Fractional type operators on Hardy spaces associated with ball quasi-Banach function spaces Pablo Rocha Abstract For 0 ≤ α < n 0\leq\alpha1 , if and only in if for any ball B\subset\mathbb{R}^{n} , |B|^{\frac{\alpha}{n}}\lesssim\|\chi_{B}\|_{X}^{(\beta-1)/\beta} , where X is a ball quasi-Banach function space and X^{\beta} denotes the \beta -convexification of X . Moreover, using extrapolation techniques, the authors also proved the H_{X}(\mathbb{R}^{n})\to H_{Y}(\mathbb{R}^{n}) boundedness of I_{\alpha} , for certain ball quasi-Banach function spaces X and Y . Recently, the author in [ 22 ] , for 0\leq\alpha0 and a ball B=B(x_{0},r) , we set \gamma B=B(x_{0},\gamma r) . The orthogonal group \mathcal{O}(n) is defined by \mathcal{O}(n):=\{A\in GL_{n}(\mathbb{R}):A^{t}=A^{-1}\} . For a measurable subset E\subset\mathbb{R}^{n} we denote |E| and \chi_{E} the Lebesgue measure of E and the characteristic function of E respectively. Given a real number s\geq 0 , we write \lfloor s\rfloor for the integer part of s . As usual we denote with \mathcal{S}(\mathbb{R}^{n}) the space of smooth and rapidly decreasing functions, with \mathcal{S}^{\prime}(\mathbb{R}^{n}) the dual space. If \beta is the multiindex \beta=(\beta_{1},...,\beta_{n}) , then |\beta|=\beta_{1}+...+\beta_{n} . Given a function g on \mathbb{R}^{n} and t>0 , we write g_{t}(x)=t^{-n}g(t^{-1}x) . Let d be a non negative integer and \mathcal{P}_{d} the subspace of L^{1}_{loc}(\mathbb{R}^{n}) formed by all the polynomials of degree at most d . Given a measurable function h , the expression h\perp\mathcal{P}_{d} stands for \int h(x)P(x)dx=0 for all P\in\mathcal{P}_{d} . Throughout this paper, C will denote a positive constant, not necessarily the same at each occurrence. 2 Preliminaries 2.1 Ball quasi-Banach function spaces In the sequel, \mathfrak{M}=\mathfrak{M}(\mathbb{R}^{n}) is the set of all measurable functions on \mathbb{R}^{n} and \mathfrak{M}_{+}=\mathfrak{M}_{+}(\mathbb{R}^{n}) is the cone of all non-negative measurable functions on \mathbb{R}^{n} . For every x\in\mathbb{R}^{n} and r>0 fixed, let B(x,r):=\{y\in\mathbb{R}^{n}:|x-y|0\}. Definition 1 . A mapping \rho:\mathfrak{M}_{+}\to[0,\infty] is called a ball quasi-Banach function norm if it satisfy the following properties: (P1) \rho(f)=0 implies that f=0 a.e.; (P2) \rho(\alpha f)=|\alpha|\rho(f) for all \alpha\in\mathbb{C} and all f\in\mathfrak{M}_{+} ; (P3) there exists C\geq 1 such that \rho(f+g)\leq C(\rho(f)+\rho(g)) for all f,g\in\mathfrak{M}_{+} ; (P4) if some f,g\in\mathfrak{M}_{+} satisfy f\leq g a.e., then \rho(f)\leq\rho(g) ; (P5) if some f_{n},f\in\mathfrak{M}_{+} satisfy f_{n}\uparrow f a.e., then \rho(f_{n})\uparrow\rho(f) ; (P6) if B\in\mathbb{B} , then \rho(\chi_{B})<\infty . Definition 2 . Let \rho be a ball quasi-Banach function norm. We then define the corresponding ball quasi-Banach function space X=X(\rho) as the set X=\{f\in\mathfrak{M}:\rho(|f|)<\infty\}. For each f\in X , define \|f\|_{X}=\rho(|f|). Remark 3 . Let (X,\|\cdot\|_{X}) be a ball quasi-Banach function space, then it is easy to check that (i) \|f\|_{X}=0 implies that f=0 a.e.; (ii) \|\alpha f\|_{X}=|\alpha|\|f\|_{X} for