Twisted commutators and internal crossed modules

DOI10.32513/ASETMJ/1932200824027MaRDI QIDQ7033610zbMATH OpenFDO

Authors Bo Shan Deval, Tim Van der Linden

Publication date 3 February 2025

Published in Advanced Studies: Euro-Tbilisi Mathematical Journal (Search for Journal in Brave)

Full work available at URL https://doi.org/10.32513/ASETMJ/1932200824027

commutator internal action internal crossed module semi-abelian category

Internal categories and groupoids (18D40) Protomodular categories, semi-abelian categories, Mal'tsev categories (18E13) 2-groups, crossed modules, crossed complexes (18G45)

Review text

Commutator theory is an active area of research in universal algebra and in categorical algebra, where different notions of commutativity and commutators were introduced and studied in a categorical context. Examples are the Huq and the Higgins commutators of subobjects, and the Smith commutator of internal equivalence relations. In the article under review, the classical Huq commutativity and the Higgins commutator are generalized, and the notion of commutator twisted by an action is applied in the study of crossed modules in a semi-abelian category.\N\NIn a pointed category with finite products, one says that two morphisms \( f:X\to Y \) and \( g:B\to Y \) with the same codomain \textbf{Huq-commute} if there exists a morphism \( \varphi_{f,g}:X\times B\to Y \) such that \( \varphi_{f,g}(1_X,0)=f \) and \( \varphi_{f,g}(0,1_B)=g \). Here \( (1_X,0) \) is the morphism induced by the universal property of the product \( X\times B \). In the category of groups, given two subgroups \( i_1:H_1\hookrightarrow G \) and \( i_2:H_2\hookrightarrow G \), \( i_1 \) and \( i_2 \) commute if and only if \( h_1h_2=h_2h_1 \) for any elements \( h_1\in H_1 \) and \( h_2\in H_2 \). In a unital category, i.e. a pointed finitely complete category such that the morphisms \( (1_X,0) \) and \( (0,1_B) \) are jointly extremally epimorphic for any objects \( X \) and \( B \), it becomes a property for two morphisms \( f \) and \( g \) to commute.\N\NBesides the Huq commutator, there is also the \textbf{Higgins commutator} that detects if two morphisms commute, or more generally measures ``how close they are to commuting when this is not the case. More precisely, given two morphisms \( f:X\to Y \) and \( g:B\to Y \) in a category that is unital and normal, i.e. pointed regular such that any regular epimorphism is a normal epimorphism, one constructs the so-called cosmash product \( \iota_{X,B}:X\diamond B\to X+B \) as the kernel of the canonical morphism \( <(1_X,0),(0,1_B)>:X+B\to X\times B \). In the category of groups, \( X\diamond B \) is generated by the elements of \( X+B \) of the form \( xbx^{-1}b^{-1} \) for elements \( x\in X \) and \( b\in B \). Then the Higgins commutator \( [f,g] \) is defined to be the image in the (regular epimorphism, monomorphism)-factorization of the composite \( <f,g>\iota_{X,B} \). In the category of groups, the Higgins commutator \( [i_1,i_2] \), where \( i_1 \) and \( i_2 \) are as above, is exactly the inclusion of the subgroup generated by the commutators \( h_1h_2(h_1)^{-1}(h_2)^{-1} \) in \( G \). It holds in general that \( f \) and \( g \) commute if and only if their commutator \( [f,g] \) is \( 0 \).\N\NIn the article under review, the authors generalize the Huq commutativity to what they call \textbf{relative commutativity}. Instead of using the cospan of the morphisms \( (1_X,0) \) and \( (0,1_B) \), they define two morphisms \( f:X\to Y \) and \( g:B\to Y \) to commute relative to the cospan of the morphisms \( k:X\to A \) and \( s:B\to A \) if there exists a \( \varphi_{f,g}:A\to Y \) such that \( \varphi_{f,g}k=f \) and \( \varphi_{f,g}s=g \) (Definition 2.1). In a normal category, the \textbf{relative commutator} \( [f,g]_{k,s} \) of \( f \) and \( g \) with respect to a jointly extremally epimorphic pair of morphisms \( k \) and \( s \) can be constructed in such a way that \( f \) and \( g \) commute relative to \( k \) and \( s \) if and only if \( [f,g]_{k,s} \) is \( 0 \) (Proposition 2.11). More precisely, the relative cosmash product \( \iota_{k,s}:X\diamond_{k,s} B\to X+B \) is defined to be the kernel of the canonical morphism \( <k,s>:X+B\to A \), and \( [f,g]_{k,s} \) is defined to be the image in the (regular epimorphism, monomorphism)-factorization of the composite \( <f,g>\iota_{k,s} \) (Definition 2.8).