T phi experiment by merav-aharoni · Pull Request #355 · qiskit-community/qiskit-experiments

merav-aharoni · 2021-08-18T13:52:49Z

Summary

New experiment Tphi. It computes the pure dephasing time and is computed by subtracting the effect of T1 from T2.

Details and comments

It is implemented here as a composite experiment with sub-experiments T1 and T2Ramsey.

yaelbh · 2021-09-29T09:56:55Z

+        self._delays_t1 = delays_t1
+        self._delays_t2 = delays_t2
+        self._unit = unit
+        self._osc_freq = osc_freq


Use options

yaelbh · 2021-09-29T09:57:21Z

+        expT2 = T2Ramsey(qubit, self._delays_t2, self._unit, self._osc_freq)
+        exps = []
+        exps.append(expT1)
+        exps.append(expT2)


exps = [expT1, expT2]

yaelbh · 2021-09-29T09:59:25Z

+        exps.append(expT2)
+
+        # Run batch experiments
+        batch_exp = super().__init__(exps)


The variable batch_exp is not used, just call super().__init__ without storing the result in a variable

yaelbh · 2021-09-29T10:00:08Z

+        # Run batch experiments
+        batch_exp = super().__init__(exps)
+
+    def run(self, backend, experiment_data, **run_options):


Don't override run

yaelbh · 2021-09-29T10:01:18Z

                    elif op.name == "delay":
                        delay = op.params[0]
-                        prob1[qubit] = prob1[qubit] * np.exp(-delay / self._t1[qubit])
+                        prob1[qubit] = prob1[qubit] * np.exp(-delay / self._t1)


The [qubit] term should remain there. Different qubits can have different T1s and the user has to provide an array of T1s when initializing the backend.

merav-aharoni · 2021-09-29T11:24:06Z

@yaelbh , thanks for the review but it is very premature. It is still WIP.

…sey backends.

…to t_phi

…ent and to CompositeExperiment. Fix in return value of CompositeAnalysis._run_analysis.

yaelbh · 2022-01-17T06:09:55Z

            experiment_data.child_data(comp_id).block_for_results()

-        return [], []
+        return experiment_data, []


… results. Added tests for Tphi

…ns instead of plain parameters for setting delays_t1 and delays_t2 in Tphi.

yaelbh · 2022-01-31T13:10:11Z

+            backend: Optional, the backend on which to run the experiment
+        """
+        # Set experiment options
+        options =  self._default_experiment_options()


Is there a reason not to do it in a way that's different from the other experiments?

If instead I do self.experiment_options.delays_t1 = delays_t1 , I get the error AttributeError: 'Tphi' object has no attribute '_experiment_options'.
Same error if I do self.set_experiment_options(delays_t1=delays_t1, delays_t2=delays_t2).
I believe the reason is that I do the super.init() later, so the base class members are not recognized yet.

Ok, so you can do it later also here. I'm just not sure if what you're doing is equivalent, so I don't want to play with it.

yaelbh · 2022-01-31T13:12:15Z

+features:
+  - |
+    New experiment Tphi. This is a composite experiment with sub-experiments
+    T1 and T2Ramsey.New classes:


Suggested change

T1 and T2Ramsey.New classes:

T1 and T2Ramsey. New classes:

I suggest to explain what's the new experiment about. The release notes are exposed to users and have to be informative for them.

Added a short description of the experiment.

yaelbh · 2022-02-01T10:04:00Z

+        options.delays_t1 = delays_t1
+        options.delays_t2 = delays_t2
+
+        self.exps = []


Maybe exps = [T1(...), T2Ramsey(...)] ?

You were right about the bug - very nice of you to think of it! See my last commit to address that case.

yaelbh · 2022-02-01T10:14:25Z

You may have a bug: what happens if the user initializes the Tphi experiment, maybe runs it, then modifies the delays of T1:

exp.set_experiment_options(delays_t1=...)

You can write a test to check (that will remain also after you fix). I think the T1 experiment will continue to run with the old delays. You probably have to override set_experiment_options, where you'll:

Call the parent's set_experiment_options, this will set delays_t1 and delays_t2 of Tphi.
If delays_t1 is in options: call <t1exp>.set_experiment_options(delays=delays_t1), and similarly for T2Ramsey.

