Flat Model Structures for Nonunital Algebras and Higher K-Theory

Reading time: 5 minute
...

📝 Original Info

  • Title: Flat Model Structures for Nonunital Algebras and Higher K-Theory
  • ArXiv ID: 0906.4735
  • Date: 2005-03-01
  • Authors: J. D. Christensen, J. P. May

📝 Abstract

We prove the existence of a Quillen Flat Model Structure in the category of unbounded complexes of h-unitary modules over a nonunital ring (or a $k$-algebra, with $k$ a field). This model structure provides a natural framework where a Morita-invariant homological algebra for these nonunital rings can be developed. And it is compatible with the usual tensor product of complexes. The Waldhausen category associated to its cofibrations allows to develop a Morita invariant excisive higher $K$-theory for nonunital algebras.

💡 Deep Analysis

Figure 1

📄 Full Content

Let A be a nonunital algebra (or ring). A classical question in Homotopy Theory is to find a 'good definition' of K-theory and cyclic type homology for this type of rings and algebras. Namely, it is always possible to embed a nonunital ring A as a two-sided ideal of a unital ring R (for instance, by choosing à = Z ⋉ A to be the ring obtained by adjoining an identity to A). Thus, it is possible to define the notions of K-theory and cyclic type homologies for A in terms of this ring Ã. But this embedding of A into a unital ring R is not unique. And therefore, the different choices of R give rise to different definitions of homology theories and K-theory for A. This problem is known in the literature as the 'excision problem' in the different theories. In [35], Wodzicki proved that if R is a unitary k-pure extension of A, then it satisfies the excision property for Cyclic, Bar or Hochschild homology if and only if A is an H-unital k-algebra, in the sense that its Bar homology HB * (A, V ) = 0 for any k-module V (see e.g. [35,Theorem 3.1]). This result extended to algebras over commutative rings his remarkable result showing that if a (nonunital) ring A satisfies the excision property in rational algebraic K-theory, then the Q-algebra A ⊗ Q verifies the excision property in Cyclic homology [35] (see also [22,32]). But this H-unitality condition is rarely satisfied in practice and thus, many authors have extended these ideas and techniques to more general settings. For instance, Cuntz and Quillen have proven in [8,9,10] that arbitrary extensions of k-algebras (k ⊃ Q a field) with k-linear section satisfy excision in periodic cyclic cohomology. And Weibel in [34] constructed a homotopy invariant algebraic K-theory satisfying excision and cohomologic descent. The obstruction for the classical K-theory excision has also been studied in [5] (see also [6]).

In the present paper, we addopt the definition of h-unitary modules given by Suslin and Wodzicki (cf. [32, 7.3(ii)]). And we define the category M(A) of hunitary left A-modules as the full subcategory of all left A-modules consisting on those modules M satisfying that Tor à n (Z, M ) = 0 (or Tor à n (k, M ) = 0, if A is an algebra over a ground field k). At first sight, this definition of M(A) depends on the choosing embedding of A as a two-sided ideal of the unital ring R. And therefore we will refer to it as the h-unitary module category associated to the pair (R, A). But we will show in the last section of the paper that this h-unitary module category is independent of R.

We prove in Theorem 3.1 that any unitary left A-module M (in particular, the regular module A) has a (unique up to isomorphism) cover by an h-unitary module (equivalently, a minimal right approximation in the sense of Auslander [3]) by an h-unitary module. And then, we show that there is a very satisfactory version of relative homological algebra in the category of unbounded complexes of h-unitary modules. Namely, we prove in Section 3 that the class of all flat h-unitary Amodules imposes a cofibrantly generated Model Structure (see [27]) in the category C(M(A)) of all complexes in M(A). Let us note that, if we consider in C( Ã) the cofibrantly generated Flat Model Structure constructed in [17], then the bounded flat complexes which are generating cofibrations form a small Waldhausen category S in the sense of [33]. And the subset T ⊂ S of generating cofibrations which are flat h-unitary is a Waldhausen subcategory of it, since any cofibration in our model structure on C(M(A)) is a cofibration in C( Ã). Therefore we can use the general construction of K-theory on a Waldhausen category (see [33, Chapter IV]) to define a K-theory on A. This K-theory is going to be excisive in the sense of [35] by our results in the last section. In particular, we deduce that the group K 0 T is independent of the choice of S (see [33, Chapter II, Theorem 9.2.2]). The explicit construction of these K-groups in terms of our model structure will be developed in a forthcoming paper. We would like to stress that our construction does not restrict just to H-unital algebras, but it applies to any nonunital algebra (or ring) without the h-unitality condition.

Unfortunately, the category of h-unitary modules is far from being a monoidal category and therefore, we cannot expect to construct a monoidal model structure on C(M(A)). We solve this problem in Section 4, where we consider the wider subcategory F(A) of Ã-Mod consisting of all firm modules in the sense of [28]. The usual tensor product of A-modules is an endofunctor in F(A). And we show in Theorem 4.1 that this tensor product in F(A) induces a unitless monoidal structure in C(F(A)) which is compatible with the model structure we have induced in C(M(A)).

At this point, the question of when the additive category of h-unitary modules is abelian naturally arises. We show in Section 2 that this category is always an accessible category in the sense of [2,24]. Actual

📸 Image Gallery

cover.png

Reference

This content is AI-processed based on open access ArXiv data.

Start searching

Enter keywords to search articles

↑↓
ESC
⌘K Shortcut