Non-asymptotic entanglement distillation studies the trade-off between three parameters: the distillation rate, the number of independent and identically distributed prepared states, and the fidelity of the distillation. We first study the one-shot ε-infidelity distillable entanglement under quantum operations that completely preserve positivity of the partial transpose (PPT) and characterize it as a semidefinite program (SDP). For isotropic states, it can be further simplified to a linear program. The one-shot ε-infidelity PPTassisted distillable entanglement can be transformed to a quantum hypothesis testing problem. Moreover, we show efficiently computable second-order upper and lower bounds for the non-asymptotic distillable entanglement with a given infidelity tolerance. Utilizing these bounds, we obtain the second order asymptotic expansions of the optimal distillation rates for pure states and some classes of mixed states. In particular, this result recovers the second-order expansion of LOCC distillable entanglement for pure states in [Datta/Leditzky, IEEE Trans. Inf. Theory 61:582, 2015]. Furthermore, we provide an algorithm for calculating the Rains bound and present direct numerical evidence (not involving any other entanglement measures, as in [Wang/Duan, Phys. Rev. A 95:062322, 2017]), showing that the Rains bound is not additive under tensor products.
I. INDRODUCTION
A. BackgroundQuantum entanglement is a striking feature of quantum mechanics and a key ingredient in many quantum information processing tasks, including teleportation [1], superdense coding [2], and quantum cryptography [3,4]. All these protocols necessarily rely on entanglement resources, especially the maximally entangled states. It is thus of great importance to develop entanglement distillation protocols to transform less useful entangled states into more suitable ones such as maximally entangled states. In general, the task of entanglement distillation aims at obtaining maximally entangled states from less-entangled bipartite states shared between two parties (Alice and Bob) and it allows them to perform local operations and classical communication (LOCC). The concept of distillable entanglement characterizes the rate at which one can asymptotically obtain maximally entangled states from a collection of identically and independently distributed (i.i.d) prepared entangled states by LOCC [5,6]. Distillation from non-i.i.d prepared states has also been considered recently [7]. Distillable entanglement is a fundamental entanglement measure which captures the resource character of entanglement. Up to now, how to calculate distillable entanglement for gen-* Electronic address: kun.fang-