# Characterization of the $H^{-1}$-norm in Evans's PDE book

I'm reading through the Sobolev Spaces section in Evans's Partial Differential Equations book, and I was stuck on a theorem characterizing the $H^{-1}$ norm. On page 299 Theorem 1 (in the second edition), he proves that for $f \in H^{-1}(U)$, $$\begin{multline}\|f\|_{H^{-1}(U)} = \inf \Big\{ \left( \int_U \sum_{i=0}^n |f^i|^2 dx \right): \langle f, v \rangle = \int_U f^0 v + \sum_{i=1}^n f^i v_{x_i} dx,\\ \;f^0,\ldots,f^n \in L^2(U), \; \forall v \in H^1(U)\Big\}. \end{multline}$$

Here $\langle \cdot, \cdot \rangle$ denotes the dual pairing of $H^{-1}$ and $H_0^1$.

To show this, he first states that given $f \in H^{-1}(U)$, we can apply to the Riesz Representation Theorem to get an element $u \in H_0^1$ such that $(u,v)_{H_0^1} = \langle f,v \rangle \; \forall v \in H_0^1$, where $(u,v)_{H_0^1} = \int Du \cdot Dv + uv dx$ is an inner product on $H_0^1$. Then we can define $f^0=u$ and $f^i = u_{x_i}$ for each $i$.

Now, if $\langle f, v \rangle = \int_U g^0 v + \sum_{i=1}^n g^i v_{x_i} dx$, for $g^0,\ldots,g^n \in L^2(U)$, then by choosing $v=u$, we can see that $$\int_U |Du|^2 +|u|^2 dx \leq \int_U \sum_{i=0}^n |g^i|^2 dx.$$

How does he get this inequality?

When you let $v=u$, you get $$\int_U |Du|^2 +|u|^2 dx = \int_U g^0 u + \sum_{i=1}^n g^i u_{x_i} dx$$ and I don't see how you can get the $\{u,u_{x_i}\}$ terms to disappear on the right side.

-

It's an application of Cauchy-Schwarz inequality(ies). Since $u$, $u_{x_i}$ and $g^i$are in $L^2(U)$ we have $\int_ug^0udx\leq \sqrt{\int_U|g^0|^2dx}\sqrt{\int_U|u|^2dx}$ and for all $i\in\{1,\ldots, n\}:\int g^iu_{x_i}\leq \sqrt{\int_U|g^i|^2dx}\sqrt{\int_U|u_{x_i}|^2dx}$, hence \begin{align*} \lVert u\rVert_{H_0^1(U)}^2&=\int_U g^0udx+\sum_{i=0}^ng^iu_{x_i}dx\\ &\leq \sqrt{\int_U|g^0|^2dx}\sqrt{\int_U|u|^2dx}+\sum_{i=1}^n\sqrt{\int_U|g^i|^2dx}\sqrt{\int_U|u_{x_i}|^2dx} \\ &\leq \sqrt{\int_U|g^0|^2dx+\sum_{i=1}^n\int_U |g^i|^2dx}\sqrt{\int_U|u|^2dx+\sum_{i=1}^n\int_U|u_{x_i}|^2dx}, \end{align*} applying this time Cauchy-Schwarz inequality for sums, namely $\displaystyle\sum_{i=0}^n a_ib_i\leq\sqrt{\sum_{i=0}^n|a_i|^2}\sqrt{\sum_{i=0}^n|b_i|^2}$, for $\displaystyle a_0=\sqrt{\int_U|g^0|^2dx}$, $\displaystyle b_0=\sqrt{\int_U|u|^2dx}$, $\displaystyle a_i=\sqrt{\int_U|g^i|^2dx}$ and $\displaystyle b_i=\sqrt{\int_U|u_{x_i}|^2dx}$ for $1\leq i\leq n$. If $u=0$, the inequality we have to show is obvious, and if it's not the case we can divide by $\displaystyle\sqrt{\int_U|u|^2dx+\sum_{i=1}^n\int_U|u_{x_i}|^2dx}$.
Thanks for the response! I don't think Evans ever mentions that $u$ is normalized to have $H_0^1$-norm 1 in the inequality, though. Are you suggested that it is implied? – user1736 Oct 14 '11 at 13:30
$u$ doesn't need to have $H_0^1$ norm, since it's not the case in general (it represents a linear functional $f$, so if $\lVert u\rVert_{H_0^1}=1$, then the "$u$" which corresponds to $2f$ won't have a $H_0^1$ norm $1$). That's reminds me a square is missing at the first line of the align. I will correct it readily. – Davide Giraudo Oct 14 '11 at 14:25