Vector-Valued Singular Integrals on Locally Doubling Spaces

We prove vector-valued boundedness of (suitable) Calderon-Zygmund operators and of the (truncated) Hardy-Littlewood maximal function on a connected locally doubling metric measure space.

💡 Research Summary

The paper addresses the problem of extending the classical Calderón‑Zygmund theory, which traditionally relies on a global doubling condition, to metric measure spaces that are only locally doubling. The authors consider a connected, locally compact metric space (X, d) equipped with a non‑zero inner regular Radon measure µ whose support equals X. For each radius R>0 they define the minimal doubling constant D_R as the smallest number satisfying µ(B(x,2r)) ≤ D_R µ(B(x,r)) for all x∈X and all 0<r≤R. The central hypothesis is that D_R is finite for some fixed R, which guarantees that every ball of radius ≤2R has finite, positive measure (Lemma 2).

The first technical contribution consists of a series of covering lemmas adapted to the locally doubling setting. Lemma 4 provides a Vitali‑type selection: given a set E of diameter <2R and a radius function r:E→(0,R], one can extract a countable subfamily of disjoint balls {B(x,r(x))} whose three‑fold dilations cover E. Lemma 5, which uses the connectedness of X, yields a Whitney‑type decomposition for any proper open set U with diam U<R: there exists a countable family {B(x,r(x))} with radii ≤R/2 such that the balls are essentially disjoint, each enlarged ball B(x,κr(x)) (κ>2) stays inside U, and the characteristic functions of the balls have uniformly bounded overlap. These lemmas replace the global doubling machinery and are the backbone of the subsequent analysis.

With these geometric tools, the authors define two vector‑valued maximal operators for a Banach‑space‑valued function f: X→B. The first, M_R f(x), is the supremum over all averages of |f| on balls containing x with radius ≤R; the second, f_M_R f(x), is the supremum of averages over balls centered at x. Lemma 8 establishes the basic analytic properties of M_R: lower semicontinuity, a weak (1,1) estimate on sets of diameter <6R, strong L^p bounds for 1<p≤∞, and the usual Lebesgue differentiation property. The constants in these estimates depend only on D_{3R} and the geometry of the space, not on any global doubling constant.

The core of the paper is a local Calderón‑Zygmund decomposition (Lemma 9). For a given integrable vector‑valued function f and a threshold α exceeding a multiple of the L^1 norm of f, the authors decompose f = g + Σ_{x∈N} h_x where:

• g is bounded (‖g‖_∞ ≤ C α) and has controlled L^1 norm (≤3‖f‖_1);
• each h_x is supported in a ball B(x,r(x)) ⊂ B(E,R/2), has zero integral, and satisfies uniform L^∞ and L^1 bounds;
• the family of balls has uniformly bounded overlap (constant D_{5·3κR});
• the total measure of the “bad” set where M_{κR} f exceeds α is small, guaranteeing that the decomposition is essentially localized. This decomposition mirrors the classical one but works entirely with balls whose radii are dictated by the local doubling constant D_R, thus avoiding any need for a global doubling condition.

Using the decomposition, the authors prove boundedness results for vector‑valued Calderón‑Zygmund operators (Theorem 14 and Corollaries 15, 17). The operators considered are standard singular integrals with kernels satisfying size and smoothness estimates adapted to the metric setting. The main theorem states that for 1<p<∞ and any Banach space B, the operator T maps L^p(µ;B) to itself boundedly, with a constant depending only on p, the kernel constants, and the local doubling constants D_R. The proof follows the classical scheme: apply the decomposition to f, estimate T on the good part g using the L^2 theory (which holds locally because each ball is doubling), and control the contribution of the bad parts h_x via the cancellation property and the bounded overlap of the supporting balls.

Finally, Theorem 18 treats the truncated Hardy–Littlewood maximal function. The authors show that the localized maximal operator M_R, when applied to vector‑valued functions taking values in mixed‑norm spaces (including ℓ^q(N) and more general spaces), satisfies the same L^p boundedness as in the classical setting. This result fills a gap in the literature, where vector‑valued maximal inequalities were previously known only under global doubling assumptions.

Throughout the paper, the authors emphasize that σ‑finiteness of µ is not required; inner regularity suffices. They also correct minor inaccuracies present in earlier works (e.g., in the proof of Corollary 15) and extend the theory to mixed‑norm spaces beyond the usual ℓ^q‑valued case. The connectedness hypothesis is shown to be essential for the Whitney‑type covering lemma, though the authors note that a weaker condition—existence of an exterior point preserving diameter—could replace full connectedness.

In summary, the paper successfully develops a local Calderón‑Zygmund framework on connected locally doubling metric measure spaces, establishing vector‑valued L^p boundedness for both singular integrals and maximal functions without invoking a global doubling condition. This broadens the applicability of harmonic analysis tools to spaces with non‑uniform geometry, such as manifolds with variable curvature, certain fractal sets, and non‑homogeneous graphs, and opens avenues for further extensions to quasi‑metric spaces, weaker connectivity assumptions, and nonlinear operators.