nLab Karoubi K-theory

Contents

Context

Cohomology

cohomology

Special and general types

Special notions

Variants

Extra structure

Operations

Theorems

Algebraic topology

Contents

Karoubi 1978 §III.4 defined topological K-theory classes equivalently by Clifford module bundles, where a Cl nCl_n-module represents a class in K nK^n and represents the trivial class if, roughly, it extends to a Cl n+1Cl_{n+1}-module.

(Over the point this is the Atiyah-Bott-Shapiro isomorphism.)

Idea

We have a sequence of Clifford algebras Cl nCl_n which are generated by nn anticommuting square roots of ±1\pm 1. The sequence is periodic up to Morita equivalence; Cl 8Cl_8 is (16)\mathbb{R}(16), the algebra of 16×1616 \times 16 real matrices, which is Morita equivalent to \mathbb{R}, and from then on it repeats every 8 with extra matrix dimensions thrown in – this is Bott periodicity. (Notice that here we treat Clifford algebras as 2\mathbb{Z}_2-graded algebras: while Cl 6Cl_6 is Morita equivalent to (8)\mathbb{R}(8) as an algebra, it is not so as a graded algebra.)

It turns out that K n(X)K^n(X) can be represented geometrically by Clifford module bundles over XX. Start with K 0K^0; we know that elements of K 0(X)K^0(X) are ‘formal differences’ VWV - W of vector bundles over XX (virtual vector bundles). We can model the formal difference VWV - W with an honest geometric object by using the 2\mathbb{Z}_2-graded vector bundle VWV \oplus W, where VV is even and WW is odd. Such a thing should represent the zero class in K-theory just when VV and WW are isomorphic; this can be rephrased as saying that there exists an odd operator ee on VWV \oplus W (hence, taking VV to WW and vice versa) such that e 2=1e^2 = 1. But this just says that VWV \oplus W has an action of the first Clifford algebra Cl 1Cl_1.

The idea is now to generalize this to K-theory in other degrees.

Definition

By default, let the ground field be the complex numbers.

Consider

Then Karoubi 78 Def. 4.11 defines the K-theory group K p,q(X)K^{p,q}(X) to be the relative Grothendieck group of this forgetful functor.

This means [K78 §2.13] that K p,q(X)K^{p,q}(X) is given by equivalence classes [E,F,α][E,F,\alpha] of triples consisting of

  • E,FClMod X p,q+1E, F \,\in\, ClMod^{p,q+1}_X

  • α:u(E)u(F)\alpha \,\colon\, u(E) \xrightarrow{\sim} u(F) an isomorphism

subject to the following equivalence relations:

  1. an isomorphism of such triples (E i,F i,α i)(E_i, F_i, \alpha_i) is a pair of isomorphism f:E 1E 2f \colon E_1 \to E_2, g:F 1F 2g \colon F_1 \to F_2 which intertwine α\alpha in that the following diagram commutes:

    u(E 1) u(f) u(E 2) α 1 α 2 u(F 1) u(g) u(F 2) \begin{array}{ccc} u(E_1) &\overset{u(f)}{\longrightarrow}& u(E_2) \\ \mathllap{{}^{\alpha_1}} \Big\downarrow && \Big\downarrow \mathrlap{{}^{\alpha_2}} \\ u(F_1) &\underset{u(g)}{\longrightarrow}& u(F_2) \end{array}
  2. an equivalence of such triples is an isomorphism between their direct sum with any triples (E,F,α)(E,F,\alpha) for which EFE \simeq F and α\alpha is homotopic, via Cl p,qCl^{p,q}-isomorphisms, to the identity morphism (“elementary triples”).

According to K78 Thm. 4.12 this definition reproduces the ordinary definition of topological K-theory groups in degree pqp-q.

Categorification

Douglas & Henriques 2011 have proposed that this description of K-theory has a good categorification that might be relevant for tmf.

Here is a report by Mike Shulman on a talk by Chris Douglas on this topic, from which also part of the above text is taken.

References

The above definition is due to

Related results and discussion:

Relation to twisted K-theory:

  • Max Karoubi: Clifford modules and twisted K-theory, Proceedings of the International Conference on Clifford algebras (ICCA7) [arXiv:0801.2794]

A categorification using conformal nets:

Last revised on December 19, 2024 at 15:23:10. See the history of this page for a list of all contributions to it.