Intersections, sums, and the Jordan-Hölder property for exact categories (Q2031522)

The Jordan-Hölder theorem is a classical result from group theory which has been taken to different contexts such as the module category of a ring. To state this result we need to consider composition series, which are defined as finite chains of subobjects \(0=G_0 \subseteq G_1 \subseteq \cdots \subseteq G_n =G\) such that each quotient \(G_i / G_{i-1}\), called composition factor, is a simple object. Two composition series are said to be equivalent if they consist of the same number of elements and the composition factors are the same except for isomorphism and permutation. The theorem states that, if an object admits a composition series, then all composition series are equivalent. We should note that there are contexts where the statement of the theorem is false (see Example 6.9 in [\textit{T. Brüstle}, \textit{S. Hassoun}, \textit{D. Langford}, and \textit{S. Roy}, J. Pure Appl. Algebra 224, No. 4, Article ID 106212, 29p. (2020; Zbl 1423.18034)] or [\textit{H. Enomoto}, ``The Jordan-Hölder property and Grothendieck monoids of exact categories'', Preprint, \url{arXiv:1908.05446}]). The aim of the article under review is to study the exact categories (see [\textit{T. Bühler}, Expo. Math. Algebra 28, 1--69 (2010; Zbl 1192.18007)]) where the statement of the theorem is fulfilled, defining what the authors call the Jordan-Hölder property. It is worth mentioning that Enomoto previously tackled this problem in [Enomoto, loc. cit.], where he characterised the exact categories with Jordan-Hölder property through an invariant called Grothendieck monoid. In addition to the Jordan-Hölder property, one of the main results of the paper is a version of the Hopkins-Levitzki theorem for exact categories. As the title of the article points out, to achieve the main results on the Jordan-Hölder property, the authors define and study the concepts of admissible intersection and sum in an exact category. These concepts are defined via admissible monomorphisms, pullbacks and pushouts, in a similar way as one would do in abelian categories. It is worth mentioning that admissible intersections and sums will not always exist for an arbitrary exact category. The categories where they do exist are called Admissible Intersection and Sum (AIS) categories. The authors prove that every AIS category satisfies the Jordan-Hölder property. The case of exact categories that are not AIS is also analysed in the article. For this, maximals and minimals on the lattice of admissible subobjects are studied. With the language developed in doing so, a diamond axiom for exact categories is defined, which turns out to be a sufficient condition for the category to satisfy the Jordan-Hölder property. Finally, the authors define Artin-Wedderburn exact categories as those in which their semisimple objects can be characterised in a certain way, which includes a notion of admissible radical. It turns out that every Artin-Wedderburn category satisfies the Jordan-Hölder property. Moreover, for a Nakayama algebra \(\Lambda\), the exact Wedderburn-Artin structures of \(\operatorname{mod}(\Lambda)\) are exactly those which satisfy the Jordan-Hölder property.