Transferring projective measurement errors between nearby states #
The summed squared distance of two PVMs on a new state is at most twice its old value plus eight times the squared vector distance. PVM normalization removes any dependence on the number of outcomes. The state vectors need not be normalized. Alice and Bob forms expose the estimate in the conventions used by the extracted Pauli guarantees.
Stated in a model (Phase 4 of planning/mipco-track.md): the old state is the model's own, the new
state φ is any vector of its Hilbert space, and the model at the new state is Ψ.withState φ.
Two PVMs have summed squared difference at most four times the state's squared norm, without an outcome-count factor.
A single operator's squared state norm splits with coefficient two when the state vector changes.
Transfer the summed squared difference of two projective measurements to a nearby state. No unit-state assumptions are needed.
Alice's local PVM error transfers between arbitrary vectors.
Transfer a bounded Alice PVM approximation to the extracted state.
Bob's local PVM error transfers with the same constants.
Transfer a bounded Bob PVM approximation to the extracted state.