Uniform results in polynomial-time security

Abstract

Most security results can be established both in the non-uniform and the uniform model of computation. Nonetheless, non-uniform results are often much easier to obtain than their uniform version. In this paper we initiate a general framework in which the classical sampling technique can be applied to obtain uniform results. Our main theorem gives sufficient conditions under which a non-uniform result can be extended to a uniform one. As a consequence, we derive the uniform version of Schrift and Shamir’s generalization of Yao’s theorem on the universality of the next-bit test.