A sharp logarithmic condition for the Hardy operator on L 1 ( 0 , ∞ ) L^{1}(0,\infty) and ℓ 1 \ell^{1}
Samson Owusu-Ensaw, Benoît F. Sehba, Ransford T. Tweneboanah · 2026
Read A sharp logarithmic condition for the Hardy operator on L 1 ( 0 , ∞ ) L^{1}(0,\infty) and ℓ 1 \ell^{1} by Samson Owusu-Ensaw, Benoît F. Sehba, Ransford T. Tweneboanah (2026) 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-2603-21252.txt.
Introduction
Owusu-Ensaw and co-authors characterize where a modified Hardy operator stays bounded on integrable functions. The classic Hardy operator is not bounded on L1 of the positive half-line, nor is its discrete form on summable sequences. They subtract a natural corrective term to build a modified operator. They then find the largest subspace on which it maps integrable functions to integrable functions. The sharp condition is a logarithmic integrability requirement, whose weight reflects obstructions at both small and large scales.
Catalog record
| CREATOR | Samson Owusu-Ensaw, Benoît F. Sehba, Ransford T. Tweneboanah |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21253 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.21252
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A sharp logarithmic condition for the Hardy operator on L 1 ( 0 , ∞ ) L^{1}(0,\infty) and ℓ 1 \ell^{1} Samson Owusu-Ensaw, Benoît F. Sehba, Ransford T. Tweneboanah Abstract . The Hardy operator is not bounded on the space of integrable functions on the positive half-line and its discrete counterpart on summable sequences. we introduce a modified Hardy operator obtained by subtracting a natural corrective term, and characterize the largest subspace of integrable functions on which this modified operator maps into integrable functions. The sharp condition is a logarithmic integrability (summability) requirement whose weight reflects obstructions on both small and large scales. 1. Introduction The operator that bears Hardy’s name arose in his 1920 paper Note on a theorem of Hilbert [ 5 ] , in which Hardy gave an elegant proof of a result on double series due to Hilbert. The continuous analogue appeared in explicit form in a 1925 paper [ 6 ] , where Hardy proved: \int_{0}^{\infty}\!\left(\frac{1}{x}\int_{0}^{x}f(t)\,dt\right)^{p}\!\,dx\;\leq\;\left(\frac{p}{p-1}\right)^{p}\int_{0}^{\infty}f(x)^{p}\,dx,\qquad 1<p<\infty,\;f\geq 0. The constant (p/(p-1))^{p} is sharp. Hardy attributed the problem to Landau and Riesz; the result was refined and extended by Hardy, Littlewood, and Pólya in their celebrated monograph [ 8 ] . The above inequality is known as Hardy’s inequality, for more on this inequality, its discrete version, and related topics, we refer the interested reader to [ 9 , 10 , 11 , 12 ] . The continuous Hardy operator is defined for measurable functions f by (1) \mathcal{Q}f(x)=\frac{1}{x}\int_{0}^{x}f(t)\,dt,\qquad x>0. We recall that for 1\leq p<\infty , the Lebesgue space L^{p}(0,\infty) consists of all measurable functions f on (0,\infty) such that \|f\|_{L^{p}}:=\left(\int_{0}^{\infty}|f(x)|^{p}\,dx\right)^{\frac{1}{p}}<\infty. The operator \mathcal{Q} is not bounded from L^{1}(0,\infty) into itself. In effect, consider for example the function f_{0}(x)=\frac{\chi_{[1,2]}(x)-\chi_{[3,4]}(x)}{x}. It is easy to see that f_{0}\in L^{1}(0,\infty) and that \mathcal{Q}f_{0}(x)=\begin{cases}0&\text{if }0<x\leq 1\\ \frac{\ln x}{x}&\text{if }1<x\leq 2\\ \frac{\ln 2}{x}&\text{if }2<x\leq 3\\ \frac{\ln 6-\ln x}{x}&\text{if }3<x\leq 4\\ \frac{\ln(3/2)}{x}&\text{if }x>4.\end{cases} It clearly follows that \mathcal{Q}f_{0}\notin L^{1}(0,\infty) . In [ 7 ] , Hardy and Littlewood made the same observation. They then proved that nevertheless, the following local result holds. Theorem 1.1 . [ 7 , Theorem 11] Let a>0 . Assume that f is a positive measurable function defined on (0,\infty) . Then the following hold. (i) If \int_{0}^{a}f(x)\log^{+}f(x)dx<\infty , then \int_{0}^{a}\mathcal{Q}f(x)dx\lesssim 1+\int_{0}^{a}f(x)\log^{+}f(x)dx. (ii) If f is decreasing and K:=\int_{0}^{a}\mathcal{Q}f(x)dx<\infty , then \int_{0}^{a}f(x)\log^{+}f(x)dx\lesssim 1+K\log^{+}K. We recall that for x>0 , \log^{+}x=\max\{0,\log x\} . The notation \log is used for function while \ln is used when the argument is a number. We aim in this note to consider a discussion of the case where the function under consideration is supported throughout the interval (0,\infty) and the integration of its transform is also considered throughout (0,\infty) . We also discuss its discrete analogue. To be more precise on the problem in consideration, knowing that \mathcal{Q} is not bounded from L^{1}((0,\infty)) to itself, we ask the following question: what is the maximal subspace of L^{1}(0,\infty) in which we can choose f so that \mathcal{Q}f\in L^{1}(0,\infty) . Our first observation