…for the sub-experiments, without storing the options for the main object. Removed _default_experiment_options, that is no longer needed

…et_experiment_options after run

… following changes in PR 633

…nto t_phi

merav-aharoni · 2022-02-03T10:37:33Z

Hi @yaelbh , I think this is ready for a (hopefully) final review.

yaelbh

One more small fix

yaelbh · 2022-02-03T12:03:40Z

+                self.component_experiment(0).set_experiment_options(delays=fields["delays_t1"])
+            elif key == "delays_t2":
+                self.component_experiment(1).set_experiment_options(delays=fields["delays_t2"])
+            else:


osc_freq is missing

Good catch! Added it + a test.

yaelbh · 2022-02-03T13:23:19Z

+        # propagate options to the sub-experiments.
+        for key in fields:
+            if key == "delays_t1":
+                self.component_experiment(0).set_experiment_options(delays=fields["delays_t1"])


There's a subtle point here, and I think what you're doing is the most appropriate.

The experiment options are exposed to the user by the property BaseExperiment.experiment_options. I understand that you wish to mask the internal implementation of a batch experiment consisting of sub-experiments. But a user who'll try tphi_exp.experiment_options.delays_t1 will get an error, because Tphi doesn't really have an option named delays_t1.

You can try to solve this by maintaining options delays_t1 and delays_t2 in Tphi, in addition to the delay options of the sub-experiments. But then what will you do if the user does tphi_exp.experiment_options.delays_t1 = .... You'll face again the issue of having to sync with the delays of the sub-experiment.

You'll keep facing similar issues if you try to override experiment_options. Possibly the only way to obtain full masking is by overriding Options, which is exaggerated. Hence the bottom line (which I already wrote in the upper line) that the current code is the most reasonable one, in my opinion.

If this were C++, I could have one of the delays_t2 (in Tphi and in T2Ramsey) fields point to the other. But as far as I know, there is no pointer concept in Python, right?

…ve test

…nto t_phi

yaelbh

Thanks @merav-aharoni, LGTM.
One last thing: check if experiment is done, as @nkanazawa1989 suggested yesterday in slack. Aftet that this can be merged, no need for further review (I'm approving now).

…nto t_phi

* Beginning implementation of T_phi * Added init files * Built circuits for T_phi * Defined class T_phi * Added run method to class Tphi * First stage of Tphi - up to run * Added test for Tphi. Added backend for testing, based on T1 and T2Ramsey backends. * black * Fixed return FakeJob from TphiBackend.run * Get sub-experiment analysis results from expdata.child_data * Added _run_analysis for TphiAnalysis. Added parameter to BatchExperiment and to CompositeExperiment. Fix in return value of CompositeAnalysis._run_analysis. * Cleaned up return value for Tphi analysis results. Computed the final results. Added tests for Tphi * Removed usage of 'units' in Tphi. Added usage of set_experiment_options instead of plain parameters for setting delays_t1 and delays_t2 in Tphi. * Removed unnecessary variable 'unit' and unnecessary imports. * Added missing 'TphiAnalysis in __init__ * Added documentation for Tphi * Added 'analysis' parameter to BatchExperiment and to CompositeExperiment * Added title and section to tphi tutorial * Fixed Tphi computation. Wrote tutorial. Added computation with uncertainties. * Fixed title of tutorial * Removed unnecessary prints and empty lines * Change of parameter merged from main * Added references for Tphi * Updated copyright to 2022. Added _set_backend to Tphi * Bug fix: parameter 'backend' was missing in calls to T1 and T2Ramsey * Cleaned up Tphi tutorial * Removed dataProcessor from default_options because it was causing a warning when loading from a device * Added release notes for tphi * Syntax changes resulting from code review * Update qiskit_experiments/library/characterization/tphi.py Co-authored-by: Yael Ben-Haim <yaelbh@il.ibm.com> * Update qiskit_experiments/library/characterization/tphi.py Co-authored-by: Yael Ben-Haim <yaelbh@il.ibm.com> * Removed two unnecessary methods, and did some cleaning following suggestions from review * Added class documentation. Removed overriding of shots * Removed unnecessary parameter that I previously added to BatchExperiment and CompositeExperiment constructors * Removed setting of shots in backend options. Changed 'block_for_results' to 'assertExperimentDone' * Fixed setting of default options in Tphi * Added links to classes in the release notes * Added info on the tphi experiment to the release notes * Implemented a set_experiment_options for tphi, that sets the options for the sub-experiments, without storing the options for the main object. Removed _default_experiment_options, that is no longer needed * Added test to check that circuits are modified correctly when using set_experiment_options after run * Added component analysis classes to the constructor of TphiAnalysis() following changes in PR 633 * Added setting of osc_freq to set_experiment_options. Added a respective test * Added to tests 'assertExperimentDone' Co-authored-by: Yael Ben-Haim <yaelbh@il.ibm.com>

