Localizing subcategories in the bootstrap category of filtered \(\mathrm{C}^\ast\)-algebras (Q2180241)

Triangulated categories are a possible framework for homological algebra and homotopy theory. They are additive and carry a suspension automorphism \(\Sigma\) and a class of distinguished diagrams of the form \(X\to Y \to Z \to \Sigma X\). They also occur in noncommutative geometry because Kasparov theory and its many equivariant versions carry a triangulated category structure. Triangulated categories may be localized at certain full subcategories. This gives another triangulated category in which the objects of the given subcategory become zero. Here it is useful to restrict attention to full subcategories that are closed under direct sums and direct summands, suspensions, and under exact triangles. Such subcategories are called localising. One approach to understand a triangulated category is to classify its localising subcategories. This has been achieved in several interesting cases. This paper adds more examples where this is possible. These examples come from equivariant Kasparov theory. Here the relevant category is the bootstrap class in the equivariant Kasparov category of \(\mathrm{C}^\ast\)-algebras over a finite totally ordered space. An object of this Kasparov category is a \(\mathrm{C}^\ast\)-algebra with a filtration by a fixed number of ideals \(0 = I_0 \subseteq I_1 \subseteq I_2 \subseteq \cdots \subseteq I_n= A\). If \(n=1\), this is just a \(\mathrm{C}^\ast\)-algebra. If \(n=2\), it is a \(\mathrm{C}^\ast\)-algebra with a fixed ideal, which is equivalent to a \(\mathrm{C}^\ast\)-algebra extension. Localising subcategories are often described by defining a notion of support for objects in the category. One hopes that the set of objects whose support is contained in a given set should be a localising subcategory, and that all localising subcategories should be of this form. For the non-equivariant Kasparov category, the support of an object is a subset of the set Spec\((\mathbb{Z})\) that consists of the prime numbers and~\(\infty\). The support of an object~\(A\) is defined as the set of elements of Spec\((\mathbb{Z})\) for which the \(K\)-theory of~\(A\) with coefficients in the appropriate finite field or in the rational numbers is non-zero. As shown by \textit{I. Dell'Ambrogio} [J. \(K\)-Theory 8, No. 3, 493--505 (2011; Zbl 1243.19003)], this gives a bijection between the localising subcategories of the bootstrap class in Kasparov theory and the set of subsets of Spec\((\mathbb{Z})\). The key point in the proof is the Universal Coefficient Theorem. It says that an object in the bootstrap class is determined uniquely up to isomorphism by its \(K\)-theory. In the more general cases considered in this paper, it turns out that the lattice of localising subcategories is not distributive. Thus they cannot correspond to subsets of a certain ``spectrum''. Instead, the resulting lattice is related to the lattice of non-crossing partitions of the \((n+1)\)-gon, where \(n\) is the number of ideals in the filtration. More precisely, the lattice of localising subcategories is a product of many copies of this lattice, one for each element of Spec\((\mathbb{Z})\). The relevant notion of support takes \(K\)-theory with coefficients in a finite field or the rational numbers for all subquotients that are defined by the given ideal filtration. Since the \(K\)-theory groups of these subquotients are related by many exact sequences, they are not independent. The paper analyses which sets of subquotients can occur as supports of localising subcategories. And this is then related to non-crossing partitions.