Chapter VIII. Borel Subgroups

FC16 — Humphreys, Linear Algebraic Groups.

This chapter contains 3 definitions, 6 theorems, 2 propositions, 3 lemmas, 14 corollaries, 1 example, 1 construction and 2 other statements.

Nobody has checked this chapter against Lean yet, so the statements below say nothing about whether they are formalized.

Recorded toolkit: complete/projective variety facts used for Borel theory

FC16-C08-U001 · §21.1; pp. 133–134 · after FC16-C01-U091, FC16-C01-U092, FC16-C04-U009

The source recalls that closed subvarieties/products/images behave well under completeness, projective varieties are complete, homogeneous spaces are quasiprojective, and complete affine varieties are finite.

Lemma: completeness transfers across bijective equivariant homogeneous morphisms

FC16-C08-U002 · §21.1; p. 133 · after FC16-C01-U090, FC16-C02-U024, FC16-C02-U025

If connected G acts transitively on irreducible X,Y and φ:X→Y is a bijective G-equivariant morphism with Y complete, then X is complete.

Theorem: Borel fixed point theorem

FC16-C08-U003 · §21.2; p. 134 · after FC16-C07-U005, FC16-C01-U090, FC16-C02-U020

A connected solvable algebraic group acting on a nonempty complete variety has a fixed point.

Definition: Borel subgroup

FC16-C08-U004 · §21.3; p. 134 · after FC16-C07-U005, FC16-C02-U006

A Borel subgroup is a maximal closed connected solvable subgroup of G; Borel subgroups of G and G° coincide.

Theorem: conjugacy of Borel subgroups and projectivity of G/B

FC16-C08-U005 · §21.3; pp. 134–135 · after FC16-C08-U004, FC16-C08-U003, FC16-C04-U003, FC16-C01-U038

For any Borel subgroup B of connected G, G/B is projective and every Borel subgroup is G-conjugate to B.

Corollary: maximal tori and maximal connected unipotent subgroups

FC16-C08-U006 · §21.3; p. 135 · after FC16-C08-U005, FC16-C07-U026

The maximal tori and maximal connected unipotent subgroups of G are precisely those occurring in Borel subgroups; each class is conjugate in G.

Definition: rank of a connected algebraic group

FC16-C08-U007 · §21.3; p. 135 · after FC16-C08-U006, FC16-C06-U016

rank G is the common dimension of maximal tori of G.

Definition: parabolic subgroup

FC16-C08-U008 · §21.3; pp. 135–136 · after FC16-C04-U009, FC16-C01-U092

A closed P≤G is parabolic when G/P is projective, equivalently complete.

Corollary: parabolic subgroups are exactly those containing a Borel subgroup

FC16-C08-U009 · §21.3; p. 135 · after FC16-C08-U008, FC16-C08-U005, FC16-C08-U003

A closed subgroup P is parabolic iff it contains a Borel subgroup; connected H is Borel iff H is solvable and G/H is projective.

Example/construction: parabolics of GLn as partial-flag stabilizers

FC16-C08-U010 · §21.3; pp. 135–136 · after FC16-C08-U009, FC16-C01-U038, FC16-C02-U005

For GL_n with standard upper triangular B, stabilizers of standard partial flags are parabolic subgroups containing B.

Corollary: Borel/parabolic/torus/unipotent subgroups under epimorphisms

FC16-C08-U011 · §21.3; p. 136 · after FC16-C02-U012, FC16-C08-U005, FC16-C08-U009, FC16-C08-U006

An epimorphism of connected algebraic groups maps Borel subgroups, parabolics, maximal tori and maximal connected unipotent subgroups onto subgroups of the same type, and every such target subgroup arises this way.

Proposition: automorphism fixing a Borel subgroup pointwise

FC16-C08-U012 · §21.4; p. 136 · after FC16-C08-U005, FC16-C01-U091

An automorphism of connected G fixing every element of a Borel subgroup is the identity.

Corollary: center versus a Borel centralizer

FC16-C08-U013 · §21.4; pp. 136–137 · after FC16-C08-U012, FC16-C08-U004

Z(G)°⊆Z(B)⊆C_G(B)=Z(G); later equality Z(B)=Z(G) is obtained.

Proposition: nilpotent Borel criteria

FC16-C08-U014 · §21.4; p. 137 · after FC16-C07-U010, FC16-C08-U005, FC16-C08-U006

If a Borel subgroup is nilpotent then it equals G; equivalently G is nilpotent iff it has a unique maximal torus.

Corollary/definition: connected centralizer of a maximal torus as Cartan subgroup

FC16-C08-U015 · §21.4; p. 137 · after FC16-C08-U014, FC16-C06-U025, FC16-C08-U006

For maximal torus T, C_G(T)° is nilpotent and self-normalizing up to identity component; it is called a Cartan subgroup at this stage.

