Sobolev space interpolation

1.3k Views Asked by At

We are attempting to show that, for $t_0,t_1\in\mathbb{R}$, and $t_\sigma=(1-\sigma)t_0+\sigma t_1$ for any $\sigma\in[0,1]$, then the following inequality holds for any $u\in H^{t_1}(\mathbb{R}^d)$:

$\|u\|_{H^{t_\sigma}(\mathbb{R}^d)}\leq\|u\|^{1-\sigma}_{H^{t_0}(\mathbb{R}^d)}\cdot \|u\|^\sigma_{H^{t_1}(\mathbb{R}^d)}$

A hint given to us was to write the LHS as a Fourier transform, but we do not see how to go about doing that nor how that would help. Any assistance in this area would be greatly appreciated. Thank you.

1

There are 1 best solutions below

4
On

We have, by definition of $H^{t_\sigma}(\def\R{\mathbf R}\R^d)$, that $$ \def\norm#1{\left\|#1\right\|}\def\F{\mathscr F} \norm{u}_{H^{t_\sigma}} = \norm{ (1 + \def\abs#1{\left|#1\right|}\abs\cdot^2)^{t_\sigma/2}\F u}_{L^2} $$ We have for $\frac 1\alpha + \frac 1\beta = \frac 12$ by Hölder \begin{align*} \norm{u}_{H^{t_\sigma}} &= \norm{ (1 + \def\abs#1{\left|#1\right|}\abs\cdot^2)^{t_\sigma/2}\F u}_{L^2}\\ &= \norm{\def\f{(1+\abs\cdot^2)}\abs{\f^{t_0/2}}^{1-\sigma}\abs{\F u}^{1-\sigma} \cdot \abs{\f^{t_1/2}}^\sigma\abs{\F u}^\sigma}_{L^2}\\ &\le \norm{\abs{\f^{t_0/2} \F u}^{1-\sigma}}_\alpha \norm{\abs{\f^{t_1/2}\F u}^{\sigma}}_\beta \end{align*} We want the first term to be $\norm{u}_{H^{t_0}}$, that is, $\alpha = \frac{2}{1-\sigma}$, then $$ \beta = \frac 1{\frac 12 - \frac 1\alpha} = \frac 1{\frac \sigma 2} = \frac 2\sigma $$ so we have \begin{align*} \norm{u}_{H^{t_\sigma}} &\le \norm{\abs{\f^{t_0/2} \F u}^{1-\sigma}}_\alpha \norm{\abs{\f^{t_1/2}\F u}^{\sigma}}_\beta\\ &= \norm{\f^{t_0/2} \F u}_{L^2}^{1-\sigma} \norm{\f^{t_1/2}\F u}_{L^2}^{\sigma}\\ &= \norm{u}_{H^{t_0}}^{1-\sigma}\norm{u}_{H^{t_1}}^\sigma \end{align*}