This paper proposes a universal distortion-free entanglement concentration protocol. The proposed protocol uses only local operations and no classical communication, but still achieves optimality in such strong senses
Entanglement concentration from many copies of unknown pure states is discussed, and we propose the protocol which not only achieves entropy rate, but also produces the perfect maximally entangled state. Our protocol is induced naturally from symmetry of n-tensored pure state, and is optimal for all the protocols which concentrates entanglement from unknown pure states, in the sense of failure probability. In the proof of optimality, the statistical estimation theory plays a key role, for concentrated entanglement gives a natural estimate of the entropy of entanglement.
Using invariance of the n-th tensored state w.r.t. the n-th symmetric group, we propose a 'variable length' universal entanglement concentration without classical communication. Like variable length data compression, arbitrary unknown states are concentrated into perfect Bell states and not approximate Bell states and the number of Bell states obtained is equal to the optimal rate asymptotically with the probability 1. One of the point of our scheme is that we need no classical communication at all. Using this method, we can construct a universal teleportation and a universal dense coding.
We propose a multipartite extension of Matsumoto and Hayashi's distortion-free entanglement concentration protocol, which takes n copies of a general multipartite state and, via local measurements and produces a maximally entangled multipartite state between local spaces of dimensions $\ensuremath{\sim}{2}^{n{E}_{i}}$, where Ei are the local entropies of the input state. However, the extended protocol is generally not universal in the sense that for the same measurement outcomes, the output state will still depend on the input state. Our main result is that when specialized to any state in the multiqubit W class, the protocol is also universal, so that as in the biparatite version, the output is a unique, maximally entangled state for each given set of measurement outcomes. Our analysis brings to the forefront an interesting family of maximally entangled multipartite states, which we term Kronecker states. A recurrence relation to obtain the coefficients of the W-class Kronecker states is also given.