Generalizations of Aczél's inequality and Popoviciu's inequality (Q2565417): Difference between revisions

The authors note that the inequality \[ \biggl(a_1^p-\sum^n_{j=2} a_j^p\biggr) \biggl(b_1^p-\sum^n_{j=2} b_j^p\biggr)\leq \biggl(a_1 b_1-\sum^n_{j=2} a_j b_j\biggr)^p \] does not always hold for \(p\geq 1\) (in particular not always for \(p>2\)) under the assumptions \(a_j>0,\, b_j>0\;(j=1,\dots,n),\;a_1^p-\sum^n_{j=2} a_j^p >0,\;b_1^p-\sum^n_{j=2} b_j^p>0\) stated in ``Analytic inequalities'' (1970; Zbl 0199.38101), pp. 58--59, of \textit{D. S. Mitrinović}. (Note: This error has been noticed also by \textit{M. Bjelica} [Math., Rev. Anal. Numér. Théor. Approximation, Anal. Numér. Théor. Approximation 19, 105--109 (1990; Zbl 0733.26011)], and by \textit{L. Losonczi} and \textit{Zs. Páles} [J. Math. Anal. Appl. 205, No. 1, 148--156 (1997; Zbl 0871.26012)]. The latter also generalized the corrected inequality. The present authors offer among others the inequality \[ \biggl(a_1^p-\sum^n_{j=2} a_j^p\biggr) \biggl(b_1^p-\sum^n_{j=2} b_j^p\biggr)\leq \biggl(n^{1-\min\{2/p,1\}}a_1 b_1-\sum^n_{j=2} a_j b_j\biggr)^p \] under the above assumptions, and generalizations.