On the stability of the quadratic mapping in normed spaces (Q6205099)

Modifying \textit{D. H. Hyers}' classical method of studying approximately additive functions [Proc. Nat. Acad. Sci. USA 27, 222-224 (1941; Zbl 0061.264)] the author proves two results concerning the following stability problem for quadratic mappings: Assuming that \(f\) maps a normed space \(E_ 1\) into a Banach space \(E_ 2\) and satisfies the inequality \[ \| f(x+y)+f(x-y)-2f(x)-2f(y)\|\leq\xi+\eta(\| x\|^ \nu+\| y\|^ \nu) \] \((x,y\in E_ 1\backslash\{0\})\) with some \(\xi,\eta\geq 0\) and \(\nu\in\mathbb{R}\) find a (possibly unique) quadratic mapping \(g:E_ 1\to E_ 2\) lying ``not far'' from \(f\). This approach to stability joints those of S. M. Ulam and D. H. Hyers as well as that of \textit{T. M. Rassias} [Proc. Amer. Math. Soc. 72, 297-300 (1978; Zbl 0398.47040)]. The author answers the question in the affirmative in the case where either \(\nu<2\), or \(\nu>2\), \(\xi=0\) and \(f(0)=0\). A modification of \textit{Z. Gajda}'s example [Internat. J. Math. Math. Sci. 14, No. 3, 431-434 (1991; Zbl 0739.39013)] shows that in the critical case \(\nu=2\) the quadratic functional equation is not stable in the considered sense.