On the smoothing theory delooping of disc diffeomorphism and embedding spaces

This paper generalizes the classical Morlet-Burghelea-Lashof-Kirby-Siebenmann smoothing theory delooping of disc diffeomorphism groups to various disc embedding spaces, establishing their equivalence to specific loop spaces of quotient classifying spaces and demonstrating how these deloopings unify Hatcher and Budney group actions into a framed little discs operad action.

The Gist

Imagine you are a master sculptor working with different types of clay: Smooth Clay (Smooth manifolds), Blocky Clay (Piecewise Linear or PL manifolds), and Rough Clay (Topological manifolds).

For a long time, mathematicians have been trying to understand the "twists and turns" (diffeomorphisms) you can make with a ball of Smooth Clay without tearing it or changing its shape fundamentally. A famous theorem from the 1970s told us that the space of all these possible twists is equivalent to a very specific, complex mathematical object involving the differences between our three types of clay.

The Big Problem:
This theorem worked great for the ball of clay itself (the disc). But what if you wanted to study how to embed (push) a smaller ball of clay inside a larger one? Or what if you wanted to study these shapes while keeping track of their "orientation" (framing)?

The old rules didn't quite fit these more complex scenarios. It was like having a perfect map for a single city, but no map for the entire highway system connecting them.

The New Discovery (The Paper's Goal):
Paolo Salvatore and Victor Turchin have updated the map. They proved that the same "twist-and-turn" logic applies not just to the ball itself, but to:

1. Embeddings: Pushing a small ball into a big one.
2. Modulo Immersions: Looking at the shape while ignoring how it was stretched (like looking at a shadow rather than the 3D object).
3. Framed Embeddings: Pushing the ball in while keeping track of its "compass direction" (framing).

They showed that all these complex spaces can be "delooped."

The Creative Analogy: The "Unwinding" Machine

To understand Delooping, imagine you have a very tangled ball of yarn (the complex space of embeddings).

• The Old View: You just looked at the tangled ball and said, "It's hard to understand."
• The New View (Delooping): The authors built a machine that unwinds the ball. They showed that if you unwind this tangled ball once, you get a simpler shape. If you unwind it again, you get an even simpler shape.

Specifically, they proved that the space of all ways to push a small disc into a big disc is mathematically equivalent to unwinding a simpler space $m+1$ times.

Think of it like this:

• The Tangled Ball: The space of all possible ways to embed a small disc into a big one.
• The Unwinding: Taking a "loop" out of the problem.
• The Result: The authors found that the tangled ball is actually just a "loop space" of a much simpler object: a space that measures the difference between how Smooth, Blocky, and Rough clay behave.

The "Magic" of the Actions (The Choreography)

Here is where it gets really cool. In math, these spaces aren't just static; they have actions (like dance moves).

• Budney's Action: Imagine you have a set of smaller discs floating inside a big one. You can move them around, shrink them, or grow them. This is like a choreography where the dancers (discs) move independently.
• Hatcher's Action: Imagine the whole stage (the big disc) rotates or twists. This is a different kind of dance move.

The authors asked: Can we combine these two dances into one super-choreography?

They proved YES. They showed that these spaces can be acted upon by a "Framed Little Discs" operad.

• The Metaphor: Imagine a dance floor where you have a group of dancers (the small discs).
  • You can move the dancers around the floor (Budney's action).
  • You can also spin the dancers themselves (Hatcher's action).
  • The authors proved you can do both at the same time in a perfectly coordinated way. It's like a dance where the dancers can shuffle their positions and spin in place simultaneously without tripping over each other.

Why Does This Matter? (The "So What?")

1. Simplifying the Complex: It turns a nightmare of high-dimensional geometry into a simpler problem of "loops" and "differences" between types of geometry. It's like realizing a complex 3D puzzle is actually just a 2D pattern wrapped around a cylinder.
2. The "Dimension 4" Mystery: In the world of math, 4 dimensions are weird. Smooth clay and Rough clay behave differently there in ways they don't in 3 or 5 dimensions. The authors had to be very careful with dimension 4, essentially saying, "The rules work everywhere, but in dimension 4, we have to look at the 'Blocky' (PL) version because the 'Rough' version is too messy."
3. Connecting the Dots: They connected the work of different mathematicians (Budney, Hatcher, Morlet) into one unified framework. It's like taking separate pieces of a jigsaw puzzle from different boxes and realizing they all fit into one giant picture.

Summary in One Sentence

Salvatore and Turchin discovered that the complex ways we can twist, turn, and embed shapes in high-dimensional space are mathematically equivalent to "unwinding" the differences between smooth, blocky, and rough geometries, and that these shapes can perform a synchronized dance of moving and spinning simultaneously.

