Discriminator projects in qiskit-experiments

[coruscating]

Description

Three improvements to the basic discriminator in qiskit-experiments:

• Allow for discriminator to output to a mixed state instead of 0 and 1
• Implement LDA and QDA in numpy, without using scikit-learn
• Discriminator for |0>, |1>, and |2> states, allow for the output of excited state promoted readout to be classified

Mentor/s

Helena Zhang (@coruscating)
Qiskit Experiments developer

Type of participant

Knows Python and LDA, QDA

Number of participants

3

Deliverable

Merged pull requests to qiskit-experiments

[HuangJunye]

Note to other advocates: this project is already matched by prior communication between mentor and mentee and so is not open for application.

[HuangJunye]

@obliviateandsurrender Can you please comment in the issue so that I can assign you?

[obliviateandsurrender]

@HuangJunye there you go.

[HuangJunye]

@obliviateandsurrender Can you please upload your presentation here by the end of today? Thank you!

[obliviateandsurrender]

@HuangJunye yes, working on it.

[obliviateandsurrender]

#22 Discriminator projects in qiskit-experiments.pdf

[obliviateandsurrender]

We have implemented two classes: ThreeLevelDiscriminator and ThreeLevelDiscriminatorAnalysis, which would perform the Excited State Promoted (ESP) readout experiments and analysis in the qiskit-experiments module. We are trying to use the currently available functionalities in the modules for its implementation as much as possible to keep things modular, namely, using QubitSpectroscopy and EFSpectroscopy modules for performing frequency sweep and using Rabi and EFRabi for performing the amplitude scan. At the same time, we have collected more IQ data from the available 5 and 7 qubits IBMQ backends to construct both supervised (classifiers and neural network architectures) and unsupervised (generalized flows) models for their classification and clustering. At the moment, we are struggling with automating data fitting for the IQ data experiments to make the functionality more user-friendly and convenient.

[HuangJunye]

@obliviateandsurrender Can you please upload the final presentation slides and link your pull requests if any here? Thank you!

[obliviateandsurrender]

Machine_Learning_based_Discrimination_for_Excited_State_Promoted_Readout.pdf
#22 Clustering Methods for Excited State Promoted Readout.pdf
@HuangJunye please find the report and the slides for the project.