Hostname: page-component-745bb68f8f-s22k5 Total loading time: 0 Render date: 2025-02-06T19:59:18.886Z Has data issue: false hasContentIssue false

Trudinger’s inequalities for Riesz potentials in Morrey spaces of double phase functionals on half spaces

Published online by Cambridge University Press:  27 December 2021

Yoshihiro Mizuta
Affiliation:
Department of Mathematics, Graduate School of Advanced Science and Engineering, Hiroshima University, Higashi-Hiroshima 739-8521, Japan e-mail: yomizuta@hiroshima-u.ac.jp
Tetsu Shimomura*
Affiliation:
Department of Mathematics, Graduate School of Humanities and Social Sciences, Hiroshima University, Higashi-Hiroshima 739-8524, Japan
Rights & Permissions [Opens in a new window]

Abstract

Our aim in this paper is to establish Trudinger’s exponential integrability for Riesz potentials in weighted Morrey spaces on the half space. As an application, we obtain Trudinger’s inequality for Riesz potentials in the framework of double phase functionals.

Type
Article
Copyright
© Canadian Mathematical Society, 2021

1 Introduction

Let ${\mathbb R}^n$ be the n-dimensional Euclidean space, and $B(x,r)$ denote the open ball in ${\mathbb R}^n$ centered at x of radius $r>0$ . We consider the Riesz potential of order $\alpha $ on the half space ${\mathbb H} = \{ x=(x',x_n) \in {\mathbb R}^{n-1} \times {\mathbb R}^1: x_n> 0\}$ defined by

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) =\int_{B(x,x_n)} |x-y|^{\alpha-n} f(y) dy \end{align*} $$

for $0<\alpha <n$ and $f\in L^1_{\textrm {loc}}({\mathbb H})$ . For $f\in L^p_{\textrm {loc}}({\mathbb H})$ with $1<p<\infty $ , Trudinger type inequalities for Riesz potentials of order $\alpha $ have been studied in the limiting case $\alpha p = n$ (see e.g., [Reference Edmunds, Gurka and Opic8Reference Edmunds and Krbec11, Reference Mizuta, Nakai, Ohno and Shimomura17, Reference Mizuta, Nakai, Ohno and Shimomura18, Reference Serrin28]).

Our first aim in this paper is to establish Trudinger’s exponential integrability for $I_{{\mathbb H},\alpha } f$ of functions f satisfying the weighted $L^p$ condition

(1.1) $$ \begin{align} \int_{\mathbb H} |f(y) y_n^\beta|^{p} dy \le 1, \end{align} $$

when $\alpha p=n$ and $\beta < (n+1)/(2p')$ , where $1/p+1/p'=1$ (see Theorem 3.1). Note that $\omega (y) = |y_n|^{\beta p}$ is not always Muckenhoupt $A_p$ weight; more precisely, $\omega $ is not Muckenhoupt $A_p$ weight when $\beta \notin (-1/p, 1/p')$ (see Remarks 2.2 and 3.3). For this purpose, we apply the technique by Hedberg in [Reference Adams and Hedberg1] using the central Hardy–Littlewood maximal function $M_{{\mathbb H}} f$ defined by

$$ \begin{align*} M_{{\mathbb H}} f(x) = \sup_{\{r> 0: B(x,r)\subset {\mathbb H}\}} \frac{1}{|B(x,r)|} \int_{B(x,r)} |f(y)| \, dy, \end{align*} $$

where $|B(x,r)|$ denotes the Lebesgue measure of $B(x,r)$ . We show the boundedness of the maximal operator $M_{{\mathbb H}}$ (Theorem 2.5), as an improvement of [Reference Mizuta and Shimomura23, Theorem  2.1]. We also give a Sobolev type inequality for $I_{{\mathbb H},\alpha } f$ of functions f satisfying (1.1) when $\alpha p<n$ and $\beta < (n+1)/(2p')$ (Theorem 3.2). Compare Theorem 3.2 with [Reference Mizuta and Shimomura23, Theorem 2.2] which is a Sobolev type inequality for the fractional maximal function.

In the previous paper [Reference Mizuta and Shimomura24, Theorem 3.4], we proved a Sobolev type inequality for $I_{{\mathbb H},,\alpha } f$ of functions f satisfying the weighted Morrey condition

(1.2) $$ \begin{align} \sup_{r> 0, x\in {\mathbb H}} \frac{r^\sigma}{|B(x,r)|}\int_{{\mathbb H}\cap B(x,r)} \left( |f(y)| y_n^\beta \right)^{p} dy \le 1, \end{align} $$

when $\alpha p<\sigma < (n+1)/2$ and $\beta < (n+1)/(2p')$ . We refer to [Reference Morrey25] and [Reference Peetre26] for Morrey spaces, which were introduced to estimate solutions of partial differential equations. See also [Reference Chiarenza and Frasca5Reference Di Fazio and Ragusa12]. Applying our discussions in Theorem 3.1, we study Trudinger’s exponential integrability for $I_{{\mathbb H},\alpha } f$ of functions f satisfying (1.2) when $\alpha p=\sigma \le n$ and $\beta < (n+1)/(2p')$ (see Theorem 4.1), as an improvement of [Reference Mizuta and Shimomura24, Theorem 3.4].

