Grover’s quantum algorithm improves any classical search algorithm. We show how random Gaussian noise at each step of the algorithm can be modeled easily because of the exact recursion formulas available for computing the quantum amplitude in Grover’s algorithm. We study the algorithm’s intrinsic robustess when no quantum correction codes are used, and evaluate how much noise the algorithm can bear with, in terms of the size of the phone book and a desired probability of finding the correct result. The algorithm loses efficiency when noise is added, but does not slow down. We also study the maximal noise under which the iterated quantum algorithm is just as slow as the classical algorithm. In all cases, the width of the allowed noise scales with the size of the phone book as $N^{-2/3}.$

• Received 22 July 1999

DOI:https://doi.org/10.1103/PhysRevA.61.012301