Therefore, a common approach is to determine a near-optimal vectoru ̅ by means of the subgradient method, an iterative procedure which generates a sequence {u^0,u^1,…}of non-negative Lagrangian multiplier vectors, where u^0 is arbitrarily defined. For t ≥ 1, each vector u^(n+1) in the sequence is obtained by its predecessor u^t by setting