A smoothing method for second order cone complementarity problem (Q1019772)

The second-order cone complementarity problem is solved. The problem is stated as follows: \[ \text{Find }z\in\mathbb{R}^n\text{ such that }\langle f(z),z\rangle= 0\text{ and }K,\;z\in K, \] where \(\langle.\rangle\) represents the Euclidean inner product, \(f: \mathbb{R}^n\to\mathbb{R}^n\) is a continuously differentiable mapping, \(K= K^{n(1)}\times\cdots\times K^{n(m)}\) with \(n(1)+\cdots+ n(m)= n\) and \(K^{n(i)}\), \(i= 1,\dots,m\) are second-order cones defined as follows: \[ K^{n(i)}= \{z_1,z_2^T)^T\in \mathbb{R}\times R^{n(i)- 1}; z_1\geq\| z_2\|\} \] with \(\| z_2\|\) denoting the Euclidean norm of \(z_2\). It is assumed that \(K^1\) is the nonnegative orthant in \(\mathbb{R}\). A smoothing Newton-type algorithm is proposed for solving the problem, its local and global convergence is investigated. Numerical results showing the effectiveness of the proposed algorithm are reported in the concluding part of the paper.