No label defined (Q4520097)

As the title shows, the book is about mean value theorems and functional equations associated to them. The best known mean value theorem may be that of Lagrange, stating that, if a real valued function \(f\) is defined and continuous on the nondegenerate real interval \([x_1,x_2]\) and differentiable on \(]x_1,x_2[,\) then there exists an \(\eta(x_1,x_2)\in ]x_1,x_2[\) such that \(\frac{f(x_1)-f(x_2)}{x_1 -x_2}=f'(\eta(x_1,x_2)).\) While the statement that \(\eta\) is in \(]x_1,x_2[\) is missing from Theorem 2.2 [and so is the continuity assumption], it is included elsewhere. The simplest related functional equation question is: for what functions \(f: I\to\mathbb{R}\) does the above equality hold with \(\eta(x_1,x_2)= \frac{x_1+x_2}{2}\) for all \(x_1\neq x_2\) in a proper real interval \(I.\) The answer is that this characterizes the polynomials of at most second degree. That remains true even if no differentiability, continuity, or other regularity assumption is made and \(f'\) is replaced by a second unknown function which can then absorb the denominator 2, so that the functional equation is now \(f(x_1)-f(x_2)=(x_1 -x_2) g(x_1+x_2)\) which (trivially) holds also for \(x_1=x_2.\) [Of course, \(x_1 +x_2\) is in general not in \(]x_1,x_2[\) anymore, even though the book often still refers to it or its two dimensional analogue as ``mean'' or ``intermediate'' value (e.g., pp. 78, 128-129, 144)]. Many other mean value theorems (Cauchy, Pompeiu, Flett, Trahan, integral, etc.), also for functions of complex or vector variables, and several functional equations associated to many of them are offered. The book presents many quite recent results but is, as the authors repeatedly point out, introductory. Indeed it should be understandable on sophomore (second year) level (though the introduction mentions ``advanced undergraduate and beginning graduate students''). Accordingly, most proofs (often lengthy) are given in full detail, mostly without reference to others. On the positive side, several elementary open problems are listed at the end of almost every chapter. [Remarks: (i) References are numbered in the bibliography but quoted just by authors and year in the rest of the book. That may not be the present authors' fault; however, there are references which do not appear in the bibliography at all, for example Schwaiger (1995) on p. 57 and Kuczma (1989) on p. 91. (ii) As domain or in conditions, often \(\mathbb{R}\) is used instead of finite intervals. That sometimes leads to unwanted results. For example, Theorem 1.7 asserts that real valued additive functions bounded from one side on \(\mathbb{R}\) are linear, which is technically correct but is satisfied only by the identically zero function. (iii) The functional equation solved in Corollary 2.2 is certainly not (2.11) quoted therein. (iv) On p. 186 the authors apparently assert that ``if the limit of \(f\) exists as \(x\) approaches 0 then \(f\) is symmetrically differentiable at 0''. If so, then the function \(f\) defined by \(f(x)=x+x^{1/2}\) for \(x\geq 0\) and by \(f(x)=-x-(-x)^{1/2}\) for \(x\leq 0\) would seem to be a counter-example. (v) Finally a quote from p. 11 without comment: ``No concrete example of a Hamel basis is known, we only know that it exists. The graph of a discontinuous additive function is not easy to draw ...''].