Maximal monotone operators and saddle functions. I (Q1084650)

The previous author's results [Rev. Roumaine Math. Pures Appl. 30, 823- 836 (1985; Zbl 0598.47059) and Nonlinear Analysis, Theory Methods Appl. 9, 1381-1399 (1985)] are generalized. The monotone operator \(T_ k\subseteq E\times E^*\), \(f\in T_ Kx:=[-f,f]\in \partial K(x,x)\), defined by means of the subdifferential concave-convex saddle function K, is studied. It is proved that a relaxed form of skew-symmetry of K is not only sufficient but also necessary condition of the maximality of \(T_ K\). These considerations are motivated by the fact that the maximal monotone operator A can be written as \(A=T_ K\) where K is a closed and skew-symmetric saddle function. (This result is proved in the 2nd part of the paper.)