Partial tensor-product functors and crossed-product functors

Julian Kranz, Timo Siebenand

Abstract

For a given discrete group $G$, we apply results of Kirchberg on exact and injective tensor products of $C^*$-algebras to give an explicit description of the minimal exact correspondence crossed-product functor and the maximal injective crossed-product functor for $G$ in the sense of Buss, Echterhoff and Willett. In particular, we show that the former functor dominates the latter.

TL;DR

This paper provides a concrete mathematical bridge between two different ways of building $C^*$-algebras from group actions: tensor-product functors and crossed-product functors. By leveraging a specific construction by Eberhard Kirchberg, the authors give explicit descriptions of "exotic" crossed products and prove that the minimal exact correspondence functor always dominates the maximal injective functor.

Introduction: The Search for Exotic Crossed Products

In the study of $C^*$-algebras, researchers often examine how a group $G$ acts on an algebra $A$. To study this interaction, they build a "crossed product" algebra. Standard methods produce either the "maximal" or "reduced" crossed product. However, for certain "non-exact" groups, the reduced crossed product behaves poorly, leading to counterexamples in the famous Baum–Connes conjecture.

This has sparked interest in "exotic" crossed products—completions that sit somewhere between the maximal and reduced versions. While several exotic functors (like the minimal exact correspondence and maximal injective functors) have been defined abstractly, their internal relationships and explicit structures remained fuzzy. This paper aims to provide a clear "picture" of these functors to understand how they relate to one another.

Methodology: The Kirchberg Tensor Product

The authors use tools from category theory and operator algebra theory. The key "ingredient" is a specific tensor-product functor discovered by Kirchberg, denoted as $-\otimes_{i,\epsilon}-$. This functor is unique because it is "injective" in its first variable and "exact" in its second.

Diagram (2) showing the interaction of the Kirchberg tensor product with various short exact sequences.

The researchers prove their results by:

  1. Embedding Theorems: Showing that these complex crossed-product functors can be realized as sub-algebras of simpler tensor products involving the reduced group $C^*$-algebra.
  2. Universal Properties: Using the defining characteristics of these functors (exactness, injectivity, and the "completely positive map" property) to show that one must be "larger" (dominate) the other.
  3. G-injective Algebras: Utilizing specific types of algebras (like $\ell^\infty(G)$) to create a more concrete representation of the minimal exact correspondence functor.

Key Insights

  • The Kirchberg Connection: The most significant "aha moment" is realizing that Kirchberg’s specific tensor product ($-\otimes_{i,\epsilon}-$) is the perfect tool to describe these exotic crossed products. It acts as the missing link between the internal structure of the group and the external structure of the algebra.
  • Functor Hierarchy: The paper establishes a clear hierarchy: the minimal exact correspondence functor is "stronger" (dominates) the maximal injective functor. This helps narrow down the search for the "right" crossed product to use in the Baum-Connes conjecture.
  • Solving Open Questions: The authors provide a positive solution to a technical question previously asked by experts in the field regarding the role of $G$-injective algebras in these constructions.

Results: Explicit Embeddings and Dominance

The authors establish several formal results regarding the structure of these functors.

Naturality diagram for crossed product functors, illustrating the relationship between maps of $C^*$-algebras and their corresponding crossed products.
  • Theorem B: Provides explicit injective homomorphisms (embeddings) for both the maximal injective and minimal exact correspondence functors. For example, $A \rtimes_{\text{inj}} G$ can be seen as a subset of $(A \rtimes_r G) \otimes_{i,\epsilon} C^*(G)$.
  • Corollary D: Formally proves the inequality $-\rtimes_{\text{inj}} G \le -\rtimes_{E_{\text{Corr}}} G$. This means the norm on the minimal exact correspondence crossed product is always greater than or equal to the norm on the maximal injective one.
  • Equivalence for Exact Groups: The paper confirms that for "well-behaved" (exact) groups, all these exotic functors collapse back into the standard reduced crossed product.

Conclusion and Future Work

The authors note that while they have shown one functor dominates the other, they haven't proven they are different for all non-exact groups. To prove that the maximal injective and minimal exact correspondence functors actually coincide, one would need to construct a crossed-product functor that is simultaneously exact and injective. Whether such a functor exists for all groups remains an open and challenging question in the field.

Summary updated

Abstract

For a given discrete group $G$, we apply results of Kirchberg on exact and injective tensor products of $C^*$-algebras to give an explicit description of the minimal exact correspondence crossed-product functor and the maximal injective crossed-product functor for $G$ in the sense of Buss, Echterhoff and Willett. In particular, we show that the former functor dominates the latter.
BibTeX
@inproceedings{kranz2021partial,
  author = {Julian Kranz and Timo Siebenand},
  title = {{Partial tensor-product functors and crossed-product functors}},
  booktitle = {Proceedings of the American Mathematical Society},
  year = {2021},
}
Figure