Further, as an application, we establish Trudinger’s inequality for $I_{{\mathbb H},\alpha } f$ in the framework of double phase functionals

(1.3) $$ \begin{align} {\Phi}(x,t) = t^p + (b(x) t)^q, \end{align} $$

where $1 < p<q$ and $b(\cdot )$ is non-negative, bounded and Hölder continuous of order $\theta \in (0,1]$ (see Theorems 5.1 and 5.2). Double phase functionals are studied by Baroni, Colombo, and Mingione [Reference Baroni, Colombo and Mingione2Reference Baroni, Colombo and Mingione3Reference Colombo and Mingione6Reference Colombo and Mingione7] regarding the regularity theory of differential equations. See [Reference Mizuta and Shimomura24, Theorem 4.1] for Sobolev’s inequality of $I_{{\mathbb H},\alpha } f$ in the framework of (1.3). We refer to [Reference Maeda, Mizuta, Ohno and Shimomura16Reference Mizuta, Ohno and Shimomura20Reference Mizuta, Ohno and Shimomura21] for related results. Other double phase problems were studied e.g., in [Reference Byun and Lee4, Reference De Filippis and Mingione13Reference Hästö and Ok15, Reference Mizuta, Nakai, Ohno and Shimomura19, Reference Mizuta and Shimomura22, Reference Ragusa and Tachikawa27].

Throughout this paper, let C denote various constants independent of the variables in question. The symbol $g \sim h$ means that $C^{-1}h\le g\le Ch$ for some constant $C>0$ .

2 Boundedness of the maximal operator in the half space

For later use, it is convenient to see the following result.

Lemma 2.1 [Reference Mizuta and Shimomura23, Lemma 2.3]

For $\varepsilon> (n-1)/2$ and $x\in {\mathbb H}$ , set

$$ \begin{align*} I(x) &= \int_{B(x,x_n)} y_n^{\varepsilon - n} dy. \end{align*} $$

Then there exists a constant $C>0$ such that

$$ \begin{align*} I(x) &\le C x_n^{\varepsilon}. \end{align*} $$

Remark 2.2 Let $\beta> (n+1)/(2p')$ . If $f(y) = |y_n|^{-a} $ , then:

  1. (1) $\displaystyle \int _{B(x,x_n)} |f(y) y_n^\beta |^p dy < \infty $ for $x\in {\mathbb H}$ when $(\beta - a)p + n> (n-1)/2$ and

  2. (2) $\displaystyle \int _{B(x,x_n)} f(y) dy = \infty $ for $x\in {\mathbb H}$ when $- a + n \le (n-1)/2$ .

If $ (n+1)/2 \le a < \beta + (n+1)/(2p)$ , then both (1) and (2) hold.

For $f\in L^1_{\textrm {loc}}({\mathbb H})$ , the central Hardy–Littlewood maximal function $M_{{\mathbb H}}f$ is defined  by

$$ \begin{align*} M_{{\mathbb H}}f(x)=\sup_{\{r>0: B(x,r)\subset {\mathbb H}\}} \frac{1}{|B(x,r)|} \int_{B(x,r)} |f(y)|\, dy. \end{align*} $$

The mapping $f \mapsto M_{{\mathbb H}} f$ is called the fractional central maximal operator.

The usual fractional maximal function $Mf$ is defined by

$$ \begin{align*} Mf(x)=\sup_{r>0} \frac{1}{|B(x,r)|} \int_{{\mathbb H}\cap B(x,r)} |f(y)|\, dy. \end{align*} $$

It is known that the maximal operator $f \to Mf$ is bounded in Morrey spaces as follows:

Lemma 2.3 [Reference Di Fazio and Ragusa12, Lemma 4]

Let $0 < \sigma \le n$ and $p> 1$ . Then there exists a constant $C> 0$ such that

$$ \begin{align*} \sup_{z\in {\mathbb R}^n, r> 0} \frac{r^\sigma}{|B(z,r)|} \int_{B(z,r)} \{M f(x)\}^p dx &\le C \sup_{z\in {\mathbb R}^n, r > 0} \frac{r^\sigma}{|B(z,r)|} \int_{B(z,r)} |f(y)|^p dy \end{align*} $$

for all measurable functions f on ${\mathbb R}^n$ .

Throughout this paper, let $1 < p < \infty $ and $1/p + 1/p' = 1$ . We extend Lemma 2.3 to  $M_{{\mathbb H}}$ . For this purpose, we prepare the following result.

Lemma 2.4 Let $ \beta < (n+1)/(2p')$ . Then there exists a constant $C> 0$ such that

$$ \begin{align*} M_{{\mathbb H}} f(x) &\le C x_n^{-\beta} \left( Mg(x) \right)^{1/p} \end{align*} $$

for all $x\in {\mathbb H}$ and measurable functions f on ${\mathbb H}$ , where $g (y) = (|f(y)| |y_n|^{\beta })^{p} \chi _{{\mathbb H}}(y)$ .

