Communication cost of simulating entanglement swapping

Nicolas Brunner, Cyril Branciard, Nicolas Gisin, Denis Rosset

Published 2011-03-25, updated 2012-09-25Version 3

Entanglement appears in two different ways in quantum mechanics, namely as a property of states and as a property of measurement outcomes in joint measurements. By combining these two aspects of entanglement, it is possible to generate nonlocality between particles that never interacted, using the protocol of entanglement swapping. We investigate the communication cost of classically simulating this process. While the communication cost of simulating nonlocal correlations of entangled states appears to be generally quite low, we show here that infinite communication is required to simulate entanglement swapping. This result is derived in the scenario of bilocality, where distant sources of particles are assumed to be independent, and takes advantage of a previous result of Massar et al. [Phys. Rev. A {\bf 63}, 052305 (2001)]. Our result implies that any classical model simulating entanglement swapping must either assume that (i) infinite shared randomness is available between any two locations in the universe, or that (ii) infinite communication takes place.