Momentum-QNG optimization algo

[borbysh]

Before submitting

Please complete the following checklist when submitting a PR:

• All new features must include a unit test.
  If you've fixed a bug or added code that should be tested, add a test to the
  test directory!

• All new functions and code must be clearly commented and documented.
  If you do make documentation changes, make sure that the docs build and
  render correctly by running make docs.

• Ensure that the test suite passes, by running make test.

• Add a new entry to the doc/releases/changelog-dev.md file, summarizing the
  change, and including a link back to the PR.

• The PennyLane source code conforms to
  PEP8 standards.
  We check all of our code against Pylint.
  To lint modified files, simply pip install pylint, and then
  run pylint pennylane/path/to/file.py.

When all the above are checked, delete everything above the dashed
line and fill in the pull request template.

---

Context:

Description of the Change:

Benefits:

Possible Drawbacks:

Related GitHub Issues:

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.70%. Comparing base (1a143d2) to head (4ac363c).
Report is 2 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #6240   +/-   ##
=======================================
  Coverage   99.70%   99.70%           
=======================================
  Files         444      445    +1     
  Lines       42236    42260   +24     
=======================================
+ Hits        42113    42137   +24     
  Misses        123      123           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dwierichs]

Hi @borbysh,
thanks a lot for opening this pull request!
I leave a first batch of small comments on the docstrings/changelog entry :)
Feel free to modify my suggestions further or reject them. Most of them are concerned with bringing the texts a bit closer to the standard PennyLane style.

Let me know how you'd like to proceed regarding tests. :) 🎉

[borbysh]

Dear David! Thank you very much for your help! I committed all the changes you proposed. Now I'm turning test_qng.py into test_momentum_qng.py) One thing remains unclear to me: where is the script that runs these tests? Sincerely yours Alex

…

On Mon, Sep 9, 2024 at 1:07 PM David Wierichs ***@***.***> wrote: ***@***.**** requested changes on this pull request. Hi @borbysh <https://github.com/borbysh>, thanks a lot for opening this pull request! I leave a first batch of small comments on the docstrings/changelog entry :) Feel free to modify my suggestions further or reject them. Most of them are concerned with bringing the texts a bit closer to the standard PennyLane style. Let me know how you'd like to proceed regarding tests. :) 🎉 ------------------------------ In doc/releases/changelog-dev.md <#6240 (comment)> : > @@ -3,7 +3,10 @@ # Release 0.39.0-dev (development release) <h3>New features since last release</h3> - +* A new class MomentumQNGOptimizer is added. It inherits the basic QNGOptimizer class and requires one additional hyperparameter (the This will lead to nicer rendering of code objects :) ⬇️ Suggested change -* A new class MomentumQNGOptimizer is added. It inherits the basic QNGOptimizer class and requires one additional hyperparameter (the +* A new class `MomentumQNGOptimizer` is added. It inherits the basic `QNGOptimizer` class and requires one additional hyperparameter (the ------------------------------ In doc/releases/changelog-dev.md <#6240 (comment)> : > Lillian M. A. Frederiksen, Christina Lee, William Maxwell, -Lee J. O'Riordan, \ No newline at end of file +Lee J. O'Riordan, +Oleksandr Borysenko We tend to do ordering by last name here :) ------------------------------ In pennylane/optimize/momentum_qng.py <#6240 (comment)> : > @@ -0,0 +1,171 @@ +# Copyright 2024 Xanadu Quantum Technologies Inc. + +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at + +# http://www.apache.org/licenses/LICENSE-2.0 + +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +"""Quantum natural gradient optimizer with momentum""" +import pennylane as qml Not needed. ⬇️ Suggested change -import pennylane as qml ------------------------------ In pennylane/optimize/momentum_qng.py <#6240 (comment)> : > + +# pylint: disable=too-many-branches +# pylint: disable=too-many-arguments +from pennylane import numpy as pnp +from pennylane.utils import _flatten, unflatten + +from .qng import QNGOptimizer + +class MomentumQNGOptimizer(QNGOptimizer): + r"""A generalization of the Quantum Natural Gradient (QNG) optimizer by considering a discrete-time Langevin equation + with QNG force. For details of the theory and derivation of Momentum-QNG, please, see: + + Oleksandr Borysenko, Mykhailo Bratchenko, Ilya Lukin, Mykola Luhanko, Ihor Omelchenko, + Andrii Sotnikov and Alessandro Lomi. + "Application of Langevin Dynamics to Advance the Quantum Natural Gradient Optimization Algorithm" + <https://arxiv.org/abs/2409.01978> This is the somewhat funky hyperref syntax for RST documents (the format for the documentation) ⬇️ Suggested change - <https://arxiv.org/abs/2409.01978> + `arXiv:2409.01978 <https://arxiv.org/abs/2409.01978>`__ ------------------------------ In pennylane/optimize/momentum_qng.py <#6240 (comment)> : > +from pennylane.utils import _flatten, unflatten + +from .qng import QNGOptimizer + +class MomentumQNGOptimizer(QNGOptimizer): + r"""A generalization of the Quantum Natural Gradient (QNG) optimizer by considering a discrete-time Langevin equation + with QNG force. For details of the theory and derivation of Momentum-QNG, please, see: + + Oleksandr Borysenko, Mykhailo Bratchenko, Ilya Lukin, Mykola Luhanko, Ihor Omelchenko, + Andrii Sotnikov and Alessandro Lomi. + "Application of Langevin Dynamics to Advance the Quantum Natural Gradient Optimization Algorithm" + <https://arxiv.org/abs/2409.01978> + + We are grateful to David Wierichs for his generous help with the multi-argument variant of the MomentumQNGOptimizer class. + + In PennyLane, the MomentumQNGOptimizer class is a subclass of the QNGOptimizer class and requires one additional The context that this is in PennyLane is always given in the documentation, so I would skip it. In RST, code-like formatting works with two backticks ``: ⬇️ Suggested change - In PennyLane, the MomentumQNGOptimizer class is a subclass of the QNGOptimizer class and requires one additional + ``MomentumQNGOptimizer`` is a subclass of the ``QNGOptimizer`` class and requires one additional ------------------------------ In pennylane/optimize/momentum_qng.py <#6240 (comment)> : > + using the ``metric_tensor_fn`` keyword argument. + + For the following cases, providing ``metric_tensor_fn`` may be useful: + + * For hybrid classical-quantum models, the "mixed geometry" of the model + makes it unclear which metric should be used for which parameter. + For example, parameters of quantum nodes are better suited to + one metric (such as the QNG), whereas others (e.g., parameters of classical nodes) + are likely better suited to another metric. + + * For multi-QNode models, we don't know what geometry is appropriate + if a parameter is shared amongst several QNodes. + + **Examples:** + + TODO Just marking this as to do via this Github comment so we don't forget :) ------------------------------ In pennylane/optimize/momentum_qng.py <#6240 (comment)> : > + See the next examples of the Pennylane code, benchmarking the Momentum-QNG optimizer + together with the basic QNG, Momentum and Adam: + QAOA: https://github.com/borbysh/Momentum-QNG/blob/main/QAOA_depth4.ipynb + VQE: https://github.com/borbysh/Momentum-QNG/blob/main/portfolio_optimization.ipynb ⬇️ Suggested change - See the next examples of the Pennylane code, benchmarking the Momentum-QNG optimizer - together with the basic QNG, Momentum and Adam: - QAOA: https://github.com/borbysh/Momentum-QNG/blob/main/QAOA_depth4.ipynb - VQE: https://github.com/borbysh/Momentum-QNG/blob/main/portfolio_optimization.ipynb + Also see the examples from the reference above, benchmarking the Momentum-QNG optimizer + against the basic QNG, Momentum and Adam: + - `QAOA <https://github.com/borbysh/Momentum-QNG/blob/main/QAOA_depth4.ipynb>`__ + - `VQE <https://github.com/borbysh/Momentum-QNG/blob/main/portfolio_optimization.ipynb>`__ ------------------------------ In pennylane/optimize/momentum_qng.py <#6240 (comment)> : > + momentum=0.9 (float): the user-defined hyperparameter :math:`\rho` + stepsize=0.01 (float): the user-defined hyperparameter :math:`\eta` Matching the order of the kwargs in __init__: ⬇️ Suggested change - momentum=0.9 (float): the user-defined hyperparameter :math:`\rho` - stepsize=0.01 (float): the user-defined hyperparameter :math:`\eta` + stepsize=0.01 (float): the user-defined hyperparameter :math:`\eta` + momentum=0.9 (float): the user-defined hyperparameter :math:`\rho` — Reply to this email directly, view it on GitHub <#6240 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANRT6235N543JBPISTE66ZTZVVXPJAVCNFSM6AAAAABN2ZUGHOVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZDEOBZGQZTQNJWGI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[dwierichs]

