0
$\begingroup$

Let

  • $D:=(0,1)$
  • $U:=L^2(D)$ and $$\phi_n(x):=\sqrt 2\sin(n\pi x)\;\;\;\text{for }n\in\mathbb N\text{ and }x\in D$$
  • $H:=H^2(D)\cap H_0^1(D)$ and $$A:=-\frac{\partial^2}{\partial x^2}u\;\;\;\text{for }u\in H$$

Since $(\phi_n)_{n\in\mathbb N}\subseteq C^\infty(D)$ is an orthonormal basis of $U$ with $$\left.\phi_n\right|_{\partial D}=0\;\;\;\text{for all }n\in\mathbb N\;,$$ $(\phi_n)_{n\in\mathbb N}\subseteq H$ with $$A\phi_n=\underbrace{\pi^2n^2}_{=:\lambda_n}\phi_n\;\;\;\text{for all }n\in\mathbb N\tag 1\;.$$

Let $r\in\mathbb R$, $$\mathfrak D(A^r):=\left\{u\in U:A^ru:=\sum_{n\in\mathbb N}\lambda_n^\alpha\langle u,\phi_n\rangle\phi_n\in U\right\}$$ and $$\left\|u\right\|_r:=\left\|A^ru\right\|\;\;\;\text{for }u\in\mathfrak D(A^r)\;.$$ I want to show that $$\left\|u\right\|_r=\sum_{n\in\mathbb N}(n\pi)^{4r}\langle u,\phi_n\rangle^2\tag 2\;.$$ We need somehow to use that $$u=\sum_{n\in\mathbb N}\langle u,\phi_n\rangle\phi_n\;\;\;\text{for all }u\in U\;,$$ but I don't know how I need to proceed in $$\left\|u\right\|_r=\int_D\left|\sum_{n\in\mathbb N}(n\pi)^{2r}\langle u,\phi_n\rangle\phi_n\right|^2{\rm d}\lambda\;\;\;\text{for all }u\in U\;.\tag 3$$

How can we prove $(2)$ and why (and that is the main part of the question) can we conclude that $\mathfrak D(A^{r/2})=H_0^r(D)$?

I've found the last statement in An Introduction to Computational Stochastic PDEs, Example 10.9, and absolutely don't understand why it holds.

$\endgroup$
  • $\begingroup$ @MoritzDoll You're right! I've confused the symbols used in the mentioned book with mine. Do you know why $\mathfrak D(A^{r/2})=H_0^r(D)$? $\endgroup$ – 0xbadf00d Feb 24 '16 at 16:26
0
$\begingroup$

$\|u\|_r = \|A^r u\| = \|\sum \langle u,ϕ_j\rangle λ_j^r ϕ_j\|$. Now using Parseval (and that the $ϕ_j$ are an orthonormal basis) we get $\|u\|_r = \sum λ_j^{2r}|\langle u, ϕ_j\rangle|^2$.

$\endgroup$
  • $\begingroup$ That was easy! Do you know why $\mathfrak D(A^{r/2})=H_0^r(D)$? $\endgroup$ – 0xbadf00d Feb 24 '16 at 16:18
  • $\begingroup$ Apply $A^{r/2}$ again and then use elliptic regularity, ie $∂^2$ is an isomorphisms between $H^s$ and $H^{s-2}$. $\endgroup$ – mcd Feb 24 '16 at 16:29
  • $\begingroup$ Apply $A^{r/2}$ to what? Unfortunately, I'm not aware of the fact that $\partial^2/\partial x^2$ is an isomorphism between $H^s$ and $H^{s-2}$. Is there an easier proof for $\mathfrak D(A^{r/2})=H_0^r(D)$ if $r\in\mathbb N$? $\endgroup$ – 0xbadf00d Feb 24 '16 at 16:35
  • $\begingroup$ For even $r$ its clear (because H^{2r} is defined such that $∂^{2r}u ∈ L^2$). For not-even $r$ it is not obvious what $A^{r/2}$ (really) is, so you need some abstract argument. $\endgroup$ – mcd Feb 24 '16 at 16:42
  • $\begingroup$ If $W^k(D)$ is the space of $k$-times weakly differentiable functions $D\to\mathbb R$, then $$H^{2r}(D)\stackrel{\text{def}}=\left\{u\in W^{2r}(D):\frac{{\rm d}^k}{{\rm d}x^k}u\in U\text{ for all }k\in\left\{0,\ldots,2r\right\}\right\}\;.$$ Could you please explain in more detail why this set is equal to $\mathfrak D(A^r)=H_0^r(D)$, for all $r\in\mathbb N$? $\endgroup$ – 0xbadf00d Feb 24 '16 at 17:32

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.