Add maximize to minimize converter by tomtuamnuq · Pull Request #114 · qiskit-community/qiskit-optimization

tomtuamnuq · 2021-04-29T07:40:39Z

Summary

Adds Maximize To Minimize converter

Fixes #41

Details and comments

Introduces a new converter class MaximizeToMinimize for QuadraticProgram as described in #41 .
It converts a problem to a minimization problem. The converter is in the default converters list of OptimizationAlgorithm. Algorithms and QuadraticProgramToQubo are now using MaximizeToMinimize.

Many algorithms converted problems internally to minimization. To avoid duplicate code MaximizeToMinimize converter is used now. Since the internal problem is now minimization, there are changes to the following results: - intermediate_fval in GroverOptimizationResult is that of the minimization QUBO. - raw_results in OptimizationResult are those of the the minimization QUBO. - ADMM optimizer logs out the minimization objective values samples in OptimizationResult are still that of the original problem. This solves issue qiskit-community#41

CLAassistant · 2021-04-29T07:40:44Z

All committers have signed the CLA.

tomtuamnuq · 2021-04-29T14:41:27Z

The installation of cvxpy by just using pip install qiskit-optimization[cvxpy] was not enough to set up the complete test on my local machine and I missed to check whether the tests were still skipped. Sorry for that.

I have checked the failed Unit Test: TestWarmStartQAOAOptimizer.test_max_cut
The problem is, that the presolver GoemansWilliamsonOptimizer is now getting a minimization qubo. Therefore, opt_result in https://github.com/Qiskit/qiskit-optimization/blob/3e21815034c72030eab17f2865895dc0d12076a4/qiskit_optimization/algorithms/warm_start_qaoa_optimizer.py#L294 is now

optimal function value: 0.0
optimal value: [0 0 0 0]
status: SUCCESS

Instead of
optimal function value: 4.0
optimal value: [0 1 1 0]
status: SUCCESS

This leads to a wrong result.

Maybe we should not add MaximizeToMinimize always to the default converters list in https://github.com/Qiskit/qiskit-optimization/blob/3e21815034c72030eab17f2865895dc0d12076a4/qiskit_optimization/algorithms/optimization_algorithm.py#L354

Then, we can alter the UnitTest to use a fixed List of converters(without MaximizeToMinimize) instead of the default list.

Maybe GoemansWilliamsonOptimizer should give a hint if it is used with a minimization problem?

woodsp-ibm

Nice job for a first contribution! You have even done a reno yaml file, nice!

t-imamichi · 2021-05-06T10:21:17Z

Could you apply black code formatter by black qiskit_optimization test.
We recently apply black to the entire code base. See #117 for details.

Use black formatter Solve conflicts: qiskit_optimization/algorithms/admm_optimizer.py qiskit_optimization/algorithms/grover_optimizer.py qiskit_optimization/algorithms/minimum_eigen_optimizer.py qiskit_optimization/algorithms/multistart_optimizer.py qiskit_optimization/algorithms/optimization_algorithm.py qiskit_optimization/algorithms/slsqp_optimizer.py test/converters/test_converters.py

t-imamichi · 2021-05-07T09:47:41Z

Sorry. Could format again by black -l 100? We added a missing config #118.

t-imamichi · 2021-05-07T09:59:51Z

I think we need to simplify each algorithm to use both OptimizationAlgorithm._convert and OptimizationAlgorithm._interpret.

For example, ADMM case

converters = [IntegerToBinary(), MaximizeToMinimize()]
original_problem = problem
problem = self._convert(problem, converters)
...
return cast(
    ADMMOptimizationResult,
    self._interpret(
        x=solution,
        converters=converters,
        problem=original_problem,
        result_class=ADMMOptimizationResult,
        state=self._state,
    ),
)

tomtuamnuq · 2021-05-07T10:30:08Z

Sorry. Could format again by black -l 100? We added a missing config #118.

Hello @t-imamichi ,
since I have merged the config file in, prior to commit with "apply make black", it should be okay.
Running black with line length spec does not change anything.

Solve conflicts: qiskit_optimization/algorithms/grover_optimizer.py test/algorithms/test_min_eigen_optimizer.py

t-imamichi · 2021-05-17T01:42:51Z

Thanks. It's almost done. It may require #136 to pass unit tests. Please wait for a while for us to fix the unit test issues.
Details #138

adekusar-drl

One thing to check out is the warm start QAOA test. Other than that, looks good to me! Thanks a lot!

Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com>

t-imamichi · 2021-05-21T16:00:11Z

Let's see the CI result. I think this PR is ready if CI passes.

t-imamichi

Thank you for your cooperation! LGTM.

tomtuamnuq added 3 commits April 29, 2021 07:22

Add MaximizeToMinimize conv with UnitTest case

ccbe011

Add reno release note for feature

38b5936

tomtuamnuq requested review from manoelmarques, pbark, stefan-woerner, t-imamichi and woodsp-ibm as code owners April 29, 2021 07:40

woodsp-ibm reviewed Apr 29, 2021

View reviewed changes

t-imamichi requested a review from adekusar-drl April 30, 2021 00:47

adekusar-drl reviewed Apr 30, 2021

View reviewed changes

tomtuamnuq added 4 commits May 1, 2021 13:10

Fix inline imports of MaximizeToMinimize

15a8faf

Fix MultistartAlgorithm solve logic

3b5542c

Remove logger in maximize_to_minimize

f3a6fc5

Extend MaximizeToMinimize test case

4e265da

tomtuamnuq added 2 commits May 7, 2021 11:05

Apply make black

18d2b53

t-imamichi and others added 5 commits May 12, 2021 15:36

Merge branch 'main' into maxmin-conv

36daf0c

Make Max2Min conv efficient

b24478d

Let algorithms use _convert

245a955

Not append max2min in _prepare_converters

cb9c7c6

Merge branch 'main' into maxmin-conv

ec9f7e3

t-imamichi reviewed May 13, 2021

View reviewed changes

Comment thread qiskit_optimization/algorithms/optimization_algorithm.py Outdated

t-imamichi reviewed May 13, 2021

View reviewed changes

Comment thread qiskit_optimization/algorithms/multistart_optimizer.py Outdated

t-imamichi changed the title ~~Maxmin conv~~ Add maximize to minimize converter May 13, 2021

tomtuamnuq added 3 commits May 16, 2021 08:49

Merge branch 'main' into maxmin-conv

0068a30

Solve conflicts: qiskit_optimization/algorithms/grover_optimizer.py test/algorithms/test_min_eigen_optimizer.py

Fix docstring

012fe39

Remove blank lines

a4a8a7a

Merge branch 'main' into maxmin-conv

a23e0ef

t-imamichi requested review from adekusar-drl, t-imamichi and woodsp-ibm May 18, 2021 15:33

adekusar-drl reviewed May 18, 2021

View reviewed changes

Comment thread qiskit_optimization/algorithms/cobyla_optimizer.py Outdated

Comment thread qiskit_optimization/algorithms/slsqp_optimizer.py Outdated

Comment thread test/algorithms/test_warm_start_qaoa.py Outdated

tomtuamnuq and others added 3 commits May 20, 2021 07:48

Merge branch 'main' into maxmin-conv

d38d641

Remove unnecessary function declarations

1c0953f

Merge branch 'main' into maxmin-conv

ecee542

t-imamichi reviewed May 21, 2021

View reviewed changes

Comment thread test/algorithms/test_warm_start_qaoa.py Outdated

t-imamichi and others added 2 commits May 21, 2021 22:55

Merge branch 'main' into maxmin-conv

232d8b9

Update test/algorithms/test_warm_start_qaoa.py

3fb2469

Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com>

t-imamichi reviewed May 21, 2021

View reviewed changes

Comment thread test/algorithms/test_warm_start_qaoa.py Outdated

Update test/algorithms/test_warm_start_qaoa.py

8fd318c

t-imamichi reviewed May 21, 2021

View reviewed changes

Comment thread test/algorithms/test_warm_start_qaoa.py Outdated

Update test/algorithms/test_warm_start_qaoa.py

3911d11

t-imamichi self-requested a review May 21, 2021 16:26

t-imamichi approved these changes May 21, 2021

View reviewed changes

Merge branch 'main' into maxmin-conv

969ec92

adekusar-drl merged commit 04adb49 into qiskit-community:main May 21, 2021

tomtuamnuq deleted the maxmin-conv branch May 22, 2021 10:42

t-imamichi mentioned this pull request May 24, 2021

GoemansWilliamsonOptimizer needs to check the sense of the input problem #143

Closed

Conversation

tomtuamnuq commented Apr 29, 2021 • edited by t-imamichi Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Details and comments

Uh oh!

CLAassistant commented Apr 29, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tomtuamnuq commented Apr 29, 2021

Uh oh!

woodsp-ibm left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

t-imamichi commented May 6, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

t-imamichi commented May 7, 2021

Uh oh!

t-imamichi commented May 7, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tomtuamnuq commented May 7, 2021

Uh oh!

Uh oh!

Uh oh!

t-imamichi commented May 17, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

adekusar-drl left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

t-imamichi commented May 21, 2021

Uh oh!

t-imamichi left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

tomtuamnuq commented Apr 29, 2021 •

edited by t-imamichi

Loading

CLAassistant commented Apr 29, 2021 •

edited

Loading

t-imamichi commented May 6, 2021 •

edited

Loading

t-imamichi commented May 7, 2021 •

edited

Loading

t-imamichi commented May 17, 2021 •

edited

Loading