We have
N1(fˇ)=N1(f),⟨f1∣f2⟩=⟨fˇ1∣fˇ2⟩(6)
for every f∈Fχ(G) and all f1 and f2 in Lχ2(G).
These formulas are consequences of the definitions.
For every g∈G and every f∈Fχ(G), the functions x↦f(g−1x) and x↦f(xg) belong to Fχ(G). They are denoted respectively by γG,χ(g)f and δG,χ(g)f. The mappings γG,χ and δG,χ are linear representations of G in Fχ(G). For every z∈Z, one has
γG,χ(gz)=χ(z)γG,χ(g),δG,χ(gz)=χ(z)δG,χ(g)(7)
Let f∈Fχ(G) and g∈G; one has
∣γG,χ(g)f∣=γG/Z(gZ)∣f∣,∣δG,χ(g)f∣=δG/Z(gZ)∣f∣(8)
where all the functions occurring in these equalities are identified with functions on G/Z.
The subspace Kχ(G) of Fχ(G) is stable under γG,χ and δG,χ. Let p be a real number ⩾1. Formulae (8) imply that the representations γG,χ and δG,χ, restricted to Kχ(G), extend to continuous isometric linear representations of G in Lχp(G), which will be denoted by γG,χ(p) and δG,χ(p). By passing to quotients, these representations also define isometric representations of G in Lχp(G), denoted in the same way.
The representations γG,χ(2) and δG,χ(2) in Lχ2(G) are unitary, and will be denoted simply by γG,χ and δG,χ, respectively, whenever no confusion with the representations in Fχ(G) is possible. One also denotes by ϱG,χ the continuous representation of G×G in Lχ2(G) or Lχ2(G) defined by
ϱG,χ(g,h)=γG,χ(g)∘δG,χ(h)=δG,χ(h)∘γG,χ(g)
for every (g,h)∈G×G(cf. Lemma 1 of V, p. 377).
The representation γG,χ on Lχ2(G) is none other than the induced representation IndZG(χ).
When Z={e}, the following lemma results from INT, VIII, p. 166, § 4, no5, Prop. 12, since G is unimodular.
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.
Nothing.
Nothing.