merav-aharoni added 5 commits July 22, 2021 15:08

Beginning implementation of T_phi

ff5e9d7

Added init files

c0f64c4

Built circuits for T_phi

fef446b

Merge branch 'main' into t_phi

1f2867c

Defined class T_phi

8582ada

coruscating added this to the Release 0.2 milestone Aug 18, 2021

merav-aharoni added 3 commits September 9, 2021 15:27

Added run method to class Tphi

94a0408

Merged with main

5b742fd

First stage of Tphi - up to run

68ad6f8

merav-aharoni modified the milestones: Release 0.2, Release 0.3 Sep 26, 2021

yaelbh requested changes Sep 29, 2021

View reviewed changes

Merge with main

0c063b1

yaelbh linked an issue Oct 20, 2021 that may be closed by this pull request

T_phi experiment #68

Closed

merav-aharoni added 10 commits January 6, 2022 11:10

Merged with main

f0c0f84

Added test for Tphi. Added backend for testing, based on T1 and T2Ram…

1b328e7

…sey backends.

black

8fa6dbb

Merged with main

fdea969

Fixed return FakeJob from TphiBackend.run

049a13f

Merge branch 'main' of github.com:merav-aharoni/qiskit-experiments in…

9db05c0

…to t_phi

Merge branch 'main' into t_phi

12292de

Merged with main

f2b6e03

Get sub-experiment analysis results from expdata.child_data

d0bc161

Added _run_analysis for TphiAnalysis. Added parameter to BatchExperim…

b1a0b9a

…ent and to CompositeExperiment. Fix in return value of CompositeAnalysis._run_analysis.

yaelbh reviewed Jan 17, 2022

View reviewed changes

merav-aharoni added 4 commits January 18, 2022 13:11

Cleaned up return value for Tphi analysis results. Computed the final…

ea8e5f1

… results. Added tests for Tphi

black and lint

7c79f02

Removed usage of 'units' in Tphi. Added usage of set_experiment_optio…

c7b8728

…ns instead of plain parameters for setting delays_t1 and delays_t2 in Tphi.

Removed unnecessary variable 'unit' and unnecessary imports.

f79d8a3

merav-aharoni added 2 commits January 31, 2022 14:46

Added links to classes in the release notes

15602b8

Merged with main

cabd8d8

yaelbh reviewed Jan 31, 2022

View reviewed changes

Added info on the tphi experiment to the release notes

1d00174

yaelbh reviewed Feb 1, 2022

View reviewed changes

merav-aharoni and others added 8 commits February 1, 2022 19:18

Implemented a set_experiment_options for tphi, that sets the options …

9ebb855

…for the sub-experiments, without storing the options for the main object. Removed _default_experiment_options, that is no longer needed

Added test to check that circtuis are modified correctly when using s…

5ab5b4c

…et_experiment_options after run

Merge branch 'main' into t_phi

0514857

Added component analysis classes to the constructor of TphiAnalysis()…

a2e56c3

… following changes in PR 633

Merge branch 'main' into t_phi

cff6af6

black

cb4fb0d

Merge branch 't_phi' of github.com:merav-aharoni/qiskit-experiments i…

feb4342

…nto t_phi

Merge branch 'main' into t_phi

d5fafc3

yaelbh requested changes Feb 3, 2022

View reviewed changes

merav-aharoni and others added 5 commits February 3, 2022 16:39

Added setting of osc_freq to set_experiment_options. Added a respecti…

b52dd6f

…ve test

Merge branch 't_phi' of github.com:merav-aharoni/qiskit-experiments i…

42e0139

…nto t_phi

Merge branch 'main' into t_phi

66333e8

Merge branch 'main' into t_phi

b20d38b

Changed variable name

bafdcc3

yaelbh approved these changes Feb 8, 2022

View reviewed changes

merav-aharoni changed the title ~~T phi~~ T phi experiment Feb 8, 2022

merav-aharoni and others added 3 commits February 8, 2022 11:20

Merge branch 'main' into t_phi

e6563ce

Added to tests 'assertExperimentDone'

9db90d8

Merge branch 't_phi' of github.com:merav-aharoni/qiskit-experiments i…

bdb0828

…nto t_phi

merav-aharoni merged commit 94b865f into qiskit-community:main Feb 8, 2022

merav-aharoni deleted the t_phi branch February 8, 2022 10:42

Conversation

merav-aharoni commented Aug 18, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Details and comments

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

merav-aharoni commented Sep 29, 2021

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yaelbh commented Feb 1, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

merav-aharoni commented Feb 3, 2022

Uh oh!

yaelbh left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yaelbh left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

merav-aharoni commented Aug 18, 2021 •

edited

Loading

yaelbh commented Feb 1, 2022 •

edited

Loading