Lemma: density criterion from a homogeneous fixed-point set

FC16-C08-U016 · §22.1; pp. 138–139 · after FC16-C04-U009, FC16-C01-U090, FC16-C01-U066

For closed connected H≤G, the union of conjugates X=⋃_g gHg^{-1} is closed if G/H is complete; if an element of H fixes only finitely many points of G/H then X contains a dense open subset of G.

Theorem: density theorem for Borel, torus, unipotent and Cartan conjugates

FC16-C08-U017 · §22.2; pp. 139–140 · after FC16-C08-U016, FC16-C08-U005, FC16-C08-U006, FC16-C06-U008, FC16-C08-U015

For connected G, the union of conjugates of a Borel subgroup is all G; the union of conjugates of a maximal torus is the semisimple locus, that of the Borel unipotent radical is the unipotent locus, and conjugates of a Cartan subgroup contain a dense open subset.

Corollary: semisimple/unipotent elements lie in connected nilpotent groups

FC16-C08-U018 · §22.2; p. 140 · after FC16-C08-U017, FC16-C07-U010

Every semisimple or unipotent element of connected G lies in a connected nilpotent subgroup.

Corollary: center of a Borel subgroup

FC16-C08-U019 · §22.2; p. 140 · after FC16-C08-U017, FC16-C08-U013

For any Borel subgroup B of connected G, Z(B)=Z(G).

Theorem: connectedness of torus centralizers

FC16-C08-U020 · §22.3; p. 140 · after FC16-C08-U003, FC16-C07-U028, FC16-C06-U025

For any torus S in connected G, C_G(S) is connected.

Corollary/definition: Cartan subgroups are centralizers of maximal tori

FC16-C08-U021 · §22.3; p. 140 · after FC16-C08-U020, FC16-C08-U015, FC16-C08-U006

The Cartan subgroups of connected G are exactly C_G(T) for maximal tori T; all are conjugate.

Corollary: commuting semisimple element and torus lie in a torus

FC16-C08-U022 · §22.3; p. 140 · after FC16-C08-U020, FC16-C07-U027

If semisimple x centralizes a torus S, then x and S are contained in a common torus.

Theorem: centralizer action on torus-fixed components of G/B

FC16-C08-U023 · §22.4; pp. 141–142 · after FC16-C08-U020, FC16-C08-U005, FC16-C08-U003

For torus S, C=C_G(S), and an irreducible component Y of the fixed-point set (G/B)^S, C acts transitively on Y.

Corollary: Borel subgroups of torus centralizers

FC16-C08-U024 · §22.4; p. 142 · after FC16-C08-U023, FC16-C08-U004

If B contains S, then B∩C_G(S) is a Borel subgroup of C_G(S), and every Borel subgroup of C_G(S) arises in this way.

Result: Cartan subgroup structure summary

FC16-C08-U025 · §22.5; p. 142 · after FC16-C08-U021, FC16-C08-U017, FC16-C07-U025

For maximal T, C_G(T)=T⋉C_u with unique maximal torus T; Cartan subgroups are conjugate, nilpotent, and their conjugates contain a dense open subset.

Lemma: identity component of Borel normalizer

FC16-C08-U026 · §23.1; p. 143 · after FC16-C08-U004, FC16-C08-U017

For Borel B, N_G(B)°=B.

Theorem: normalizer theorem for Borel subgroups

FC16-C08-U027 · §23.1–23.2; pp. 143–145 · after FC16-C08-U026, FC16-C08-U005, FC16-C08-U017

Every Borel subgroup is self-normalizing: N_G(B)=B.

Corollary: Borel subgroups are maximal abstract solvable subgroups among containing groups

FC16-C08-U028 · §23.1; p. 143 · after FC16-C08-U027, FC16-C08-U004

A Borel subgroup is maximal among solvable subgroups of G, without assuming competitors connected or closed.

Corollary: parabolic subgroups are connected and self-normalizing

FC16-C08-U029 · §23.1; pp. 143–144 · after FC16-C08-U027, FC16-C08-U009

If P is parabolic then P=N_G(P), hence P is connected.

Corollary: parabolics containing a common Borel are not nontrivially conjugate

FC16-C08-U030 · §23.1; p. 144 · after FC16-C08-U029, FC16-C08-U005

If parabolics P,Q both contain B and are G-conjugate, then P=Q.

Corollary: Borel subgroup normalizes exactly its unipotent radical

FC16-C08-U031 · §23.1; p. 144 · after FC16-C08-U027, FC16-C07-U029, FC16-C08-U006

For Borel B with U=R_u(B), N_G(U)=B.

Construction/result: variety of Borel subgroups

FC16-C08-U032 · §23.3; p. 145 · after FC16-C08-U027, FC16-C08-U005, FC16-C04-U009

The set 𝔅 of all Borel subgroups is identified G-equivariantly with G/B by xBx^{-1}↔︎xB; Borel subgroups containing H correspond to H-fixed points on G/B.