Solvability of monotone tensor complementarity problems

The tensor complementarity problem is a special instance in the class of nonlinear complementarity problems, which has many applications in multi-person noncooperative games, hypergraph clustering problems and traffic equilibrium problems. Two most important research issues are how to identify the solvability and how to solve such a problem via analyzing the structure of the involved tensor. In this paper, based on the concept of monotone mappings, we introduce a new class of structured tensors and the corresponding monotone tensor complementarity problem. We show that the solution set of the monotone tensor complementarity problem is nonempty and compact under the feasibility assumption. Moreover, a necessary and sufficient condition for ensuring the feasibility is given via analyzing the structure of the involved tensor. Based on the Huber function, we propose a regularized smoothing Newton method to solve the monotone tensor complementarity problem and establish its global convergence. Under some mild assumptions, we show that the proposed algorithm is superlinearly convergent. Preliminary numerical results indicate that the proposed algorithm is very promising.

Automated summary

This paper introduces the tensor complementarity problem and its applications in various fields. By analyzing the structure, a monotone tensor complementarity problem is proposed along with the properties of its solution set, and the conditions for feasibility are given. Furthermore, a regularized smoothing Newton method is proposed to solve the problem, and its global convergence and superlinear convergence are proven.