Optimal deterministic quantum algorithm for the promised element distinctness problem

The element distinctness problem is to determine whether a string of N elements, i.e. x=(x1,…,xN) where xi∈{1,…,M} and M≥N, contains two elements of the same value (a.k.a colliding pair), for which Ambainis proposed an optimal quantum algorithm. The idea behind Ambainis' algorithm is to first reduce the problem to the promised version in which x is promised to contain at most one colliding pair, and then design an algorithm A requiring O(N2/3) queries based on quantum walk search for the promise problem. However, A is probabilistic and may fail to give the right answer. We thus, in this work, design a deterministic quantum algorithm for the promise problem which never errs and requires O(N2/3) queries. This algorithm is proved optimal. Technically, we modify the quantum walk search operator on quasi-Johnson graph to have arbitrary phases, and then use Jordan's lemma as the analyzing tool to reduce the quantum walk search operator to the generalized Grover's operator. This allows us to utilize the recently proposed fixed-axis-rotation (FXR) method for deterministic quantum search, and hence achieve 100% success.