$A$ is a symmetric tensor in $L^\infty(\Omega)^{n\times n}$, with $\Omega$ a bounded domain in $\mathbb{R}^n$.

Prove that

$\displaystyle\int_\Omega (Av)\cdot (Av)\,dx\leq \int_\Omega (Av)\cdot (\|A\|_{L^\infty(\Omega)}v)\,dx$

where $v\in L^2(\Omega)^n$.

  • $\begingroup$ I edited my question with a specific space where live the vector. $\endgroup$ – yemino Aug 16 '16 at 21:52

This boils down to the question, whether for symmetric $A$ $$ \|Ax\|_2^2\le \|A\|_\infty \cdot x^TAx $$ holds for all $x$.

This is not true without positive definiteness of $A$ (take $A=-I$).

It is true for positive definite $A$: $$ \|Ax\|_2^2 =\|A^{1/2}A^{1/2}x\|_2^2 \le \|A^{1/2}\|_2^2 \cdot \|A^{1/2}x\|_2^2 \le \|A^{1/2}\|_2^2\cdot x^TAx. $$ Since $A^{1/2}$ is symmetric positive definite $$ \|A^{1/2}\|_2^2 = \lambda_\max(A^{1/2})^2 = \lambda_\max(A) \le \|A\|_\infty, $$ which is the claim. Note that we can use any matrix norm on $\mathbb R^{n,n}$ instead of $\|\cdot\|_\infty$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.