Proof Let f be a non-negative measurable function on ${\mathbb H}$ . For $0 < r < x_n/2$ ,

$$ \begin{align*} \int_{B(x,r)} y_n^{-\beta p'} dy &\le C x_n^{-\beta p'} r^n \end{align*} $$

and for $x_n/2 < r < x_n$ and $-\beta p' + n> (n-1)/2$

$$ \begin{align*} \int_{B(x,r)} y_n^{-\beta p'} dy &\le C x_n^{-\beta p' + n} \le C x_n^{-\beta p'} r^n \end{align*} $$

by Lemma 2.1. Hence, we have by Hölder’s inequality

$$ \begin{align*} \int_{B(x,r)} f(y) dy &\le \left(\int_{B(x,r)} y_n^{-\beta p'} dy \right)^{1/p'} \left(\int_{B(x,r)} (f(y) y_n^{\beta})^{p} dy \right)^{1/p} \\ &\le Cx_n^{-\beta} r^{n/p'} \left(\int_{B(x,r)} (f(y) y_n^{\beta})^{p} dy \right)^{1/p} , \end{align*} $$

so that

$$ \begin{align*} M_{{\mathbb H}} f(x) &\le C x_n^{-\beta} \sup_{0 < r < x_n} \left( \frac{1}{|B(x,r)|} \int_{B(x,r)} (f(y) y_n^{\beta})^{p} dy \right)^{1/p}, \end{align*} $$

as required.▪

By Lemmas 2.3 and 2.4, we obtain the following result, which is an improvement of [Reference Mizuta and Shimomura23, Theorem 2.1].

Theorem 2.5 Let $ \beta < (n+1)/(2p')$ and $0 < \sigma \le n$ . Then there exists a constant $C> 0$ such that

$$ \begin{align*} & \sup_{r>0,z\in {\mathbb H}} \frac{r^\sigma}{|B(z,r)|} \int_{{\mathbb H}\cap B(z,r)} | x_n^\beta M_{{\mathbb H}} f(x) |^{p} dx \\[3pt] & \quad \le C\sup_{r>0,z\in {\mathbb H}} \frac{r^\sigma}{|B(z,r)|}\int_{{\mathbb H}\cap B(z,r)} |f(y) y_n^\beta|^{p} dy \end{align*} $$

for all measurable functions f on ${\mathbb H}$ .

Proof Let f be a measurable function on ${\mathbb H}$ , and take q such that $1 < q < p$ and $\beta < (n+1)/(2q')$ . Lemma 2.4 with p replaced by q and Lemma 2.3 give

$$ \begin{align*} & \sup_{r>0,z\in {\mathbb H}} \frac{r^\sigma}{|B(z,r)|} \int_{{\mathbb H}\cap B(z,r)} | x_n^\beta M_{{\mathbb H}} f(x) |^{p} dx \\[3pt] & \quad \le C\sup_{r>0,z\in {\mathbb H}} \frac{r^\sigma}{|B(z,r)|} \int_{{\mathbb H}\cap B(z,r)} |Mg(x)|^{p/q} dx \\[3pt] & \quad \le C \sup_{r>0,z\in {\mathbb H}} \frac{r^\sigma}{|B(z,r)|} \int_{{\mathbb H}\cap B(z,r)} |g(y)|^{p/q} dy \\[3pt] & \quad = C\sup_{r>0,z\in {\mathbb H}} \frac{r^\sigma}{|B(z,r)|}\int_{{\mathbb H}\cap B(z,r)} |f(y) y_n^\beta|^{p} dy, \end{align*} $$

where $g (y) = ( |f(y)| |y_n|^{\beta })^{q} \chi _{{\mathbb H}}(y)$ .▪

3 Trudinger’s inequality for Riesz potentials in $L^p$

For $0<\alpha <n$ and $f\in L^1_{\textrm {loc}}({\mathbb H})$ , let us consider the Riesz potential of order $\alpha $ on ${\mathbb H}$ defined by

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) =\int_{B(x,x_n)} |x-y|^{\alpha-n} f(y) dy. \end{align*} $$

We are now ready to show Trudinger’s exponential integrability for Riesz potentials on ${\mathbb H}$ .

Theorem 3.1 Let $ \alpha p =n$ and $ \beta < (n+1)/(2p')$ . Then there exist constants $c_1>0, c_2>0$ such that

$$ \begin{align*} \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left( \left|x_n^{\beta} I_{{\mathbb H},\alpha} f(x)/c_1 \right|{}^{p'} \right) dx &\le c_2 \end{align*} $$

for all $R> 0$ and measurable functions f satisfying (1.1).

Proof Let $\alpha p = n$ and f be a non-negative measurable function on ${\mathbb H}$ satisfying (1.1). Write

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &=\int_{B(x,r)} |x-y|^{\alpha-n} f(y) dy + \int_{B(x,x_n)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) dy \\[3pt] &= T_1(x) + T_2(x). \end{align*} $$

