The assignment-fixing conjecture for quadratic assignment solvers

Let a quadratic assignment problem (QAP) instance be given, and suppose that a few assignments are identified as belonging to a high-quality solution. Consider permanently fixing those assignments and then solving the remaining problem. Assignment-fixing conjecture. If fixing these assignments and solving the remainder is a relatively effective strategy for the solver, then the max–min ant system (MMAS) will perform well relative to the breakout local search algorithm (BMA). The conjecture proposes a possible explanation for differences in algorithm performance across QAP instance classes, but the supplied text does not establish the effectiveness of the strategy or the predicted relative performance.