Let $u\in H^k(\mathbb{R}^n)$ a function with complex-valued.
Question 1: If $u\in H^k(\mathbb{R}^n)$ and $D^\alpha u\in L_2(\mathbb{R}^n)$, by definition for each $|\alpha|\le k,$ then we have
$$\|D^\alpha u\|_{L_2}=\|\widehat{D^\alpha u}\|_{L_2}=\|(iy)^\alpha\widehat{u}\|_{L_2},$$
by properties of Fourier transform.
So is correct to afirm, for multiindex $\alpha=k e_i,$ $\{e_j\}$ basis of $\mathbb{R}^n$, that
$$\int_{\mathbb{R}^n}|y|^{2k}|\widehat{u}|^2dy=\int_{\mathbb{R}^n}|D^k u|^2 dx?$$ Cause in Evans book I found a inequality for this setup (chapter 5, section 8).
Question 2: How can I show that previous equation implies that
$$\int_{\mathbb{R}^n}(1+|y|^{k})^2|\widehat{u}|^2dy\le C\|u\|^2_{H^k} ?$$
Someone has a hint for me?
Thanks.
Well for the full norm you need all derivatives up to order $k$ not just the highest one, so apply the rule for general $|\alpha|\le k$, and sum the result. For example you need the $L^2$ norm in there, this will give you the $1$ since $\|u\|_2^2=\|\hat u\|_2^2$.