A successful state transfer (or teleportation) experiment must perform better than the benchmark set by the `best' measure and prepare procedure. We consider the benchmark problem for the following families of states: (i) displaced thermal equilibrium states of given temperature; (ii) independent identically prepared qubits with completely unknown state. For the first family we show that the optimal procedure is heterodyne measurement followed by the preparation of a coherent state. This procedure was known to be optimal for coherent states and for squeezed states with the `overlap fidelity' as figure of merit. Here we prove its optimality with respect to the trace norm distance and supremum risk. For the second problem we consider n i.i.d. spin-1/2 systems in an arbitrary unknown state $\rho$ and look for the measurement-preparation pair $(M_{ n},P_{ n})$ for which the reconstructed state $\omega_{ n}:=P_{ n}\circ M_{ n} (\rho^{ \otimes n})$ is as close as possible to the input state, i.e. $\|\omega_{ n}- \rho^{ \otimes n}\|_{ 1}$ is small. The figure of merit is based on the trace norm distance between input and output states. We show that asymptotically with $n$ the this problem is equivalent to the first one. The proof and construction of $(M_{ n},P_{ n})$ uses the theory of local asymptotic normality developed for state estimation which shows that i.i.d. quantum models can be approximated in a strong sense by quantum Gaussian models. The measurement part is identical with `optimal estimation', showing that `benchmarking' and estimation are closely related problems in the asymptotic set-up.
Categories: quant-ph cond-mat.stat-mech physics.atom-ph