First note that

$$ \begin{align*} T_1(x) &\le C r^\alpha M_{\mathbb H} f(x). \end{align*} $$

Next, we have by Hölder’s inequality for $0 < r < x_n/2$

$$ \begin{align*} T_{21}(x) &= \int_{B(x,x_n/2)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) dy \\[3pt] &\le C x_n^{-\beta} \int_{B(x,x_n/2)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) y_n^\beta dy \\[3pt] &\le C x_n^{-\beta} \left( \int_{B(x,x_n/2)\setminus B(x,r)} |x-y|^{(\alpha-n)p'} dy \right)^{1/p'} \\[3pt] & \quad \times \left( \int_{B(x,x_n/2)\setminus B(x,r)} \{f(y) y_n^\beta\}^p dy \right)^{1/p} \\[3pt] &\le C x_n^{-\beta} \left( \log (x_n/r) \right)^{1/p'}, \end{align*} $$

since $\alpha p = n$ . Moreover, we have by Hölder’s inequality and Lemma 2.1 for $x_n/2 \le r < x_n$

$$ \begin{align*} T_{22}(x) &= \int_{B(x,x_n)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) dy \\[3pt] &\le C x_n^{\alpha-n} \int_{B(x,x_n)\setminus B(x,r)} f(y) dy \\[3pt] &\le C x_n^{\alpha-n} \left( \int_{B(x,x_n)\setminus B(x,r)} y_n^{-\beta p'} dy \right)^{1/p'} \left( \int_{B(x,x_n)\setminus B(x,r)} \{ f(y) y_n^\beta\} ^p dy \right)^{1/p} \\[3pt] &\le C x_n^{\alpha-n} \left( x_n^{-\beta p' + n} \right)^{1/p'} \left( \int_{B(x,x_n)\setminus B(x,r)} \{ f(y) y_n^\beta\} ^p dy \right)^{1/p} \\[3pt] &\le C x_n^{-\beta}, \end{align*} $$

since $-\beta p' + n> (n-1)/2$ and $\alpha p = n$ . Therefore,

$$ \begin{align*} T_2(x) &\le C x_n^{-\beta} \left\{ \log (e+(x_n/r)) \right\}^{1/p'} , \end{align*} $$

so that

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &\le C r^\alpha M_{\mathbb H} f(x) + C x_n^{-\beta} \left\{ \log (e+(x_n/r)) \right\}^{1/p'} \end{align*} $$

for every $r> 0$ . Letting $r = \{x_n^{\beta } M_{\mathbb H} f(x)\}^{-1/\alpha }$ , we obtain

$$ \begin{align*} x_n^{\beta} I_{{\mathbb H},\alpha} f(x) &\le C+ C \left(\log (e+x_n \{x_n^{\beta} M_{\mathbb H} f(x)\}^{1/\alpha}) \right)^{1/p'}. \end{align*} $$

Hence, there exists a constant $c_1>0$ such that

$$ \begin{align*} x_n^{\beta} I_{{\mathbb H},\alpha} f(x) &\le c_1 \left( \log (e+x_n^{\alpha p} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p ) \right)^{1/p'} , \end{align*} $$

so that

$$ \begin{align*} \exp \left[ \left\{x_n^{\beta} I_{{\mathbb H},\alpha} f(x)/c_1 \right\}^{p'} \right] &\le e + x_n^{n} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p, \end{align*} $$

since $\alpha p = n$ . Now it follows from Theorem 2.5 that

$$ \begin{align*} & \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left[ \left\{x_n^{\beta} I_{{\mathbb H},\alpha} f(x)/c_1 \right\}^{p'} \right] dx \\ &\le e + \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} x_n^{n} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p dx \\ &\le e + C \int_{{\mathbb H}\cap B(0,R)} \{x_n^{\beta} M_{\mathbb H} f(x)\}^{p} dx \\ &\le c_2 \end{align*} $$

for $R> 0$ .▪

In the same manner as the previous proof, we obtain Sobolev’s inequality in weighted $L^p$ spaces.

Theorem 3.2 (cf. [Reference Mizuta and Shimomura23, Theorem 2.2])

Let $1/p^* = 1/p - \alpha /n> 0$ and $ \beta < (n+1)/ (2p')$ . Then there exists a constant $C>0$ such that

$$ \begin{align*} \int_{{\mathbb H}} \left| x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right|{}^{p^*} dx \le C \end{align*} $$

for all measurable functions f satisfying (1.1).

In fact, as in the proof of Theorem 3.1, we have by Hölder’s inequality

$$ \begin{align*} T_{21}(x) \le C x_n^{-\beta} r^{\alpha-n/p} \end{align*} $$

for $0 < r < x_n/2$ , and

$$ \begin{align*} T_{22}(x) \le C x_n^{-\beta} x_n^{\alpha - n/p} \end{align*} $$

for $x_n/2 \le r < x_n$ . Hence,

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &\le C r^\alpha M_{\mathbb H} f(x) + C x_n^{-\beta} r^{\alpha - n/p} \end{align*} $$

for every $r> 0$ . Letting $r = \{x_n^{\beta } M_{\mathbb H} f(x)\}^{-p/n}$ , we obtain

$$ \begin{align*} x_n^{\beta} I_{{\mathbb H},\alpha} f(x) &\le C\{x_n^{\beta} M_{\mathbb H} f(x)\}^{1-\alpha p/n} \\ &= C \{x_n^{\beta} M_{\mathbb H} f(x)\}^{p/p^*}. \end{align*} $$

