete Riemannian manifold with uniform lower bounds on both Ricci curvature and injectivity radius admits an $L^\infty$-close bi-Lipschitz smooth metric that preserves two-sided Ricci curvature bounds and a uniform positive lower bound on injectivity radius.

The Gist

Imagine you have a piece of fabric that represents the shape of the universe (a mathematical "manifold"). In this paper, the author is dealing with a piece of fabric that is a bit rough, crinkled, and maybe even has some sharp kinks or weird bumps. However, this fabric has two very important safety features:

1. It's never too tight: No matter where you look, you can always stretch a small circle on the fabric without it folding back on itself immediately (this is the Injectivity Radius).
2. It never curves inward too sharply: The fabric doesn't have any "saddle" points that pinch it into a sharp negative curve; it's generally flat or curves outward gently (this is the Ricci Curvature).

The problem is that this fabric is not smooth. It's a bit jagged. In the world of physics and geometry, we usually prefer our fabrics to be perfectly smooth so we can do calculations (like predicting how light travels or how gravity works).

The Big Question:
Can we take this rough, jagged fabric and "iron it out" to make it perfectly smooth, without changing its overall shape too much? And crucially, can we do this while making sure the fabric still has those two safety features (it doesn't get too tight, and it doesn't get too curvy)?

The Answer:
Yes! Maja Gwóźdź proves that you can.

The "Ironing" Process (The Analogy)

Think of the rough fabric as a crumpled map. You want to smooth it out so you can read the streets clearly, but you don't want to stretch the map so much that the distance between New York and London becomes the same as the distance between New York and Paris. You also don't want to smooth it so aggressively that the paper tears or the roads disappear.

Here is how the author does it, step-by-step:

1. The "Zoom and Resize" Trick
First, the author takes the fabric and resizes it. Imagine zooming in on a specific part of the map so that the smallest "safe circle" you can draw is exactly 1 unit big. This makes the math easier to handle. It's like putting the fabric on a standard-sized table.

2. The "Softening" Filter
Next, the author applies a mathematical "smoothing filter." Think of this like running a gentle, high-tech iron over the fabric.

• The Catch: You can't just iron it however you want. If you iron too hard, you might stretch the fabric until it's thin and weak. If you iron too lightly, the wrinkles stay.
• The Solution: The author uses a special technique (called "controlled smoothing") that acts like a smart iron. It knows exactly how much pressure to apply. It smooths out the sharp kinks but ensures the fabric doesn't stretch or shrink by more than a tiny, controlled amount (this is the Bi-Lipschitz part—meaning the distances stay roughly the same).

3. Checking the Safety Features
After the ironing, the author has to check if the fabric is still safe.

• Did we tear it? No, because the "smoothing" was gentle enough.
• Is it still flat enough? Yes. The author uses a clever trick involving the "volume" of the fabric. They prove that because the original fabric had a certain amount of "bulk" (volume) in every small area, the smoothed version must also have enough bulk.
• The "Cheeger-Gromov-Taylor" Safety Net: This is a fancy mathematical rule that says: "If a piece of fabric has enough bulk and isn't curving wildly, then it can't have any sharp, dangerous folds." The author uses this rule to prove that the smoothed fabric still has a healthy "Injectivity Radius" (it won't fold on itself).

4. The Result
The final product is a perfectly smooth fabric that:

• Looks almost exactly like the original rough one (you can't tell the difference with the naked eye).
• Is still safe to walk on (no sharp folds).
• Has predictable curvature (no weird, unpredictable bumps).

Why Does This Matter?

In the real world, we often deal with data or shapes that are "noisy" or "rough." Maybe we are trying to model the shape of a galaxy, or the structure of a protein, or the curvature of spacetime near a black hole. These models often start out rough or incomplete.

This paper gives mathematicians and physicists a guaranteed recipe to take a rough, imperfect model and turn it into a smooth, usable one, without breaking the fundamental laws of the shape they are studying. It answers a specific question asked by other mathematicians (the "Bandara Question") and confirms that we can always "smooth out" the universe's rough edges, provided the universe isn't too weird to begin with.

In short: You can iron out the wrinkles of the universe without stretching the fabric or tearing the seams.

Technical Summary

Here is a detailed technical summary of the paper "Bi-Lipschitz Smoothing under Ricci and Injectivity Bounds" by Maja Gwóźdź.

1. Problem Statement

The paper addresses Question 2 from the Morgan–Pansu list of open problems (proposed by L. Bandara at the 2018 "Modern Trends in Differential Geometry" conference). The core problem concerns the regularization of Riemannian manifolds with specific geometric lower bounds.

The Question:
Given a complete Riemannian manifold $(M^n, g)$ satisfying:

1. A uniform lower bound on the injectivity radius: $\text{inj}(M, g) \geq \ell > 0$.
2. A uniform lower bound on the Ricci curvature: $\text{Ric}_g \geq k g$ for some $k > 0$.

Does there exist a smooth Riemannian metric $h$ on $M$ that is bi-Lipschitz close to $g$ (i.e., $C^{-1}g \leq h \leq Cg$) while simultaneously satisfying:

• A uniform positive lower bound on the injectivity radius: $\text{inj}(M, h) \geq L > 0$.
• A two-sided bound on the Ricci curvature: $|\text{Ric}_h| \leq K$.

Crucially, the constants $C, L, K$ must depend only on the dimension $n$, the injectivity radius bound $\ell$, and the Ricci curvature bound $k$.

