Bourbaki

Definition 1

Tag 038K, permanent. Cite it as [Bourbaki, Tag 038K].

Théories spectrales, chapter V, § 1. Représentations unitaires

Let ϱ\varrho be a continuous linear representation of a topological group G in a topological vector space E. A continuous linear representation π\pi of G in F is a subrepresentation of ϱ\varrho if F is a closed subspace of E and if, for every g∈Gg\in G, the space F is stable under ϱ(g)\varrho (g) and π(g)\pi (g) is the endomorphism of F deduced from ϱ(g)\varrho (g) by passing to subspaces.

Let ϱ\varrho be a continuous linear representation of a topological group G in a topological vector space E. The closure of {0}\{0\} in E is a subrepresentation of ϱ\varrho. More generally, the closure of a subspace stable under the endomorphisms ϱ(g)\varrho (g) is a subrepresentation of ϱ\varrho.

A subrepresentation π\pi of ϱ\varrho is determined in a unique manner by a closed subspace F, stable under all the endomorphisms ϱ(g)\varrho (g). This latter condition will often be stated in the form “F is a G-invariant subspace of E”. One will then also say that F defines a subrepresentation of ϱ\varrho, or sometimes, by abuse of language, that F is a subrepresentation of ϱ\varrho, or of E.

Let F be a subspace of E defining a subrepresentation π\pi of ϱ\varrho. For every g∈Gg\in G, the linear mapping ϱ(g)\varrho (g) defines, by passing to the quotient, an endomorphism ϱ~(g)\widetilde{\varrho}(g) of E/FE/F. The mapping g↦ϱ~(g)g\mapsto \widetilde{\varrho}(g) is a continuous linear representation of G in E/FE/F and the canonical projection of E onto E/FE/F is a G-morphism. One says that ϱ~\widetilde{\varrho} is the quotient representation of G on E/FE/F; it is also denoted by ϱ/π\varrho /\pi.

The subspace of the elements of E invariant under the action of G is a trivial subrepresentation of ϱ\varrho. It is denoted by ϱG\varrho^G or else EGE^G.

Let A be a subset of E. The closed subspace F generated by the elements ϱ(g)x\varrho (g)x, where g∈Gg\in G and x∈Ax\in A, is a subrepresentation of ϱ\varrho, called the subrepresentation of ϱ\varrho generated by A; if F = E, one says that A generates ϱ\varrho. If the subrepresentation F is finite-dimensional, one says that the subset A is G-finite.

Suppose that A is reduced to a single element x∈Ex\in E. If A generates ϱ\varrho, one says that xx is a cyclic vector of ϱ\varrho and that ϱ\varrho is a cyclic representation; one says that xx is a G-finite vector if A is G-finite.

The set of G-finite vectors of ϱ\varrho is a vector subspace of E stable under ϱ\varrho; it is not necessarily closed.

Let (ϱi)i∈I(\varrho_i)_{i\in I} be a family of continuous linear representations of G in locally convex topological K-vector spaces (Ei)i∈I(E_i)_{i\in I}. Let E be the direct sum space of the spaces EiE_i. The mapping g↦(ϱi(g))g\mapsto (\varrho_i(g)) is a linear representation of G in E, called the sum, or direct sum, of the representations ϱi\varrho_i. If I is finite, it is continuous.

If the representations ϱi\varrho_i are all equal to a representation ϱ\varrho for every i∈Ii\in I, one says that the direct sum of the representations ϱi\varrho_i is the sum of Card(I) copies of the representation ϱ\varrho, and it is also denoted by ϱCard(I)\varrho^{Card(I)}, or Card(I) ϱ\varrho.

Let (ϱi)i∈I(\varrho_i)_{i\in I} and (πj)j∈J(\pi_j)_{j\in J} be finite families of continuous linear representations of G in topological K-vector spaces (Ei)i∈I(E_i)_{i\in I} and (Fj)j∈J(F_j)_{j\in J}, respectively. The canonical isomorphism of K-vector spaces

HomK(⨁i∈IEi,⨁j∈JFj)→⨁(i,j)∈I×J_K(\bigoplus_{i\in I}E_i,\bigoplus_{j\in J}F_j)\rightarrow \bigoplus_{(i,j)\in I\times J} HomK(Ei,Fj)_K(E_i,F_j) (A, II, p. 13, Corollary 1) induces, by passing to the subspaces, an isomorphism

(1) HomG(⨁i∈Iϱi,⨁j∈Jπj)→⨁(i,j)∈I×J_G(\bigoplus_{i\in I}\varrho_i,\bigoplus_{j\in J}\pi_j)\rightarrow \bigoplus_{(i,j)\in I\times J} HomG(ϱi,πj)_G(\varrho_i, \pi_j)

which is likewise said to be canonical.

Let ϱ1\varrho_1 and ϱ2\varrho_2 be linear representations of G in topological K-vector spaces E1E_1 and E2E_2. For u∈L(E1;E2)u\in \mathscr{L}(E_1; E_2) and g∈Gg\in G, one sets ϱ(g)u=ϱ2(g)∘u∘ϱ1(g−1)\varrho (g)u=\varrho_2(g)\circ u\circ \varrho_1(g^{-1}). The mapping g↦ϱ(g)g\mapsto \varrho (g) is a linear representation of G in L(E1;E2)\mathscr{L}(E_1; E_2). The space of the elements invariant under this representation coincides with HomG(ϱ1,ϱ2)_G(\varrho_1, \varrho_2).

Let ϱ\varrho be a continuous linear representation of G in a locally convex space E. Recall (INT, VIII, p. 131, § 2, no2^o2) that the contragredient representation ϱ˘\breve{\varrho} of ϱ\varrho is the linear representation of G in the dual E′E' of E defined by ϱ˘(g)=tϱ(g−1)\breve{\varrho}(g) =^t\varrho (g^{-1}).

Bourbaki sets the proof directly after the statement, and the corpus does not separate them, so what is above is the statement and its proof together.

Cited by

Nothing.

Cites

Nothing.

Machine translation of the English, by gpt-5.4, gpt-5-6-mini, against glossary version 34. Not checked by a person. That model is a cut down version of the one the rest of this language was written by.