all \alpha\in\mathbb{C} and all f\in X ; (iii) there exists C\geq 1 such that \|f+g\|_{X}\leq C(\|f\|_{X}+\|g\|_{X}) for all f,g\in X ; (iv) if f\in\mathfrak{M} , g\in X are such that |f|\leq|g| a.e., then f\in X and \|f\|_{X}\leq\|g\|_{X} ; (v) if 0\leq f_{n}\uparrow f a.e., then either f\notin X and \|f_{n}\|_{X}\uparrow\infty , or f\in X and \|f_{n}\|_{X}\uparrow\|f\|_{X} ; (vi) if B\in\mathbb{B} , then \chi_{B}\in X . A ball quasi-Banach function space X is called a ball Banach function space if the constant C , appearing in (iii) of Remark 3 , is equal to 1 , and for any ball B\in\mathbb{B} there exists a positive constant C_{(B)} , depending on B , such that \int_{B}|f(x)|dx\leq C_{(B)}\|f\|_{X}, for all f\in X . Thus, every Banach function space is a ball Banach function space (see [ 4 , Definition 1.3] ). An interesting work on the properties of (quasi)-Banach functions spaces can be found in [ 18 ] . Definition 4 . (See [ 4 ] ) For any ball Banach function space X , the associate space (also called the Köthe dual) X^{\prime} is defined by setting X^{\prime}:=\left\{f\in\mathfrak{M}:\|f\|_{X^{\prime}}:=\sup\{\|fg\|_{L^{1}(\mathbb{R}^{n})}:g\in X,\|g\|_{X}=1\}<\infty\right\} where \|\cdot\|_{X^{\prime}} is called the associate norm of \|\cdot\|_{X} . Lemma 5 . ( [ 26 , Proposition 2.3] and [ 34 , Lemma 2.6] ) Let X be a ball Banach function space. Then X^{\prime} is also a ball Banach function space and X coincides with its second associate space X^{\prime\prime} . In other words, a function f belongs to X if and only if it belongs to X^{\prime\prime} and, in that case, \|f\|_{X}=\|f\|_{X^{\prime\prime}}. Definition 6 . A function f\in X is said to have absolutely continuous quasi-norm in X if \|f\chi_{E_{j}}\|_{X}\downarrow 0 for every sequence \{E_{j}\}_{j=1}^{\infty} satisfying E_{j}\supset E_{j+1} for all j\in\mathbb{N} and \bigcap_{j=1}^{\infty}E_{j}=\emptyset . The set of all functions in X of absolutely continuous quasi-norm is denoted by X_{a} . If X=X_{a} , then the space X itself is said to have an absolutely continuous quasi-norm. Taking into account that the orthogonal group \mathcal{O}(n) induces an action on functions by f_{A}(x)=f(A^{-1}x) , where A\in\mathcal{O}(n) , we introduce the following definition. Definition 7 . We say that a ball quasi-Banach function space X is \mathcal{O}(n) -invariant if for every f\in X and A\in\mathcal{O}(n) , f_{A}\in X with \|f_{A}\|_{X}=\|f\|_{X} . Remark 8 . It is easy to check that if X is a \mathcal{O}(n) -invariant ball Banach function space, then X^{\prime} is \mathcal{O}(n) -invariant. Remark 9 . We point out that not all ball quasi-Banach function space is \mathcal{O}(n) -invariant. For instance, the mixed-norm spaces are not \mathcal{O}(n) -invariant (see [ 3 ] ). Definition 10 . Let X be a ball quasi-Banach function space and p\in(0,\infty) . The p -convexification X^{p} of X is defined by setting X^{p}:=\{f\in\mathfrak{M}:|f|^{p}\in X\} equipped with the quasi-norm \|f\|_{X^{p}}:=\||f|^{p}\|^{1/p}_{X} . Remark 11 . If X is a \mathcal{O}(n) -invariant ball quasi-Banach function space and p\in(0,\infty) , then X^{p} is \mathcal{O}(n) -invariant. Definition 12 . Let X be a ball quasi-Banach function space and p\in(0,\infty) . The space X is said to be p -convex if there exists a positive constant C such that, for any \{f_{j}\}_{j=1}^{\infty}\subset X^{1/p} , \left\|\sum_{j=1}^{\infty}|f_{j}|\right\|_{X^{1/p}}\leq C\sum_{j=1}^{\infty}\left\|f_{j}\right\|_{X^{1/p}}. In particular, when C=1 , X is said to be strictly p -convex. 