On The Rigidity Of The Roots Of Power Series With Constrained Coefficients
Jacob Kewarth · 2025
Read On The Rigidity Of The Roots Of Power Series With Constrained Coefficients by Jacob Kewarth (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-2505-18317.txt.
Introduction
Jacob Kewarth studies in 2025 the roots of power series whose coefficients are constrained to a given set. The set of all such roots changes as the allowed coefficient set changes. Kewarth asks how deep complex roots can reach into the unit disc, and when the root set is connected.
He gives an estimate for that depth and criteria for connected or disconnected root sets. For two finite symmetric integer sets that both contain one, equal root sets force agreement of elements in a certain range. The result shows a surprising rigidity in how coefficient sets determine their roots.
Catalog record
| CREATOR | Jacob Kewarth |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-18317
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On The Rigidity Of The Roots Of Power Series With Constrained Coefficients Jacob Kewarth Abstract . Here we consider the set Σ S \Sigma_{S} of roots of power series whose coefficients lie in a given set S S and how such sets of roots vary as the set S S varies. We give an estimate of the depth that complex roots can reach into the disc, offer some criterion for the set of roots to be connected or disconnected, and show that for two finite symmetric sets S S and T T of integers containing 1 1 , if Σ S = Σ T \Sigma_{S}=\Sigma_{T} then all of their elements between 1 1 and 2 max ( S ) + 1 2\sqrt{\max(S)}+1 must agree. 1. Introduction 1.1. History and Outline Given a finite (or bounded) set S let \Sigma_{S} denote the set of roots in \mathbb{D} of power series with coefficients in S . Dynamical interest in such sets was started by Barnsley and Harrington (see [BHa] ) who studied the family of iterated function systems (IFS) z\mapsto\lambda z+1 and z\mapsto\lambda z-1 for various choices of |\lambda|<1 . They found that an attractor A_{\lambda} was connected if and only if \lambda\in\Sigma_{\{0,\pm 1\}} , dubbing the set the “Mandelbrot set for pairs of linear maps”. The set \Sigma_{\{0,\pm 1\}} was shown to be connected and locally connected by Bousch (see [Bo1], [Bo2] ). In [Ba] Bandt proved that \Sigma_{\{0,\pm 1\}} has at least one (non-trivial) hole and in [CKW] Calegari, Koch, and Walker showed that there are infinitely many holes. [SX] Solomyak and Xu were able to certify a whole ‘wedge’ of \Sigma_{\{0,\pm 1\}} near the imaginary axis for which polynomial roots are interior points of \Sigma_{\{0,\pm 1\}} and [CKW] showed in fact that interior points of \Sigma_{\{0,\pm 1\}} are dense away from the real axis. In [Ti] Tiozzo related the set \Sigma_{\{\pm 1\}} to the set of Galois conjugates of entropies of real quadratic polynomials. Many variations have since appeared studying the set of roots \Sigma_{S} for different sets of coefficients S . In [BY] the set \Sigma_{S} was studied for sets S whose elements lie on the unit circle, [BBBP] studied the problem for the set S=[-g,g] with the additional assumption that the power series first coefficient is always 1. [Na] gave a broad criterion for the set of roots of power series starting with 1 to be connected and locally connected. Motivated by interest in the family of IFS z\mapsto\lambda z+\xi^{j}_{n} (where \xi_{n} is the nth root of unity e^{2\pi i/n} ) the set of roots of power series starting with 1 whose coefficients are \frac{\xi_{n}^{j}-\xi_{n}^{k}}{1-\xi_{n}} have been studied extensively (see [BHu] , [HI] , and [Na] ) and most notably has been shown to be connected, locally connected, and regular-closed for each n\geqslant 3 . The object of this paper is to study the rigidity of the set of such roots. We ask how different sets of coefficients yield different sets of roots. Notice that if S=cT for some c\neq 0 then \Sigma_{S}=\Sigma_{T} and so we further ask if such a criterion is necessary for \Sigma_{S} to be equal to \Sigma_{T} . Since the roots set is invariant by scaling of the set of coefficients, we mostly consider sets S which are normalized, that