Fourier Transform of Anisotropic Mixed-norm Hardy Spaces with Applications to Hardy-Littlewood Inequalities

Let \vec{p}\in{(0,1]}^n, r\in(\max\{p_+,1\},\infty ], s be as in (?) and f\in{H_A^{\vec{p}}({\mathbb{R}^n})}. Without loss of generality, we may assume that \|f\|_{{H_A^{\vec{p}}({\mathbb{R}^n})}}>0. Then, by Lemma 3.7 and Definition 3.2(2), we find that there exist a sequence \{\lambda_i\}_{i\in{\mathbb N}}\subset\mathbb{C} and a sequence of (\vec{p},r,s)-atoms \{a_i\}_{i\in{\mathbb N}}, supported respectively in \{B^{(i)}\}_{i\in{\mathbb N}}\subset\mathfrak{B}, such that

and

Taking the Fourier transform on the both sides of (Proof), we have

Note that a function f\in L^1({\mathbb{R}^n}) implies that \widehat{f} is well defined in {\mathbb{R}^n}, so does \widehat{a_i} for any i\in{\mathbb N}. From Lemmas 3.4 and 3.6, and (Proof), it follows that, for any x\in{\mathbb{R}^n},

Therefore, for any x\in{\mathbb{R}^n}, the function

is well defined pointwisely and

We next show the continuity of the function F on {\mathbb{R}^n}. If we can prove that F is continuous on any compact subset of {\mathbb{R}^n}, then the continuity on {\mathbb{R}^n} is obvious. Indeed, for any compact subset E, there exists a positive constant K, depending only on A and E, such that \rho_*(x)\le K holds for every x\in E. By this and (Proof), we conclude that, for any x\in E,

Thus, the summation \sum_{i\in{\mathbb N}}\lambda_i\widehat{a_i}(\cdot) converges uniformly on E. This, together with the fact that, for any i\in{\mathbb N}, \widehat{a_i}(x) is continuous, implies that F is also continuous on any compact subset E and hence on {\mathbb{R}^n}.

Finally, to complete the proof of Theorem 3.1, by (Proof) and (Proof), we only need to show that

For this purpose, from Lemma 3.4 and the definition of Schwartz functions, we deduce that, for any \varphi\in{\mathcal S}({\mathbb{R}^n}) and i\in{\mathbb N},

where B^*_0 is the unit dilated ball with respect to A^*. This implies that there exists a positive constant C such that \left|\int_{{\mathbb{R}^n}}\widehat{a_i}(x)\varphi(x)\,dx\right|\le C holds true uniformly for any i\in{\mathbb Z}. Combining this, Lemma 3.6 and (Proof), we have

Therefore, for any \varphi\in{\mathcal S}({\mathbb{R}^n}),

This finishes the proof of (Proof) and hence of Theorem 3.1.

Without loss of generality, we may take x_0=\mathbf{0}. This implies that \mathop\mathrm{\,supp\,}(D_{A}^{i_0}a)\subset B_0. Then, for any \alpha\in{\mathbb Z}_+^n with |\alpha|\leq s and x\in{\mathbb{R}^n}, we easily obtain

Let P be the degree s-|\alpha| Taylor expansion polynomial of the function e^{-2\pi\imath \xi\cdot x} centered at the origin \mathbf{0} respect to \xi. Applying Definition 3.2(iii), the Taylor remainder theorem and the Hölder inequality, we have

The other estimate can be directly deduced from the Hölder inequality, namely,

This finishes the proof of Lemma 3.3.

Let a be a (\vec{p},r,s)-atom supported in x_0+B_{i_0} with some x_0\in{\mathbb{R}^n} and i_0\in{\mathbb Z}. Without loss of generality, we may assume x_0=\mathbf{0}. By Lemma 3.3 with \alpha=(\overbrace{0,\ldots,0}^{n\ \rm times}), we find that, for any x\in{\mathbb{R}^n},

Next, we show that

Indeed, there exists a K\in{\mathbb Z} large enough such that, if i_0\in(K,\infty )\cap{\mathbb Z}, then

On the other hand, if i_0\in(-\infty ,K], by [hlyy20, Lemma 6.8], we conclude that, for any \varepsilon\in(0,1),

Letting \varepsilon\to0, we have

Thus, (Proof) holds true. From this and (Proof), it follows that, for any x\in{\mathbb{R}^n},

We next prove (3.4) by considering two cases: \rho_*(x)\leq b^{-i_0} and \rho_*(x)>b^{-i_0}.