Now it follows from Theorem 2.5 that

$$ \begin{align*} \int_{{\mathbb H}} \left\{ x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right\}^{p^*} dx &\le C \int_{{\mathbb H}} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p dx \\[2pt] &\le C \int_{{\mathbb H}} |y_n^{\beta} f(y)|^p dy. \end{align*} $$

Remark 3.3 Let $\beta + \alpha - b + n/q \le a < \beta -b+ n/p$ and $(n-1)/q < (n-1)/p < b$ . If $f(y) = |y_n|^{-a} |y|^{-b} \chi _{B(0,1)}(y)$ , where $\chi _E$ denotes the characteristic function of E, then:

  1. (1) $\displaystyle \int _{{\mathbb H}\cap B(0,1)} |f(y) y_n^\beta |^p dy < \infty $ when $-bp + (n-1) < 0$ and $(-a+\beta )p + (-b p + n-1) + 1> 0$ ;

  2. (2) $\displaystyle I_\alpha f(x) = \int _{\mathbb H} |x-y|^{\alpha -n} f(y) dy = \infty $ for all $x\in {\mathbb H}$ when $a \ge 1$ ;

  3. (3) $\displaystyle I_{{\mathbb H},\alpha } f(x) \ge C x_n^{\alpha -a} |x|^{-b} $ for all $x\in {\mathbb H}\cap B(0,1)$ ;

  4. (4) $\displaystyle \int _{{\mathbb H}\cap B(0,1)} \{x_n^\beta I_{{\mathbb H},\alpha } f(x) \}^q dx = \infty $ when $-bq + (n-1) < 0$ and $(\beta -a + \alpha ) q + (-b q+ n-1) + 1 \le 0$ .In particular, it happens that

    $$ \begin{align*} \int_{{\mathbb H}\cap B(0,1)} \{x_n^\beta I_{{\mathbb H},\alpha} f(x) \}^q dx = \infty, \end{align*} $$
    when $q> p^*$ .

For (3), it suffices to see that

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &\ge \int_{B(x,x_n/2)} |x-y|^{\alpha-n} f(y) dy \\[2pt] &\ge C x_n^{-a}|x|^{-b} \int_{B(x,x_n/2)\cap B(0,1)} |x-y|^{\alpha-n} dy \\[2pt] &\ge C x_n^{-a+\alpha} |x|^{-b}. \end{align*} $$

4 Trudinger’s inequality for Riesz potentials in Morrey spaces

In this section, we are concerned with Trudinger’s exponential integrability in weighted Morrey spaces.

Theorem 4.1 Let $\alpha p = \sigma \le n$ and $ \beta < (n+1)/(2p')$ . Then there exist constants $c_1>0, c_2>0$ such that

$$ \begin{align*} \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left(|x_n^{\beta} I_{{\mathbb H},\alpha} f(x)/c_1| \right) dx &\le c_2 \end{align*} $$

for all $R> 0$ and measurable functions f on ${\mathbb H}$ satisfying (1.2).

Proof Let f be a non-negative measurable function on ${\mathbb H}$ satisfying (1.2). Write

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &= \int_{B(x,r)} |x-y|^{\alpha-n} f(y) dy + \int_{B(x,x_n)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) dy \\[2pt] &= T_1(x) + T_2(x). \end{align*} $$

By (1.2), we have for $0 < r < x_n/2$

$$ \begin{align*} T_{21}(x) &= \int_{B(x,x_n/2)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) dy \\[2pt] &\le C x_n^{-\beta} \int_{B(x,x_n/2)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) y_n^\beta dy \\[2pt] &\le C x_n^{-\beta} \int_r^{x_n/2} t^{\alpha-1} \left( \frac{1}{|B(x,t)|} \int_{B(x,t)} f(y) y_n^\beta dy \right) dt \\[2pt] &\le C x_n^{-\beta} \int_r^{x_n/2} t^{\alpha-1} \left( \frac{1}{|B(x,t)|} \int_{B(x,t)} \{f(y) y_n^\beta\}^p dy \right)^{1/p} dt \\[2pt] &\le C x_n^{-\beta} \int_r^{x_n/2} t^{-1} dt \\[2pt] &\le C x_n^{-\beta} \log (x_n/r), \end{align*} $$

since $\alpha p = \sigma $ . Moreover, as in the proof of Theorem 3.1, by Hölder’s inequality and Lemma 2.1, we have for $x_n/2 \le r < x_n$

$$ \begin{align*} T_{22}(x) &= \int_{B(x,x_n)\setminus B(x,r)} |x-y|^{\alpha-n} f(y) dy \\[2pt] &\le C x_n^{\alpha-n} \int_{B(x,x_n)\setminus B(x,r)} f(y) dy \\[2pt] &\le C x_n^{\alpha-n} \left( x_n^{-\beta p' + n} \right)^{1/p'} \left( \int_{B(x,x_n)\setminus B(x,r)} \{ f(y) y_n^\beta\} ^p dy \right)^{1/p} \\[2pt] &\le C x_n^{\alpha-n} \left( x_n^{-\beta p' + n} \right)^{1/p'} (x_n^{n-\sigma})^{1/p} \\[2pt] &= C x_n^{-\beta}, \end{align*} $$

since $-\beta p' + n> (n-1)/2$ and $\alpha p = \sigma $ . Therefore,

$$ \begin{align*} T_2(x) &\le C x_n^{-\beta} \log (x_n/r) , \end{align*} $$

so that

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &\le C r^\alpha M_{\mathbb H} f(x) + C x_n^{-\beta} \log (e +(x_n/r)) \end{align*} $$

for every $r> 0$ . Letting $r = \{x_n^{\beta } M_{\mathbb H} f(x)\}^{-1/\alpha }$ , we obtain

$$ \begin{align*} x_n^{\beta} I_{{\mathbb H},\alpha} f(x) &\le C+ C \log (e+x_n \{x_n^{\beta} M_{\mathbb H} f(x)\}^{1/\alpha}). \end{align*} $$

