1
$\begingroup$

Is it true that $$\| \Delta u \|_{L^p(\Bbb R^d)} + \| u \|_{L^p(\Bbb R^d)}\quad\text{and}\quad\| u \|_{W^{2,p}(\Bbb R^d)}$$ are equivalent norms?

This results is pretty easy and straightforward for $p=2$ using techniques via Fourier transform and Plancherel. But what could we use in place of Fourier transform when $p\neq 2?$

Please prove or disprove or provide me with some good reference where I can fine.

I fact I need to show that the domain of the generator of the Gauss-Weierstrass semigroup in $L^p(\Bbb R^d)$ is $W^{2,p}(\Bbb R^d)$. This result is counterpart for the case $p=2$ where the domain of the generator is $W^{2,2}(\Bbb R^d)$.

$\endgroup$
3
  • 4
    $\begingroup$ This is well known. You can start, for instance, with the textbook of Gilbarg and Trudinger. $\endgroup$ Commented Dec 19, 2017 at 19:27
  • $\begingroup$ why a down vote here? is this question stupid? $\endgroup$
    – Guy Fsone
    Commented Dec 19, 2017 at 20:21
  • 2
    $\begingroup$ The question is perfectly reasonable, but the timing is off. Let's say by seventy years or so. $\endgroup$ Commented Dec 21, 2017 at 16:36

1 Answer 1

5
$\begingroup$

First, the classical Calderón–Zygmund estimate gives $$\left\|D^2 u\right\|_{L^p(\mathbb{R}^d)}\leq C(p,d) \left\|\Delta u\right\|_{L^p(\mathbb{R}^d)}.$$ By interpolation, $\left\|u\right\|_{W^{2,p}(\mathbb{R})}$ is equivalent to $$\left\|\Delta^2 u\right\|_{L^p(\mathbb{R}^d)}+\left\|u\right\|_{L^p(\mathbb{R}^d)}.$$ For detail proof you can look up Gilbarg-Trudinger Chapter 9 for $L^p$ estimate, and Adams for Sobolev spaces.

$\endgroup$
2
  • $\begingroup$ How do we manage the gradient terms? $\endgroup$
    – Guy Fsone
    Commented Dec 21, 2017 at 6:20
  • 2
    $\begingroup$ In Adams' book Sobolev space, lemma 5.6 shows that the gradient term can be bounded by zero order and second order terms. $\endgroup$ Commented Dec 21, 2017 at 14:05

You must log in to answer this question.

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