5
$\begingroup$

Let $V$, $W$ be vector bundles over a compact Riemannian manifold $M$ and let $F$ be a smooth elliptic operator of order $k$ from $V$ to $W$. "Standard elliptic theory" then gives us the following two estimates:

  1. For any $l \geq 0$ and $\alpha \in (0,1)$ there exists a constant $C>0$ such that $$\|v\|_{C^{k+l,\alpha}} \leq C \left( \|F(v)\|_{C^{l,\alpha}}+\|v\|_{C^0} \right)$$ for all $v \in C^{k+l,\alpha}(V)$.

  2. For any $l \geq 0$ and $\alpha \in (0,1)$ there exists a constant $D>0$ such that $$\|v\|_{C^{k+l,\alpha}} \leq D \|F(v)\|_{C^{l,\alpha}}$$ for all $v \in C^{k+l,\alpha}(V)$ and $v \perp \text{Ker}F$.

Similar estimates exist for the $L^p$-theory, i.e. take the $L^p_{k}$-norm instead of the $C^{k,\alpha}$-norm for $p \in (1,\infty)$. I am interested in both estimates.

Question: Are there situations in which $C$ or $D$ has been computed explicitly?

I have a personal interest in the Laplace operator acting on $1$-forms.

I know there is some work for operators acting on functions on domains in $\mathbb{R}^n$, namely the references listed in the introduction of Kouta Sekine, Kazuaki Tanaka, Shin'ichi Oishi: Inverse norm estimation of perturbed Laplace operators and corresponding eigenvalue problems. But that's not for compact manifolds, and only for trivial vector bundles.

$\endgroup$
6
  • $\begingroup$ Do you mean the best possible constant or a sufficiently large explicit constant? $\endgroup$
    – Deane Yang
    Commented Feb 3, 2023 at 14:42
  • $\begingroup$ A sufficiently large explicit constant. The estimate need not be sharp. $\endgroup$
    – user505117
    Commented Feb 3, 2023 at 17:18
  • $\begingroup$ If you work your way through the proof of any of these inequalities, it is always possible at each step to figure out an explicit constant. Usually, you don’t need an actual formula. You just need to know exactly what parameters the constant depends on. This is true, for example, when you are using these estimates to solve a nonlinear PDE. Is this the context of your question? $\endgroup$
    – Deane Yang
    Commented Feb 3, 2023 at 17:52
  • $\begingroup$ @DeaneYang I want to give a numerically verified proof of an analysis statement. For this, I have an approximate solution to a PDE. It is explicit and I can calculate its failure to satisfy the PDE, which is small. If the inverse of the linearisation of the PDE is not too large, then I can show that the approximate solution can be perturbed to a genuine solution, which is close to the approximate solution. Some numerically verified proofs like this exist in the literature, and they use explicit (though not sharp) elliptic estimates. $\endgroup$
    – user505117
    Commented Feb 3, 2023 at 21:23
  • $\begingroup$ This then is for some specific Riemannian manifolds? $\endgroup$
    – Deane Yang
    Commented Feb 3, 2023 at 22:04

1 Answer 1

2
$\begingroup$

The closest I found:

In Michael Plum: Explicit $H_2$ Estimates and Pointwise Bounds for Solutions of Second-Order Elliptic Boundary Value Problems the following explicit estimate is given (equation 3): $$||u||_{L^\infty,\Omega} \leq K ||Lu||_{L^2,\Omega}+K_B ||Bu||_{L^\infty,\partial \Omega}.$$ Here, $L$ denotes an order $2$ differential operator like the Laplacian, $\Omega$ denotes a domain in $\mathbb{R}^2$ or $\mathbb{R}^3$ and $\partial \Omega$ its boundary, and $B$ is the boundary operator of Dirichlet, Neumann, or mixed type.

Edit: Other examples are Theorem 1.1 in Batu Güneysu, Stefano Pigola: Quantitative $C^1$-estimates on manifolds, and Theorem 2.1 in Batu Güneysu, Stefano Pigola: Nonlinear Calderón–Zygmund inequalities for maps.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .