Lie Algebra Decomposition Classes for Reductive Algebraic Groups in Arbitrary Characteristic
Joel Summerfield · 2025
Read Lie Algebra Decomposition Classes for Reductive Algebraic Groups in Arbitrary Characteristic by Joel Summerfield (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-01440.txt.
Introduction
Joel Summerfield studies how the Lie algebra of a reductive algebraic group breaks into decomposition classes, a finer sorting of the way its elements fall into orbits. Earlier work handled this mostly under limits on the field's characteristic. Summerfield works over algebraically closed fields of any characteristic at all.
He extends a dimension formula for these classes. He also adds Levi-type decomposition classes to handle the trouble that appears in bad characteristic. The study also develops Lusztig-Spaltenstein induction for non-nilpotent orbits. It shows this induction does not depend on the chosen parabolic subgroup. It finishes the covering relation for the closure ordering of classes in good characteristic. The results tie the geometry of orbits more tightly to the arithmetic of the field.
Catalog record
| CREATOR | Joel Summerfield |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 167 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-01440
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Lie Algebra Decomposition Classes for Reductive Algebraic Groups in Arbitrary Characteristic Joel Summerfield Abstract . Decomposition classes provide a way of partitioning the Lie algebras of an algebraic group into equivalence classes based on the Jordan decomposition. In this paper, we investigate the decomposition classes of the Lie algebras of connected reductive algebraic groups, over algebraically closed fields of arbitrary characteristic. We extend some results previously proved under restrictions on the characteristic, and introduce Levi-type decomposition classes to account for some of the difficulties encountered in bad characteristic. We also establish properties of Lusztig-Spaltenstein induction of non-nilpotent orbits, extending the known results for nilpotent orbits. 1. Introduction In [ BK79 , §5.2] , Borho and Kraft introduced Zerlegungsklassen (decomposition classes) as a tool for studying Schichten der Lie-Algebra (sheets of a Lie algebra). They considered a (connected) semisimple algebraic group G of adjoint type, over an algebraically closed field of characteristic 0 , acting via the adjoint action on its Lie algebra \mathfrak{g}=\Lie G . For an arbitrary element x\in\mathfrak{g} , with Jordan decomposition x=x_{\mathrm{s}}+x_{\mathrm{n}} , define \C_{G}x_{\mathrm{s}}\coloneqq\mathopen{}\mathclose{{\left\{g\in G\penalty\ \middle|\penalty\ g\cdot x_{\mathrm{s}}=x_{\mathrm{s}}}}\right\} . Another element y\in\mathfrak{g} then has ähnliche Jordanzerlegung (similar Jordan decomposition) if there exists g\in G such that the Jordan decomposition of g\cdot y=y^{\prime}_{\mathrm{s}}+y^{\prime}_{\mathrm{n}} satisfies \C_{G}y^{\prime}_{\mathrm{s}}=\C_{G}x_{\mathrm{s}} and \mathopen{}\mathclose{{\left(\C_{G}x_{\mathrm{s}}}}\right)\cdot y^{\prime}_{\mathrm{n}}=\mathopen{}\mathclose{{\left(\C_{G}x_{\mathrm{s}}}}\right)\cdot x_{\mathrm{n}} . This yields an equivalence relation on \mathfrak{g} , whose corresponding equivalence classes are decomposition classes. Useful properties of these decomposition classes were then established in later parts of [ BK79 ] and [ Bor81 ] . In [ Spa82 , §1.2] , Spaltenstein extended the idea as follows. They considered a connected reductive algebraic group G , over an algebraically closed field of arbitrary characteristic, again acting via the adjoint action on \mathfrak{g}=\Lie G . Elements x=x_{\mathrm{s}}+x_{\mathrm{n}} and y=y_{\mathrm{s}}+\penalty y_{\mathrm{n}}\in\mathfrak{g} were said to be equivalent in \mathfrak{g} if there exists g\in G such that \fc_{\mathfrak{g}}\mathopen{}\mathclose{{\left(g\cdot x_{\mathrm{s}}}}\right)=\fc_{\mathfrak{g}}y_{\mathrm{s}}\coloneqq\mathopen{}\mathclose{{\left\{z\in\mathfrak{g}\penalty\ \middle|\penalty\ \mathopen{}\mathclose{{\left[y_{\mathrm{s}},z}}\right]=0}}\right\} and g\cdot x_{\mathrm{n}}=y_{\mathrm{n}} . This again yields an equivalence relation on \mathfrak{g} , whose corresponding equivalence classes were named packets by Spaltenstein. This definition generalises the concept of decomposition classes introduced in [ BK79 ] , and coincides with them in the characteristic 0 case. Spaltenstein then established properties of packets in [ Spa82 ] and [ Spa84 ] . As with Borho and Kraft, packets were introduced by Spaltenstein to study the maximal irreducible subsets of \mathfrak{g} consisting of equal-dimension orbits, known as sheets. Spaltenstein demonstrated in [ Spa82 ] that some of the properties from [ BK79 ] generalised immediately to good characteristic, using the fact that connected stabilisers of semisimple elements of \mathfrak{g} are Levi subgroups of G . Moreover, certain properties related to nilpotent orbits were also shown in [ Spa82 ] to hold in the classical cases in bad characteristic (see § 6.7 for details). In [ Bro98 , §3] , Broer also considered decomposition classes, working under the assumption that G is the adjoint group of a semisimple Lie algebra \mathfrak{g} , over an algebraically closed field. They primarily used the additional assumption that the characteristic is very good, and generalised further results from [ BK79 ] relating to the closures of decomposition classes, as well as establishing that decomposition classes are smooth. Further results on decomposition classes can be found in [ Bro98a ] and [ TY05 ] (both assuming characteristic 0 ), [ PS18 ] (assuming the Standard Hypotheses), and [ Amb25 ] (partly assuming good characteristic), amongst other places. This paper first defines decomposition classes for an arbitrary algebraic group G , over an algebraically closed field \mathbb{K} of arbitrary characteristic, acting on its Lie algebra \mathfrak{g}=\Lie G via the adjoint action. For any x\in\mathfrak{g} , we define its connected stabiliser \operatorname{\mathrm{C}}^{\,\circ}_{G}x\subseteq G to be the identity component of \operatorname{\mathrm{C}}^{\,\circ}_{G}x\coloneqq\mathopen{}\mathclose{{\left\{g\in G\penalty\ \middle|\penalty\ g\cdot x=x}}\right\} , and let x=x_{\mathrm{s}}+x_{\mathrm{n}} denote its Jordan decomposition. Decomposition classes are then the equivalence classes of \mathfrak{g} under the relation x\sim y , which holds if and only if there exists g\in G such that \operatorname{\mathrm{C}}^{\,\circ}_{G}\mathopen{}\mathclose{{\left(g\cdot x_{\mathrm{s}}}}\right)=\operatorname{\mathrm{C}}^{\,\circ}_{G}y_{\mathrm{s}} and g\cdot x_{\mathrm{n}}=y_{\mathrm{n}} . The decomposition class containing x\in\mathfrak{g} is denoted \operatorname{\mathfrak{J}}_{G}x , and we prove that each decomposition class has constant stabiliser dimension, and constant centraliser dimension. From § 2.2 onwards, we assume that G is a connected reductive algebraic group, and establish that our definition of decomposition classes coincides with the definition of packets used by Spaltenstein. We prove some initial properties of decomposition classes in Theorem 2.10 , including that they are G -stable, \mathbb{K}^{\times} -stable, irreducible, and constructible sets which form a finite partition of \mathfrak{g} . Then we turn our attention to decomposition varieties, which are defined as the Zariski-closures of decomposition classes. We equip the set of decomposition classes \mathfrak{D}[G] with the closure order, where \operatorname{\mathfrak{J}}_{G}x\preceq\operatorname{\mathfrak{J}}_{G}y if and only if \operatorname{\mathfrak{J}}_{G}x\subseteq\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{} . In § 3 , we explore how the structure of decomposition classes are affected by central surjections, which leads us to the following preservation result. Theorem 1 . Suppose \varphi\colon G\rightarrow H is a separable central surjection of connected reductive algebraic groups. For any x\in\mathfrak{g} , let \check{x}\coloneqq\mathrm{d}\varphi(x)\in\mathfrak{h} . Then \operatorname{\mathfrak{J}}_{G}x\mapsto\operatorname{\mathfrak{J}}_{H}\check{x} defines a bijection \mathfrak{D}[G]\rightarrow\mathfrak{D}[H] , with the following properties : \mathrm{(i)} \mathrm{d}\varphi\colon\mathfrak{g}\rightarrow\mathfrak{h} restricts to a surjection \operatorname{\mathfrak{J}}_{G}x\rightarrow\operatorname{\mathfrak{J}}_{H}\check{x} . \mathrm{(ii)} Preservation of closure : \mathrm{d}\varphi\mathopen{}\mathclose{{\left(\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{}}}\right)=\macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{} . \mathrm{(iii)} Preservation of the partial order : \operatorname{\mathfrak{J}}_{G}x\preceq\operatorname{\mathfrak{J}}_{G}y if and only if \operatorname{\mathfrak{J}}_{H}\check{x}\preceq\operatorname{\mathfrak{J}}_{H}\check{y} . \mathrm{(iv)} \dim\operatorname{\mathfrak{J}}_{G}x=\dim\ker\mathrm{d}\varphi+\dim\operatorname{\mathfrak{J}}_{H}\check{x} . Note that (within the generality in which we are working) the stabiliser dimension \dim\C_{G}x and the centraliser dimension \dim\fc_{\mathfrak{g}}x do not necessarily coincide for arbitrary x\in\mathfrak{g} . The fibres of the stabiliser dimension map \dim\C_{G}\colon\mathfrak{g}\rightarrow\mathbb{N} are referred to as stabiliser level sets , and the irreducible components of non-empty stabiliser level sets are called stabiliser sheets . Analogously, the fibres of the centraliser dimension map \dim\C_{G}\colon\mathfrak{g}\rightarrow\mathbb{N} are referred to as centraliser level sets , and the irreducible components of non-empty centraliser level sets are called centraliser sheets . We then use level set to refer to a subset of \mathfrak{g} which is (at least one of) a stabiliser level set or a centraliser level set, and sheet to refer to a subset of \mathfrak{g} which is (at least one of) a stabiliser sheet or a centraliser sheet. This is an important departure from the literature (see § 4 for more details) which allows us to uniformly prove results regarding both situations. Since decomposition classes have constant stabiliser dimension, they are each contained in a unique stabiliser level set; analogously, each decomposition class is contained in a unique centraliser level set. Given a level set \mathfrak{g}_{\langle m\rangle} of \mathfrak{g} , we let \mathfrak{D}_{\langle m\rangle}[G]\coloneqq\mathopen{}\mathclose{{\left\{\mathfrak{J}\in\mathfrak{D}[G]\penalty\ \middle|\penalty\ \mathfrak{J}\subseteq\mathfrak{g}_{\langle m\rangle}}}\right\} denote the set of decomposition classes contained in \mathfrak{g}_{\langle m\rangle} . Theorem 2 . Suppose \mathfrak{g}_{\langle m\rangle} is a level set of \mathfrak{g} , and \mathfrak{J}\in\mathfrak{D}_{\langle m\rangle}[G] . \mathrm{(i)} \mathfrak{J} is a dense subset of an irreducible component of \mathfrak{g}_{\langle m\rangle} if and only if \mathfrak{J} is maximal in \mathfrak{g}_{\langle m\rangle} ( with respect to the closure order ) . \mathrm{(ii)} The irreducible components of \mathfrak{g}_{\langle m\rangle} are in bijection with the decomposition classes which are maximal in \mathfrak{g}_{\langle m\rangle} ( with respect to the closure order ) , via \overline{\mathfrak{J}}\cap\mathfrak{g}_{\langle m\rangle}\mathrel{\reflectbox{$\mapsto$}}\mathfrak{J} . \mathrm{(iii)} If \mathfrak{J} coincides with an irreducible component of \mathfrak{g}_{\langle m\rangle} , then