Petz map and Python’s lunch

Abstract

We look at the interior operator reconstruction from the point of view of Petz map and study its complexity. We show that Petz maps can be written as precursors under the condition of perfect recovery. When we have the entire boundary system its complexity is related to the volume/action of the wormhole from the bulk operator to the boundary. When we only have access to part of the system, Python’s lunch appears and its restricted complexity depends exponentially on the size of the subsystem one loses access to.