Residual spectral algorithm for solving monotone equations on a Hilbert space (Q2444223)

The paper describes an algorithm to numerically solve a nonlinear monotone equation on a Hilbert space. The new algorithm generates an iterations from an initial point that can be easily implemented in practice because of its simplicity for building the search direction. The step-length is computed in order to guarantee the convergence of the method. In addition, the algorithm can be used in the non-differentiable case. As an application, the algorithm is used for solving the Lyapunov matrix equation. In this case, the method is very competitive because it needs low storage memory and it does not require large matrix factorizations. The paper ends with some numerical experiments for a set of test problems to illustrate the efficacy of the proposed algorithms.