Symmetric M-ary phase discrimination using quantum-optical probe states

We present a theoretical study of minimum error probability discrimination, using quantum-optical probe states, of M optical phase shifts situated symmetrically on the unit circle. We assume ideal lossless conditions and full freedom for implementing quantum measurements and for probe-state selectio...

Full description

Bibliographic Details
Main Authors: Shapiro, Jeffrey H., Nair, Ranjith, Yen, Brent J., Guha, Saikat, Pirandola, Stefano
Other Authors: Massachusetts Institute of Technology. Research Laboratory of Electronics
Format: Article
Language:en_US
Published: American Physical Society 2012
Online Access:http://hdl.handle.net/1721.1/73840
https://orcid.org/0000-0002-6094-5861
Description
Summary:We present a theoretical study of minimum error probability discrimination, using quantum-optical probe states, of M optical phase shifts situated symmetrically on the unit circle. We assume ideal lossless conditions and full freedom for implementing quantum measurements and for probe-state selection, subject only to a constraint on the average energy, i.e., photon number. In particular, the probe state is allowed to have any number of signal and ancillary modes and to be pure or mixed. Our results are based on a simple criterion that partitions the set of pure probe states into equivalence classes with the same error probability performance. Under an energy constraint, we find the explicit form of the state that minimizes the error probability. This state is an unentangled but nonclassical single-mode state. The error performance of the optimal state is compared with several standard states in quantum optics. We also show that discrimination with zero error is possible only beyond a threshold energy of (M−1)/2. For the M=2 case, we show that the optimum performance is readily demonstrable with current technology. While transmission loss and detector inefficiencies lead to a nonzero erasure probability, the error rate conditional on no erasure is shown to remain the same as the optimal lossless error rate.