Bell inequalities with communication assistance

In this paper, we consider the possible correlations between two parties using local machines and shared randomness with an additional amount of classical communication. This is a continuation of the work initiated by Bacon and Toner [Phys. Rev. Lett. 90, 157904 (2003)] who characterized the correlation polytope for $2\times 2$ measurement settings with binary outcomes plus one bit of communication. Here, we derive a complete set of Bell inequalities for $3\times 2$ measurement settings and a shared bit of communication. When the communication direction is fixed, nine Bell inequalities characterize the correlation polytope, whereas when the communication direction is bidirectional, 143 inequalities describe the correlations. We then prove a tight lower bound on the amount of communication needed to simulate all no-signaling correlations for a given number of measurement settings.

Figure 1

Alice and Bob have respective black boxes ${\mathcal{M}}_A$ and ${\mathcal{M}}_B$. The input-output correlations are given by the conditional probabilities $p\left(ab\right|ij)$.

Figure 2

(a) An Alice $\to$ Bob fixed-direction communication scheme in which $\kappa \left(i\right)$ is sent from Alice to Bob. (b) A bidirectional communication scheme sending data in both directions. Note that model (b) contains both Alice $\to$ Bob and Bob $\to$ Alice fixed-direction communication schemes.