Absolute and Unconditional Convergence of Series of Ergodic Averages and Lebesgue Derivatives
Bryan Johnson, Joseph Rosenblatt · 2025
Read Absolute and Unconditional Convergence of Series of Ergodic Averages and Lebesgue Derivatives by Bryan Johnson, Joseph Rosenblatt (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-2501-09202.txt.
Introduction
Bryan Johnson and Joseph Rosenblatt wrote a math paper in two thousand twenty-five on series built from averages. They ask when such series converge in the strong senses called absolute and unconditional. The series come from processes in ergodic theory and in harmonic analysis. Cases include Cesaro averages and Lebesgue derivatives. The authors sort out which of these series have the strong forms of convergence. The work is a focused piece on convergence for averaged processes.
Catalog record
| CREATOR | Bryan Johnson, Joseph Rosenblatt |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 125 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-09202
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Absolute and Unconditional Convergence of Series of Ergodic Averages and Lebesgue Derivatives Bryan Johnson, Joseph Rosenblatt Abstract . We consider when there is absolute or unconditional convergence of series of various types of stochastic processes. These processes include differences of averages in ergodic theory and harmonic analysis, like the classical Cesàro average in ergodic theory and Lebesgue derivatives in harmonic analysis. 2020 Mathematics Subject Classification Primary: 28D05, 42A61; Secondary: 37A46 1. Introduction Unconditional convergence of a series in a Banach space has a number of different equivalent formulations. We are interested specifically in series of the form \sum\limits_{k=1}^{\infty}M_{n_{k+1}}f-M_{n_{k}}f where \left(n_{k}\right) is a non-decreasing sequence of whole numbers, and M_{n} denotes an average, like the usual average in ergodic theory. It turns out that unconditional convergence of such differences depends on spectral properties of the averages. Also of interest in this paper will be series of the form \sum\limits_{k=1}^{\infty}D_{\epsilon_{k+1}}f-D_{\epsilon_{k}}f where D_{\epsilon}f is defined to be the average of f over a small interval. Section 2 contains some basic definitions and characterizations of unconditional convergence, weak unconditional convergence, and absolute convergence that will be used. It will be shown that for any sequence \left(\epsilon_{k}\right) going to zero, absolute convergence of the series of operators \sum\limits_{k=1}^{\infty}D_{\epsilon_{k+1}}f-D_{\epsilon_{k}}f fails in general whether viewed as operators on L^{2}(\mathbb{R}) or on L^{2}(\mathbb{Z}) . Similarly, it will also be shown that for any sequence of integers \left(n_{k}\right) going to \infty , absolute convergence of the series of operators \sum\limits_{k=1}^{\infty}M_{n_{k+1}}f-M_{n_{k}}f also fails in general. Furthermore, it is shown that for any sequence of integers \left(n_{k}\right) going to infinity, there is an f in L^{1}(X) such that weak unconditional convergence of \sum\limits_{k=1}^{\infty}M_{n_{k+1}}f-M_{n_{k}}f fails in L^{1}(X) . Section 3 contains an interesting spectral characterization for determining when \sum\limits_{k=1}^{\infty}M_{n_{k+1}}f-M_{n_{k}}f is unconditionally convergent for all f in L^{2}(X) . It also contains examples of when this characterization holds and when it fails. Section 4 is about sequences of differences using probability measures. It is noted that there is also a spectral characterization like the one in Section 3 for determining unconditional convergence in this situation. Among other things, it is