On the arithmetic and harmonic averages in nondiscrete setting (Q2853299)

Let for \(x\left( t\right) \in L_{\log }^{1}\left(\mathbb{R}_{+}\right) \) NEWLINE\[NEWLINE \sigma \left( T\right) :=\frac{1}{T}\int_{0}^{T}x\left( t\right) dt\text{, \;\;}T>0, NEWLINE\]NEWLINE and for \(t^{-1}x\left( t\right) \in L_{\log }^{1}\left( [1,\infty )\right) \) NEWLINE\[NEWLINE \tau \left( T\right) :=\frac{1}{\log T}\int_{1}^{T}\frac{x\left( t\right) }{t }dt\text{, \;\;}T>1. NEWLINE\]NEWLINE The author proves the following theorems:NEWLINENEWLINE{Theorem 1.} The finite limit NEWLINE\[NEWLINE \lim_{T\rightarrow \infty }\sigma \left( T\right) =\xi \tag{1}NEWLINE\]NEWLINE exists if and only if NEWLINE\[NEWLINE \lim_{T\rightarrow \infty }\frac{1}{T}\int_{T}^{2T}x\left( t\right) dt=\xi . NEWLINE\]NEWLINENEWLINENEWLINE{Theorem 2.} The finite limit NEWLINE\[NEWLINE \lim_{T\rightarrow \infty }\tau \left( T\right) =\xi \tag{2} NEWLINE\]NEWLINE exists if and only if NEWLINE\[NEWLINE \lim_{T\rightarrow \infty }\frac{1}{T}\int_{e^{T}}^{e^{2T}}\frac{ x\left( t\right) }{t}dt=\xi . NEWLINE\]NEWLINENEWLINENEWLINELet a function \(x\left( t\right) \) be such that \(\left( t\log t\right) ^{-1}x\left( t\right) \in L_{\log }^{1}\left( [e,\infty )\right) \) and let NEWLINE\[NEWLINE \tau _{2}\left( T\right) :=\frac{1}{\log \log T}\int_{e}^{T}\frac{x\left( t\right) }{t\log t}dt\text{, \;\;}T>e. NEWLINE\]NEWLINE The author obtains the following results:NEWLINENEWLINE{Theorem 3.} The finite limit NEWLINE\[NEWLINE \lim_{T\rightarrow \infty }\tau _{2}\left( T\right) =\xi \tag{3}NEWLINE\]NEWLINE exists if and only if NEWLINE\[NEWLINE \lim_{T\rightarrow \infty }\frac{1}{T} \int_{e^{e^{T}}}^{e^{e^{2T}}}\frac{x\left( t\right) }{t\log t}dt=\xi . NEWLINE\]NEWLINE {Theorem 4.} Assume that \(x\left( t\right) \in L_{\log }^{1}\left(\mathbb{R}_{+}\right) \) and \(t^{-1}x\left( t\right) \in L_{\log }^{1}\left( [1,\infty )\right) \). Then the implication \(\left( 1\right) \Longrightarrow \left( 2\right) \) holds true. The converse implication is not true in general.NEWLINENEWLINE{Theorem 5.} Assume that \(t^{-1}x\left( t\right) \in L_{\log }^{1}\left( [1,\infty )\right) \) and \(\left( t\log t\right) ^{-1}x\left( t\right) \in L_{\log }^{1}\left( [e,\infty )\right) \). Then the implication \(\left( 2\right) \Longrightarrow \left( 3\right) \) holds true. The converse implication is not true in general.