2.2 Maximal functions and additional assumptions For 0\leq\alpha0,\phi\in\mathcal{F}_{N}\right\}. (5) Definition 15 . ( [ 26 , Theorem 3.1 - (ii) with b=n/r+1 ] ) Let X be a ball quasi-Banach function space such that the Hardy-Littlewood maximal operator M is bounded on X^{1/r} for some r\in(0,\infty) . For N\geq\lfloor n/r+3\rfloor , define the Hardy space H_{X}(\mathbb{R}^{n}) associated with X as H_{X}(\mathbb{R}^{n})=\left\{f\in\mathcal{S}^{\prime}(\mathbb{R}^{n}):\|\mathcal{M}^{0}_{N}f\|_{X}<\infty\right\}, (6) Fixed N\geq\lfloor n/r+3\rfloor , we consider \|f\|_{H_{X}(\mathbb{R}^{n})}:=\|\mathcal{M}^{0}_{N}f\|_{X} . Before establishing the finite Hardy space H^{X,q,s}_{fin}(\mathbb{R}^{n}) , we recall the definition of X -atom. Definition 16 . Let X be a ball quasi-Banach function space satisfying ( 3 ) for some 0<\theta0 such that M(w)(x)\leq Cw(x),\,\,\,\,\,a.e.\,x\in\mathbb{R}^{n}, the best possible constant is denoted by [w]_{\mathcal{A}_{1}} . Equivalently, a weight w\in\mathcal{A}_{1} if there exists C>0 such that for every ball B \frac{1}{|B|}\int_{B}w(x)dx\leq C\,ess\inf_{x\in B}w(x). (7) Remark 21 . If w\in\mathcal{A}_{1} and 00 such that for every ball B \left(\frac{1}{|B|}\int_{B}w(x)dx\right)\left(\frac{1}{|B|}\int_{B}[w(x)]^{-\frac{1}{p-1}}dx\right)^{p-1}\leq C. It is well known that \mathcal{A}_{p_{1}}\subset\mathcal{A}_{p_{2}} for all 1\leq p_{1}0 such that for every ball B \left(\frac{1}{|B|}\int_{B}[w(x)]^{q}dx\right)^{1/q}\left(\frac{1}{|B|}\int_{B}[w(x)]^{-p^{\prime}}dx\right)^{1/p^{\prime}}\leq C<\infty. For p=1 , we say that a weight w\in\mathcal{A}_{1,q} if there exists C>0 such that for every ball B \left(\frac{1}{|B|}\int_{B}[w(x)]^{q}dx\right)^{1/q}\leq C\,ess\inf_{x\in B}w(x). When p=q , this definition is equivalent to w^{p}\in\mathcal{A}_{p} . A weight w satisfies the reverse Hölder inequality with exponent s>1 , denoted by w\in RH_{s} , if there exists C>0 such that for every ball B , \left(\frac{1}{|B|}\int_{B}[w(x)]^{s}dx\right)^{\frac{1}{s}}\leq C\frac{1}{|B|}\int_{B}w(x)dx; the best possible constant is denoted by [w]_{RH_{s}} . Remark 24 . We observe that if w\in RH_{s} , then by Hölder’s inequality, w\in RH_{t} for all 10 such that \|f\|_{H^{p_{0}}(w)}\leq C\|f\|_{H^{X,p_{0},q,d}_{fin}(w)}, for all f\in H^{X,p_{0},q,d}_{fin}(w) . To obtain the opposite inequality, fix f\in H^{X,p_{0},q,d}_{fin}(w) , then f is supported on a ball B=B(0,K) for some K>1 . Let \widetilde{B}=B(0,4K) . We assume \|f\|_{H^{p_{0}}(w)}=1 , and observe that the space L^{p_{0}}(w) , with w\in\mathcal{A}_{1} , is a ball quasi-Banach function space such that the assumption ( 3 ) holds true for any \theta,s\in(0,1] , \theta0 such that \|f\|_{H^{X,p_{0},q,d}_{fin}(w)}\leq C , under the assumption \|f\|_{H^{p_{0}}(w)}=1 , and thus the proposition will be followed from the homogeneity of \|\cdot\|_{H^{X,p_{0},q,d}_{fin}(w)} . Since this estimate is similar to the proof of [ 8 , Lemma 7.11] , we will give the main steps of its proof adapting them to our context. We have, by the proof of [ 29 , Theorem 1, on p. 118] , that f=\sum_{j,k}^{\infty}\lambda_{j,k}\|\chi_{B_{j,k}}\|_{X}^{-1}\|\chi_{B_{j,k}}\|_{X}a_{j,k}=\sum_{j,k}^{\infty}\lambda_{j,k}a_{j,k}, where every a_{j,k} is an (X,\infty,d_{X}) -atom supported on the ball B_{j,k} (we recall that, by Remark 13 , d_{X} can be taken such that d_{X}\geq\lfloor n(p_{0}^{-1}-1)\rfloor , and so \widetilde{a}_{j,k}:=\|\chi_{B_{j,k}}\|_{X}\,a_{j,k} is an (\infty,d_{X}) -atom in the sense of [ 29 , p. 111] ), the \lambda_{j,k} ’s are non-negative scalars, and \left\|\sum_{j,k}^{\infty}\left(\frac{\lambda_{j,k}}{\|\chi_{B_{j,k}}\|_{X}}\right)^{p_{0}}\chi_{B_{j,k}}\right\|_{L^{1}(w)}^{1/p_{0}}\leq C\|f\|_{H^{p_{0}}(w)}=C. (9) Therefore, by taking into account ( 8 ), we can write (as in [ 8 , p. 484] ) f=h+\ell=k_{w}^{-1}(k_{w}h)+(\ell-\ell_{i})+\sum_{F_{i}}\lambda_{j,k}a_{j,k}, where k_{w} is a constant given by k_{w}=\frac{c^{-1}w(B)^{1/p_{0}}}{\|\chi_{\widetilde{B}}\|_{X}}, k_{w}h is an X -atom, i is chosen large enough so that (\ell-\ell_{i}) is an X -atom, and the last sum is a finite sum of X -atoms. Then, \|f\|_{H^{X,p_{0},q,d}_{fin}(w)}^{p_{0}}\leq k_{w}^{-p_{0}}\frac{w(\widetilde{B})}{\|\chi_{\widetilde{B}}\|_{X}^{p_{0}}}+\frac{w(\widetilde{B})}{\|\chi_{\widetilde{B}}\|_{X}^{p_{0}}}+\left\|\sum_{j,k}^{\infty}\left(\frac{\lambda_{j,k}}{\|\chi_{B_{j,k}}\|_{X}}\right)^{p_{0}}\chi_{B_{j,k}}\right\|_{L^{1}(w)}. By Remark 22 , the first term is bounded by a constant. By ( 9 ), the last term is bounded by the constant C^{p_{0}} . Finally, to bound the middle term, we apply [ 34 , Lemma 2.5] , with X^{1/p_{0}} , and obtain w(\widetilde{B})=\int_{\widetilde{B}}w(x)dx\leq C\|\chi_{\widetilde{B}}\|_{X^{1/p_{0}}}\|w\|_{(X^{1/p_{0}})^{\prime}}\leq C\|\chi_{\widetilde{B}}\|_{X}^{p_{0}}. Thus, \|f\|_{H^{X,p_{0},q,d}_{fin}(w)}\leq C, as we wished to prove. ∎ Lemma 29 . Let X be a ball quasi-Banach function space and let 0\frac{np_{0}}{\alpha} , then for \frac{1}{q_{0}}=\frac{1}{p_{0}}-\frac{\alpha}{n} \int_{\mathbb{R}^{n}}|T_{\alpha,m}a(x)|^{q_{0}}w(x)dx\leq C|B|^{\frac{\alpha}{n}q_{0}}\|\chi_{B}\|_{X}^{-q_{0}}\sum_{i=1}^{m}w_{A_{i}^{-1}}(B), b) If \alpha=0 and w\in\mathcal{A}_{1}\cap RH_{(q/p_{0})^{\prime}} , then \int_{\mathbb{R}^{n}}|T_{0,m}a(x)|^{p_{0}}w(x)dx\leq C\|\chi_{B}\|_{X}^{-p_{0}}\sum_{i=1}^{m}w_{A_{i}^{-1}}(B), Proof. The proof is analogous to that of [ 20 , Lemma 3.2] , but now considering the X -atoms in Definition 16 . ∎ We end this section with the following proposition, which is crucial to get our main result presented in the next section. Proposition 30 . Let 0\leq\alpha\max\left\{1,\frac{p_{0}n}{\alpha}\right\} , X^{1/p_{0}} and Y^{1/q_{0}} are ball Banach function spaces such that (Y^{1/q_{0}})^{\prime}=((X^{1/p_{0}})^{\prime})^{p_{0}/q_{0}} and the Hardy-Littlewood operator M is bounded on (Y^{1/q_{0}})^{\prime} , then T_{\alpha,m} can be extended to a