Nonlinear control simulation on a vector machine (Q1123555)

Numerical solution of large sized nonlinear optimal control problems by vector processing and by scalar processing is considered. The nonlinear problems are solved iteratively by transformation into linear quadratic control problems. Two algorithms are investigated: A Riccati form algorithm and a stacked form algorithm obtained by introducing Lagrange multipliers. Benchmarks are carried out. The results indicate that the Riccati form algorithm is superior to the stacked form algorithm for medium and large sized control problems. Only for small problems the stacked form algorithm has some advantage.