(2/2) Add parameter sweep support for Pauli string measurements with readout mitigation #7569

ddddddanni · 2025-08-12T06:00:58Z

Previous PR is #7435
This PR allow user to run measuring_pauli_strings in sweep mode. In this mode, the function uses parameterized circuits and sweeps parameters for both Pauli measurements and readout benchmarking

eliottrosenberg

I tested this on hardware, and it seems to have a bug. For example, compare the last line of the output of this cell to the output of this cell. result and result2 come from the couple of cells above those. Reproduced again futher down in that colab.

eliottrosenberg

Ok, I found the difference. If you start with the following circuit:

0: ───────────────────────H───@───H───
                              │
1: ───────────────H───@───H───@───────
                      │
2: ───H───────@───H───@───────────────
              │
3: ───H───@───@───────────────────────
          │
4: ───H───@───────H───@───────────────
                      │
5: ───────────────H───@───H───@───────
                              │
6: ───────────────────────H───@───H───

and then measure X(q(0))*X(q(1))*X(q(2))*X(q(3))*X(q(4))*X(q(5))*X(q(6)), the sweep version gives you

0: ───────────────────────H───@───H───PhX(phi0/2 - 1/2)^theta0───M('m')───
                              │                                  │
1: ───────────────H───@───H───@───────PhX(phi1/2 - 1/2)^theta1───M────────
                      │                                          │
2: ───H───────@───H───@───────────────PhX(phi2/2 - 1/2)^theta2───M────────
              │                                                  │
3: ───H───@───@───────────────────────PhX(phi3/2 - 1/2)^theta3───M────────
          │                                                      │
4: ───H───@───────H───@───────────────PhX(phi4/2 - 1/2)^theta4───M────────
                      │                                          │
5: ───────────────H───@───H───@───────PhX(phi5/2 - 1/2)^theta5───M────────
                              │                                  │
6: ───────────────────────H───@───H───PhX(phi6/2 - 1/2)^theta6───M────────

with the basis change all in the same moment, whereas in the batch version, you get

0: ───────────────────────────────H───────────@───H───────────Ry(-0.5π)───M('m')───
                                              │                           │
1: ───────────────H───────────@───H───────────@───Ry(-0.5π)───────────────M────────
                              │                                           │
2: ───H───────@───H───────────@───Ry(-0.5π)───────────────────────────────M────────
              │                                                           │
3: ───H───@───@───Ry(-0.5π)───────────────────────────────────────────────M────────
          │                                                               │
4: ───H───@───────H───────────@───Ry(-0.5π)───────────────────────────────M────────
                              │                                           │
5: ───────────────H───────────@───H───────────@───Ry(-0.5π)───────────────M────────
                                              │                           │
6: ───────────────────────────────H───────────@───H───────────Ry(-0.5π)───M────────

with the basis change done as early as possible. In this case, you get a higher-fidelity result with the basis change done as early as possible because you are insensitive to the dephasing that happens between the final basis change gate and the measurement, but that may not always be the intended behavior. Maybe we could add an insert_strategy argument so that users have the choice of whether to do the basis change all in the same penultimate moment or as early as possible.

eliottrosenberg · 2025-08-25T17:30:29Z

Thanks, @ddddddanni! It looks like it is printing out the circuits for debugging, and you might want to remove that.

review-notebook-app · 2025-08-25T17:43:39Z

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

ddddddanni · 2025-08-25T18:37:10Z

Thanks, @ddddddanni! It looks like it is printing out the circuits for debugging, and you might want to remove that.

Done!

codecov · 2025-09-01T23:16:33Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.37%. Comparing base (6816824) to head (121286c).
⚠️ Report is 7 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #7569    +/-   ##
========================================
  Coverage   99.37%   99.37%            
========================================
  Files        1078     1082     +4     
  Lines       96166    96761   +595     
========================================
+ Hits        95563    96157   +594     
- Misses        603      604     +1

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

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

eliottrosenberg

There are some things that @ddddddanni and I will keep investigating, but I think we can go ahead and merge this. I just want to clarify that this still has shuffling for the run_batch case, correct? Thanks!

ddddddanni · 2025-09-15T23:12:37Z

There are some things that @ddddddanni and I will keep investigating, but I think we can go ahead and merge this. I just want to clarify that this still has shuffling for the run_batch case, correct? Thanks!

Sure! I have re-enabled shuffling function

ddddddanni · 2025-09-15T23:13:01Z

@NoureldinYosri can you please review this PR? Thank you!!

NoureldinYosri · 2025-09-23T03:38:55Z

cirq-core/cirq/contrib/paulistring/pauli_string_measurement_with_readout_mitigation.py

    return operations


+def _pauli_strings_to_basis_change_with_sweep(


I'm having a déjà vu moment, didn't you write a similar method before?

I have a giant PR https://github.com/quantumlib/Cirq/pull/7358/files which you reviewed long time ago. I break that PR into smaller ones: #7435 and this pr, per you requested.
For question you asked in #7358: https://screenshot.googleplex.com/6HHTPQdj3R6Vwfy and https://screenshot.googleplex.com/9kw8sR9Q8g4BnMF - each qubit can belong to many input pauli strings. Since input pauli strings are qubit-wise commuting, it's safe to do so. For example, input pauli strings could be {X(q_0)I(q_1), I(q_0)X(q_1)}, and this function returns params_dict{q_0:x basis change, q_1: x basis change}.