bounded operator H_{X}(\mathbb{R}^{n})\to Y . b) If for some 0\max\left\{1,p_{0}\left(1+2^{n+3}\|M\|_{(X^{1/p_{0}})^{\prime}}\right)\right\}, (11) then T_{0,m} can be extended to a bounded operator H_{X}(\mathbb{R}^{n})\to X . Proof. The operator T_{\alpha,m} is well defined on the elements of H^{X,q/p_{0},d}_{fin}(\mathbb{R}^{n}) . So given f\in H^{X,q/p_{0},d}_{fin}(\mathbb{R}^{n}) , from Lemma 5 , we have that \|T_{\alpha,m}f\|_{Y}^{q_{0}}=\||T_{\alpha,m}f|^{q_{0}}\|_{Y^{1/q_{0}}}=\sup\int|T_{\alpha,m}f(x)|^{q_{0}}|g(x)|dx, where the supremum is taken over all g\in(Y^{1/q_{0}})^{\prime} such that \|g\|_{(Y^{1/q_{0}})^{\prime}}\leq 1 . Now we utilize the Rubio de Francia iteration algorithm with respect to (Y^{1/q_{0}})^{\prime} . Given a function g , define \mathcal{R}g(x)=\sum_{i=0}^{\infty}\frac{M^{i}g(x)}{2^{i}\|M\|_{(Y^{1/q_{0}})^{\prime}}^{i}}, where M^{0}g=g and, for i\geq 1 , M^{i}g=M\circ\cdot\cdot\cdot\circ Mg denotes i iterates of the Hardy-Littlewood maximal operator. The function \mathcal{R}g satisfies: (1)\,|g(x)|\leq\mathcal{R}g(x) for all x\in\mathbb{R}^{n} ; (2)\,\|\mathcal{R}g\|_{(Y^{1/q_{0}})^{\prime}}\leq C\|g\|_{(Y^{1/q_{0}})^{\prime}} ; (3)\,\mathcal{R}g\in\mathcal{A}_{1} and [\mathcal{R}g]_{\mathcal{A}_{1}}\leq 2\|M\|_{(Y^{1/q_{0}})^{\prime}} . by these properties and since (Y^{1/q_{0}})^{\prime}=((X^{1/p_{0}})^{\prime})^{p_{0}/q_{0}} and H^{X,q/p_{0},d}_{fin}(\mathbb{R}^{n})=H^{p_{0},q/p_{0},d}_{fin}(\mathcal{R}g) as sets, Proposition 30 - a) gives \int|T_{\alpha,m}f(x)|^{q_{0}}|g(x)|dx\leq\int|T_{\alpha,m}f(x)|^{q_{0}}\mathcal{R}g(x)dx =\|T_{\alpha,m}f\|_{L^{q_{0}}(\mathcal{R}g)}^{q_{0}}\leq C\sum_{i=1}^{m}\|f\|_{H^{p_{0}}\left([(\mathcal{R}g)_{A_{i}^{-1}}]^{\frac{p_{0}}{q_{0}}}\right)}^{q_{0}} =C\sum_{i=1}^{m}\left(\int[\mathcal{M}_{N}f(x)]^{p_{0}}[(\mathcal{R}g)_{A_{i}^{-1}}(x)]^{\frac{p_{0}}{q_{0}}}dx\right)^{\frac{q_{0}}{p_{0}}} by Hölder’s inequality (see [ 34 , Lemma 2.5] ), we have \leq C\|[\mathcal{M}_{N}f]^{p_{0}}\|_{X^{1/p_{0}}}^{\frac{q_{0}}{p_{0}}}\sum_{i=1}^{m}\left\|[(\mathcal{R}g)_{A_{i}^{-1}}]^{\frac{p_{0}}{q_{0}}}\right\|_{(X^{1/p_{0}})^{\prime}}^{\frac{q_{0}}{p_{0}}} =C\|\mathcal{M}_{N}f\|_{X}^{q_{0}}\sum_{i=1}^{m}\left\|[(\mathcal{R}g)_{A_{i}^{-1}}]\right\|_{((X^{1/p_{0}})^{\prime})^{p_{0}/q_{0}}} since X is \mathcal{O}(n) -invariant (see Remarks 8 and 11 ) and (Y^{1/q_{0}})^{\prime}=((X^{1/p_{0}})^{\prime})^{p_{0}/q_{0}} , we have =C\|f\|_{H_{X}(\mathbb{R}^{n})}^{q_{0}}\sum_{i=1}^{m}\|\mathcal{R}g\|_{(Y^{1/q_{0}})^{\prime}} \leq C\|f\|_{H_{X}(\mathbb{R}^{n})}^{q_{0}}\left\|g\right\|_{(Y^{1/q_{0}})^{\prime}}. Thus \|T_{\alpha,m}f\|_{Y}\leq C\|f\|_{H_{X}(\mathbb{R}^{n})}, for all f\in H^{X,q/p_{0},d}_{fin}(\mathbb{R}^{n}) . Then, the part a) of the theorem follows from Corollary 20 . Finally, to prove the part b) of the theorem (i.e.: the case \alpha=0 ), we proceed as in the proof of a) , by considering p_{0}=q_{0} , X=Y , Proposition 30 - b) and verifying that Rg\in\mathcal{R}H_{(q/p_{0})^{\prime}} . Indeed, ( 11 ), Remark 24 and Lemma 25 give \mathcal{R}g\in RH_{(q/p_{0})^{\prime}} . Thus b) follows and with them, the theorem. ∎ Corollary 32 . Let 