is, for which the minimum positive element of S is 1 . Our main result towards this goal is the following: Theorem 6.2 . Let S and T be finite non-empty symmetric normalized sets of integers. If \Sigma_{S}=\Sigma_{T} and k is an integer in (1,2\sqrt{\max(S)}+1) then k\in S if and only if k\in T . Our approach to studying rigidity is to first study how the geometry of \Sigma_{S} depends on and varies with S . The main technical tool will be a series of criteria we develop in Section 3 to verify whether an element z\in\mathbb{D} belongs to \Sigma_{S} . Pictures of \Sigma_{S} (where the elements of S do not have big gaps) tend to display two prominent features. Firstly, the real line contains an interval which seems to extend far deeper into the unit disc than any complex roots can reach, and secondly there appears to be an annulus contained in \Sigma_{S} . Using the criteria of Section 3 we certify both claims for appropriate sets S . In particular, if we consider: Definition 1.1 . If S is a non-empty normalized finite set of real numbers we define: \rho_{inn}(S)=\inf\{R>0\mid A_{R}\subseteq\Sigma_{S}\} where A_{R} is the annulus \{R\leqslant|z|<1\} . When discussing coefficient sets which S which are normalized it helps to consider how far apart consecutive elements of S are, we call the total gap of S the maximum such distance. Then when the total gap of S is small enough we prove that the subset \Sigma_{S}^{1} of \Sigma_{S} of roots of power series with coefficients in S which start with 1 contains an annulus giving a upper bound on \rho_{inn} . Theorem 3.9 . Let S be a symmetric non-empty finite set of real numbers with total gap 1 and let M=\max_{s\in S}|s| . Then \Sigma^{1}_{S} contains the annulus \frac{1}{\sqrt{M}}\leqslant|z|<1 . In particular, we have that: \rho_{inn}(S)\leqslant\frac{1}{\sqrt{M}} In Section 4 we consider the largest possible annulus containing the complex roots of \Sigma_{S} . The reason for avoiding real roots is that real roots of \Sigma_{S} tend to get far deeper than complex roots. Indeed, real roots may get as deep at \frac{1}{M+1} whereas it is well known that complex roots may not be smaller than \frac{1}{\sqrt{M}+1} . This causes the real roots to stick out giving the appearance of an ‘antenna’. Definition 1.2 . If S is a non-empty normalized finite set of real numbers we define: \rho_{out}=\sup\{R>0\mid(\Sigma_{S}\setminus\mathbb{R})\subseteq A_{R}\} where A_{R} is the annulus \{R\leqslant|z|<1\} . Remark . Note that by definition we have \rho_{out}\leqslant\rho_{inn} . To approach the problem, we first develop a general inequality on the size of all roots of a power series which may be of independent interest: Theorem 4.1 . If S is a (possibly infinite) bounded non-empty set of (possibly complex) numbers and P(z)=\sum_{j=0}^{\infty}p_{j}z^{j} is a power series with coefficients in S and with p_{0}=1 then for any roots \lambda_{1},...,\lambda_{k}\in\mathbb{D} of P(z) , we have that: \prod_{j=1}^{k}(|\lambda_{j}|^{-1}-1)\leqslant\sup_{j}|p_{j}|. In particular, P(z) does not have k roots \lambda_{1},...,\lambda_{k} all with |\lambda_{j}|<\frac{1}{\sup_{j}|p_{j}|^{1/k}+1} . Using this inequality we given an alternative proof of the well known estimate that \rho_{out}\geqslant\frac{1}{\sqrt{M}+1} . The remainder of the first part of section 4 is dedicated to showing this estimate is never best possible. Namely, we will prove the following: Theorem 4.4 . Let S be a finite non-empty symmetric normalized set of real numbers and let M=\max(S) . Then there is no sequence of numbers \lambda_{n}\in\Sigma_{S}\setminus\mathbb{R} such that |\lambda_{n}|\rightarrow\frac{1}{\sqrt{M}+1} . Moreover, we have that: \frac{1}{\sqrt{M}+1}<\rho_{out}(S)\leqslant\frac{1}{\sqrt{M+k}} for k=\max\{s\in S\mid 1\leqslant s<2\sqrt{M}+1\} . The second part of section 4 gives a brief estimates on how deep roots can get at each angle \theta . In Section 5 we use the work of [Na] to give the following classification for connectedness: Theorem 5.1 . If S is a normalized symmetric set of integers containing 0 then: (1) If the total gap of S is 3 or more then \Sigma_{S} is disconnected. (2) If the