Page 11 - Demo

Multiple shots are needed to complete the synthetic ensemble for reconstruction
• From one shot, a reconstruction would be impossible – it would be subject to the same errors as the NPA itself!
• For a set of shots, however, the errors “average away”, making a reconstruction possible
• Vary 𝐵()* between shots as in deterministic calibration (upper plot)
• Assuming all shots are sampling the same distribution, place them into a single ensemble, “collapsing” the 𝐵()* dimension
• This step is crucial – it helps the reconstruction “average away” the systematic errors from un- calibrated channels
• Separate out some data for testing – needed for validating the source reconstruction
Presenter: Gabriel Player
11