Strategic oscillation

Choose a boundary or paired search regimes, move-selection rules for the relevant sides or directions, and a policy controlling reversal or excursion depth. Drive search toward and beyond a boundary, or rebound from it at specified depths, then reverse or change regimes so subsequent trajectories explore alternative approaches. Glover describes tabu conditions that compel crossings/returns or graduated penalties whose weights change with position and direction; a fixed sine wave or single depth schedule is not required. For the later illustrated feasibility instance, while feasible choose an objectively best improving feasible move when one exists, otherwise an objectively best available move, which may enter infeasibility. While infeasible, restrict to moves that improve the infeasibility measure and select the objectively best among them; if none improves infeasibility, first restrict to the least deterioration of that measure and then select by objective quality. These two ranking levels are not interchangeable. Keep feasible solution quality distinct from temporary infeasible search states.