Note on the geometric-arithmetic mean inequality

From MaRDI portal





The authors prove that the geometric-arithmetic mean inequality is asymptotically equivalent with very high probability. They study the inequality in the form: \[ \sqrt{\frac{1}{n}\sum_{1}^{n}\lambda _{i}^{2}}\geq \left( \prod_{1}^{n}\lambda _{i}\right) ^{1/n},\;\lambda _{i}>0\,. \] For \(x=\left( x_{i}\right) _{1}^{n}\in S^{n-1}\) , i.e., \(\sum_{1}^{n}x_{i}^{2}=1\) , it becomes \(\left( \prod_{1}^{n}\left| x_{i}\right| \right) ^{1/n}\leq \frac{1}{ \sqrt{n}}\) . The reverse inequality is proved for \(x\in S^{n-1}\) as follows: \(\left( \prod_{1}^{n}\left| x_{i}\right| \right) ^{1/n}\geq \frac{\theta }{ \sqrt{n}}\) with the probability above \(1-\left( C/\sqrt{\theta }\right) ^{n}\), for some absolute constant \(C>0\) (\(C=\text{1,6}\) suffices) and \(0<\theta <1\) . For the entire collection see [Zbl 1009.00019].











This page was built for publication: Note on the geometric-arithmetic mean inequality

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4424446)