Fractional maximal functions and mean oscillation on bounded doubling metric measure spaces (Q6075493)

Let \((X, d, \mu)\) be a doubling metric measure space. For a function \(f\in L^1_{\text{loc}}(X,\mu)\) and \(\alpha\geq 0\), define the following maximal function: \[ M^{\alpha} f(x)=\sup_{B\ni x} \frac{\text{rad}(B)^{\alpha}}{\mu(B)}\int_B |f|d\mu, \] where the supremum is taken over all balls \(B \subset X\) containing the point \(x \in X\) satisfying \(0 < \text{rad}(B) < 2 \operatorname{diam}(X)\). When \(\alpha=0\), \(Mf := M^0 f\) is the standard (uncentered) Hardy-Littlewood maximal function. When \(\alpha>0\), then \(M^{\alpha}f\) is known as the fractional maximal function. The measure \(\mu\) is said to satisfy a lower mass bound with lower mass bound exponent \(Q>0\) if there exists some constant \(0 < c_L \leq 1\) such that \[ \mu(B(x,r))\geq c_L r^Q \] for all \(x\in X\) and \(0<r< 2\operatorname{diam}(X)\). The measure \(\mu\) is said to satisfy an annular decay condition if there exist an exponent \(0<\beta \leq 1\) and a constant \(C_{\beta}\geq 1\) such that \[ \mu(B(x,R)\setminus B(x,r))\leq C_{\beta} \bigg(\frac{R-r}{R}\bigg)^{\beta}\mu(B(x,R)) \] for all \(x\in X\) and \(0<r\leq R<\infty\). The main results of the paper are the following two theorems. Theorem 1. Let \((X, d, \mu)\) be a doubling metric measure space. Then the Hardy-Littlewood maximal operator \(M\) is bounded from \(\text{BMO}(X)\) to \(\text{BLO}(X)\). Furthermore, if \(\mu\) satisfies an annular decay condition, then \(M\) maps \(\text{VMO}(X)\) to itself. Theorem 2. Let \((X, d, \mu)\) be a doubling metric measure space which is bounded with lower mass bound exponent \(Q\). Then for every \(\alpha\in(0,Q)\) the fractional maximal operator \(M^{\alpha}\) is bounded from \(\text{BMO}(X)\) to \(\text{BLO}(X)\). Furthermore, if \(\mu\) satisfies an annular decay condition, then \(M^{\alpha}\) maps \(\text{VMO}(X)\) to itself. From these theorems, it is deduced the following corollary for the Euclidean case. Corollary. For every \(\alpha\in [0,n)\) and every cube \(E\subset \mathbb{R}^n\), \(M^{\alpha}\) is bounded from \(\text{BMO}(E)\) to \(\text{BLO}(E)\). Furthermore, \(M^{\alpha}\) maps \(\text{VMO}(E)\) to itself. This corollary complements the known \(\alpha=0\) results in the literature by extending the BMO-BLO result to the fractional maximal function. For the VMO-VMO result, however, this is new even for \(\alpha=0\). It is also shown using examples that in the above results the action of \(M^{\alpha}\) is not continuous.