For every measure μ∈M1(G), the functions FG(μ) and FG(μ) are continuous and bounded. The mappings FG:μ↦FG(μ) and FG:μ↦FG(μ) are continuous morphisms of the involutive algebra M1(G) into the involutive algebra of bounded continuous functions on G (Example 1 of I, p. 99).
Let μ∈M1(G). For every χ∈G, one has
∣F(μ)(χ)∣=∣∫G⟨χ,x⟩dμ(x)∣⩽∥μ∥1(7)
therefore the Fourier transform of μ is bounded. Analogously, one verifies that F(μ) is bounded.
If χ tends to χ0 in G, the function χ on G tends to χ0 uniformly on every compact while remaining bounded by the constant function 1 which belongs to L1(G,μ). By the Lebesgue theorem (INT, IV, §3, no7, th. 6), it follows that F(μ)(χ) tends to F(μ)(χ0). Hence F(μ) is continuous. The same is true of F(μ).
For every χ∈G, the mapping μ↦χ(μ)=∫⟨χ,x⟩dμ(x) is a hermitian character of M1(G) (Lemma 1 of II, p. 202). This implies that F and F are morphisms of involutive algebras from M1(G) into the involutive algebra of bounded continuous functions on G. Inequality (7) proves that these morphisms are continuous.
The Fourier transform of G (resp. the Fourier cotransform of G) is the mapping μ↦FG(μ) (resp. the mapping μ↦FG(μ)) of M1(G) into Cb(G).
Let us note some useful formulae for μ∈M1(G),x∈G and χ∈G:
F(μ)(χ)=F(μ)(χ−1)=F(μ)(χ)(8)
∥F(μ)∥∞=∥F(μ)∥∞⩽∥μ∥1(9)
F(εx)(χ)=⟨χ,x⟩
(10)
F(εx)(χ)=⟨χ,x⟩
(in particular F(εe)=F(εe)=1),
F(εx∗μ)(χ)=⟨χ,x⟩F(μ)(χ)
(11)
F(εx∗μ)(χ)=⟨χ,x⟩F(μ)(χ)
F(χ⋅μ)=εχ∗F(μ)
(12)
F(χ⋅μ)=εχ−1∗F(μ)
Formulae (8), (9), (10) and (11) follow from the definitions. Let us prove the first of formulae (12), the second being analogous. For every ξ in G one has the equalities
F(χ⋅μ)(ξ)=∫G⟨ξ,x⟩⟨χ,x⟩dμ(x)=∫G⟨ξχ−1,x⟩dμ(x)
=F(μ)(ξχ−1)=(εχ∗F(μ))(ξ)
Let us note moreover that for every χ∈G and every measures μ and ν in M1(G), one has
(εχ∗F(μ))(εχ∗F(ν))=εχ∗(F(μ)F(ν))(13)
since the two members of this equality are functions on G whose value at ξ∈G is
F(μ)(ξχ−1)F(ν)(ξχ−1)
Let H be a locally compact commutative group and let φ:G→H be a continuous morphism. Let μ∈M1(G). The image measure φ(μ) is defined (INT, V, §6, no1, remark 1), and one gets FH(φ(μ))=FG(μ)∘φ(cf. INT, V, §6, no4, prop. 7).
By restriction to the subalgebra L1(G) of M1(G), one obtains the definition of the Fourier transform and of the Fourier cotransform on L1(G). One thus has for f∈L1(G) and χ∈G:
Let σ be an automorphism of G and Δ the module of σ (INT, VII, §1, no4, def. 4). For f∈L1(G), one has
F(f∘σ)=Δ−1F(f)∘σ−1(16)
(cf. loc. cit., formula (31)).
If one identifies G and X(L1(G)) (prop. 1 of II, p. 202), the Fourier cotransform is nothing but the Gelfand transform of the Banach algebra L1(G) (I, p. 7, def. 5).
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.