1
$\begingroup$

Reading this recent thread I was reminded of a related problem I still haven't solved so I post it here in hopes of a positive result.


Let $V_0 \subset H_0$ and $V_1 \subset H_1$ be separable Hilbert spaces with embeddings continuous, compact and dense. Let $$\phi_{1 \to 0}:H_1 \to H_0$$ be a bounded linear (bijective) map with bounded linear inverse $\phi_{0 \to 1}$. Suppose also that $$\phi_{1 \to 0}:V_1 \to V_0$$ is also bounded with the bounded inverse $\phi_{0 \to 1}$.

Let $\{b_j\}$ be an orthonormal basis for $H_0$ which is also orthogonal in $V_0$ and define the sets $$B^n_0 := \text{span}(b_1, ..., b_n)\quad \text{and}\quad B^n_1 = \phi_{0 \to 1}(B^n_0) = \text{span}(\phi_{0 \to 1}(b_1), ..., \phi_{0 \to 1}(b_n))$$

The orthogonal projection operator $P^n_0:H_0 \to B^n_0$ is defined to satisfy $(P^n_0u-u, u_n)_{H_0}=0$ for all $u_n \in B^n_0$, for $u \in H_0$. It's bounded in $H_0$ and let's assume that$$\lVert P^n_0u \rVert_{V_0} \leq \lVert u \rVert_{V_0}\quad\text{for all $u \in V_0$.}$$

Question: Define $P^n_1:H_1 \to B^n_1$ by $$(P^n_1v-v, v_n)_{H_1} = 0\quad\text{for all $v_n \in B^n_1$.}$$

Does this operator also satisfy $$\lVert P^n_1v \rVert_{V_1} \leq C\lVert v \rVert_{V_1}\quad\text{for all $v \in V_1$,}$$ for a constant $C$ not depending on $n$?


Remarks:

  • Morally I feel like it should since there is a map between the two spaces which is continuous/bounded.

  • Note that $\phi_{1 \to 0}(P^n_1v)=(\phi_{0 \to 1})^{-1}(P^n_1v) \in B^n_0$, so there is some $w \in V_0$ such that $P^n_0w = \phi_{1 \to 0}(P^n_1v)$, then by continuity of the maps we get $$\lVert P^n_1v \rVert_{V_1} \leq C\lVert w \rVert_{V_1}$$ but I can't get a bound on the $w$. Maybe it can be chosen in a way to get my estimate to work..

  • It would be enough for me to consider the model case where the spaces $H$ are $L^2$-like spaces with $(u,v)_{H_0} := \int_\Omega uv$ and $(a,b)_{H_1} := \int_{\Omega} h\phi_{1 \to 0}a \phi_{1 \to 0}b$ where $h$ is a smooth weight (and $\Omega$ is bounded).

$\endgroup$

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.