Means to an End

From MaRDI portal





This is a joint review with [\textit{R. P. Kubelka}, Missouri J. Math. Sci. 15, No. 2 (2003; Zbl 1097.26020)]. The ratio \((A_n(\underline a)/G_n(\underline a))^n\) is greater than \(1\) and increases with \(n\) and so has a limit. If this limit is finite then the simpler expression \(A_n(\underline a)/G_n(\underline a)\) tends to \(1\); the question remains what happens if the limit is infinite. A simple observation based on the Stirling formula shows that if \(a_n = n, n\geq 1\), then \(A_n(\underline a)/G_n(\underline a)\to e/2\). The author, having made this observation, extends this to the case \(a_{n} = n^{s}, n\geq 1,\, s\in {\mathbb R}\), when the limit is shown to be \(e^s/(s+1)\). In the second paper the same result is shown to hold if \( a_n\) is a positive polynomial in \(n\), and also to hold if the means are integral means. The methods of proof are elementary being based on Riemann sums and the asymptotic behaviour of polynomials..











This page was built for publication: Means to an End

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2747360)