The Takeaway: They didn't just solve a puzzle; they built a new lens that makes the entire landscape of high-dimensional shapes much easier to see and understand.

Technical Summary

Here is a detailed technical summary of the paper "On the Smoothing Theory Delooping of Disc Diffeomorphism and Embedding Spaces" by Paolo Salvatore and Victor Turchin.

1. Problem Statement and Context

The paper addresses a fundamental problem in differential topology and homotopy theory: understanding the homotopy type of spaces of diffeomorphisms and embeddings of discs relative to their boundaries. Specifically, it seeks to generalize and refine the celebrated Morlet-Burghelea-Lashof-Kirby-Siebenmann smoothing theory theorem.

• The Classical Result: The classical theorem establishes an equivalence for the group of diffeomorphisms of the $n$-disc relative to the boundary, $\text{Diff}_{\partial}(D^n)$, as an $(n+1)$-fold loop space:
  $$\text{Diff}_{\partial}(D^n) \simeq \Omega^{n+1}(\text{PL}_n / O_n) \quad (\text{and } \simeq \Omega^{n+1}(\text{TOP}_n / O_n) \text{ for } n \neq 4).$$
  Here, $\text{PL}_n$, $\text{TOP}_n$, and $O_n$ represent the groups of piecewise linear, topological, and orthogonal automorphisms of $\mathbb{R}^n$ (germs at the origin), respectively.

• The Gap: While this result identifies the homotopy type, it does not fully capture the rich algebraic structure of these spaces. Recent work by Ryan Budney introduced an $E_{n+1}$-action (an action of the little $(n+1)$-discs operad) on $\text{Diff}_{\partial}(D^n)$. The authors ask: Is the classical delooping compatible with this operadic action? Furthermore, does this delooping generalize to spaces of smooth embeddings $\text{Emb}_{\partial}(D^m, D^n)$ and framed embeddings $\text{Emb}^{\text{fr}}_{\partial}(D^m, D^n)$, and do these spaces admit compatible operadic structures?

• The Dimension 4 Obstacle: A significant technical hurdle in smoothing theory is the dimension $n=4$, where the relationship between smooth, PL, and topological structures is anomalous (e.g., the existence of exotic $\mathbb{R}^4$). The paper aims to clarify which results hold for $n=4$ and which require restriction to $n \neq 4$.

2. Methodology

The authors employ a sophisticated blend of smoothing theory, operad theory, and homotopy pullback techniques. Their methodology involves several key steps:

1. Simplicial Models and Germs: They work within the category of simplicial sets to rigorously define mapping spaces (diffeomorphisms, homeomorphisms, embeddings) using "germs" of maps near the origin or boundary. This allows for precise handling of local flatness and framing.
2. Intermediate Framed Spaces: A core innovation is the introduction of intermediate spaces of framed homeomorphisms and framed embeddings.
   • They define $\text{Homeo}^{O_n}_{\partial}(N)$ as the space of triples $(f, F, H)$, where $f$ is a homeomorphism, $F$ is a lift to the tangent bundle (or microbundle), and $H$ is a path connecting $F$ to the linearization of $f$.
   • They prove that for $n \neq 4$, the inclusion $\text{Diff}_{\partial}(N) \hookrightarrow \text{Homeo}^{O_n}_{\partial}(N)$ is a homotopy equivalence. This bridges the gap between smooth and topological categories while preserving the framing structure necessary for operadic actions.
3. Homotopy Cartesian Squares: The authors utilize homotopy pullback squares (homotopy cartesian squares) to relate spaces of diffeomorphisms, homeomorphisms, and formal immersions. They restate classical smoothing theory results (Burghelea-Lashof) in this modern homotopy-theoretic language to establish equivalences between smooth and topological/PL spaces.
4. Operadic Actions: They explicitly construct actions of framed operads (specifically the overlapping discs operad $R_{m+1}$ and its framed versions $E^{O_m}_{m+1}$) on these spaces. They demonstrate that the delooping maps are not just equivalences of spaces but equivalences of algebras over these operads.
5. Scanning and Immersion Theory: They utilize Smale-Hirsch type immersion theory and scanning maps to relate embedding spaces to loop spaces of Stiefel manifolds (e.g., $V_{n,m} = O_n/O_{n-m}$) and their topological/PL counterparts.

3. Key Contributions and Results

The paper presents three main theorems (A, B, and C) that generalize and refine previous results.