2. Methodology

The proof relies on a combination of geometric analysis tools, specifically controlled smoothing, volume comparison, and injectivity radius estimates. The methodology proceeds in the following logical steps:

A. Scaling and Compactness

1. Rescaling: The author first rescales the metric $\bar{g} = \ell^{-2}g$ to normalize the injectivity radius bound to $\text{inj}(M, \bar{g}) \geq 1$.
2. Compactness: Since $k > 0$, the Bonnet–Myers theorem implies that $(M, g)$ has finite diameter. Combined with completeness, this ensures $M$ is compact. This compactness is essential for applying Croke's volume lower bound.
3. Ricci Normalization: Under the rescaling, the Ricci curvature satisfies $\text{Ric}_{\bar{g}} \geq 0$.

B. Harmonic Coordinate Control

The proof utilizes the framework of controlled smoothing developed by Petersen, Wei, and Ye [2].

1. Weak Harmonic Norms: The author establishes that the rescaled manifold $(M, \bar{g})$ admits a bound on its "weak harmonic $L^{1,p_0}$-norm" (where $p_0 = 2n$). This is derived using the lower Ricci bound and the injectivity radius bound.
2. Morrey Estimates: Using Morrey–Sobolev inequalities (Lemmas 3 and 4), the $L^{1,p_0}$ control is upgraded to a $C^{0,\alpha}$ (Hölder) control of the metric coefficients in harmonic coordinates.
3. Smoothing Theorem: The Petersen–Wei–Ye smoothing theorem is applied. This theorem guarantees that if a metric has bounded weak harmonic norms, there exists a smooth metric $\bar{h}$ that is bi-Lipschitz close to the original metric and possesses uniform $C^{2,\alpha}$ bounds in harmonic coordinates.

C. Deriving Geometric Bounds for the Smoothed Metric

Once the smoothed metric $\bar{h}$ is constructed, the author derives the required geometric bounds:

1. Curvature Bounds: Using the $C^{2,\alpha}$ bounds on the metric coefficients in harmonic coordinates, Lemma 5 (a standard elliptic estimate) is applied to bound the Riemann curvature tensor $|Rm_{\bar{h}}|_{\bar{h}}$. Since the Ricci tensor is a contraction of the Riemann tensor, this yields a two-sided bound $|\text{Ric}_{\bar{h}}|_{\bar{h}} \leq \bar{K}$.
2. Volume Lower Bound:
   • Croke's theorem [3] provides a universal lower bound for the volume of balls in $(M, \bar{g})$ based on the injectivity radius.
   • The bi-Lipschitz equivalence between $\bar{g}$ and $\bar{h}$ allows the transfer of this volume lower bound to the smoothed metric $\bar{h}$.
3. Injectivity Radius Lower Bound:
   • The Cheeger–Gromov–Taylor estimate [4] is applied to $(M, \bar{h})$. This estimate relates the injectivity radius to the volume of small balls and the sectional curvature bounds.
   • Since $\bar{h}$ has bounded sectional curvature (derived from the Riemann bound) and a uniform volume lower bound, the Cheeger–Gromov–Taylor estimate yields a uniform positive lower bound for $\text{inj}(M, \bar{h})$.

D. Rescaling Back

Finally, the metric is rescaled back to the original scale ($h = \ell^2 \bar{h}$). The bi-Lipschitz constants, injectivity radius, and curvature bounds are adjusted according to the scaling laws (Lemma 2) to produce the final constants $C, L, K$.

3. Key Contributions

• Resolution of an Open Problem: The paper provides a definitive affirmative answer to Question 2 in the Morgan–Pansu list, confirming that manifolds with lower Ricci and injectivity bounds can be approximated by smooth metrics with two-sided Ricci bounds and preserved injectivity radius.
• Explicit Constants: The proof demonstrates that the resulting constants ($C, L, K$) depend only on the dimension $n$, the initial injectivity radius $\ell$, and the initial Ricci bound $k$.
• Synthesis of Techniques: The work effectively bridges the gap between geometric analysis (Ricci bounds) and smoothing techniques (controlled smoothing in harmonic coordinates), showing how volume comparison theorems (Croke) and injectivity radius estimates (Cheeger–Gromov–Taylor) can be used to control the geometry of the smoothed metric.

4. Main Results

Theorem 1 (Smoothing under inj and Ric lower bounds):
For every integer $n \geq 2$ and constants $\ell, k > 0$, there exist constants $C, L, K > 0$ (depending only on $n, \ell, k$) such that for any complete Riemannian manifold $(M^n, g)$ with $\text{inj}(M, g) \geq \ell$ and $\text{Ric}_g \geq k g$, there exists a smooth complete metric $h$ satisfying:

1. Bi-Lipschitz Equivalence: $C^{-1} g \leq h \leq C g$.
2. Injectivity Radius: $\text{inj}(M, h) \geq L$.
3. Two-Sided Ricci Bound: $|\text{Ric}_h|_h \leq K$.

5. Significance

This result is significant for several reasons in differential geometry:

• Regularity of Singular Spaces: It provides a robust method to approximate "rough" or singular geometric structures (defined by weak curvature bounds) with smooth structures without losing control over fundamental geometric quantities like injectivity radius.
• Compactness Arguments: Such smoothing results are often prerequisites for compactness theorems (e.g., Gromov-Hausdorff convergence) where one needs to work with smooth approximations to apply standard analytical tools.
• Geometric Analysis: It validates the use of smoothing techniques in contexts where both lower Ricci curvature and lower injectivity radius are present, a common setting in the study of Einstein manifolds and Ricci flow limits.

The paper successfully demonstrates that the combination of Ricci lower bounds (which control volume growth) and injectivity radius lower bounds (which prevent topological collapse) is sufficient to construct a smooth approximation with bounded curvature (both upper and lower), a non-trivial result since smoothing typically risks creating large curvature spikes.