qiskit-experiments
qiskit-experiments copied to clipboard
Qiskit-Extensions
P1 analysis for decay analysis
Summary
This PR upgrades Decay common analysis class to support single point scan.
Details and comments
If one sets the analysis option raw_data_mode=True the analysis class will skip the curve fit and store each single data point as individual analysis result entry under the name "MyExperiment_single_point".
Note that current implementation is really inefficient, because of tight coupling of data processing to analysis. This should be updated after the refactoring of analysis so that it can call data processing and _run_analysis as consecutive callbacks, i.e. separation of processing from analysis.
What's special in Decay compared to other types of analysis? We could have a new analysis class, inheriting directly from CurveAnalysis, that outputs the raw data.
Because decay is monotonic to xval. I've never seen P1 type experiment with non-monotonic data. Probably @oliverdial knows other examples. (If any, it might be better to implement as mix-in)
Non-monotonic decay in T1 traces happens pretty often if you have a near-resonant TLS; P1 will oscillate for the first 10s of us. (Confession; I haven't looked at this PR, just providing the example as per request :) )
Thanks! The background of this work is that some people wanted to monitor only single (or few) point of T1 curve. We have other examples such as Stark shift spectroscopy https://arxiv.org/abs/2105.15201, or survival probability of RB as a cost function of heavy optimization task https://arxiv.org/abs/2003.05952
However I've never seen this approach is used with non-decay type experiment. So I believe it should be sufficient to support this analysis (i.e. omit fit and return data point) only for decay analysis.