0<\alpha\max\left\{1,\frac{p_{0}n}{\alpha}\right\} , X^{1/p_{0}} and Y^{1/q_{0}} are ball Banach function spaces such that (Y^{1/q_{0}})^{\prime}=((X^{1/p_{0}})^{\prime})^{p_{0}/q_{0}} and the Hardy-Littlewood operator M is bounded on (Y^{1/q_{0}})^{\prime} . (iv - b ) There exists 0\max\left\{1,p_{0}\left(1+2^{n+3}\|M\|_{(X^{1/p_{0}})^{\prime}}\right)\right\} . Weighted Lebesgue spaces. Given 01 , then L^{p}(w) is a ball Banach function space with (L^{p}(w))^{\prime}=L^{p^{\prime}}(w^{1-p^{\prime}}) , where \frac{1}{p}+\frac{1}{p^{\prime}}=1 . In [ 26 , Section 7.1] , it was pointed out that a weighted Lebesgue space may not be a Banach function space. Given 0\leq\alpha\max\left\{1,\frac{p_{0}n}{\alpha}\right\} or \widetilde{q}>\max\left\{1,p_{0}\left(1+2^{n+3}\|M\|_{(X^{1/p_{0}})^{\prime}}\right)\right\} , according to the case. Finally, since w(x)=|x|^{a}\in\mathcal{A}_{1} for -n0:\kappa_{p(\cdot)}(f/\lambda)\leq 1\right\}<\infty. The amount \|f\|_{p(\cdot)} is known as the Luxemburg norm of f with respect to p(\cdot) . By [ 26 , Section 7.8] , we have that the couple (L^{p(\cdot)}(\mathbb{R}^{n}),\|\cdot\|_{p(\cdot)}) is a ball quasi-Banach function space. If p_{-}>1 , then L^{p(\cdot)}(\mathbb{R}^{n}) is a ball Banach function space with (L^{p(\cdot)}(\mathbb{R}^{n}))^{\prime}=L^{p^{\prime}(\cdot)}(\mathbb{R}^{n}) , where \frac{1}{p(x)}+\frac{1}{p^{\prime}(x)}=1 for all x . An exponent p(\cdot):\mathbb{R}^{n}\to(0,\infty) is said to be globally log-Hölder continuous if there exist positive constants C and p_{\infty} such that |p(x)-p(y)|\leq\frac{-C}{\log(|x-y|)},\,\,\text{for}\,\,|x-y|\leq 1/2 and |p(x)-p_{\infty}|\leq\frac{C}{\log(e+|x|)},\,\,\text{for all}\,\,x\in\mathbb{R}^{n}. From now on, we consider the space L^{p(\cdot)}(\mathbb{R}^{n}) with p(\cdot) globally log-Hölder continuous. (i) For X:=L^{p(\cdot)} , the assumption ( 3 ) holds true for any s\in(0,1] and \theta\in(0,\min\{s,p_{-}\}) (indeed, we apply [ 16 , Lemma 2.4] on L^{p(\cdot)/\theta} , with u=s/\theta and |f_{j}|^{\theta} instead of f_{j} ); the assumption ( 4 ) holds true for any r\in(0,\min\{1,p_{-}\}) and \widetilde{q}\in(\max\{1,p_{+}\},\infty) (this follows by duality, [ 7 , Theorem 1.5] and Remark 14 ). (ii) By [ 6 , Proposition 2.18 and Theorem 2.61] , for any s\in(0,p_{-}) , (L^{p(\cdot)})^{1/s}=L^{p(\cdot)/s} and the space L^{p(\cdot)} is strictly s -convex . (iii) By dominated convergence Theorem, we have that if \{f_{j}\}\subset X and f_{j}\to 0 a.e., then \kappa_{p(\cdot)}(f_{j})\to 0 . So, from [ 6 , Theorem 2.69] , it follows \|f_{j}\|_{p(\cdot)}\to 0 . Then, the quasi-norm \|\cdot\|_{p(\cdot)} is absolutely continuous . If the exponent p(\cdot) is radial (i.e.: for any A\in\mathcal{O}(n) fixed, p(Ax)=p(x) for all x\in\mathbb{R}^{n} ), then the quasi-norm \|\cdot\|_{p(\cdot)} is \mathcal{O}(n) - invariant . (iv - a and b ) For any s\in(0,p_{-}) , the Hardy-Littlewood maximal operator M is bounded on L^{p(\cdot)/s} (see [ 7 , Theorem 1.5] ), and so also on (L^{p(\cdot)/s})^{\prime}=L^{(p(\cdot)/s)^{\prime}} , since the exponent (p(\cdot)/s)^{\prime} results globally log-Hölder continuous with ((p(\cdot)/s)^{\prime})_{-}>1 . Given 