rigetti / forest-benchmarking

A library for quantum characterization, verification, validation (QCVV), and benchmarking using pyQuil.
Apache License 2.0
52 stars 23 forks source link

Tomography should account for readout imperfections #87

Open marcusps opened 5 years ago

marcusps commented 5 years ago

We should not assume readout to be perfect, and we should offer at least 2 possibilities:

  1. use the readout error mitigation in PyQuil's operator estimation
  2. estimate the POVMs and use them to update the estimation process

Generally 2 can be done in an ad hoc way (pretend that those POVM estimate are error free), but we can also account for that in some reconstruction methods (linear GST, for example).

marcusps commented 5 years ago

@joshcombes @kylegulshen Thoughts?

joshcombes commented 5 years ago

Good idea. I'd recommend for it not to be on by default. Lets make sure people understand how bad things are by default.