NoureldinYosri · 2025-09-23T03:42:08Z

cirq-core/cirq/contrib/paulistring/pauli_string_measurement_with_readout_mitigation.py

+        basis_change_circuits = []
+        input_circuit_unfrozen = input_circuit.unfreeze()
+        for pauli_strings in pauli_string_groups:
+            basis_change_circuit = circuits.Circuit(


I think that we don't really need an insert_strategy, we can just do

circuits.Circuit.from_moments( *input_circuit, _pauli_strings_to_basis_change_ops(pauli_strings, qid_list), ops.measure(*qid_list, key="m"), )

We want to give users choices of whether to do the basis change all in the same penultimate moment or as early as possible. This was discussed https://screenshot.googleplex.com/4LPKcw4eYCSvP85.

NoureldinYosri · 2025-09-23T03:42:22Z

cirq-core/cirq/contrib/paulistring/pauli_string_measurement_with_readout_mitigation.py

+            basis_change_circuit = circuits.Circuit(
+                input_circuit_unfrozen,
+                _pauli_strings_to_basis_change_ops(pauli_strings, qid_list),
+                ops.measure(*qid_list, key="m"),


lets use a more meaningful key name

NoureldinYosri · 2025-09-23T03:42:40Z

cirq-core/cirq/contrib/paulistring/pauli_string_measurement_with_readout_mitigation.py

+        ]
+        measurement_op = ops.M(*qid_list, key="m")
+
+        parameterized_circuit = circuits.Circuit(


.from_moments

Same above, we may want to allow users to control the insert strategy of basis changes.

NoureldinYosri · 2025-09-23T03:43:31Z

cirq-core/cirq/contrib/shuffle_circuits/shuffle_circuits_with_readout_benchmarking.py

    sweep_params = []
    for bitstr in random_bitstrings:
-        sweep_params.append({exp: bit for exp, bit in zip(exp_symbols, bitstr)})
+        sweep_params.append({exp.name: bit for exp, bit in zip(exp_symbols, bitstr)})


how about str(exp)?

github-actions bot added the size: L 250< lines changed <1000 label Aug 12, 2025

ddddddanni mentioned this pull request Aug 12, 2025

pauli_string_measurement_with_readout_mitigation_test is flaky #7485

Closed

ddddddanni marked this pull request as ready for review August 14, 2025 03:31

ddddddanni requested review from vtomole and a team as code owners August 14, 2025 03:31

ddddddanni requested a review from maffoo August 14, 2025 03:31

eliottrosenberg self-requested a review August 14, 2025 16:23

eliottrosenberg requested changes Aug 16, 2025

View reviewed changes

eliottrosenberg requested changes Aug 21, 2025

View reviewed changes

ddddddanni requested review from mrwojtek, wcourtney and verult as code owners August 25, 2025 17:43

github-actions bot added size: XL lines changed >1000 and removed size: L 250< lines changed <1000 labels Aug 25, 2025

ddddddanni force-pushed the sweep-p branch from c7486a5 to 4ce6f22 Compare August 25, 2025 18:26

github-actions bot added size: L 250< lines changed <1000 and removed size: XL lines changed >1000 labels Aug 25, 2025

github-actions bot added size: M 50< lines changed <250 size: XL lines changed >1000 and removed size: L 250< lines changed <1000 size: M 50< lines changed <250 labels Sep 1, 2025

ddddddanni force-pushed the sweep-p branch from b52fef4 to bc518f4 Compare September 1, 2025 22:54

github-actions bot added size: L 250< lines changed <1000 and removed size: XL lines changed >1000 labels Sep 1, 2025

ddddddanni added 3 commits September 1, 2025 15:56

initial version

c6f7cbd

Fix test and lint

9a28839

Allow user to specify insert strategy for pauli circuit creation

7addb7e

ddddddanni added 3 commits September 1, 2025 15:56

Add insert_strategy

02ddd0f

remote debug print

1ba8d2c

(for testing) unshuffle circuits in batch mode

0540d07

ddddddanni force-pushed the sweep-p branch from bc518f4 to 0540d07 Compare September 1, 2025 22:57

Having the sweep's key as a string instead of a symbol

cffc694

eliottrosenberg approved these changes Sep 4, 2025

View reviewed changes

ddddddanni added 3 commits September 10, 2025 18:48

Merge branch 'main' into sweep-p

338fde3

Revert the changes that were used for testing only

d5ba4fb

Merge branch 'main' into sweep-p

3ece94a

NoureldinYosri requested changes Sep 23, 2025

View reviewed changes

Address comments

121286c

NoureldinYosri approved these changes Sep 23, 2025

View reviewed changes

		return operations


		def _pauli_strings_to_basis_change_with_sweep(

(2/2) Add parameter sweep support for Pauli string measurements with readout mitigation #7569

Are you sure you want to change the base?

(2/2) Add parameter sweep support for Pauli string measurements with readout mitigation #7569

Conversation

ddddddanni commented Aug 12, 2025

Uh oh!

eliottrosenberg left a comment

Choose a reason for hiding this comment

Uh oh!

eliottrosenberg left a comment

Choose a reason for hiding this comment

Uh oh!

eliottrosenberg commented Aug 25, 2025

Uh oh!

review-notebook-app bot commented Aug 25, 2025

Uh oh!

ddddddanni commented Aug 25, 2025

Uh oh!

codecov bot commented Sep 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

eliottrosenberg left a comment

Choose a reason for hiding this comment

Uh oh!

ddddddanni commented Sep 15, 2025

Uh oh!

ddddddanni commented Sep 15, 2025

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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

codecov bot commented Sep 1, 2025 •

edited

Loading