Local uniqueness and convergence of iterative methods for nonsmooth variational inequalities (Q1910059): Difference between revisions

The authors study the nonsmooth variational inequality problem: find an \(x \in K\) such that \[ (y - x)^T F(x) \geq 0 \quad \forall y \in K, \tag{1} \] where \(F : \mathbb{R}^n \to \mathbb{R}^n\) is a locally Lipschitzian and semismooth mapping. Using nonsmooth analysis, they characterize various monotonicity properties of locally Lipschitzian mappings in finite-dimensional spaces and also give some sufficient conditions to guarantee the local uniqueness of solutions to (1). Two local iterative algorithms to solve (1) are presented. One of the algorithms, starting from an \(x_0 \in K\) sufficiently close to the solution \(x^*\), generates the sequence \(\{x_k\}\), where \[ (y - x)^T \bigl( F(x_k) + V_k (x - x_k) \bigr) \geq 0 \quad \forall y \in K, \bigl( V_k \in \partial F (x_k) \bigr) \] and \(\partial F\) denotes the generalized subdifferential of \(F\). It is proved that if \(V^*\) is positive definite for all \(V^* \in \partial F(x^*)\) and \(F\) is semismooth at \(x^*\) then \(\{x_k\}\) converges superlinearly to \(x^*\).