Asymptotic greediness of the Haar system in the spaces \(L_p[0,1]\), \(1<p<\infty \) (Q2187718)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Asymptotic greediness of the Haar system in the spaces \(L_p[0,1]\), \(1<p<\infty \)
scientific article

    Statements

    Asymptotic greediness of the Haar system in the spaces \(L_p[0,1]\), \(1<p<\infty \) (English)
    0 references
    3 June 2020
    0 references
    The Haar system is an unconditional basis of \(L^p[0,1]\) that plays a central role in harmonic analysis and its applications. As demonstrated with particular clarity in [\textit{T. P. Hytönen}, Ann. Math. (2) 175, No. 3, 1473--1506 (2012; Zbl 1250.42036)], all Calderón-Zygmund operators (and thus a wide range of solution maps to PDE problems) can be constructed from basic operations on Haar systems. An influential result of \textit{D. L. Burkholder} [Stud. Math. 91, No. 1, 79--83 (1988; Zbl 0652.42012)] shows that, as a function of \(p\), the unconditionality constant of the Haar system behaves like \(p^{*} = \max \{p,p/(p-1)\}\). Many (optimal) quantitative results in harmonic analysis involve similar ideas (see, e.g., [Hytönen, loc.\,cit.]). Burkholder's method of proof, in particular, has been extensively developed in recent years as part of a general theory called the {Bellman function method} (see [\textit{F. Nazarov} et al., in: Systems, approximation, singular integral operators, and related topics. Proceedings of the 11th international workshop on operator theory and applications, IWOTA 2000, Bordeaux, France, June 13--16, 2000. Basel: Birkhäuser. 393--423 (2001; Zbl 0999.60064)]). The paper under review addresses a natural question, raised by Hytönen: does the greedy constant of the Haar basis also grow linearly with \(p^{*}\) as \(p^{*}\) tends to infinity? Recall that the greedy constant of a basis \((h_{j})_{j \in \mathbb{N}}\) is the infimum of all \(C>0\) such that \[ \|f-\mathcal{G}_{m}(f)\| \leq C \left\|f - \sum \limits _{j \in A} a_{j}h_{j} \right\| \] for all \(f \in L^{p}\), all \(m \in \mathbb{N}^{*}\), all finite subsets \(A\subset \mathbb{N}\) of cardinality \(m\), and all \((a_{j})_{j \in A} \in \mathbb{R}^{m}\), where \(\mathcal{G}_{m}(f)\) denotes the \(m\)-th greedy approximation of \(f\) (i.e., the sum of \(m\) terms with the largest coefficients in the basis expansion). Qualitatively, a~basis has finite greedy constant if and only if it is unconditional and democratic (in the sense that \(\|\sum _{j \in A} h_{j}\| \leq C_{d} \|\sum _{j \in B} h_{j}\|\) for some \(C_{d}>0\) and all finite \(A,B \subset \mathbb{N}\) of equal cardinality). The question thus has two interesting quantitative components: determining how the democracy constant \(C_{d}\) depends on \(p^{*}\), and determining how the greedy constant depends on both the unconditionality constant and the democracy constant. Known results for this second component give a linear growth in the lower bound, but only a quadratic growth in the upper bound. This is true even if one replaces democracy by super-democracy or symmetry for largest coefficients (see Section~1 of the paper). The authors nonetheless prove that the greedy constant of the Haar basis does grow linearly in \(p^{*}\). To do so, they exploit an extra property called {bi-democracy}: \[ \left\|\sum _{j \in A} h_{j} \right\|_{p} \cdot \left\|\sum _{k \in B} h_{k} \right\|_{p'} \leq C_{b} m , \quad |A|=|B|=m. \] Their proof has two parts. The first part is abstract (valid for all bi-greedy bases in Banach spaces) and shows that the greedy constant of a bi-democratic basis is bounded by a linear combination of the unconditionality constant and the bi-democratic constant. This result nicely extends previous works in the democratic case, such as [\textit{P. M. Berná} et al., Rev. Mat. Complut. 30, No. 2, 369--392 (2017; Zbl 1375.41020)]. Its proof involves a~natural separation of the indices that appear in the greedy approximation from those that do not (see Theorem~2.3). The second part shows that the bi-democratic constant of the Haar basis is proportional to~\(p^{*}\). For the lower bound, this is proven by selecting appropriate intervals inductively (starting with \(I_{1}=[0,1)\) then picking \(I_{j+1}\) to be the left half of \(I_{j}\)). For the upper bound, the authors use two key properties of the Haar basis: the dyadic intervals form a partition of \([0,1)\), and the dual functional of the Haar functions in \(L^{p}\) are the Haar functions in \(L^{p'}\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    Haar basis
    0 references
    greedy basis
    0 references
    unconditional basis
    0 references
    democratic basis
    0 references
    \(L_p\) spaces
    0 references
    Lebesgue-type inequalities
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references