1: \begin{abstract}

2: In this paper we consider a non-monotone (mixed) variational inequality model with (nonlinear) convex conic constraints.

3: Through developing an equivalent Lagrangian function-like primal-dual saddle-point system for the VI model in question, we introduce an

4: augmented Lagrangian primal-dual method, to be called ALAVI in the current paper, for solving a general constrained VI model. Under an assumption, to be called the primal-dual variational coherence condition in the paper, we prove the convergence of ALAVI. Next, we show that many existing generalized monotonicity properties are sufficient---though by no means necessary---to imply the above mentioned coherence condition, thus are sufficient to ensure convergence of ALAVI. Under that assumption, we further show that ALAVI has in fact an $o(1/\sqrt{k})$ global rate of convergence where $k$ is the iteration count. By introducing a new gap function, this rate further improves to be $O(1/k)$ if the mapping is monotone. %ergodic rate of convergence respectively.

5: Finally, we show that under a metric subregularity condition, even if the VI model may be non-monotone the local convergence rate of ALAVI improves to be linear. Numerical experiments on some randomly generated highly nonlinear and non-monotone VI problems show practical efficacy of the newly proposed method. 

6: 

7: %even though the mapping is non-monotone.

8: 

9: %primal-dual variational coherent-condition (weaker than monotonicity) for (VIP), we prove that the entire sequence generated by ALAVI converges to one solution of (VIP). Furthermore, we introduce a KKT stationary measure and a new gap function for (VIP), and provide the iteration complexity ($O(1/\sqrt{k})$ nonergodic and $O(1/k)$ ergodic rate). Finally, we show that ALAVI exhibits R-linear convergence rate under metric subregularity of KKT mapping of (VIP).

10: \end{abstract}

11: