A quantum game of telephone

As quantum networks scale, it becomes increasingly difficult to equip every node with the local entanglement sources required for standard Ancilla-Assisted Process Tomography (AAPT). This paper investigates whether one can characterize unknown quantum channels in a multinode network by using previously probed links as ancillae for subsequent ones, effectively creating a 'quantum game of telephone.'

The authors implement a three-node quantum network using polarization-entangled photon pairs distributed via fiber. They compare two inference strategies:

1. Sequential Inference: A step-by-step approach where each channel is estimated using the results of the previous one.
2. Global Inference: A simultaneous estimation method that evaluates all network measurements at once, enforcing cross-measurement consistency.

They test these methods by introducing controlled noise (depolarization) into the first channel and observing how well the protocol recovers the characteristics of both the noisy upstream channel and the subsequent, near-unitary channels.

The study reveals that sequential estimation is highly susceptible to error compounding, where noisy channels create parameter degeneracies that obscure the true state of the network. However, the global inference approach provides a significant advantage: by incorporating downstream near-unitary channels as boundary constraints, the protocol can 'backpropagate' information to resolve ambiguities in earlier, noisier channels. This allows for accurate characterization even when intermediate states are completely depolarized, effectively bypassing the strict full-rank requirements of standard AAPT.