Function spaces of CW homotopy type are Hilbert manifolds

Jaka Smrekar, Atsushi YamashitaProceedings of the American Mathematical Society 2008 · DOI

Abstract

Let $X$ be a countable CW complex and $Y$ an Absolute Neighborhood Retract (ANR) for metric spaces. Let $Y^X$ denote the space of continuous maps from $X$ to $Y$ equipped with the compact-open topology. This paper demonstrates that, under mild restrictions, the following properties are equivalent: (1) $Y^X$ is an $\ell^2$-manifold, (2) $Y^X$ is an ANR, and (3) $Y^X$ has the homotopy type of a CW complex. These results bridge the gap between abstract homotopy theory and the concrete geometric structure of infinite-dimensional function spaces.

Introduction

In the field of topology, the space of all continuous functions between two spaces, $Y^X$, is a fundamental object of study. When the domain $X$ is compact, previous research—most notably by Katsuro Sakai in 1988—established that these function spaces often behave like Hilbert manifolds. These are infinite-dimensional spaces that locally resemble the Hilbert space $\ell^2$.

However, when the domain $X$ is non-compact, such as an infinite grid or a complex network modeled as a countable CW complex, the local properties of the mapping space become significantly more complex. In many cases, the mapping space may lose basic properties like local path-connectedness. This paper addresses the conditions under which these non-compact mapping spaces retain a "nice" geometric structure, specifically identifying when they function as ANRs or Hilbert manifolds.

Methodological Framework

The authors utilize several sophisticated tools from geometric topology to characterize these spaces:

  • Toruńczyk’s Characterization: This serves as the primary test for Hilbert manifolds. It requires a space to be an ANR and satisfy the Discrete Approximation Property (DAP), which ensures that maps into the space can be "wiggled" to make their images discrete.
  • Cauty’s Theorem: This deep result characterizes ANRs as metrizable spaces where every open set possesses the homotopy type of a CW complex.
  • Homotopy Theory of Fibrations: To handle non-compact domains, the authors decompose $X$ into an increasing sequence of finite subcomplexes. By treating restriction maps as fibrations, they can pass topological properties from finite-dimensional approximations to the full infinite-dimensional mapping space.

Key Insights: The "All-or-Nothing" Equivalence

The most significant contribution of this work is the discovery of a striking equivalence between local geometric properties and global homotopy properties. For the mapping spaces under consideration, the following are equivalent:

  1. Being an ANR (a local geometric property).
  2. Having the homotopy type of a CW complex (a global shape property).
  3. Being semilocally contractible (a local connectivity property).

In general topology, these properties are rarely equivalent. However, the inherent structure of function spaces forces them to coincide. Furthermore, the authors demonstrate that countability is essential; if the mapping space $Y^X$ is metrizable, the domain $X$ must almost certainly be a countable CW complex.

Main Results and Applications

Theorem 1.2: The Main Characterization

If $X$ is a countable CW complex and $Y$ is a separable, complete ANR with no isolated points, then the mapping space $Y^X$ is an $\ell^2$-manifold if and only if it has the homotopy type of a CW complex. This result implies a high degree of structural rigidity: if two pairs of spaces are homotopy equivalent, their respective mapping spaces are not merely homotopy equivalent, but actually homeomorphic.

Theorem 4.1: Algebraic Criteria for the Circle

The authors provide a practical application for maps into the circle $S^1$. They prove that $(S^1)^X$ is a Hilbert manifold if and only if the first cohomology group $H^1(X; \mathbb{Z})$ is finitely generated. This effectively translates a difficult question of infinite-dimensional topology into a straightforward calculation in algebraic topology.

Limitations and Future Directions

While this work significantly generalizes Sakai’s earlier results from compact to countable domains, several challenges remain:

  • Non-separable Spaces: The results do not easily extend to non-separable range spaces $Y$, where the mapping space may fail to be homogeneous (looking different at different points).
  • Characterization Difficulty: The authors note that a complete characterization of which pairs $(X, Y)$ result in a CW homotopy type for $Y^X$ is likely impossible, as it intersects with some of the most difficult unsolved problems in homotopy theory.
  • General Domains: The behavior of mapping spaces where the domain $X$ is an uncountable or non-metrizable complex remains largely unexplored and is expected to be significantly more chaotic.

Summary updated

Abstract

Let <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper X"> <mml:semantics> <mml:mi>X</mml:mi> <mml:annotation encoding="application/x-tex">X</mml:annotation> </mml:semantics> </mml:math> </inline-formula> be a countable CW complex and <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper Y"> <mml:semantics> <mml:mi>Y</mml:mi> <mml:annotation encoding="application/x-tex">Y</mml:annotation> </mml:semantics> </mml:math> </inline-formula> an ANR (for metric spaces) and let <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper Y Superscript upper X"> <mml:semantics> <mml:msup> <mml:mi>Y</mml:mi> <mml:mi>X</mml:mi> </mml:msup> <mml:annotation encoding="application/x-tex">Y^X</mml:annotation> </mml:semantics> </mml:math> </inline-formula> denote the space of continuous maps from <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper X"> <mml:semantics> <mml:mi>X</mml:mi> <mml:annotation encoding="application/x-tex">X</mml:annotation> </mml:semantics> </mml:math> </inline-formula> to <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper Y"> <mml:semantics> <mml:mi>Y</mml:mi> <mml:annotation encoding="application/x-tex">Y</mml:annotation> </mml:semantics> </mml:math> </inline-formula> with the compact-open topology. We show that, under mild restrictions, the following are equivalent: (1) <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper Y Superscript upper X"> <mml:semantics> <mml:msup> <mml:mi>Y</mml:mi> <mml:mi>X</mml:mi> </mml:msup> <mml:annotation encoding="application/x-tex">Y^X</mml:annotation> </mml:semantics> </mml:math> </inline-formula> is an <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="script l squared"> <mml:semantics> <mml:msup> <mml:mi> ℓ </mml:mi> <mml:mn>2</mml:mn> </mml:msup> <mml:annotation encoding="application/x-tex">\ell ^2</mml:annotation> </mml:semantics> </mml:math> </inline-formula> -manifold, (2) <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper Y Superscript upper X"> <mml:semantics> <mml:msup> <mml:mi>Y</mml:mi> <mml:mi>X</mml:mi> </mml:msup> <mml:annotation encoding="application/x-tex">Y^X</mml:annotation> </mml:semantics> </mml:math> </inline-formula> is an ANR, (3) <inline-formula content-type="math/mathml"> <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" alttext="upper Y Superscript upper X"> <mml:semantics> <mml:msup> <mml:mi>Y</mml:mi> <mml:mi>X</mml:mi> </mml:msup> <mml:annotation encoding="application/x-tex">Y^X</mml:annotation> </mml:semantics> </mml:math> </inline-formula> has the homotopy type of a CW complex. We also give a few interesting examples and applications.
BibTeX
@inproceedings{smrekar2008function,
  author = {Jaka Smrekar and Atsushi Yamashita},
  title = {{Function spaces of CW homotopy type are Hilbert manifolds}},
  booktitle = {Proceedings of the American Mathematical Society},
  year = {2008},
  doi = {10.1090/s0002-9939-08-09584-1},
}
Figure