Expand qiskit-neko testing

[mtreinish]

Description

The qiskit-neko project: https://github.com/mtreinish/qiskit-neko is a new effort to add a proper integration test suite to the overall qiskit project. It's designed to be run in CI to ensure that a proposed change to any qiskit project doesn't break backwards compatibility or break an interface that another qiskit project is currently using. The idea behind the project is to provide feedback before we merge a pull request if an interface is potentially broken by a change and to prevent that from being merged (so it is never released). However, this project is still in it's early stages and needs to be expanded and put into CI configurations for qiskit projects. This project is to help expand the test coverage in qiskit-neko so that we have a representative set of tests and then to potentially help get it properly running in CI for all of qiskit.

Deliverables

Pull requests to qiskit-neko and potentially to other qiskit projects to start getting qiskit-neko running in CI.

Mentors details

• Mentor 1
  • Name: Matthew Treinish
  • GitHub ID: mtreinish
  • What they do: Qiskit core developer

Number of mentees

3

Type of mentees

• Mentor 1
  • Required:
  • Experience using qiskit
  • Desire to contribute to an open source software project
  • Nice to have:
  • Familiarity with software testing
  • Experience with python (specifically unittest)

[tgag17]

Checkpoint 1 ppt: #10 Expand qiskit-neko testing.pdf

[HuangJunye]

@tgag17 Can you please provide your checkpoint 2 updates? Instructions are given in the slack channel.

[tgag17]

@HuangJunye Sorry for the delay. I will update the checkpoint 2 asap.

[tgag17]

Update: Checkpoint 2

Qiskit-neko is being developed to incorporate CI coverage in Qiskit. These tests are used primarily for two purposes: backwards compatibility testing for Qiskit to validate the changes proposed to any Qiskit project do not break functionality from the previous releases and to validate that functionality works as expected with different providers.

Before the start of the project, a few tests were already present in qiskit-neko. These covered the domains of circuits, experiments and the qiskit-nature classes. As a part of qamp 2022, I have till now added a few tests covering the qiskit-machine learning classes.

Particularly, by taking reference from the tutorials available, the following tests have been added:

• Quantum Neural Networks (Using OPFlow): Constructs a single qubit single-layer quantum neural network using the OPFlowQNN class and tests the output against the expected output.
• Quantum Neural Networks (Using TwoLayer): Constructs a multi-qubit multi-layer quantum neural network using the TwoLayerQNN class and tests the output against the expected output.
• Variational Quantum Classifier: Constructs a quantum classifier to classify and provide labels to the points of a toy model. Tests the output accuracy against the expected output.

The PR for the tests is still under review: (https://github.com/mtreinish/qiskit-neko/pull/3) These tests provide an initial test coverage for the quantum-machine-learning module.

Future Work:

In the next steps, the focus will be shifted to creating Github actions for CI and integrating qiskit-neko with the qiskit repository.

[tgag17]

Final Checkpoint ppt: #10 Expand qiskit-neko testing.pdf