Case 1). \rho_{*}(x)\leq b^{-i_0}. In this case, note that \rho_*((A^*)^{i_0}x)\leq 1. From (Proof), Lemma 3.5 and the fact that

we deduce that, for any x\in{\mathbb{R}^n} satisfying \rho_*(x)\leq b^{-i_0},

This shows (3.4) for Case 1).

Case 2). \rho_*(x)>b^{-i_0}. In this case, note that \rho_*((A^*)^{i_0}x)>1. Using (Proof), Lemma 3.5 again and the fact that

it is easy to see that, for any x\in{\mathbb{R}^n} satisfying \rho_*(x)>b^{-i_0},

which completes the proof of (3.4) and hence of Lemma 3.4.

Observe that, for any \{\lambda_i\}_{i\in{\mathbb N}}\subset\mathbb{C} and \gamma\in(0,1],

By this and the inverse Minkovski inequality, we know that

Letting N\to\infty, we obtain the desired inequality as in Lemma 3.6.

Let \vec{p}\in(0,1]^n, r\in(\max\{p_+,1\},\infty ], s be as in (?) and f\in{H_A^{\vec{p}}({\mathbb{R}^n})}. Then, by Lemma 3.7 and Definition 3.2(2), we find that there exists a sequence \{\lambda_i\}_{i\in{\mathbb N}}\subset\mathbb{C} and a sequence of (\vec{p},r,s)-atoms, \{a_i\}_{i\in{\mathbb N}}, supported, respectively, in \{B^{(i)}\}_{i\in{\mathbb N}}\subset\mathfrak{B} such that

and

Furthermore, from the proof of Theorem 3.1, it follows that the function

is continuous and satisfies that \widehat{f}=F in {\mathcal S}'({\mathbb{R}^n}).

Thus, to show Theorem 4.1, we only need to prove (4.1) holds true for the function F as in (Proof). To do this, observe that, for any (\vec{p},r,s)-atom a supported in x_0+B_{k_0} with some x_0\in{\mathbb{R}^n} and k_0\in{\mathbb Z}, when \rho_*(x)\leq b^{-k_0}, (Proof) holds true. This, together with the fact that

implies that

For any x\in{\mathbb{R}^n}, we get the following inequality by (Proof):

Moreover, by (3.4) and the fact \sum_{i\in{\mathbb N}}|\lambda_i|<\infty, we know that the dominated convergence theorem can be applied to the right side of (Proof). Combining this and (Proof), we deduce that

which completes the proof of Theorem 4.1.

Let \vec{p}\in(0,1]^n and f\in{H_A^{\vec{p}}({\mathbb{R}^n})}. Then, by Lemma 3.7 and Definition 3.2(2), we find that there exist a sequence \{\lambda_i\}_{i\in{\mathbb N}}\subset\mathbb{C} and a sequence of (\vec{p},2,s)-atoms \{a_i\}_{i\in{\mathbb N}}, supported respectively, in \{B^{(i)}\}_{i\in{\mathbb N}}\subset\mathfrak{B} such that

and

To prove Theorem 4.3, it suffices to show that (4.3) holds true for the function F as in (Proof). For this purpose, by the fact that \underline{p}\le p_+\le1, the inverse Minkovski inequality and (Proof), we have

On another hand, from (Proof), the fact that p_+\in(0,1], (Proof) and the Fatou lemma, it follows that

Next, we devote to proving the following uniform estimate for all (\vec{p},2,s)-atoms, namely,

where M is a positive constant independent of a. Assume that (Proof) holds true for the moment. Combining this, (Proof) and (Proof), we conclude that

This is the desired conclusion (4.3).

Thus, the rest of the whole proof is to show the assertion (Proof). Indeed, for any (\vec{p},2,s)-atom a supported in a dilated ball x_0+B_{i_0} with some x_0\in{\mathbb{R}^n} and i_0\in{\mathbb Z}, it is easy to see that

where B^*_0 is the unit dilated ball with respect to A^*.

Let \theta be a fixed positive constant such that

Then, to deal with {\rm I}_1, by (Proof), we know that

As for the estimate of {\rm I}_2, by the Hölder inequality, the Plancherel theorem, the fact that 0<p_-\le p_+\le1 and the size condition of a, we obtain

This finishes the proof of (Proof) and hence of Theorem 4.3.