will be that for an integrable function f , \mathcal{Q}f\in L^{1}(0,\infty) only if the integral of f is zero. This means in particular that for a non-negative integrable function f , we cannot have \mathcal{Q}f\in L^{1}(0,\infty) . Thus, for the problem to still be considered for functions f in L^{1}(0,\infty) , there is a need to perturb f to obtain a function \tilde{f} with the properties: \tilde{f}\in L^{1}(0,\infty) and \int_{0}^{\infty}\tilde{f}(x)dx=0 . We will see that this is equivalent (for an appropriate perturbation) to considering the boundedness on L^{1}(0,\infty) of the operator \mathcal{H}f(x)=\mathcal{Q}f(x)-\frac{1}{x+1}\int_{0}^{\infty}f(t)dt. Our result can be stated as follows: If f is non negative and integrable, then \mathcal{H}f\in L^{1}((0,\infty)) if and only if \int_{0}^{\infty}f(x)\ln\left(2+x+\frac{1}{x}\right)dx<\infty, with equivalence of the norms. One may ask if there is any relation between the Hardy and Littlewood local result and ours. The answer is quite simple: their result is not only local but also, we are not considering the same operator. Also, because of the necessity for an integrable function to have mean zero for its image by the Hardy’s operator to be in L^{1}(0,\infty) , one cannot expect Theorem 1.1 to be global. Let us say a few words about the discrete results. The discrete Hardy operator \Gamma acts on a sequence a=(a_{k})_{k\geq 1} of real numbers by forming its running Cesàro means, (\Gamma a)_{n}=\frac{1}{n}\sum_{k=1}^{n}a_{k},\qquad n\geq 1. The operator \Gamma is bounded on \ell^{p} (the space of p -summable sequences) for every 1<p<\infty with the sharp constant (p/(p-1))^{p} (the discrete counterpart of Hardy’s classical inequality). As in the continuous case, the boundedness at the endpoint p=1 fails: the sequence (1/(k(k+1)))_{k} is summable, yet its image under \Gamma has general term 1/(n+1) , which is not summable. The first discrete result of this paper shows that this failure is not an accident of the example: if a summable sequence a has a summable image \Gamma a , then the sum of a must be zero. This necessary condition motivates the introduction of the discrete modified operator (\widetilde{\Gamma}a)_{n}=\frac{1}{n}\sum_{k=1}^{n}a_{k}-\frac{1}{n+1}\sum_{k=1}^{\infty}a_{k},\qquad n\geq 1, obtained by subtracting the corrective term \frac{1}{n+1}\sum_{k=1}^{\infty}a_{k} , in exact analogy with the continuous construction. The discrete main result identifies the sharp condition under which \widetilde{\Gamma}a\in\ell^{1} : for a sequence of positive terms, this holds if and only if \sum_{k=1}^{\infty}a_{k}\ln(k+1)<\infty . The logarithmic weight \ln(k+1) is the discrete counterpart of the weight \ln(1+t) that governs the large-scale behavior in the continuous result; note that, unlike the continuous case, there is no small-scale logarithmic obstruction for sequences, since the index k is bounded away from zero. The proof relies on a splitting of \widetilde{\Gamma}a into two parts, analogous to the decomposition used in the continuous case, and on a comparison between partial harmonic sums and their logarithmic asymptotic via the Euler–Mascheroni constant. As a consequence, the \ell^{1} -norm of \widetilde{\Gamma}a is equivalent to \gamma\sum_{k}a_{k}+\sum_{k}a_{k}\ln(k+1) , providing a quantitative version of the characterization. We are motivated not only by Hardy-Littlewood local result Theorem 1.1 , but also by some recent and old results of the same vein obtained for some other operators: the Hilbert’s integral operator, the Bergman projection, and the maximal functions (see [ 1 , 2 , 3 , 4 , 13 ] ). In the next section, we study the problem for the continuous Hardy operator while in Section 3, we discuss the discrete analogue results. Throughout this text, for two positive quantities A and B , the notation A\lesssim B (resp. A\gtrsim B ) means that A\leq CB ( A\geq CB ) for some universal constant C>0 . If both A\lesssim B and B\lesssim A hold, we write A\approx B . 2. The result for the continuous Hardy operator The key observation is that \mathcal{Q}f\notin L^{1}((0,\infty)) whenever \int_{0}^{\infty}f\,dt\neq 0 , because the integrand behaves like a positive constant divided by x near infinity. The following proposition makes this precise. Proposition 2.1 . Suppose f\in L^{1}(0,\infty) and \mathcal{Q}f\in L^{1}(0,\infty) . Then (2) \int_{0}^{\infty}f(t)\,dt=0. Proof. For x>0 , multiply both sides of ( 1 ) by x : (3) x\,\mathcal{Q}f(x)=\int_{0}^{x}f(t)\,dt. Since f\in L^{1}((0,\infty)) , the right-hand side of ( 3 ) converges to \int_{0}^{\infty}f(t)\,dt as x\to\infty . In particular, (4) \lim_{x\to\infty}x\,|\mathcal{Q}f(x)|=\left|\int_{0}^{\infty}f(t)\,dt\right|=:\ell. Suppose for contradiction that \ell\neq 0 ; without loss of generality assume \ell=1 (replace f by