Hence, there exists a constant $c_1>0$ such that

$$ \begin{align*} x_n^{\beta} I_{{\mathbb H},\alpha} f(x) &\le c_1\log (e+x_n^{\alpha p} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p ), \end{align*} $$

so that

$$ \begin{align*} \exp \left(x_n^{\beta} I_{{\mathbb H},\alpha} f(x)/c_1 \right) &\le e + x_n^{\alpha p} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p. \end{align*} $$

Hence, in view of Theorem 2.5, we obtain

$$ \begin{align*} & \frac{1}{|B(0,R)|} \int_{{\mathbb H}\cap B(0,R)} \exp \left(x_n^{\beta} I_{{\mathbb H},\alpha} f(x)/c_1 \right) dx \\[2pt] &\le e + C \frac{1}{|B(0,R)|} \int_{{\mathbb H}\cap B(0,R)} x_n^{\sigma} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p dx \\[2pt] &\le e + C \frac{R^{\sigma}}{|B(0,R)|} \int_{{\mathbb H}\cap B(0,R)} \{x_n^{\beta} M_{\mathbb H} f(x)\}^{p} dx \\[2pt] &\le c_2 \end{align*} $$

for $R> 0$ .▪

In the same manner as the previous proof, we obtain Sobolev’s inequality in weighted Morrey spaces, which is an improvement of [Reference Mizuta and Shimomura24, Theorem 3.4].

Theorem 4.2 Let $1/p_{\sigma } = 1/p - \alpha /\sigma> 0$ , $0 < \sigma \le n$ and $ \beta < (n+1)/(2p')$ . Then there exists a constant $C>0$ such that

$$ \begin{align*} \frac{r^\sigma}{|B(z,r)|}\int_{{\mathbb H}\cap B(z,r)} \left|x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right|{}^{p_{\sigma}} dx \le C \end{align*} $$

for all $z\in {\mathbb H}$ , $r> 0$ and measurable functions f on ${\mathbb H}$ satisfying (1.2).

In fact, as in the proof of Theorem 4.1, we have by Hölder’s inequality

$$ \begin{align*} T_{21}(x) \le C x_n^{-\beta} r^{\alpha-\sigma/p} \end{align*} $$

for $0 < r < x_n/2$ , and

$$ \begin{align*} T_{22}(x) \le C x_n^{-\beta} x_n^{\alpha - \sigma/p} \end{align*} $$

for $x_n/2 \le r < x_n$ . Hence,

$$ \begin{align*} I_{{\mathbb H},\alpha} f(x) &\le C r^\alpha M_{\mathbb H} f(x) + C x_n^{-\beta} r^{\alpha - \sigma/p} \end{align*} $$

for every $r> 0$ . Letting $r = \{x_n^{\beta } M_{\mathbb H} f(x)\}^{-p/\sigma }$ , we obtain

$$ \begin{align*} x_n^{\beta} I_{{\mathbb H},\alpha} f(x) &\le C\{x_n^{\beta} M_{\mathbb H} f(x)\}^{1-\alpha p/\sigma} \\[2pt] &= C \{x_n^{\beta} M_{\mathbb H} f(x)\}^{p/p_{\sigma}}. \end{align*} $$

Now it follows from Theorem 2.5 that

$$ \begin{align*} \frac{r^\sigma}{|B(z,r)|}\int_{{\mathbb H}\cap B(z,r)} \left\{ x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right\}^{p_{\sigma}} dx &\le C \frac{r^\sigma}{|B(z,r)|}\int_{{\mathbb H}\cap B(z,r)} \{x_n^{\beta} M_{\mathbb H} f(x)\}^p dx \\[2pt] &\le C \frac{r^\sigma}{|B(z,r)|}\int_{{\mathbb H}\cap B(z,r)} |y_n^{\beta} f(y)|^p dy \end{align*} $$

for all $z\in {\mathbb H}$ and $r> 0$ .

5 Double phase functionals

In this section, we consider the double phase functional

$$ \begin{align*} \Phi(x,t) = t^{p} + (b(x) t)^{q}, \end{align*} $$

where $1 < p<q$ and $b(\cdot )$ is non-negative, bounded and Hölder continuous of order $\theta \in (0,1]$ [Reference Maeda, Mizuta, Ohno and Shimomura16].

We obtain Trudinger’s inequality for $I_{{\mathbb H},\alpha } f$ in weighted Morrey spaces of the double phase functional $\Phi (x,t)$ using Theorem 4.1.

