Nonnegative inverse eigenvalue problems with partial eigendata (Q664539)

The paper focuses on the inverse problem of constructing a real nonnegative matrix from the given partial eigendata. The canonical form can be stated as follows: \textit{NIEP}: Construct a nontrivial \(n\times n\) nonnegative matrix \(A\) from a set of measured partial eigendata \({(\lambda_k,x_k)}^p_{k=1}\,(p\leq n)\). The first section is of introductory nature. In the second section the authors give a sufficient and necessary condition for the solvability of the nonlinear inverse eigenvalue problem \textit{NIEP} by neglecting the nonnegative requirement and discuss the associated best approximation problem with respect to an 'à priori' analytical nonnegative matrix. In the third section the authors review some preliminary definitions and basic results about nonnegative programming, they reformulate the \textit{NIEP} as a monotone complementary problem (MCP) and propose a generalized Newton-type method for solving a nonsmooth equation equivalent to the MCP, establishing the global and quadratic convergence analysis. In the fourth section they extend the proposed Newton-type method to the symmetric nonnegative inverse problem and to the cases of lower bounds and of prescribed entries. The efficiency of the proposed method for solving the \textit{NIEP} is demonstrated in the fifth section. Ten numerical tests performed using MATLAB 7.10 are presented.