0\leq\alpha\max\left\{1,\frac{p_{0}n}{\alpha}\right\} or \widetilde{q}>\max\left\{1,p_{0}\left(1+2^{n+3}\|M\|_{(X^{1/p_{0}})^{\prime}}\right)\right\} , according to the case. Now, we have that H_{X}(\mathbb{R}^{n})=H^{p(\cdot)}(\mathbb{R}^{n}) is the variable Hardy spaces with exponent p(\cdot) defined in [ 16 ] . Finally, if p(\cdot) is radial and globally log-Hölder continuous, then Theorem 31 (and so also Corollary 32 ) applies with X=L^{p(\cdot)} and Y=L^{q(\cdot)} and recover [ 20 , Theorem 1.1] . Lorentz spaces. Given 00:|\{x:|f(x)|>s\}|\leq t\} . By [ 26 , Section 7.3] , the couple \left(L^{p,q}(\mathbb{R}^{n}),\|\cdot\|_{L^{p,q}}\right) is a ball quasi-Banach function space, whose quasi-norm \|\cdot\|_{L^{p,q}} satisfies \||g|^{r}\|_{L^{p,q}}=\|g\|_{L^{pr,qr}}^{r} for all 0\max\left\{1,\frac{p_{0}n}{\alpha}\right\} or \widetilde{q}>\max\left\{1,p_{0}\left(1+2^{n+3}\|M\|_{(X^{1/p_{0}})^{\prime}}\right)\right\} , according to the case. Finally, H_{X}(\mathbb{R}^{n})=H^{p,q}(\mathbb{R}^{n}) is the Hardy-Lorentz space defined in [ 1 ] . Then, Theorem 31 (and so also Corollary 32 ) applies on such X and Y . Orlicz spaces. A function \Phi:[0,\infty)\to[0,\infty) is called an Orlicz function if (1) it is non-decreasing and satisfies \lim_{t\to 0^{+}}\Phi(t)=\Phi(0)=0 , \Phi(t)>0 for all t>0 and \lim_{t\to\infty}\Phi(t)=\infty ; (2) the function x\to\Phi(|f(x)|) is measurable for every measurable function f on \mathbb{R}^{n} . Given an Orlicz function \Phi , define the modular associated with \Phi by \kappa_{\Phi}(f):=\int_{\mathbb{R}^{n}}\Phi(|f(x)|)\,dx. Then, the Orlicz space L^{\Phi}(\mathbb{R}^{n}) is defined to be the set of all measurable functions f on \mathbb{R}^{n} such that \|f\|_{\Phi}:=\inf\left\{\lambda>0:\kappa_{\Phi}(f/\lambda)\leq 1\right\}<\infty. The amount \|f\|_{\Phi} is known as the Luxemburg norm of f with respect to \Phi . We say that two Orlicz function \Phi and \Psi are equivalent, denote \Phi\sim\Psi , if \Psi(t/C)\leq\Phi(t)\leq\Psi(Ct) for some C\geq 1 and all t\geq 0 , and so L^{\Phi}=L^{\Psi} with \|\cdot\|_{\Phi}\approx\|\cdot\|_{\Psi} . For any s\in(0,\infty) fixed, we set \Phi_{s}(t):=\Phi(t^{s}) for all t>0 . Then, \Phi_{s} is an Orlicz function and for any measurable function f , one has \|f\|_{\Phi}^{s}=\||f|^{s}\|_{\Phi_{1/s}} . An Orlicz function \Phi:[0,\infty)\to[0,\infty) is said to be of positive lower (respectively, positive upper) type p with p\in(0,\infty) if there exists a positive constant C , depending on p , such that, for any t>0 and r\in(0,1] (respectively, r\in[1,\infty) ), \Phi(rt)\leq Cr^{p}\Phi(t). (12) When \Phi is an Orlicz function with positive lower type p_{\Phi}^{-} and positive upper type p_{\Phi}^{+} , by [ 26 , p. 92] , the couple (L^{\Phi},\|\cdot\|_{\Phi}) is a ball quasi-Banach space. Moreover, if \Phi is an Orlicz function of positive lower (resp., positive upper) type p_{\Phi}^{-} (resp., type p_{\Phi}^{+} ), then \Phi_{s} is of positive lower (resp., positive upper) type sp_{\Phi}^{-} (resp., type sp_{\Phi}^{+} ). A function \Phi:[0,\infty)\to[0,\infty) is called a Young function if \Phi is convex, left-continuous, \lim_{t\to 0^{+}}\Phi(t)=\Phi(0)=0 , and \lim_{t\to\infty}\Phi(t)=\infty . From the convexity and \Phi(0)=0 , it follows that any Young function is non-decreasing. If \Phi is a young function, from [ 13 , Lemma 3.2.2 (b) and Theorem 3.3.7 (b)] , it follows that the couple (L^{\Phi},\|\cdot\|_{\Phi}) is a Banach space. If an Orlicz function \Phi is also a Young function, then \Phi is a bijective continuous function from [0,\infty) onto [0,\infty) . For a Young function \Phi , we define \Phi^{-1} and its complementary function \widetilde{\Phi} on [0,\infty) by \Phi^{-1}(s):=\inf\{t\geq 0:\Phi(t)>s\} and \widetilde{\Phi}(t):=\sup\{ts-\Phi(s):s\in[0,\infty)\}, respectively. Then, the Köthe dual (L^{\Phi})^{\prime}=L^{\widetilde{\Phi}} with comparable norms (see [ 13 , Lemma 2.4.2 and Theorem 3.4.6] ). Moreover, by [ 19 , Property 1.6] , we have that s\leq\Phi^{-1}(s)(\widetilde{\Phi})^{-1}(s)\leq 2s,\,\,\,\,s\geq 0. (13) If \Phi is a Young-Orlicz function, then \Phi^{-1} is the usual inverse function of \Phi . An Orlicz function \Phi is called an N -function if it is a continuous and convex function such that \lim_{t\to 0^{+}}\frac{\Phi(t)}{t}=0,\,\,\,\,\,\,\text{and}\,\,\,\,\,\,\lim_{t\to\infty}\frac{\Phi(t)}{t}=\infty. (14) From [ 33 , Lemma 2.16] we have that if \Phi is an Orlicz function of positive lower type p_{\Phi}^{-}\in(1,\infty) and positive upper type p_{\Phi}^{+} , then there exists an Orlicz N -function \Psi equivalent to \Phi . Thus, without loss of generality, we may always assume that an Orlicz function \Phi of positive lower type p_{\Phi}^{-} and positive upper type p_{\Phi}^{+} , with 11 , being \Theta an Orlicz function, by [ 26 , Theorem 7.12] , the Hardy-Littlewood maximal M is bounded on L^{\Theta} . Finally, Remark 14 implies that ( 4 ) holds true for X=L^{\Phi} , any s\in(0,p_{\Phi}^{-}) and any q\in(\max\{1,p_{\Phi}^{+}\},\infty) . (ii) Given an Orlicz function \Phi with positive lower type p_{\Phi}^{-} and positive upper type p_{\Phi}^{+} , and s\in(0,p_{\Phi}^{-}) , by [ 33 , Lemma 2.16] , there exists a Young-Orlicz function \Psi equivalent to \Phi_{1/s} (i.e. L^{\Phi_{1/s}}=L^{\Psi} with comparable quasi-norms), since \|\cdot\|_{\Psi} is a norm, we can assume that X=L^{\Phi} is strictly s -convex for s\in(0,p_{\Phi}^{-}) . (iii) It is clear that \|\cdot\|_{\Phi} is \mathcal{O}(n) - invariant . Since \lim_{t\to 0^{+}}\Phi(t)=\Phi(0)=0 , by dominated convergence Theorem, we have for every sequence \{f_{j}\}\subset L^{\Phi} , with f_{j}\downarrow 0 , that \kappa_{\Phi}(f_{j})\to 0 and so \|f_{j}\|_{\Phi}\to 0 . Let’s see this, suppose that 0<\kappa_{\Phi}(f_{j})\to 0 . Given 0<\epsilon<1 and C>0 as in ( 12 ), we have 0\max\left\{1,\frac{p_{0}n}{\alpha}\right\} or q>\max\left\{1,p_{0}\left(1+2^{n+3}\|M\|_{(X^{1/p_{0}})^{\prime}}\right)\right\} , according to the case. Finally, H_{X}(\mathbb{R}^{n})=H^{\Phi}(\mathbb{R}^{n}) is the Orlicz-Hardy space defined in [ 17 ] . Then, Theorem 31 (and so also Corollary 32 ) applies on such X and Y . Pablo Rocha, Instituto de Matemática (INMABB), Departamento de Matemática, Universidad Nacional del Sur (UNS)-CONICET, Bahía Blanca, Argentina. e-mail: pablo.rocha@uns.edu.ar