20.07.2013 Views

Estimation optimale du gradient du semi-groupe de la chaleur sur le ...

Estimation optimale du gradient du semi-groupe de la chaleur sur le ...

Estimation optimale du gradient du semi-groupe de la chaleur sur le ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

S. Albeverio, A. Kosyak / Journal of Functional Analysis 236 (2006) 634–681 643<br />

<br />

<br />

<strong>de</strong>t C<br />

= exp −<br />

(2π) m 1<br />

exp(tEpq)<br />

2<br />

∗ C exp(tEpq)x, x <br />

dx = dμBpq(t)(x),<br />

where (Bpq(t)) −1 = Cpq(t) = exp(tEpq) ∗ C exp(tEpq) (we note that <strong>de</strong>t C = <strong>de</strong>t Cpq(t)).<br />

Hence, using (2) we get<br />

We shall prove that<br />

H μ LI+tEpq <br />

B ,μB =<br />

<strong>de</strong>t Cpq(t) + C<br />

2<br />

<strong>de</strong>t Cpq(t) <strong>de</strong>t C<br />

<strong>de</strong>t 2 Cpq(t)+C<br />

2<br />

1/4<br />

<br />

=<br />

<strong>de</strong>t C<br />

<strong>de</strong>t Cpq(t)+C<br />

2<br />

1/2<br />

. (3)<br />

= <strong>de</strong>t C + t2<br />

4 cppA q q(C), (4)<br />

where A p q (C), 1 p,q m, <strong>de</strong>note the cofactors of the matrix C corresponding to the row p<br />

and the column q. Wehave<br />

hence<br />

<strong>de</strong>t Cpq(t)+C<br />

2<br />

<strong>de</strong>t C<br />

<br />

and finally, using (3) we get<br />

where<br />

H μ mLI+tEpq m<br />

B ,μB =<br />

B (n) = <br />

1r,sm<br />

= <strong>de</strong>t C + t2<br />

<strong>de</strong>t C<br />

<strong>de</strong>t Cpq(t)+C<br />

2<br />

∞<br />

n=q+1<br />

4 cppA q q(C)<br />

= 1 +<br />

<strong>de</strong>t C<br />

t2<br />

4 cppbqq,<br />

1/2<br />

<br />

= 1 + t2<br />

4 cppbqq<br />

−1/2 H μ LI+tEpq<br />

B (n)<br />

<br />

,μB (n) =<br />

∞<br />

n=q+1<br />

b (n)<br />

rs Ers and C (n) := B (n) −1 = <br />

<br />

1 + t2<br />

4 c(n)<br />

1r,sm<br />

pp b(n) qq<br />

c (n)<br />

rs Ers.<br />

−1/2<br />

,<br />

So using the properties of the Hellinger integral for two Gaussian mea<strong>sur</strong>es we conclu<strong>de</strong> that<br />

m LI+tEpq m<br />

μB ⊥ μB ∀t ∈ R \{0} ⇔<br />

∞<br />

n=q+1<br />

<br />

1 + t2<br />

4 c(n)<br />

⇔ S L m<br />

pq μB =∞.<br />

pp b(n) qq<br />

−1/2<br />

= 0

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

Saved successfully!

Ooh no, something went wrong!