2
$\begingroup$

I am trying to understand this paper, and have some basic question, and hope this is OK for the MO.

Let $f\in \mathcal{S}(\mathbb R^d)$ (Schwartz Space).

We know that $\widehat{\nabla f}(\xi)= 2 \pi i \xi \hat{f} (\xi). $ We define $$\widehat{|\nabla| f^{s}} (\xi) = (2 \pi |\xi|)^s \hat{f} (\xi), (s\in \mathbb R).$$

We may now define that $$(- \Delta) ^{s/2}f= |\nabla | f^s, (s\in \mathbb R).$$

We define Little wood-Paley projection associated to $-\Delta$ as follows: Let
$\phi:[0, \infty) \to \infty$ be a smooth functin such that $\phi(\lambda) =1$ for $0 \leq \lambda \leq 1$ and $ \phi(\lambda)=0$ for $ \lambda \geq 2.$

For each dyadic number $N \in 2^{\mathbb Z},$ we define $\phi_N(\lambda) = \phi(\lambda/N)$ and $\psi_N(\lambda) = \phi_N(\lambda)- \phi_{N/2}(\lambda).$ We define the Little wood-Paley projection as follos:

$$\widehat{P_Nf} = \psi_N \hat{f}.$$

(And so $P_Nf= (\psi_N)^{\vee} \ast f$)

Now we define $\mathcal{L}_a$ as the Friedrich extension of the operator $-\Delta + \frac{a}{|x|^2}$ (initially defined on $C_c^{\infty}(\mathbb R^d\setminus \{0\})$) See Section 1.1 for details

Authors (in the same paper page 16) define the Little wood-Paley projections associated to $\mathcal{L}_a$ as follows: $$P_N^a:= \psi_N(\sqrt{\mathcal{L}_a})$$

My Basic Questions are: (1) How should I interpret $\psi_N(\sqrt{\mathcal{L}_a})f$ explicitly? (Do we have explicit formula like: $P_Nf= (\psi_N)^{\vee} \ast f$ ?) (2) How should I use Spectral Theorem to define $m(\sqrt{\mathcal{L}_a})f$ for $m:(0, \infty)\to \mathbb C$ some nice?

$\endgroup$
3
  • $\begingroup$ For any self-adjoint operator $L$ and any measurable function $f$ on the spectrum, $f(L)$ is defined, via the spectral theorem, but to make it more explicit, you'd have to know more about the spectral representation (which for the Laplacian can be implemented using the FT, but it won't be as easy for $L_a$). However, these topics seem too basic for this site; math.stackexchange.com is the right place to ask (more concrete questions tend to work better, though). $\endgroup$ Commented May 22, 2017 at 16:26
  • $\begingroup$ @CR: Thanks. Do I need to find spectral representation to define $\psi_N(\sqrt{\mathcal L}_a)f$ precisely? ( Any reference? where I could find this..., I think, paper I am reading authors have not mentioned explicit formula for this...) PS: I am wondering without explicit formula(representation), how one could get the further insights.. $\endgroup$
    – XYZ
    Commented May 23, 2017 at 5:11
  • $\begingroup$ For example, you could work with a spectral representation (= unitary transformation so that $L_a$ becomes multiplication by the variable in the new Hilbert space $\bigoplus L^2(\mathbb R, \rho_n)$), and then $\psi_N(\sqrt{L_a})$ is multiplication by $\psi_N(\sqrt{t})$ there. I don't think this can be worked out very explicitly in your example (though I might be wrong about this). If you want to read up on the general background, then you can use any book that discuss the spectral theorem in Hilbert spaces. $\endgroup$ Commented May 23, 2017 at 15:56

0

You must log in to answer this question.

Browse other questions tagged .