Equivariant embedding of metrizable $G$-spaces in linear $G$-spaces

Aasa Feragen

TL;DR

This paper tackles the challenge of embedding complex, symmetric spaces (called $G$-spaces) into simpler, linear spaces while preserving their symmetry. The authors prove that any such space can be "nicely" embedded into a linear space, ensuring that the embedded space has a well-behaved neighborhood within the linear space. This result has important implications for understanding the structure of these symmetric spaces, showing that certain types of "retracts" are equivalent.

What Problem Does This Paper Address?

In topology, embedding a space within another space is a fundamental tool for understanding its properties. When dealing with spaces that have symmetries (e.g., a sphere with rotations acting on it, where the group $G$ represents the rotations), we often want to preserve these symmetries during the embedding process. This is known as an "equivariant embedding."

The paper focuses on a specific class of spaces called $\mathcal{M}_G$, which are proper, metrizable $G$-spaces with metrizable orbit spaces. Think of orbit spaces as the "shapes" you see if you collapse all points related by the symmetry group $G$ into single points. The challenge is to embed these $\mathcal{M}_G$ spaces into "linear $G$-spaces" – spaces that are both linear (like vector spaces) and have a symmetry action.

Previous work by Antonyan and Elfving had established some results in this direction, but they often relied on assumptions about the group $G$ being linear or the spaces having specific properties like being finite-dimensional. This paper aims to generalize these results to a broader class of Lie groups and $G$-spaces, specifically addressing the question of whether any $\mathcal{M}_G$ space can be embedded into a linear $G$-space such that the embedded space has a "nice" $G$-neighborhood within the linear space. This "niceness" is crucial for further topological constructions like extensions and retractions.

Methods

The paper employs a combination of established topological techniques and novel constructions. Key methods include:

  • Cartan G-spaces and Slices: The authors build upon the theory of Cartan $G$-spaces, which are $G$-spaces with certain local properties related to their symmetry. A crucial concept is that of a "slice," which is a local cross-section of the orbits of the $G$-action. These slices are fundamental for understanding the local structure of $G$-spaces.
  • Tubular Coverings: The paper utilizes the concept of a tubular covering, which is a collection of neighborhoods that are "tubular" around the orbits. The authors show that for spaces in $\mathcal{M}_G$ with a paracompact orbit space, a countable and locally finite tubular covering exists. This is achieved by leveraging the fact that $G$ has countably many compact conjugacy classes of subgroups.
  • Isovariant Maps and G-Homeomorphisms: A significant portion of the paper is dedicated to proving a key lemma (Lemma 4.2) that establishes when an isovariant $G$-map between two Cartan $G$-spaces is a $G$-homeomorphism. This relies on the induced map on the orbit spaces being a homeomorphism.
  • Construction of Linear G-Spaces: The core of the proof of the main theorem involves constructing a suitable linear $G$-space. This is done by first considering the case of a "global slice" (where the entire space is a single orbit) and then generalizing to arbitrary $\mathcal{M}_G$ spaces. This construction involves products of locally convex linear $G$-spaces and embeddings into Banach $G$-spaces.
  • Dugundji Extension Theorem: For the applications section, the authors utilize the equivariant Dugundji extension theorem, a powerful tool for extending maps from subsets to larger sets while preserving equivariance.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Generalization of Embedding Theorems: The paper successfully generalizes existing equivariant embedding theorems to a broader class of Lie groups and $G$-spaces, removing previous linearity assumptions on the group.
  • The Importance of G-Neighborhoods: A central theme is the requirement that the embedded space not only lies within a linear $G$-space but also possesses a $G$-neighborhood within that space that also belongs to the class $\mathcal{M}_G$. This is a crucial technical condition for further topological arguments.
  • Equivalence of G-ANE and G-ANR: A significant corollary is the demonstration that for the class of spaces $\mathcal{M}_G$, the notions of $G$-equivariant Absolute Neighborhood Extensor (G-ANE) and $G$-equivariant Absolute Neighborhood Retractor (G-ANR) are equivalent. This simplifies the classification of these spaces.
  • Connection to G-CW Complexes: The paper also shows that any $G$-ANE for $\mathcal{M}_G$ has the $G$-homotopy type of a $G$-CW complex, providing a more concrete model for these spaces.

Key Results

  • Theorem 5.1: The main result states that for any Lie group $G$ and any $G$-space $X \in \mathcal{M}_G$, there exists a $G$-embedding $e: X \to L$ into a locally convex linear $G$-space $L$. Furthermore, $e(X)$ is a closed subset of a $G$-invariant convex subset $C$ of $L$, and $e(X)$ has a $G$-neighborhood $V$ in $C$ such that $V \in \mathcal{M}_G$.
  • Corollary 6.1: Any $G$-ANE for $\mathcal{M}_G$ is $G$-homotopy equivalent to some $G$-CW complex.
  • Section 6.2: The paper proves that the classes $G$-ANE-$\mathcal{M}_G$ and $G$-ANR-$\mathcal{M}_G$ are equivalent.

Limitations and Future Work

  • Assumptions on $\mathcal{M}_G$: The results are specific to the class $\mathcal{M}_G$ (proper metrizable $G$-spaces with metrizable orbit spaces). While this is a broad class, further work might explore embeddings into other types of spaces or for different classes of $G$-spaces.
  • "Nice" Embeddings: The paper focuses on achieving a "nice" neighborhood. Further research could explore the properties of the embedding itself and the structure of the ambient linear $G$-space $L$.
  • Specific Lie Groups: While the paper generalizes beyond linear Lie groups, specific properties of certain Lie groups might lead to more refined embedding results or constructions.

The paper's findings directly pave the way for further investigations into the structure and classification of equivariant topological spaces, particularly in the context of non-compact Lie group actions. The equivalence of G-ANE and G-ANR is a significant simplification, and the connection to G-CW complexes provides a concrete way to study these spaces.

Summary updated

Abstract

Given a Lie group $G$ we study the class $\mathcal {M}_G$ of proper metrizable $G$-spaces with metrizable orbit spaces, and show that any $G$-space $X \in \mathcal {M}_G$ admits a closed $G$-embedding into a convex $G$-subset $C$ of some locally convex linear $G$-space, such that $X$ has some $G$-neighborhood in $C$ which belongs to the class $\mathcal {M}_G$. As a corollary we see that any $G$-ANR for $\mathcal {M}_G$ is a $G$-ANE for $\mathcal {M}_G$.
BibTeX
@inproceedings{feragen2008equivariant,
  author = {Aasa Feragen},
  title = {{Equivariant embedding of metrizable $G$-spaces in linear $G$-spaces}},
  booktitle = {Proceedings of the American Mathematical Society},
  year = {2008},
  doi = {10.1090/s0002-9939-08-09307-6},
}

Loading related content...

Figure