Bourbaki

Proposition 3

Tag 02NA, permanent. Cite it as [Bourbaki, Tag 02NA].

Théories spectrales, chapter II, § 3. Sous-espaces invariants

The algebra L1(G)L^1(G) satisfies the Ditkin condition (I, p. 92, def. 2).

Let χ\chi be a character of L1(G)L^1(G). Distinguish two cases according as χ\chi is zero or nonzero. If χ\chi is zero, it must be verified that for every function f∈L1(G)f\in L^1(G), there exists a sequence (fn)n⩾1(f_n)_{n\geqslant 1} in L1(G)L^1(G) such that F(fn)\mathscr{F}(f_n) vanishes outside a compact subset of G^\widehat{G} and such that fn∗ff_n*f tends to ff in L1(G)L^1(G). The existence of such a sequence follows from Lemma 1 above and Proposition 8 of II, p. 211.

Suppose now that χ\chi is nonzero, hence χ∈X(L1(G))=G^\chi \in \mathsf{X}(L^1(G)) =\widehat{G} (Proposition 1 of II, p. 202). Let f∈L1(G)f\in L^1(G) be such that GL1(G)(f)(χ)=\mathscr{G}_{L^1(G)}(f)(\chi ) = F‾(f)(χ)=0\overline{\mathscr{F}}(f)(\chi ) = 0. It is a question of proving the existence of a sequence (fn)n⩾1(f_n)_{n\geqslant 1} in L1(G)L^1(G) such that f∗fnf*f_n converges to ff in L1(G)L^1(G) and such that F‾(fn)\overline{\mathscr{F}}(f_n) vanishes in a neighbourhood of χ\chi. We may suppose that ∥f∥1=1\|f\|_1= 1. By translation in G^\widehat{G}, we reduce to the case where χ=e\chi =e.

Let KnK_n be a compact subset of G such that

∫G−Kn∣f(x)∣dx⩽1n \int_{G-K_n}|f(x)|dx\leqslant \frac{1}{n}

Let un∈L1(G)u_n\in L^1(G) be a function ⩾0\geqslant 0 such that ∥un∥1=1\|u_n\|_1= 1 and

∥f−f∗un∥1⩽1n \|f-f*u_n\|_1\leqslant \frac{1}{n}

(cf. Proposition 8 of II, p. 211, (iii)). According to Lemma√\surd 2, there exists a function jnj_n in L1(G)L^1(G) such that ∥jn∥1⩽2‾\|j_n\|_1\leqslant \overline{2}, whose Fourier cotransform is equal to 1 in a neighbourhood of ee, and moreover such that ∥jn−jn∗εx∥1⩽n−1\|j_n-j_n*\varepsilon_x\|_1\leqslant n^{-1} for every x∈Knx\in K_n. Put

fn=un−jn∗un f_n=u_n-j_n*u_n

We shall prove that the sequence (fn)n⩾1(f_n)_{n\geqslant 1} has the required properties. First of all, we have

F(fn)=F(un)−F(jn)F(un)=(1−F(jn))F(un) \mathscr{F}(f_n) =\mathscr{F}(u_n)-\mathscr{F}(j_n)\mathscr{F}(u_n) = (1-\mathscr{F}(j_n))\mathscr{F}(u_n)

hence the Fourier transform of fnf_n vanishes in a neighbourhood of χ=e\chi =e. On the other hand,

1

∥f∗fn−f∥1⩽∥f∗un−f∥1+∥f∗jn∥1∥un∥1⩽+∥f∗jn∥1 \|f*f_n-f\|_1\leqslant \|f*u_n-f\|_1+\|f*j_n\|_1\|u_n\|_1\leqslant +\|f*j_n\|_1
n n

Now, for almost every y∈Gy\in G, we have

(f∗jn)(y)=∫Gf(x)jn(x−1y)dx=∫Gf(x)(jn(x−1y)−jn(y))dx (f*j_n)(y) =\int_Gf(x)j_n(x^{-1}y)dx=\int_Gf(x)(j_n(x^{-1}y)-j_n(y))dx

since, by assumption, we have F(f)(e)=∫Gf(x)dx=0\mathscr{F}(f)(e) =\int_Gf(x)dx= 0. Hence

∥f∗jn∥1⩽∫G∣f(x)∣∥jn∗εx−jn∥1dx \|f*j_n\|_1\leqslant \int_G|f(x)| \|j_n*\varepsilon_x-j_n\|_1dx
=∫∣f(x)∣∥jn∗εx−jn∥1dx =\int|f(x)| \|j_n*\varepsilon_x-j_n\|_1dx
+∫Kn∣f(x)∣∥jn∗εx−jn∥1dx +\int^{K_n}|f(x)| \|j_n*\varepsilon_x-j_n\|_1dx
G−KnG-K_n

1∫∫1\int\int 5 ⩽∣f(x)∣dx+4∣f(x)∣dx⩽\leqslant |f(x)|dx+ 4|f(x)|dx\leqslant.

nKnG−Knnn_{K_n}G-K_{_n}n

Finally, ∥f∗fn−f∥1⩽6n−1\|f*f_n-f\|_1\leqslant 6n^{-1} and therefore f∗fnf*f_n converges in ff in L1(G)L^1(G), as desired.

Applying prop. 5, we obtain the following result:

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, gpt-5-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.