\N\NSemi-abelian categories provide a good categorical context to capture some of the essential homological properties of the categories of groups, of rings, of associative algebras and of Lie algebras. In particular, semi-abelian categories are unital and normal. The classical notion of action of a group on another group can be generalized to the semi-abelian setting. An action of an object \( B \) on an object \( X \) is given by a morphism \( \xi:B\flat X\to X \) satisfying two equations, where \( B\flat X \) is the kernel of the canonical morphism \( <1_B,0>:B+X\to B \). Semi-abelian categories allow for a semidirect product construction, generalizing the one for groups, that yields an equivalence of categories between internal actions and (equivalence classes of) split short exact sequences. Given an internal action \( \xi:B\flat X \to X \), the cospan of the morphisms \( k \) and \( s \) in the associated split short exact sequence \( X\xrightarrow{k} X\rtimes_\xi B \overset{p}{\underset{s}\rightleftarrows} B \) is jointly extremally epimorphic. Hence the authors specialize their concepts of relative commutativity and commutator to \textbf{twisted commutativity and commutator}. More precisely, they say that \( f:X\to Y \) and \( g:B\to Y\) \( \xi \)-commute if they commute relative to \( k \) and \( s \), and they write \( [f,g]_\xi \) for the relative commutator \( [f,g]_{k,s} \) (Definition 2.15). In fact, for \( k \) and \( s \) to be induced by an action is a property rather than additional structure.\N\NIn the \textit{Introduction}, the authors state that their motivation for developing a theory of commutativity twisted by an action arose from recent investigations in semi-abelian cohomology, in particular on the relation between [\textit{D. Rodelo} and \textit{T. Van der Linden}, Adv. Math. 287, 31--108 (2016; Zbl 1333.18022)] and [\textit{C. S. Simeu} and \textit{T. van der Linden}, Theory Appl. Categ. 36, 379--422 (2021; Zbl 1467.18019)]) [\( \ldots \)]. In the section \textit{4 Further work}, they elaborate on more possible directions of future research such as the link to the Peiffer commutator in [\textit{A.S. Cigoli} et al., Homology Homotopy Appl. 19, No. 1, 181--207 (2017; Zbl 1388.18001)], and higher-order twisted commutators. After defining the generalized versions of commutativity and commutators, the authors continue to study their fundamental properties and some of their instances in the section \textit{2 Relative commutativity and commutators}. In the section \textit{3 Internal crossed modules}, they characterize crossed modules in a semi-abelian category satisfying the so-called ``Smith is Huq property in terms of twisted commutators.\N\NThe notions of precrossed module and crossed module of groups can be generalized to any semi-abelian category \( \mathscr{C} \) in such a way that the category of internal precrossed modules is equivalent to the category of internal reflexive graphs, and that the category of internal crossed modules is equivalent to the category of internal groupoids, as it is the case for groups. In Theorem 3.11, it is proven that a morphism \( \partial:X\to B \) and an action \( \xi:B\flat X\to X \) yield a precrossed module if and only if the twisted commutator \( [\delta,1_B]_\xi \) vanishes. Corollary 3.16 states that \( \delta \) and \( \xi \) yield a crossed module if and only if \( [\delta,1_B]_\xi=0 \) and \( [k,e\delta]_{c^{X,X}}=0 \), where \( e \) is the split monomorphism in the split short exact sequence associated to the action \( \xi \), and \( c^{X,X} \) is the internal conjugation action of \( X \), under the assumption that \( \mathscr{C} \) satisfies the ``Smith is Huq property, i.e. two internal equivalence relations commute in the sense of Smith if and only if their normalizations commute in the sense of Huq, or equivalently the categories of internal star-multiplicative graphs and of internal groupoids are equivalent. The condition for an internal reflexive graph to admit a star-multiplicative graph structure can generally be expressed via twisted commutators. In the subsection \textit{3.2 Two special cases of crossed modules}, the results obtained so far are applied, and the known facts that the crossed modules with monomorphic \( \partial \) are precisely the normal subobjects \( X \) of \( B \) with induced internal conjugation action, and that the crossed modules with regular epimorphic \( \partial \) are precisely the central extensions are recovered.

Cites work

Cited in

(2)

This page was built for publication: Twisted commutators and internal crossed modules

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7033610)