Toland-singer formula cannot distinguish a global minimizer from a choice of stationary points^* (Q4239759)

The paper deals with DC programming problems in \(R^n\). One obtains characterizations of global solutions in the case of difference of coercive differentiable convex functions and in the case of difference of a coercive differentiable convex function and a sublinear one by using excess functions; the excess function of \(f:X\to R\) is defined by \(E_f(x):=f(x)-f'(x,x)\), where \(f'(x,x)\) is the directional derivative of \(f\) at \(x\in X\) in the direction \(x\), supposed to exist. An application is given to the minimization of a quotient of non-negative coercive differentiable convex functions.