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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

586 A. Porretta, L. Véron / Journal of Functional Analysis 236 (2006) 581–591<br />

is boun<strong>de</strong>d at infinity; however, this assumption does not inclu<strong>de</strong> the case when g has a slow<br />

growth at infinity (such as g(r) ≡ r(ln r) α with α>2) and is not so general as (2.3).<br />

Theorem 2.2. Assume that (2.2)–(2.4) hold. Then any solution u of (2.1) is radial and ∂u/∂r > 0<br />

in BR \{0}.<br />

The following preliminary result is a consequence of more general results in [3,11,12,18].<br />

However, we provi<strong>de</strong> here a simp<strong>le</strong> self-contained proof for the radial case.<br />

Lemma 2.1. Let h be a convex increasing function satisfying the Kel<strong>le</strong>r–Osserman condition<br />

<br />

+∞<br />

a<br />

s<br />

ds<br />

√ < ∞, H(s)= h(t) dt, (2.13)<br />

H(s)<br />

for some a>0. Then the prob<strong>le</strong>m<br />

<br />

−v + h(v) = 0 in BR,<br />

lim|x|→R v(x) =∞<br />

has a unique solution.<br />

a<br />

(2.14)<br />

Proof. Since h is increasing, there exist a maximal and a minimal solution v and v, which are<br />

both radial, so that it is enough to prove that v = v. To this purpose, observe that if v is radial<br />

we have (v ′ rN−1 ) ′ = rN−1 h(v) so that, since v ′ (0) = 0, and rep<strong>la</strong>cing H by H ˜ = H − H(min v)<br />

which is nonnegative on the range of values of v,wehave<br />

which yields<br />

Define now<br />

(v ′ r N−1 ) 2<br />

2<br />

<br />

=<br />

0<br />

r<br />

s 2(N−1) h(v)v ′ ds r 2(N−1) ˜<br />

H(v)<br />

0 v ′ <br />

< 2H(v). ˜<br />

(2.15)<br />

<br />

w = F(v)=<br />

v<br />

∞<br />

ds<br />

.<br />

2H(s) ˜<br />

A straightforward computation, and condition (2.13), show that w solves the prob<strong>le</strong>m<br />

<br />

w = b(w)(|Dw| 2 − 1) in BR,<br />

w = 0 on∂BR,<br />

(2.16)

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

Saved successfully!

Ooh no, something went wrong!