It is now seen that one should have lim
T →∞
I
1
T = lim
T →∞
I
′ 1
T =
C
1+ β
α,d
Z
R
d
Z
R
d
p
1
x − y|x|
−γ
| y|
−d−α1+β
d x d y Z
R
d
ϕzdz
1+ β
χ
1+ β
3.97 Note that the integrals are finite by 3.21, 3.28 and 2.16. The justification of 3.97 requires
some work, but we omit it for brevity. iii As d =
γ, we must keep the measure µ
γ
in its original form 1.8. Arguing as in the proof of ii and taking into account 2.14, instead of 3.96 we obtain
I
′ 1
T = 1
log T Z
1
Z
R
d
Z
R
d
p
1
x − y T
sd α
1 + |x T
s α
|
d
×
Z
T
1−s
−1
Z
R
d
p
u
y − z T
−sα
ϕzχu + 1T
s−1
dzdu
1+ β
d y d x ds. Since
lim
T →∞
1 log T
Z
R
d
p
1
x − y T
sd α
1 + |x T
s α
|
d
d x = s 1
α σS
d−1
p
1
y, it can be shown, with some effort, that
lim
T →∞
I
1
T = lim
T →∞
I
′ 1
T =
C
1+ β
α,d
1 2
α σs
d−1
Z
R
d
p
1
y| y|
−d−α1+β
d y Z
R
d
ϕzdz
1+ β
χ
1+ β
0. Again, we omit the remaining parts of the proof.
3.7 Proof of Theorem 2.6
We only give an outline of the proof. The following lemma is constantly used.
Lemma Let ϕ ∈ S
R
d
, ϕ ≥ 0. a If d
α2 + β1 + β, then the functions Gϕ, GGϕ
1+ β
and GG ϕ
2
are bounded. b If d
α1 + ββ, then additionally Gϕ
1+ β
and GG ϕ
1+ β
1+ β
are integrable and bounded. c If
α γ ≤ d and d α2 + ββ − γβ, then additionally to the properties in a, Z
R
d
GG ϕ
1+ β
x 1
1 + |x|
γ
d x ∞.
1359
This Lemma follows easily from 1.12 and 3.25-3.28.
Proof of part a of the theorem. As before we consider µ
γ
d x = |x|
−γ
d x. In 3.9 we substitute u
′
= s − u, then s
′
= T − sT and, finally, x
′
= x T
1 α
s
−1α
, obtaining I
1
T =
Z
R
d
Z
1
Z
R
d
p
1
x − ys
−1α
T
−1α
Z
T 1−s
T
u
ϕ yχ s +
u T
du
1+ β
s
−γα
|x|
−γ
d y dsd x see 2.22. It is easily seen that by part b of the Lemma we have
lim
T →∞
I
1
T = Z
R
d
p
1
x|x|
−γ
d x Z
1
s
−γα
χ
1+ β
sds Z
R
d
Gϕ y
1+ β
d y. 3.98
For β 1 this is exactly log Eexp{−C〈 e
X , ϕ ⊗ ψ〉}, where X is the limit process described in the
theorem. Moreover, in this case 3.14 and boundedness of G ϕ easily imply
I
2
T ≤ C T
1−γα1−21+β
→ 0. For
β = 1 we use 3.10 and 3.4, obtaining I
2
T = I
′ 2
T − I
′′ 2
T − V
2 I
′′′ 2
T , where
I
′ 2
T = H
T
F
2 T
Z
T
Z
R
d
Z
R
d
p
s
x − y|x|
−γ
d x ϕ yχs
Z
T s
T
u−s
ϕ yχ u
T
dud y ds, I
′′ 2
T = H
T
F
2 T
Z
R
d
Z
T
Z
R
d
p
T −s
x − yϕ yχ
T
T − s ×
Z
s
T
s−u
ϕχT − uv
T
·, u y|x|
−γ
dud y dsd x I
′′′ 2
T = H
T
F
2 T
Z
R
d
Z
T
Z
R
d
p
T −s
x − yϕ yχ
T
T − s Z
s
T
s−u
v
2 T
·, u y|x|
−γ
dud y dsd x. Substituting u
′
= u − s, s
′
= sT , and then x
′
= x T
1 α
s
1 α
and using part a of the Lemma and 2.22 we have
lim
T →∞
I
′ 2
T = Z
R
d
p
1
x|x|
−γ
d x Z
1
s
−γα
χ
2
sds Z
R
d
ϕ yGϕ yd y. 3.99
Applying 3.5, 2.22 and the Lemma above we get I
′′ 2
T ≤ C H
T
F
3 T
Z
R
d
Z
T
Z
R
d
p
s
x − y|x|
−γ
ϕ yGϕGϕ yd y dsd x ≤
C
1
T
−121−γα
→ 0, 1360
and, analogously, I
′′′ 2
T ≤ C H
T
F
3 T
Z
R
d
Z
T
Z
R
d
p
s
x − y|x|
−γ
ϕ yGGϕ
2
yd y dsd x ≤
C
2
T
−121−γα
→ 0. This and 3.98, 3.99 imply that for
β = 1 the limit of V 2I
1
T + I
2
T is exactly log Eexp{−C〈 e
X , ϕ ⊗ ψ}. Similar estimations, together with the Lemma, yield 3.16 and 3.17.
This completes the proof of part a of the theorem.
Proof of part b of the theorem. Following the general scheme one can show
lim
T →∞
I
1
T = Z
R
d
p
1
x|x|
−α
d x χ
1+ β
Z
R
d
Gϕ
1+ β
yd y, lim
T →∞
I
2
T = c
β
Z
R
d
p
1
x|x|
−α
d x χ
2
Z
R
d
ϕ yGϕ yd y, and 3.16 and 3.17 recall that c
β
is defined by 2.24. This is accomplished by an argument similar to the one used in part a. Due to the criticality
γ = α, the integrals R
T
. . . ds in 3.9- 3.11 require a different treatment. They are split into
R
1
. . . ds + R
T 1
. . . ds; the first summand converges to zero, and in the second one we use the substitution s
′
= log s log T . Here, again, we use repeatedly the Lemma above together with the easily checked fact that
sup
T 2
1 log T
Z
R
d
Z
T
T
s
hx|x|
−α
dsd x ∞
for any integrable and bounded function h recall that d α. We omit details.
Proof of part c of the theorem. Recall that the case γ d has been proved in [BGT6]. For
α γ ≤ d we use the Lemma part c is particularly important. We show lim
T →∞
I
1
T = Z
R
d
GG ϕ
1+ β
xµ
γ
d xχ
1+ β
0, lim
T →∞
I
2
T = c
β
Z
R
d
G ϕGϕxµ
γ
d xχ
2
0, 3.16 and 3.17. Here
µ
γ
is either given by 1.8 or, for γ d one can take µ
γ
d x = |x|
−γ
d x. Again, the details are omitted.
3.8 Proof of Theorem 2.8