Local constants and Bohr’s phenomenon for Banach spaces of analytic polynomials
A. Defant, D. Galicer, M. Mansilla, M. Mastyło, S. Muro · 2025
Read Local constants and Bohr’s phenomenon for Banach spaces of analytic polynomials by A. Defant, D. Galicer, M. Mansilla, M. Mastyło, S. Muro (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-26326.txt.
Introduction
Andreas Defant and co-authors build tools to pin down how three classic constants behave for spaces of multivariable polynomials, in a 2025 paper. The constants gauge how well a space can be projected, how stable its basic building blocks are, and how those traits grow with dimension and degree. The spaces hold all polynomials whose terms live inside a chosen index set, measured by their peak size on a sphere.
Their framework yields sharp growth estimates for the constants. They also apply it to the Bohr radius, a classic measure from complex analysis, for broad classes of sequence spaces. A fine study of Lorentz spaces follows, with optimal results across a wide range of settings.
Catalog record
| CREATOR | A. Defant, D. Galicer, M. Mansilla, M. Mastyło, S. Muro |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 164 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-26326
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Local constants and Bohr’s phenomenon for Banach spaces of analytic polynomials A. Defant, D. Galicer, M. Mansilla, M. Mastyło, S. Muro Abstract . The primary aim of this work is to develop methods that provide new insights into the relationships between fundamental constants in Banach space theory–specifically, the projection constant, the unconditional basis constant and the Gordon-Lewis constant–for the Banach space 𝒫 J ( X n ) \mathcal{P}_{J}(X_{n}) of multivariate analytic polynomials. This class consists of all polynomials whose monomial coefficients vanish outside the set of multi-indices J J , and it is equipped with the supremum norm on the unit sphere of the finite-dimensional Banach space X n = ( ℂ n , ∥ ⋅ ∥ ) X_{n}=(\mathbb{C}^{n},\|\cdot\|) . We establish a general framework for proving quantitative results on the asymptotic optimal behavior of these constants, which depend on both the dimension of the space and the degree of the polynomials. Using the tools developed, we derive asymptotic estimates of the Bohr radius for general Banach sequence lattices. Additionally, we apply our results to the asymptotic study of local constants and the Bohr radius within finite-dimensional Lorentz sequence spaces, which requires a refined analysis of the combinatorial structure of the associated index sets. As a consequence, we obtain optimal results across a broad range of parameters. 2020 Mathematics Subject Classification 46B07, 46B28, 46E15, 32A05, 32A15 Introduction The study of local properties of infinite-dimensional Banach spaces, which are intricately linked to the structure of finite-dimensional subspaces, is fundamental to both the local and general theory of Banach spaces. This area of research is driven by various aspects of modern analysis. One of the core challenges in local Banach space theory is understanding the asymptotic behavior of certain fundamental constants that have significant applications. The goal of this work is to establish estimates for key constants in Banach space theory–specifically the projection constant, the unconditional basis constant, and the Gordon-Lewis constant–within the context of the finite-dimensional Banach spaces \mathcal{P}_{J}(X_{n}) of polynomials. In this setting, \mathcal{P}_{J}(X_{n}) denotes the Banach space of multivariate analytic polynomials P defined on finite-dimensional Banach spaces X_{n}=(\mathbb{C}^{n},\|\cdot\|) , where the monomial coefficients of P vanish outside the set of multi-indices J . This space is equipped with the supremum norm over the unit sphere of X_{n} . Additionally, we explore the relationships between the involved constants. Furthermore, we aim to unify these concepts in the study of the Bohr radius for unit spheres of finite-dimensional Banach spaces and demonstrate the application of our general results to a significant class of finite-dimensional spaces generated by the scale of Lorentz spaces. To accomplish these objectives, we will employ techniques from various fields of analysis, including local Banach space theory, probability theory, complex analysis, Banach lattice theory, and combinatorics. To clarify these aims, we will present key definitions and essential technical tools that naturally arise in our investigation. The projection constant. The projection constant, in particular, is one of the most significant concepts in modern Banach space theory, closely related to the classical problem of describing complemented subspaces of Banach spaces—a problem that has been extensively studied since the inception of abstract operator theory. Recall that if X is a closed subspace of a Banach space Y , then the relative projection constant of X in Y is defined by \displaystyle\boldsymbol{\lambda}(X,Y) \displaystyle=\inf\big\{\|P\|:\,\,P\in\mathcal{L}(Y,X),\,\,P|_{X}=\id_{X}\big\}\,, where \id_{X} denotes the identity operator on X , and \mathcal{L}(U,V) represents the Banach space of all bounded linear operators between Banach spaces U and V with the supremum norm. Here, \mathcal{L}(U):=\mathcal{L}(U,U) , and we adopt the convention that \inf\varnothing=\infty . The (absolute) projection constant of X is expressed as: \boldsymbol{\lambda}(X):=\sup\,\,\boldsymbol{\lambda}(I(X),Y)\,, where the supremum is taken over all Banach spaces Y and isometric embeddings I\colon X\to Y . Note that if X is a finite-dimensional Banach space and X_{1} is a subspace of some C(K) -space isometric to X , then (see, e.g., [ 61 , III.B.5 Theorem] ): \displaystyle\boldsymbol{\lambda}(X)=\boldsymbol{\lambda}(X_{1},C(K))\,. Thus, determining \boldsymbol{\lambda}(X) is equivalent to finding the norm of a minimal projection from C(K) onto X_{1} . General bounds for projection constants of various finite-dimensional Banach spaces have been studied by many authors. The most fundamental general upper bound is due to Kadets and Snobar [ 39 ] : For every finite-dimensional Banach space X , we have (1) \boldsymbol{\lambda}(X)\leq\sqrt{\text{dim}X}. Regarding the Kadets-Snobar estimate ( 1 ), we also mention a remarkable result of Pisier [ 54 , Corollary 10.8] , who constructed an infinite dimensional Banach space X such that \|P\|\geq\delta\sqrt{\text{rank}\,P} for some \delta\in(0,1) and all finite rank projections P\colon X\to X . König [ 42 ] proved that the Kadets-Snobar estimate ( 1 ) is asymptotically best possible. More precisely, he showed that there exists a sequence (X_{n_{k}})_{k=1}^{\infty} of finite-dimensional real Banach spaces such that \text{dim}(X_{n_{k}})=n_{k} , where n_{k}\to\infty as k\to\infty , and \lim_{k\to\infty}\frac{\boldsymbol{\lambda}(X_{n_{k}})}{\sqrt{n_{k}}}=1\,. Another surprising result in this direction is due to Figiel, Lindenstrauss and Milman [ 32 ] . It states that there is a universal constant c\in(0,1) such that every n -dimensional Banach space E contains a subspace F such that \boldsymbol{\lambda}(F)\geq c\,\sqrt{n} . The unconditional basis constant. The unconditional basis constant of a basis (e_{i})_{i\in I} of a Banach space E is given by the infimum over all K>0 such that for any finitely supported family (\alpha_{i})_{i\in I} of scalars and for any finitely supported family (\varepsilon_{i})_{i\in I} with |\varepsilon_{i}|=1,\,i\in I we have \Big\|\sum_{i\in I}\varepsilon_{i}\alpha_{i}e_{i}\Big\|\leq K\Big\|\sum_{i\in I}\alpha_{i}e_{i}\Big\|\,. We denote the unconditional basis constant of (e_{i})_{i\in I} by \boldsymbol{\chi}((e_{i})_{i\in I})=\boldsymbol{\chi}((e_{i})_{i\in I};X) . We also write \boldsymbol{\chi}((e_{i})_{i\in I})=+\infty , whenever (e_{i})_{i\in I} is not unconditional, and say that (e_{i})_{i\in I} is a 1 -unconditional basis, whenever \boldsymbol{\chi}((e_{i})_{i\in I})=1 . The unconditional basis constant \boldsymbol{\chi}(X) of X is defined to be the infimum of \boldsymbol{\chi}((e_{i})_{i\in I}) taken over all possible unconditional basis (e_{i})_{i\in I} of X . The Gordon-Lewis constant. A significant milestone in the study of unconditional basis constants was achieved by Gordon and Lewis in [ 36 ] , who introduced the constant now known as the G-L constant. In light of our study of this constant for the space of polynomials, we need to introduce some additional concepts to properly present the original definition of this important concept in the context of local Banach space theory. Given Banach spaces X , Y and 1\leq p\leq\infty , an operator u\in\mathcal{L}(X,Y) is said to be p -factorable whenever there exist a measure space (\Omega,\Sigma,\mu) and operators v\in\mathcal{L}(X,L_{p}(\mu)) , w\in\mathcal{L}(L_{p}(\mu),Y^{\ast\ast}) , satisfying the following factorization \kappa_{Y}u\colon X\stackrel{{\scriptstyle v}}{{\longrightarrow}}L_{p}(\mu)\stackrel{{\scriptstyle w}}{{\longrightarrow}}Y^{\ast\ast}\,; here, as usual, \kappa_{Y}\colon Y\to Y^{\ast\ast} is the canonical injection. In this case the \gamma_{p} -norm of the p -factorable operator u is given by (2) \gamma_{p}(u)=\inf\|v\|\|w\|\,, where the infimum is taken over all possible factorizations. We are mainly interested in the norms \gamma_{p} for operators acting between finite dimensional Banach spaces X and Y . In this case, the infimum in ( 2 ) is realized considering all possible factorizations of the more simple form {\lx@inpgf@ignorespaces X} {\lx@inpgf@ignorespaces Y\,\,,} {\lx@inpgf@ignorespaces\ell_{p}^{n}} \scriptstyle{\lx@inpgf@ignorespaces v} \scriptstyle{\lx@inpgf@ignorespaces u} \scriptstyle{\lx@inpgf@ignorespaces w} where n is arbitrary. An operator u\in\mathcal{L}(X,Y) is said to be 1 -summing if there is a constant C>0 such that for each choice of finitely many x_{1},\ldots,x_{N}\in X one has \sum_{j=1}^{N}\|ux_{j}\|_{Y}\leq C\sup\Big\{\sum_{j=1}^{N}|x^{\ast}(x_{j})|\,:\,\,\|x^{\ast}\|_{X^{\ast}}\leq 1\Big\}\,. By \pi_{1}(u\colon X\to Y)