27.01.2014 Views

Stein's method, Malliavin calculus and infinite-dimensional Gaussian

Stein's method, Malliavin calculus and infinite-dimensional Gaussian

Stein's method, Malliavin calculus and infinite-dimensional Gaussian

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Proof of Lemma 10.1 (see also [66, Lemma 2]). Without loss of generality, we can<br />

assume that H = L 2 (Z; Z; ), where (A; Z) is a Polish space, <strong>and</strong> is a -…nite <strong>and</strong> non-atomic<br />

measure. Thus, we can write<br />

Z<br />

hDF; DGi H = p q hI p 1 (f); I q 1 (g)i H<br />

= p q I p 1 f(; z) I q 1 g(; z) (dz)<br />

It follows that<br />

=<br />

E<br />

" <br />

a<br />

8<br />

><<br />

>:<br />

If r < p q then<br />

Z p^q<br />

X1<br />

p 1<br />

= p q r!<br />

A<br />

r<br />

r=0<br />

p^q<br />

X1<br />

p 1<br />

= p q r!<br />

r<br />

r=0<br />

Xp^q<br />

p 1<br />

= p q (r 1)!<br />

r 1<br />

r=1<br />

#<br />

1<br />

2<br />

q hDF; DGi H<br />

a 2 + p 2 P p<br />

p 1<br />

r=1 (r 1)!2<br />

r 1<br />

q 1<br />

r<br />

Z<br />

q 1<br />

r<br />

q 1<br />

r 1<br />

2 q 1<br />

<br />

I p+q 2 2r f(; z)e r g(; z) (dz)<br />

<br />

I p+q 2 2r (f e r+1 g)<br />

I p+q<br />

2r (f e r g):<br />

r 1 2(p + q 2r)!kf e r gk 2 H (p+q 2r) if p < q;<br />

(10.13)<br />

(a p!hf; gi H p) 2 + p 2 P p 1<br />

p 1 4(2p<br />

r=1 (r 1)!2<br />

r 1<br />

2r)!kf e r gk 2 if p = q:<br />

H (2p 2r)<br />

kf e r gk 2 H (p+q 2r) kf r gk 2 H (p+q 2r) = hf p r f; g q r gi H 2r<br />

If r = p < q, then<br />

kf p r fk H 2rkg q r gk H 2r<br />

1 2 kf p r fk 2 H 2r + kg q r gk 2 H 2r <br />

:<br />

kf e p gk 2 H (q p) kf p gk 2 H (q p) kfk 2 H p kg q p gk H 2p:<br />

If r = p = q, then f e p g = hf; gi H p: By plugging these last expressions into (10.13), we deduce<br />

immediately the desired conclusion.<br />

The combination of the results presented in this section with Theorem 10.1 lead to the<br />

following statement, which is a collection of the main …ndings contained in the papers by Peccati<br />

<strong>and</strong> Tudor [73] <strong>and</strong> Nualart <strong>and</strong> Ortiz-Latorre [66].<br />

Theorem 10.2 (See [66, 73]) Fix d 2 <strong>and</strong> let C = fC(i; j) : i; j = 1; :::; dg be a d d<br />

positive de…nite matrix. Fix integers 1 q 1 : : : q d . For any n 1 <strong>and</strong> i = 1; : : : ; d, let f (n)<br />

i<br />

belong to H q i<br />

. Assume that<br />

F (n) = (F (n)<br />

1 ; : : : ; F (n)<br />

d<br />

) = (I q1 (f (n)<br />

1 ); : : : ; I qd (f (n)<br />

d<br />

)) n 1;<br />

is such that<br />

lim<br />

n!1<br />

(n)<br />

E[F<br />

i<br />

F (n)<br />

j<br />

] = C(i; j); 1 i; j d: (10.14)<br />

Then, as n ! 1, the following four assertions are equivalent:<br />

52

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!