On interpolation of reflexive variable Lebesgue spaces on which the Hardy-Littlewood maximal operator is bounded (Q2671462)

Let \(L^{0}\left(\mathbb{R}^{d}\right)\) denote the space of all (equivalence classes of) Lebesgue measurable complex-valued functions on \(\mathbb{R}^{d}\) with the toplogy of convergence in measure on sets of finite measure. Let \(p(\cdot):\mathbb{R}^{d}\rightarrow \left[ 1,\infty \right]\) be a measurable a.e. finite function. By \(L^{p(\cdot)}\left(\mathbb{R}^{d}\right)\) we denote the set of all functions \(f\in L^{0}\left(\mathbb{R}^{d}\right)\) such that \[ I_{p(\cdot)}\left( \frac{f}{\lambda}\right) :=\iint\limits_{\mathbb{R}^{d}}\left\vert \frac{f(x)}{\lambda}\right\vert^{p(x)}dx<\infty \] for some \(\lambda >0\). \(L^{p(\cdot)}\left(\mathbb{R}^{d}\right)\) becomes a Banach space if we endow it with the Luxemburg-Nakano norm \[ \left\Vert f\right\Vert_{p(\cdot)}:=\inf \left\{ \lambda > 0: I_{p(\cdot)}\left( \frac{f}{\lambda}\right) \leq 1\right\} . \] It is called variable Lebesgue space. Let \(1\leq q<\infty\) and \(f\in L_{\mathrm{loc}}^{q}\left(\mathbb{R}^{d}\right)\). The \(q\)-th maximal operator is defined by \[ \left( M_{q}f\right) (x):=\sup_{x\in Q}\left( \frac{1}{\left\vert Q\right\vert}\iint\limits_{Q}\left\vert f(y)\right\vert^{q}dy\right)^{\frac{1}{q}}, \] where the supremum is taken over all cubes \(Q\subset\mathbb{R}^{d}\) containing \(x\). Note that \(M_{1}\) is the usual Hardy-Littlewood maximal operator. Denote by \(B_{M}\left(\mathbb{R}^{d}\right)\) the set of measurable a.e. finite functions \(p(\cdot):\mathbb{R}^{d}\rightarrow \left[ 1,\infty \right]\) such that the Hardy-Littlewood maximal operator is bounded on \(L^{p(\cdot)}\left(\mathbb{R}^{d}\right)\). The main result of this paper is the following Theorem~1.3. Theorem 1.3. Let \(p(\cdot):\mathbb{R}^{d}\rightarrow \left[ 1,\infty \right]\) be a measurable function satisfying \(1<p_{-}\leq p_{+}<\infty\). Then \(p(\cdot)\in B_{M}\left(\mathbb{R}^{d}\right)\) if and only if for every \(q\in \left( 1,\infty \right)\), there exists a number \(\Theta_{p(\cdot),q}\in \left( 0,1\right)\) such that for every \(\theta \in (0,\Theta_{p(\cdot),q}]\) the variable exponent \(r(\cdot)\) defined by \[ \frac{1}{p(x)}=\frac{\theta}{q}+\frac{1-\theta}{r(x)},\ x\in\mathbb{R}^{d}, \] belongs to \(B_{M}\left(\mathbb{R}^{d}\right)\). In Section 2, the authors formulate an interpolation lemma which implies the proof of the sufficiency portion of Theorem~1.3. In Section~3, they show that if \(p(\cdot)\in B_{M}\left(\mathbb{R}^{d}\right)\) satisfies \(1<p_{-}\leq p_{+}<\infty\), then the variable exponents \(\left( \frac{1}{t}\left( \frac{p(\cdot)}{s}\right)^{\prime}\right)^{\prime}\) belong to \(B_{M}\left(\mathbb{R}^{d}\right)\) for all \(s,t\geq 1\) sufficiently close to \(1\). Based on this result, they complete the proof of the necessity portion of Theorem~1.3 in Section~4.