Thanks for adding the tests, @borbysh .
I left suggestions where to change the optimizer to the new class.
Then we'll have to adapt the tests to actually expect the right behaviour (namely the one of MomentumQNGOptimizer instead of QNGOptimizer)

The tests are run automatically in the continuous integration, which you can see at the bottom of the pull request overview.
In order to run them locally, install the developer's requirements via

pip install -r requirements-dev.txt

Afterwards, you can run the tests with pytest:

python -m pytest tests/optimizer/test_momentum_qng.py

[borbysh]

Dear David!
I made corrections according to your advises.
Regards
Alex

[borbysh]

Dear Pietropaolo!
Thank you very much for your review and suggestions!
I accepted all the changes you proposed.
Regards,
Alex

[borbysh]

Dear Pietropaolo,
thank you very much for your review and comments!
I made some changes according to your suggestions.
Regards
Alex
P.S. Looks like a duplicate comment, but the more - the better)

[PietropaoloFrisoni]

@borbysh Awesome! I am happy to approve once we have a green CI :)

I think you need to pull from the master branch and resolve the conflicts in the changelog file before the tests start. If you have issues I can take care of it myself (just let me know)

[dwierichs]

I took the liberty to perform the merge :)

[borbysh]

I'm a bit confused. I only changed "m_t" to "metric_tensor" in the left-hand side in line 126. I didn't alter the right-hand side and it passed tests 3 weeks ago)

[PietropaoloFrisoni]

I also took the liberty to fix one line of code with black to pass the test : )

Congrats again @borbysh !🎉

[borbysh]

Thank you very much, David Wierichs and Pietropaolo Frisoni!
I wouldn't manage to ride this beast without your help!

[dwierichs]

@borbysh You wanna push the "squash and merge" button? :)

[borbysh]

With a great pleasure, If you tell me where it is!

[dwierichs]

There should be a large green button above the comment box here.

[borbysh]

I'm in panic. Can't find it

[dwierichs]

Haha, no worries :D
Maybe I'm missing something and you can't perform the action 🤔 I'll ask.

[Alex-Preciado]

No worries, I'm merging this PR now. Thank you so much for your contribution @borbysh 🚀

[PietropaoloFrisoni]

@borbysh It's the 'Squash and merge' green tab on the very bottom. I can merge the PR for you if you wish

Edit: Looks like it's no longer necessary : )

[borbysh]

Thank you very much, guys! I owe you some good drinks!