Theorem 5.1 Let $0 < \sigma \le n$ , $1/q = 1/p - \theta /\sigma $ , $1/p_\sigma = 1/p - \alpha /\sigma> 0$ and $1/q_\sigma = 1/q - \alpha /\sigma = 0$ . Suppose $ \beta < (n+1)/(2p')$ . Then there exist constants $c_1>0, c_2>0$ such that

$$ \begin{align*} & \frac{R^\sigma}{|B(0,R)|} \int_{{\mathbb H}\cap B(0,R)} \left|x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right|{}^{p_\sigma} dx \\[2pt] & \quad + \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left( |x_n^\beta b(x) I_{{\mathbb H},\alpha} f(x)/c_1| \right) dx \le c_2 \end{align*} $$

for all $R>0$ and measurable functions f on ${\mathbb H}$ satisfying

(5.1) $$ \begin{align} \sup_{x\in {\mathbb H},r>0} \frac{r^\sigma}{|B(x,r)|}\int_{{\mathbb H}\cap B(x,r)} \Phi\left(y, |f(y)| y_n^\beta \right) dy \le 1. \end{align} $$

Proof Let f be a non-negative measurable function on ${\mathbb H}$ satisfying (5.1). First, we see from Theorem 4.2 that

$$\begin{align*}\sup_{r>0: x\in {\mathbb H}} \frac{r^\sigma}{|B(x,r)|}\int_{{\mathbb H}\cap B(x,r)} (z_n^\beta I_{{\mathbb H},\alpha} f(z))^{p_\sigma} dz \le C, \end{align*}$$

since $\alpha p<\sigma $ .

Note that

$$ \begin{align*} & b(x) I_{{\mathbb H},\alpha} f(x) \\[2pt] & \quad = \int_{B(x,x_n)} \{b(x)-b(y)\} |x-y|^{\alpha-n} f(y) dy + \int_{B(x,x_n)} b(y)|x-y|^{\alpha-n}f(y) dy \\[2pt] & \quad \le C \int_{B(x,x_n)} |x-y|^{\alpha+\theta-n} f(y) dy + \int_{B(x,x_n)} |x-y|^{\alpha-n} b(y)f(y) dy \\[2pt] & \quad = C I_{{\mathbb H},\alpha+\theta} f(x)+I_{{\mathbb H},\alpha} [bf](x), \end{align*} $$

when $x\in {\mathbb H}$ . We find by Theorem 4.1

$$ \begin{align*} \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left(|x_n^{\beta} I_{{\mathbb H},\alpha+\theta} f(x)/c_1| \right) dx &\le c_2 \end{align*} $$

and

$$ \begin{align*} \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left(|x_n^{\beta} I_{{\mathbb H},\alpha} [bf](x)/c_1| \right) dx &\le c_2 \end{align*} $$

for all $R>0$ since $1/q_\sigma = 1/q - \alpha /\sigma = 1/p - (\alpha +\theta )/\sigma = 0$ . Thus, we complete the proof.▪

We obtain the following theorem using Theorem 3.1.

Theorem 5.2 Let $1/q = 1/p - \theta /n$ , $ \alpha p <n=\alpha q$ and $ \beta < (n+1)/(2p')$ . Then there exist constants $c_1>0, c_2>0$ such that

$$ \begin{align*} & \int_{{\mathbb H}} \left|x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right|{}^{p^*} dx \\ & {} + \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left( \left|x_n^{\beta} b(x) I_{{\mathbb H},\alpha} f(x)/c_1 \right|{}^{q'} \right) dx \le c_2 \end{align*} $$

for all $R>0$ and measurable functions f satisfying

(5.2) $$ \begin{align} \int_{{\mathbb H}} \Phi\left(y, |f(y)| y_n^\beta \right) dy \le 1. \end{align} $$

Proof Let f be a non-negative measurable function on ${\mathbb H}$ satisfying (5.2). Then Theorem 3.2 gives

$$ \begin{align*} \int_{{\mathbb H}} \left|x_n^{\beta} I_{{\mathbb H},\alpha} f(x) \right|{}^{p^*} dx \le C, \end{align*} $$

since $\alpha p<n$ . Further, Theorem 3.1 gives

$$ \begin{align*} \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left( \left|x_n^{\beta} I_{{\mathbb H},\alpha+\theta} f(x)/c_1 \right|{}^{p'} \right) dx &\le c_2 \end{align*} $$

and

$$ \begin{align*} \frac{1}{|B(0,R)|}\int_{{\mathbb H}\cap B(0,R)} \exp \left( \left|x_n^{\beta} I_{{\mathbb H},\alpha} [bf](x)/c_1 \right|{}^{q'} \right) dx &\le c_2 \end{align*} $$

for all $R>0$ since $ (\alpha +\theta ) p=n=\alpha q$ and $1/p' < 1/q'$ . Thus, we obtain the result.▪

