Hyperstability of a functional equation (Q1046931)

The main result of the paper is the following. Let \(\alpha,\varepsilon \in\mathbb{R}\) be fixed, \(\alpha<0\) and \(\varepsilon\geq 0\). Then, the function \(f:\,]0,1[\,\to\mathbb{R}\) satisfies the inequality \[ \left| f(x)+(1-x)^{\alpha}f\left(\frac{y}{1-x}\right)-f(y)-(1-y)^{\alpha} f\left(\frac{x}{1-y}\right)\right| \leq \varepsilon \tag{1} \] for all \((x,y)\in D\doteq \{(x,y)\in\mathbb{R}^{2}:x,y,x+y\in \,]0,1[\,\}\) if, and only if, there exist \(a,b \in\mathbb{R}\) such that \[ f(x)= ax^{\alpha}+b((1-x)^{\alpha}-1). \qquad\qquad (x\in \,]0,1[\,) \] This result has the somewhat surprising consequence that the parametric fundamental equation of information (that can be obtained from \((1)\) with \(\varepsilon=0\) and plays an important role in characterizing information measures) is hyperstable, that is, the solutions of the stability inequality \((1)\) and the parametric fundamental equation of information are the same. As a corollary, it is also proved that the system of equations that defines \(\alpha\)-recursive and semi-symmetric information measures is stable.