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.

The assumption tells us that the form<br />

D. Serre / Journal of Functional Analysis 236 (2006) 409–446 415<br />

Q(G) + 2φ(G)· Z +|Z| 2<br />

takes non-negative values when Z ∈ R l and G has rank one. This amounts to saying that the<br />

quadratic form<br />

Q(G) − φ(G) 2<br />

is non-negative over the cone of rank-one matrices of size (d − 1) × n. Since we have min{d −<br />

1,n} 2, this implies ([16, Theorem 2.3], see also [18]) that there exists a null-form Q0 over<br />

M(d−1)×n(R) such that Q+(G) := Q(G) −|φ(G)| 2 + Q0(G) is non-negative. Hence the form<br />

is non-negative. ✷<br />

W(F)+ Q0(G) = Q+(G) + φ(G)+ p(Y) 2<br />

Examp<strong>le</strong>. Let us consi<strong>de</strong>r the energy of an isotropic e<strong>la</strong>stic material, given by (5), which is<br />

convex if and only if λ 0 and 2λ + μd 0, a condition that <strong>de</strong>pends on the dimension. Rankone<br />

convexity holds if and only if λ 0 and 2λ + μ 0. It is easy to see that in this <strong>la</strong>tter case,<br />

there exists a null-form Q0 such that W + Q0 is convex; however, this fact is use<strong>le</strong>ss when the<br />

physical domain has a non-trivial boundary, as in our case. The meaningful statement is that<br />

when λ 0 and λ + μ 0 (an intermediate assumption if d 3)), W(G,Y) is non-negative<br />

when G has rank one, and therefore there exists a tangential null-form that can be ad<strong>de</strong>d to W to<br />

make it convex. For instance, in the extreme case μ =−λ (say that λ = 2), then<br />

W(F)= 1<br />

T<br />

F + F<br />

2<br />

2 − 2(Tr F) 2<br />

= (F12 − F21) 2 + (F23 + F32) 2 + (F13 + F31) 2 + (F33 − F11 − F22) 2<br />

+ 4(F12F21 − F11F22). (7)<br />

The <strong>la</strong>st term of the right-hand si<strong>de</strong> is a TNF, and the rest is non-negative.<br />

2.2. Convex stored energies in general<br />

For more general energy <strong>de</strong>nsities, it may happen that W is convex, even though W is not<br />

convexifiab<strong>le</strong> in the sense given above. To study the convexity of W in a systematic way, we<br />

perform a Fourier transform v = Fyu with respect to the tangential variab<strong>le</strong>s. This has the effect<br />

to <strong>de</strong>coup<strong>le</strong> the tangential frequencies. The following <strong>le</strong>mma is obvious, except for the notations:<br />

we extend W to Mn(C) as a sesquilinear form. Thus W keeps real values. Additionally, we use<br />

the same <strong>de</strong>composition F = (G, Y ) as above.

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

Saved successfully!

Ooh no, something went wrong!