References

Adams, D. R. and Hedberg, L. I., Function spaces and potential theory, Springer-Verlag, Berlin, Heidelberg, 1996.CrossRefGoogle Scholar
Baroni, P., Colombo, M., and Mingione, G., Regularity for general functionals with double phase. Calc. Var. Partial Differ. Equat. 57(2018), no. 2, Paper No. 62.CrossRefGoogle Scholar
Baroni, P., Colombo, M., and Mingione, G., Non-autonomous functionals, borderline cases and related function classes. St Petersburg Math. J. 27(2016), 347379.CrossRefGoogle Scholar
Byun, S. S. and Lee, H. S., Calderón-Zygmund estimates for elliptic double phase problems with variable exponents. J. Math. Anal. Appl. 501(2021), 124015.CrossRefGoogle Scholar
Chiarenza, F. and Frasca, M., Morrey spaces and Hardy-Littlewood maximal function. Rend. Mat. 7(1987), 273279.Google Scholar
Colombo, M. and Mingione, G., Regularity for double phase variational problems. Arch. Rat. Mech. Anal. 215(2015), 443496.CrossRefGoogle Scholar
Colombo, M. and Mingione, G., Bounded minimizers of double phase variational integrals. Arch. Rat. Mech. Anal. 218(2015), 219273.CrossRefGoogle Scholar
Edmunds, D. E., Gurka, P., and Opic, B., Double exponential integrability, Bessel potentials and embedding theorems, Studia Math. 115(1995), 151181.Google Scholar
Edmunds, D. E., Gurka, P., and Opic, B., Sharpness of embeddings in logarithmic Bessel-potential spaces. Proc. Royal Soc. Edinburgh. 126(1996), 9951009.CrossRefGoogle Scholar
Edmunds, D. E. and Hurri-Syrjänen, R., Sobolev inequalities of exponential type. Israel. J. Math. 123(2001), 6192.CrossRefGoogle Scholar
Edmunds, D. E. and Krbec, M., Two limiting cases of Sobolev imbeddings. Houston J. Math. 21(1995), 119128.Google Scholar
Di Fazio, G. and Ragusa, M. A., Commutators and Morrey spaces, Boll. Un. Mat. Ital. 7(1991), no. 5-A, 323332.Google Scholar
De Filippis, C. and Mingione, G., On the regularity of minima of non-autonomous functionals. J. Geom. Anal. 30(2020), no. 2, 15841626.CrossRefGoogle Scholar
De Filippis, C. and Oh, J., Regularity for multi-phase variational problems. J. Differ. Equat. 267(2019), no. 3, 16311670.CrossRefGoogle Scholar
Hästö, P. and Ok, J., Calderón-Zygmund estimates in generalized Orlicz spaces. J. Differ. Equat. 267(2019), no. 5, 27922823.CrossRefGoogle Scholar
Maeda, F. -Y., Mizuta, Y., Ohno, T., and Shimomura, T., Sobolev’s inequality inequality for double phase functionals with variable exponents. Forum Math. 31(2019), 517527.CrossRefGoogle Scholar
Mizuta, Y., Nakai, E., Ohno, T., and Shimomura, T., An elementary proof of Sobolev embeddings for Riesz potentials of functions in Morrey spaces ${L}^{1,\nu, \beta }(G)$ . Hiroshima Math. J. 38(2008), 425436.CrossRefGoogle Scholar
Mizuta, Y., Nakai, E., Ohno, T., and Shimomura, T., Boundedness of fractional integral operators on Morrey spaces and Sobolev embeddings for generalized Riesz potentials. J. Math. Soc. Japan 62(2010), 707744.CrossRefGoogle Scholar
Mizuta, Y., Nakai, E., Ohno, T., and Shimomura, T., Campanato-Morrey spaces for the double phase functionals with variable exponents. Nonlinear Anal. 197(2020), 111827.CrossRefGoogle Scholar
Mizuta, Y., Ohno, T., and Shimomura, T., Sobolev’s inequalities for Herz-Morrey-Orlicz spaces on the half space. Math. Ineq. Appl. 21(2018), 433453.Google Scholar
Mizuta, Y., Ohno, T., and Shimomura, T., Boundedness of fractional maximal operators for double phase functionals with variable exponents. J. Math. Anal. Appl. 501(2021), 124360.Google Scholar
Mizuta, Y. and Shimomura, T., Hardy-Sobolev inequalities in the unit ball for double phase functionals. J. Math. Anal. Appl. 501(2021), 124133.CrossRefGoogle Scholar
Mizuta, Y. and Shimomura, T., Sobolev type inequalities for fractional maximal functions and Green potentials in half spaces. Positivity 25(2021), 11311146.CrossRefGoogle Scholar
Mizuta, Y. and Shimomura, T., Sobolev type inequalities for fractional maximal functions and Riesz potentials in half spaces. Studia Math., to appear.Google Scholar
Morrey, C. B., On the solutions of quasi-linear elliptic partial differential equations. Trans. Amer. Math. Soc. 43(1938), 126166.CrossRefGoogle Scholar
Peetre, J., On the theory of ${\mathbf{\mathcal{L}}}_{p,\lambda }$ spaces. J. Funct. Anal. 4(1969), 7187.CrossRefGoogle Scholar
Ragusa, M. A. and Tachikawa, A., Regularity for minimizers for functionals of double phase with variable exponents. Adv. Nonlinear Anal. 9(2020), no. 1, 710728.CrossRefGoogle Scholar
Serrin, J., A remark on Morrey potential. Contemp. Math. 426(2007), 307315.CrossRefGoogle Scholar