Efficient Quantum State Identity Testing
We study the following quantum state identity problem: given access to copies of $n$ unknown pure quantum states, the goal is to determine whether every pair has fidelity at least $1-\varepsilon$ or some pair has fidelity at most $\varepsilon$. This problem relaxes the identical-or-orthogonal promise considered in previous work. For any fixed overlap parameter $\varepsilon\in(0,1/4)$, we present two quantum algorithms with complementary guarantees on circuit depth and sample complexity. The first runs in constant depth using parallel swap tests and takes $O(\log(n/δ)\log(2/δ))$ copies of each input state to succeed with probability at least $1-δ$. The second uses the Schur transform to reduce the sample complexity to $O(\log(n/δ))$, through an analysis relating Jucys--Murphy elements to Schur sampling. We also prove a matching sample complexity lower bound of $Ω(\log(n/δ))$ for any quantum algorithm with failure probability at most $δ\le 1/2-c$, for any constant $c>0$, establishing that the second algorithm achieves the optimal dependence on both $n$ and $δ$ for fixed $\varepsilon$.