Theorem A: Delooping of Diffeomorphisms

• Result: For $n \neq 4$, the space $\text{Diff}_{\partial}(D^n)$ is equivalent to $\Omega^{n+1}(\text{TOP}_n/O_n)$ and $\Omega^{n+1}(\text{PL}_n/O_n)$ as $E_{n+1}$-algebras (specifically $E^{O_n}_{n+1}$-algebras).
• Significance: This confirms that the Budney $E_{n+1}$-action is compatible with the classical smoothing theory delooping. For $n=4$, they establish an equivalence of spaces (but not of $E_1$-algebras) to $\Omega^5(\text{PL}_4/O_4)$.

Theorem B: Delooping of Embedding Spaces

• Result: For $1 \le m \le n$(with restrictions on$n=4$ and codimension 2), the authors establish deloopings for:
  1. Standard Embeddings: $\text{Emb}_{\partial}(D^m, D^n) \simeq \Omega^m \text{hofiber}(V_{n,m} \to V^t_{n,m})$.
  2. Framed Embeddings: $\text{Emb}^{\text{fr}}_{\partial}(D^m, D^n) \simeq \Omega^{m+1}(O_n \backslash \text{TOP}_n / \text{TOP}_{n,m})$.
  3. Embeddings Modulo Immersions: $\text{Emb}_{\partial}(D^m, D^n) \simeq \Omega^{m+1}(\text{TOP}_n / \text{TOP}_{n,m})$.
• Compatibility: These equivalences are shown to be compatible with the $E^{O_m \times O_{n-m}}_{m+1}$-action (a framed little discs action).
• Codimension 2 Exception: For $n - m = 2$, the equivalences hold only for the union of path-components corresponding to invertible elements in $\pi_0$ (i.e., knots with homotopy circle complements).
• Dimension 4: For $n=4$, the results hold for the subspace of PL-trivial knots (denoted by a $\star$), as the TOP delooping is not available in dimension 4.

Theorem C: Combination of Actions

• Result: The paper constructs a model for the framed embedding space that combines Budney's $E_{m+1}$-action (from the source disc) and Hatcher's $O_{m+1}$-action (pre-composition).
• Equivalence: $\text{Emb}^{\text{fr}}_{\partial}(D^m, D^n) \simeq O_{m+1} \Omega^{m+1}(\text{tEmb}(S^m, S^n) // O_{n+1})$.
• Significance: This unifies two distinct group actions into a single framed little discs operad action ($E^{O_{m+1}}_{m+1}$), providing a unified algebraic structure for these embedding spaces.

Corollaries and Applications

• Finiteness of Homotopy Groups: The authors use the delooping to analyze the homotopy groups of $\text{TOP}_{n,n-2}$ and $\text{PL}_{n,n-2}$. They show that for $n \ge 5$, these groups have non-finitely generated homotopy groups, implying they are not equivalent to $O_2$.
• PL Triviality in Dimension 4: They prove that among smooth knots $D^2 \hookrightarrow D^4$, only the trivial knot is PL-trivial. This implies $\pi_0 \text{Emb}_{\partial}(D^2, D^4)_{\star} = 0$.
• Conjecture on Boavida-Weiss: They discuss the relationship between their smoothing theory delooping and the operadic delooping via the Goodwillie-Weiss manifold calculus, conjecturing that the Boavida-Weiss map is an equivalence.

4. Significance and Impact

1. Unification of Structures: The paper successfully unifies the classical smoothing theory (which relates smooth and topological structures) with modern operadic homotopy theory (which describes the algebraic structure of configuration spaces). It proves that the "delooping" is not just a homotopy equivalence of spaces but an equivalence of structured algebras.
2. Resolution of Compatibility: It resolves the question of whether the Budney $E_{n+1}$-action is compatible with the Morlet delooping, affirming that it is for $n \neq 4$.
3. Handling Dimension 4: The paper provides a nuanced analysis of the dimension 4 case, distinguishing between smooth, PL, and topological categories where standard smoothing theory fails. It clarifies exactly which parts of the theory survive in dimension 4 (specifically for PL-trivial knots).
4. New Tools for Embedding Spaces: By introducing framed intermediate spaces and explicit operadic actions, the authors provide a robust framework for studying embedding spaces, which are central to knot theory, manifold calculus, and the study of diffeomorphism groups.
5. Foundational for Future Research: The results serve as a bridge between the classical work of Kirby, Siebenmann, and Lashof and the modern work of Willwacher, Randal-Williams, and Boavida de Brito, offering a precise target for future investigations into the rational homotopy of these spaces and their relation to Feynman diagrams and graph complexes.

In summary, Salvatore and Turchin have revitalized smoothing theory by embedding it into the framework of operads, providing a comprehensive and structurally rich description of disc diffeomorphisms and embeddings across all dimensions (with specific caveats for $n=4$).