A method for converting a class of univariate functions into d. c. functions (Q1304397)

The paper presents a procedure for d.c. representation of univariate functions \(f: [a,b]\subset \mathbb{R}\to \mathbb{R}\) of the class P-C\(^2\) on \([a,b]\) which consists, by definition, of functions \(f(x)\) such that: 1) \(f''(x)\) is continuous for all but finitely many points in the open interval \((a,b)\); 2) \(f_-'(x)\) and \(f_x(x)'\) are finite everywhere in \((a,b).\) By d.c. representation of a function \(f:[a,b]\to \mathbb{R}\) is meant a couple of convex functions \(p(x), q(x)\) on \([a,b]\) such that \(f(x)= p(x)-q(x)\;\forall x\in [a,b].\) Results similar to those presented in this paper have been obtained earlier, see the reviewer's book [Convex analysis and global optimization (Kluwer, 1998; Zbl 0904.90156)].