1: \begin{abstract}
2: Steepest descent preconditioning is considered for the recently proposed nonlinear generalized minimal
3: residual (N-GMRES) optimization algorithm for unconstrained nonlinear optimization.
4: Two steepest descent preconditioning variants are proposed.
5: The first employs a line search, while the second employs a predefined small step.
6: A simple global convergence proof is provided for the N-GMRES optimization algorithm with the first steepest descent
7: preconditioner (with line search), under mild standard conditions on the objective function and the line
8: search processes. Steepest descent preconditioning for N-GMRES optimization is also motivated by relating it to
9: standard non-preconditioned GMRES for linear systems in the case of a standard quadratic optimization
10: problem with symmetric positive definite operator.
11: Numerical tests on a variety of model problems show that the N-GMRES optimization algorithm is able to very
12: significantly accelerate convergence of stand-alone steepest descent optimization.
13: Moreover, performance of steepest-descent
14: preconditioned N-GMRES is shown to be competitive with standard nonlinear
15: conjugate gradient and limited-memory Broyden-Fletcher-Goldfarb-Shanno methods for the model problems considered.
16: These results serve to theoretically and numerically establish steepest-descent preconditioned N-GMRES as a general optimization method for unconstrained nonlinear optimization, with performance that appears
17: promising compared to established techniques.
18: In addition, it is argued that the real potential of the N-GMRES optimization framework lies in the fact
19: that it can make use of problem-dependent nonlinear preconditioners that are more powerful than steepest descent
20: (or, equivalently, N-GMRES can be used as a simple wrapper around any other iterative optimization process to seek acceleration of that process), and this potential is illustrated with a further application example.
21: \end{abstract}
22: