1: $, and thus guarantees the convergence to a unique optimum even when $