Open nkanazawa1989 opened 2 years ago
@KathrinKoenig @schulta Can you please comment in the issue so that I can assign you?
@HuangJunye Thank you in advance!
@HuangJunye Sorry for being late and thank you in advance
Slides for Checkpoint 1. (Updated on Oct. 7) #12 Benchmarking Crosspoint Calibration.pdf
Graph and report for checkpoint 2:
Boxplot_calibrations.pdf
Checkpoint 2_Report_Benchmarking Crosspoint Calibration.pdf
Slides for final checkpoint part 2 #12_Benchmarking_Crosspoint_Calibration_Results_final.pdf
Description
This project is associated with qubit calibration, qiskit-experiment, and Qiskit pulse.
To fine tune the control pulse parameters, we conventionally use the ping-pong calibration technique. This amplifies the error of control parameters in the form of rotation angle error by repeating the unit sequence. However, the observed rotation angle error cannot be directly converted into the error of control parameter, and we often use empirical formula to update the control parameter, provided that the error is sufficiently small. Thus we often perform a closed-loop type experiment to repeat the whole sequence until the rotation angle error goes below a certain threshold.
On the other hand, we have an alternative technique (sometime) called cross point calibration. Though this is not quite standard in our software stack, it can directly estimate the optimal control parameter at arbitrary precision.
In this project, you benchmark the cross point technique against the conventional ping-poing approach. If you find the technique is worth implementing, this can officially be implemented in the qiskit experiment, or we create a third party library of cross point calibration. Calibration code will be provided.
Mentor/s
@nkanazawa1989, Qiskit Experiment / Terra developer / researcher
Type of participant
Number of participants
1
Deliverable
calibration library, conference presentation (if result is enough scientific and you are interested in, but no guarantee and no budget support)