shown here that if the spectrum of \mu is contained in a Stolz region, then for all f in L^{2}(X) , unconditional convergence of \sum\limits_{k=1}^{\infty}\mu^{n_{k+1}}f-\mu^{n_{k}}f occurs for all sequences \left(n_{k}\right) going to infinity. If the measure \mu is only assumed to be strictly aperiodic however, it turns out that we are only guaranteed unconditional convergence for certain sequences \left(n_{k}\right) going to infinity. There is also a result which shows that if \mu is only assumed to be strictly aperiodic, and \left(w_{k}\right) is any sequence of reals going to infinity, then unconditional convergence of the weighted differences \sum\limits_{k=1}^{\infty}w_{k}(\mu^{n_{k+1}}f-\mu^{n_{k}}f) can fail when 2n_{k}{}^{p}\geq n_{k+1}\geq n_{k}{}^{p} for p>1 . Section 5 shows that the spectral characterizations for unconditional convergence of \sum\limits_{k=1}^{\infty}D_{\epsilon_{k+1}}f-D_{\epsilon_{k}}f for f in L^{2}(\mathbb{T}),L^{2}(\mathbb{R}) and H_{2}(\mathbb{T}) are all the same. Also presented here are examples comparable to those given in Section 2 for which this spectral characterization holds and fails. Lastly, it is noted that if E_{\epsilon}f is defined to be (D_{\epsilon}\circ D_{\epsilon})f , then unconditional convergence of \sum\limits_{k=1}^{\infty}E_{\epsilon_{k+1}}f-E_{\epsilon_{k}}f for f in L^{2}(\mathbb{T}) will occur for all sequences of \epsilon_{k} going to zero. Section 6 uses interpolation to show that in the case of \left(n_{k}\right) being lacunary, not only is \sum\limits_{k=1}^{\infty}M_{n_{k+1}}f-M_{n_{k}}f unconditionally convergent in L^{p} -norm for all f\in L^{p}(X) for p=2 , as is shown in Section 2 , but this also holds for all p,1<p<\infty . In Section 1 , the case of p=1 was already shown to never hold for any increasing n_{k} . It is also shown that for any sequence of positive numbers \left(\epsilon_{k}\right) satisfying \epsilon_{k+1}\leq r\epsilon_{k} where r<1 and for any 1<p<\infty then for any f\in L^{p} , the series \sum\limits_{k=1}^{\infty}D_{\epsilon_{k+1}}f-D_{\epsilon_{k}}f is unconditionally convergent in L^{p} -norm, extending the p=2 version of this which was proved in Section 4 . Section 7 examines what happens when the series have terms like the ergodic averages themselves, in general and for specific dynamical systems. Now the absolute and unconditional convergence is typically only a first Baire category phenomenon. But it is not so clear when unconditional convergence can happen with absolute convergence failing. Section 8 gives a list a few of the questions that have come up when pursuing the research in this article. These questions, and their answers, might bring into sharper focus what some of the issues are that come up when considering the results given here. 2. Unconditional Convergence A series \sum\limits_{k=1}^{\infty}f_{k} in a Banach space X is said to be unconditionally convergent if for all permutations \pi of \{1,2,3,\ldots\} the series \sum\limits_{k=1}^{\infty}f_{\pi(k)} converges in the norm topology. The series is weakly unconditionally convergent if for all elements f^{*} in the dual space of X , the series \sum\limits_{k=1}^{\infty}f^{*}\left(f_{k}\right) converges unconditionally. We have the following two theorems. See Lindenstrauss and Tzafriri [10] and Day [7] for the background information. Theorem 2.1 . For a series \sum\limits_{k=1}^{\infty}f_{k} in a Banach space, the following are equivalent: (a) \sum\limits_{k=1}^{\infty}f_{k} is unconditionally convergent, (b) \quad\sum\limits_{k=1}^{\infty}c_{k}f_{k} converges for all sequences \left(c_{k}\right),\left|c_{k}\right|\leq 1 for all k=1,2,3,\ldots , (c) \quad\lim_{m,n\rightarrow\infty}\sup_{\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=m}^{n}c_{k}f_{k}\right\|=0 . Remark : As a consequence of this, if \sum\limits_{k=1}^{\infty}f_{k} is unconditionally convergent, then we would have that \sup_{M\geq 1}\sup_{\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=1}^{M}c_{k}f_{k}\right\| is bounded and \sup_{\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=1}^{\infty}c_{k}f_{k}\right\| is bounded. ∎ Theorem 2.2 . For a series \sum\limits_{k=1}^{\infty}f_{k} in a Banach space, the following are equivalent: (a) \sum\limits_{k=1}^{\infty}f_{k} is weakly unconditionally convergent, (b) \quad\sup_{M\geq 1}\sup_{\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=1}^{M}c_{k}f_{k}\right\| is bounded. Generally, weakly unconditionally convergent series are not unconditionally convergent. However, the two properties are equivalent in reflexive Banach spaces. In particular, for the reflexive spaces L^{p}(X,m),1<p<\infty , they are equivalent. See for example Theorem 5 in Bessaga and Pelczyński [ 7 ] . Hence, we have the following result. Theorem 2.3 . If \sum\limits_{k=1}^{\infty}f_{k} is a series in L^{p}(X,m),1<p<\infty , then the following are equivalent: (a) \sum\limits_{k=1}^{\infty}f_{k} is unconditionally convergent, (b) \sup_{M\geq 1}\sup_{\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=1}^{M}c_{k}f_{k}\right\|_{p} is bounded. Suppose that \left(T_{k}\right) is a sequence of bounded linear operators on L^{p}(X,m) . Then the above gives this homogeneous version of unconditional convergence. Theorem 2.4 . The following are equivalent when 1<p<\infty : (a) for all f\in L^{p}(X,m),\sum\limits_{k=1}^{\infty}T_{k}f is unconditionally convergent, (b) there is a constant C such that for all f\in L^{p}(X,m) , \sup_{M\geq 1,\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f\right\|_{p}\leq C\|f\|_{p} Proof. Clearly, for a fixed f\in L^{p}(X,m) , (b) implies (b) in Theorem 2.3 , and so \sum\limits_{k=1}^{\infty}T_{k}f is unconditionally convergent by Theorem 2.3 . Conversely, assume (a). Then for each f , \left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f\right\|_{p} is bounded independently of M and \left(c_{k}\right),\left|c_{k}\right|\leq 1 . Consider the set B_{L}=\left\{f\in L^{p}:\sup_{M\geq 1,\left|c_{k}\right|\leq 1}\left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f\right\|_{p}\leq L\right\} This is a closed set in the norm topology because it is the intersection over all M and \left(c_{k}\right),\left|c_{k}\right|\leq 1 of the sets B_{L}\left(M,\left(c_{k}\right)\right)=\left\{f\in L^{p}:\left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f\right\|_{p}\leq L\right\} and each of these sets is closed in the norm topology. So by the Baire Category Theorem, there exists some L_{0} such that B_{L_{0}} has non-empty interior. That is, for some f_{0}\in L^{p} , some \delta_{0}>0 , if \|f\|_{p}\leq\delta_{0} , then f+f_{0}\in B_{L_{0}} . Hence, if \|f\|_{p}\leq\delta_{0} , then for all M\geq 1 and \left(c_{k}\right),\left|c_{k}\right|\leq 1 , \displaystyle\left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f\right\|_{p} \displaystyle\leq\left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f_{0}\right\|_{p}+\left\|\sum\limits_{k=1}^{M}c_{k}T_{k}\left(f+f_{0}\right)\right\|_{p} \displaystyle\leq 2L_{0}. Hence, if 0\neq f\in L^{p} , then for M\geq 1 and \left(c_{k}\right),\left|c_{k}\right|\leq 1 , \left\|\sum\limits_{k=1}^{M}c_{k}T_{k}\left(\frac{f\delta_{0}}{\|f\|_{p}}\right)\right\|_{p}\leq 2L_{0} That is, \left\|\sum\limits_{k=1}^{M}c_{k}T_{k}f\right\|_{p}\leq\left(\frac{2L_{0}}{\delta_{0}}\right)\|f\|_{p} ∎ A series \sum\limits_{k=1}^{\infty}x_{k} in a Banach