English

A Criterion for ${\rm Q}$-tensors

Optimization and Control 2023-04-18 v1

Abstract

A tensor A{\mathcal A} of order mm and dimension nn is called a Q{\rm Q}-tensor if the tensor complementarity problem has a solution for all qRn{\bf q} \in {\mathbb R}^{n}. This means that for every vector q{\bf q}, there exists a vector u{\bf u} such that u0,w=Aum1+q0, and uTw=0{\bf u} \geq {\bf 0},{\bf w} = {\mathcal A}{\bf u}^{m-1}+{\bf q} \geq {\bf 0},~\text{and}~ {\bf u}^{T}{\bf w} = 0. In this paper, we prove that within the class of rank one symmetric tensors, the Q{\rm Q}-tensors are precisely the positive tensors. Additionally, for a symmetric Q{\mathrm Q}-tensor A{\mathcal A} with rank(A)=2rank({\mathcal A})=2, we show that A{\mathcal A} is an R0{\mathrm R}_{0}-tensor. The idea is inspired by the recent work of Parthasarathy et al. \cite{Parthasarathy} and Sivakumar et al. \cite{Sivakumar} on Q{\rm Q}-matrices.

Keywords

Cite

@article{arxiv.2304.08119,
  title  = {A Criterion for ${\rm Q}$-tensors},
  author = {Sonali Sharma and K. Palpandi},
  journal= {arXiv preprint arXiv:2304.08119},
  year   = {2023}
}
R2 v1 2026-06-28T10:08:03.409Z