Approximate quantum error correction with covariant codes

[joint work with: Victor Albert, John Preskill (Caltech), Sepehr Nezami, Grant Salton, Patrick Hayden (Stanford University), and Fernando Pastawski (Freie Universität Berlin)]

Quantum error correction and symmetries are concepts that are relevant in many physical systems, such as in condensed matter physics and in holographic quantum gravity, and play a central role in error correction of reference frames [Hayden et al., arXiv:1709.04471]. I will show that codes that are covariant with respect to a continuous local symmetry necessarily have a limit to their ability to serve as approximate error-correcting codes against erasures at known locations. This is because the environment necessarily gets information about the global logical charge of the encoded state due to the covariance of the code. Our bound vanishes either in the limit of large individual subsystems, or in the limit of a large number of subsystems; in either case there exist codes which approximately achieve the scaling of our bound and become good covariant error-correcting codes. Our results can be interpreted as an approximate version of the Eastin-Knill theorem, quantifying to which extent it is not possible to carry out universal computation approximately on encoded states. In the context of holographic AdS/CFT, our approach provides some insight on time evolution in the bulk versus time evolution on the boundary. We expect further implications for symmetries in holography and in condensed matter physics.