Abstract
This paper presents a new algorithm for computing the non-Archimedean ε in DEA models. It is shown that this algorithm is polynomial-time of O(n), where n is the number of decision making units (DMUs). Also it is proved that using only inputs and outputs of DMUs, the non-Archimedean ε can be found such that, the optimal values of all CCR models, which are corresponding to all DMUs, are bounded and an assurance value is obtained.