The constituents of a restriction to a normal subgroup form one orbit #
Let N be a normal subgroup of G and let ρ be an irreducible representation of G on V.
Given one minimal N-stable subspace to start from, restricting ρ to N breaks it into
irreducible constituents
(TauCeti.Representation.isSemisimpleRepresentation_comp_subtype_of_isAtom; for a
finite-dimensional V such a subspace always exists, and then no hypothesis is needed, which is
TauCeti.Representation.isSemisimpleRepresentation_comp_subtype). This file identifies which
constituents occur: they are the translates of any one of them, up to isomorphism of
N-representations. Equivalently, G permutes the isotypic components of the restriction
transitively.
The argument is short because the two halves it needs are already in place. The translates
TauCeti.Representation.conjSubrep ρ g σ of a minimal N-stable subspace σ are again minimal
(TauCeti.Representation.isAtom_conjSubrep_iff) and they span
(TauCeti.Representation.iSup_conjSubrep_eq_top); and a simple submodule of a module spanned by
simple submodules is isomorphic to one of them (Mathlib's Submodule.linearEquiv_of_sSup_eq_top).
The spanning statement is needed in the lattice of k[N]-submodules of the restriction viewed as a
k[N]-module, where the isotypic machinery lives, rather than in the lattice of k-subspaces where
it is proved; that reading is TauCeti.Representation.iSup_asSubmodule_conjSubrep_eq_top, which
lives in TauCeti/RepresentationTheory/Induction/Clifford/Basic.lean because the semisimplicity
proof there needs it too.
No invertibility of Nat.card N is used: irreducibility of the ambient representation replaces
Maschke's theorem throughout, exactly as in
TauCeti/RepresentationTheory/Induction/Clifford/Basic.lean. The single-orbit statements need no
finiteness either; finiteness of G is assumed only by
TauCeti.Representation.finite_isotypicComponents, which counts the isotypic components through the
translates. Finite-dimensionality of V is needed only to know that a minimal N-stable subspace
exists at all: each of the orbit and isotypic-component statements below takes such a subspace as a
hypothesis, except TauCeti.Representation.exists_isAtom_forall_nonempty_linearEquiv_conjSubrep,
the packaged form of the theorem, which assumes finite-dimensionality and produces one.
Main statements #
TauCeti.Representation.exists_nonempty_linearEquiv_conjSubrep: Clifford's theorem, single orbit form. Any two minimalN-stable subspaces of an irreducible representation are translates of one another, up to isomorphism ofN-representations.TauCeti.Representation.exists_isAtom_forall_nonempty_linearEquiv_conjSubrep: the same statement packaged for a finite-dimensional irreducible representation, where the minimal subspace to translate is produced rather than assumed.TauCeti.Representation.isotypicComponents_eq_range: the isotypic components of the restriction are exactly the components of the translates of a given minimalN-stable subspace, soGacts transitively on them;Representation.conjSubrepIsotypicComponentpackages each such component, andTauCeti.Representation.finite_isotypicComponentsreads off from the same hypothesis that there are finitely many of them whenGis finite.TauCeti.Representation.isIsotypicOfType_asSubmodule_iff: the restriction is isotypic exactly when every translate of one constituent is isomorphic to it — the case in which the inertia group of the constituent is all ofG.
Implementation notes #
An N-constituent is presented as an atom of the lattice of N-subrepresentations, as in
Basic.lean, and "isomorphic as representations of N" is spelled as a k[N]-linear equivalence
of the associated submodules of Representation.asModule (ρ.comp N.subtype). That is what the
isotypic API of Mathlib/RingTheory/SimpleModule/Isotypic.lean speaks, and it is the same relation
as an isomorphism of the subrepresentations, transported along
Subrepresentation.subrepresentationSubmoduleOrderIso. That an atom of the one lattice is a simple
module in the other is Subrepresentation.isSimpleModule_asSubmodule_iff, in
TauCeti/RepresentationTheory/Subrepresentation.lean; it uses neither normality nor irreducibility.
Reading the translate through TauCeti.Representation.conjSubrepEquiv turns each statement below
into the classical one about the conjugate representation {}^g V.
References #
This file proves the single-orbit half of the Clifford's theorem milestone of Layer 5 of
TauCetiRoadmap/RepresentationTheory/InductionRestriction/README.md: "its irreducible
N-constituents form a single G-orbit". That the constituents share one multiplicity is the
other half, proved in TauCeti/RepresentationTheory/Induction/Clifford/Multiplicity.lean from the
single-orbit statements below. What remains of the milestone is the packaged decomposition
Res_N W ≅ e · ⨁ᵢ {}^{gᵢ} V indexed by a transversal of the inertia group, which is not proved
here.
The mathematics is the classical argument of C. W. Curtis and I. Reiner, Representation Theory of Finite Groups and Associative Algebras, §49.
Every irreducible N-constituent is a translate of any fixed one. If σ is a minimal
N-stable subspace of an irreducible representation ρ and T is any simple k[N]-submodule of
the restriction, then T is isomorphic, as a representation of N, to the translate
conjSubrep ρ g σ for some g : G.
The translates of σ are simple and span, so T sits inside a sum of simple modules and is
therefore isomorphic to one of them.
Clifford's theorem, single-orbit form. Any two minimal N-stable subspaces of an
irreducible representation are translates of one another, up to isomorphism of representations of
N: the irreducible constituents of the restriction to N form a single G-orbit.
Clifford's theorem, single-orbit form, packaged. The restriction to N of a
finite-dimensional irreducible representation has a minimal N-stable subspace, and every minimal
N-stable subspace is a translate of it up to isomorphism of representations of N.
G permutes the isotypic components of the restriction transitively. The isotypic
components of the restriction to N of an irreducible representation are exactly the components of
the translates of one minimal N-stable subspace.
Given one minimal N-stable subspace, a restriction along a normal subgroup of a finite group
has finitely many isotypic components. There is one component for each translate of that
subspace, so at most one for each element of G.
The restriction is isotypic exactly when one constituent is fixed by all of G. The
right-hand side says that the inertia group of σ is all of G; Clifford's theorem then leaves the
restriction with a single isomorphism class of constituents.
The isotypic component cut out by the translate of a fixed constituent.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The underlying submodule of the component indexed by g is the isotypic component of the
translated constituent conjSubrep ρ g σ.