Vector subdifferentials via recession mappings^∗^∗this research was supported by grants from M.U.R.S.T (Itall) and the australian research council.$ef: (Q4351789)

A vector subdifferential is defined for a class of directionally differentiable mappings \(f: X\to Y\), where \(X\) and \(Y\) are infinite-dimensional, and \(Y\) is an order-complete vector lattice. The construction is based on the deconvolution \[ (f^*_-g)(x):= \sup\{f(x+ d)- g(d): d\in X\}; \] note that \(^*f(x):= (f^*_-f)(x)\) is the recession function. Then the recession subdifferential is \[ \partial^*f(x):= \{A\in L(X, Y): (\forall v\in X)Av\leq{^*F_x}(v)\}, \] where the directional derivative \(F_x(d):= f'(x, d)\) is positively homogeneous. A calculus is developed for these subdifferentials; they are also shown to be equivalent to subdifferentials of Nesterov type. They are applied to obtain first-order necessary optimality conditions for a class of vector optimality problems, involving either proper or weak minimality concepts.