— Under the hypothesis Φ(0)−ℜΦ≤ε+q, we have
μ(T′−Cr)≤3(ε+r−2Tr(q/h))
for every r>0.
One writes Tr(q/h) for the trace of q with respect to h (cf. Annex, No. 1). The formula (55) is trivial when Tr(q/h) is infinite. We assume henceforth that Tr(q/h) is finite, hence that h(x)=0 implies q(x)=0 for x∈T.
Let a1,…,an be elements of T, and D the set of x′∈T′ such that ∑j=1n⟨aj,x′⟩2>1. For every real t≥0 we have 3(1−e−t/2)≥0, and we even have
3(1−e−t/2)≥3(1−e−1/2)≥3(1−(49)−1/2)=1
for t>1, because e>49. Applying these inequalities to t=∑j=1n⟨aj,x′⟩2, we obtain
μ(D)≤3∫T′(1−exp(−21j=1∑n⟨aj,x′⟩2))dμ(x′).
Let γ be the measure on R having density t↦(2π)−1/2e−t2/2 with respect to Lebesgue measure. By Lemma 2 of No. 4,
∫Reiutdγ(t)=e−u2/2
for all real u. Consequently,
1−exp(−21j=1∑n⟨aj,x′⟩2)=∫⋯∫(1−ei∑j=1n⟨aj,x′⟩tj)dγ(t1)⋯dγ(tn)
for all x′∈T′. The function of x′,t1,…,tn to be integrated in the second member is continuous and is bounded above in absolute value by 2, and the measures μ and γ are bounded; one can therefore integrate the two members of (57) with respect to dμ(x′) and interchange the integrations with respect to μ and γ; one obtains
∫T′(1−exp(−21j=1∑n⟨aj,x′⟩2))dμ(x′)=∫⋯∫(Φ(0)−Φ(j=1∑ntjaj))dγ(t1)⋯dγ(tn).
Since q is a quadratic form on T, there exist real numbers qjk such that
q(j=1∑ntjaj)=j,k∑qjktjtk for
t1,…,tn real; in particular,
qjj=q(aj) for
1≤j≤n. Moreover, the integral
∫Rtndγ(t) has the values 1, 0, 1 for
n=0,1,2, respectively (No. 4, Lemma 1). From this, one deduces immediately
∫⋯∫(ε+q(j=1∑ntjaj))dγ(t1)…dγ(tn)=ε+j=1∑nq(aj). Now, the first member of (58) and
Φ(0) are real numbers; one can therefore replace
Φ by
ℜΦ in the second member of (58). The inequality
Φ(0)−ℜΦ≤ε+q and the formulas (56), (58) and (59) then imply
μ(D)≤3(ε+j=1∑nq(aj)). Let us fix the number
r>0. Since the quadratic form
h is positive, there exist a basis
(e1,…,en) of
T and an integer
m between 0 and
n such that
h(j=1∑ntjej)=j=1∑mtj2 for
t1,…,tn real (Annex, No. 1, Prop. 2). It is then immediate that
Cr consists of the
x′∈T′ such that
j=1∑m⟨ej,x′⟩2≤r2,j=m+1∑n⟨ej,x′⟩2=0. For every integer
l≥1, let
Dl be the set of
x′∈T′ satisfying the inequality
j=1∑m⟨r−1ej,x′⟩2+j=m+1∑n⟨lej,x′⟩2>1. One sees easily that the sequence
(Dl)l≥1 is increasing with union
T′−Cr, whence
μ(T′−Cr)=l→∞limμ(Dl).
But by (60),
μ(Dl)≤3(ε+j=1∑mr−2q(ej)+j=m+1∑nl2q(ej));
for j=m+1,…,n we have h(ej)=0, therefore q(ej)=0. Moreover, Tr(q/h)=∑j=1mq(ej) (Annex, No. 1, Prop. 2). The relation (55) then follows from (61) and (62).
Q.E.D.
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.