Qcnn example

[ihpcdingwj]

Checklist:

• Reviewers confirm new code works as expected.
• Tests are passing.
• Coverage does not decrease.
• Documentation is updated.

[scarrazza]

What is the difference between this and PR #695?

[jf-kong]

What is the difference between this and PR #695?

This PR is for an example (in the folder examples/qcnn_classifier) using the QCNN model based on PR #729. This is based on a previous suggestion by @MatteoRobbiati to have an extended example/tutorial to illustrate the use of the QCNN model. PR #695 is an outdated PR (prior to #729) that has since been closed.

[MatteoRobbiati]

Can you please merge the main branch on this, in order to avoid conflicts due to the already merged qcnn branch?

[codecov]

Codecov Report

Patch and project coverage have no change.

Comparison is base (0a84cb2) 100.00% compared to head (a772fff) 100.00%.

Additional details and impacted files

@@            Coverage Diff            @@
##            master      #834   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           48        48           
  Lines         6595      6595           
=========================================
  Hits          6595      6595           

Flag	Coverage Δ
unittests	100.00% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[MatteoRobbiati]

Thank you very much for this, it is very useful for a better understanding of the model you implemented.

Let me add some suggestions:

1. this is not the first example in qibo based on an IPython implementation. The qPDF is another one. Personally, I think the notebook approach is very useful in terms of visualization of the problem (e.g. in the documentation), but I think this folder starts to be a bit messy.
   According to this, I purpose two solutions: the first one is to remove the implementation part from README.md (I mean from ## How to use the QCNN class), which is not needed since already shown into the notebook. On this way this example will follow the line of the qPDF one. The second is to change this example into a Python script as is done for all the examples in qibo and to move this notebook to the tutorials repository, in order to be used into the documentation.
   I prefer this second option, since notebooks are useful tools but very difficult to handle with several hands.
2. please select only one of the two notebooks and fix it: there are some unused packages and modules there (e.g. gates, numpy etc.) and the qcnn package must be loaded properly (from qibo.models.qcnn import QuantumCNN);
3. add your example here.

Some small suggestions follow.

[MatteoRobbiati]

Remove the space between [...] and (...) in order to make te link work.

[MatteoRobbiati]

Please add link to the documentation when it is available (where you mention the docs).

[scarrazza]

@ihpcdingwj is this ready for a second review pass?

[MatteoRobbiati]

What is the status here @ihpcdingwj?

Let me summarize some remaining tasks:

• remove one of the two notebooks. You have two versions of the same example;
• fix the imports in the notebook: some packages are unused;
• fix the qcnn import by setting from qibo.models.qcnn import QuantumCNN;
• please follow the general convention: first Standard library imports, then related third party imports, then
  local application/library specific imports.
• add your example here;
• since we decided to keep the jupyter notebook form, please remove the part of README.md which is at this point reduntant: the ##How to use the QCNN class is not needed if already described in the notebook.

[MatteoRobbiati]

I did some small fix to the notebook (like passing circuit.draw() to the print function, etc),
fixed the link2path in the README.md and changed the name of the notebook into qcnn_demo.ipynb.

The last suggestion is to explicitly write in the README.md that scikit-learn is needed to run your example. It should be better to link the official installation page.

I also suggest to rebase this branch on the master.

[ihpcdingwj]

I did some small fix to the notebook (like passing circuit.draw() to the print function, etc), fixed the link2path in the README.md and changed the name of the notebook into qcnn_demo.ipynb.

The last suggestion is to explicitly write in the README.md that scikit-learn is needed to run your example. It should be better to link the official installation page.

I also suggest to rebase this branch on the master.

Thanks @MatteoRobbiati for the editing. We have made the changes suggested. Could you also clarify how do we proceed with "rebase"?

[MatteoRobbiati]

Thanks @MatteoRobbiati for the editing. We have made the changes suggested. Could you also clarify how do we proceed with "rebase"?

Thank you for the modification.
With rebase I mean this git operation, but I think in this case we don't have conflicts.

[jf-kong]

Thanks @MatteoRobbiati for the editing. We have made the changes suggested. Could you also clarify how do we proceed with "rebase"?

Thank you for the modification. With rebase I mean this git operation, but I think in this case we don't have conflicts.

Thanks @MatteoRobbiati. Have rewritten history :)

[MatteoRobbiati]

Thanks @MatteoRobbiati. Have rewritten history :)

Perfect! I think we are ready to merge :)