Adding CPMG protocol #1057

andrea-pasquale · 2024-12-11T08:14:17Z

Closes #896.
Here is an example: http://login.qrccluster.com:9000/M0By3hjUTAakgT8Jg1VlXA==
I've also fixed a factor 2 in the spin_echo protocol.

codecov · 2024-12-11T08:14:52Z

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.02%. Comparing base (d7745ae) to head (1a8f9f7).
Report is 82 commits behind head on 0.2.

Additional details and impacted files

@@            Coverage Diff             @@
##              0.2    #1057      +/-   ##
==========================================
- Coverage   97.03%   97.02%   -0.02%     
==========================================
  Files          98       99       +1     
  Lines        7895     7921      +26     
==========================================
+ Hits         7661     7685      +24     
- Misses        234      236       +2

Flag	Coverage Δ
unittests	`97.02% <100.00%> (-0.02%)`	⬇️

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

Files with missing lines	Coverage Δ
src/qibocal/protocols/__init__.py	`100.00% <100.00%> (ø)`
src/qibocal/protocols/coherence/cpmg.py	`100.00% <100.00%> (ø)`
src/qibocal/protocols/coherence/spin_echo.py	`100.00% <100.00%> (ø)`
...rc/qibocal/protocols/coherence/spin_echo_signal.py	`100.00% <100.00%> (ø)`
src/qibocal/protocols/coherence/t1.py	`100.00% <ø> (ø)`
src/qibocal/protocols/coherence/t2.py	`100.00% <100.00%> (ø)`
src/qibocal/protocols/coherence/utils.py	`98.11% <100.00%> (+0.77%)`	⬆️
src/qibocal/protocols/coherence/zeno.py	`100.00% <100.00%> (ø)`

... and 1 file with indirect coverage changes

Edoardo-Pedicillo · 2024-12-12T10:15:49Z

doc/source/protocols/cpmg/cpmg.rst

+In this section we show how to run the dynamical decoupling sequence CPMG.
+
+The CPMG sequence consists in applying N equally spaced :math:`\pi` pulses
+within two :math:`\pi` pulses. By increasing the number of :math:`\pi` pulses :math:`T_2`


within two :math:`\pi` pulses.

It should be pi/2, right ?

Edoardo-Pedicillo · 2024-12-12T11:51:19Z

src/qibocal/protocols/coherence/cpmg.py

+    return figures, fitting_report
+
+
+cpmg = Routine(_acquisition, _fit, _plot, _update)


The _update is the spin_echo one, it means the t2_spin_echo value is updated. This is the case when N = 1, but generally speaking the the $T2^{(N)}$ is sensitive to different frequencies noise, so we should differentiate the cases.

Yes, I was thinking about this. Perhaps at this point it is easier to just remove the update for this protocol. What do you think?

Edoardo-Pedicillo · 2024-12-12T11:52:34Z

src/qibocal/protocols/coherence/spin_echo.py

Can we delete this routine to reduce code duplication ?

Actually, the spin-echo protocol doesn't implement the CPMG sequence but the CP sequence, which is slightly different.

qibocal/src/qibocal/protocols/coherence/spin_echo.py

Line 44 in db00d75

sequence, delays = dynamical_decoupling_sequence(platform, targets, kind="CP")

But yes, generally I agree the two protocols are exactly the same apart from the sequence type.

Edoardo-Pedicillo · 2024-12-12T12:15:29Z

src/qibocal/protocols/coherence/utils.py

+        sequence.append((qd_channel, rx90_pulse))
+
+        for i in range(n):
+            sequence.append((qd_channel, drive_delays[2 * i]))
+            sequence.append((ro_channel, ro_delays[2 * i]))
+            sequence.append((qd_channel, pulse))
+            sequence.append((qd_channel, drive_delays[2 * i + 1]))
+            sequence.append((ro_channel, ro_delays[2 * i + 1]))
+
+        sequence.append((qd_channel, rx90_pulse))
+
+        sequence.append(
+            (
+                ro_channel,
+                Delay(duration=2 * rx90_pulse.duration + n * pulse.duration),
+            )


The sequence should be the following

I don't see the t/2N delays after/before the first/last pulse.

The sequence can be re-written as
RX90 - (wait(t/2N) RY wait(t/2N))^N RX90
which is exactly what I am doing in the for loop.

I've specified that the delay should have also a factor 1/2N in the acquisition

qibocal/src/qibocal/protocols/coherence/cpmg.py

Line 61 in db00d75

values=wait_range / 2 / params.n,

Edoardo-Pedicillo · 2024-12-12T12:19:46Z

doc/source/protocols/cpmg/cpmg.rst

+
+The CPMG sequence consists in applying N equally spaced :math:`\pi` pulses
+within two :math:`\pi` pulses. By increasing the number of :math:`\pi` pulses :math:`T_2`
+should increase until reaching the :math:`2 T_1` limit.


This behaviour depends on the kind of noise.

I am not sure about this. Generally by increasing the number of pi pulses according to here I believe that we should be able to mitigate noises of the type 1/f. Then I don't know whether we could have other noises which are not 1/f but I don't know if we can detect them through a CPMG sequence.
Did you have something else in mind?

Did you have something else in mind?

You could have a white noise but I think the conclusions are the same. I just suggest to add what you have written in this comment in the docs.

Then I don't know whether we could have other noises which are not 1/f but I don't know if we can detect them through a CPMG sequence.

By changing the number of pulses increase the sensitivity to higher frequencies, so I would say yes

andrea-pasquale

Thanks for the comments @Edoardo-Pedicillo.
I will try to reduce code duplication :)

andrea-pasquale · 2024-12-12T13:18:12Z

src/qibocal/protocols/coherence/cpmg.py

+    return figures, fitting_report
+
+
+cpmg = Routine(_acquisition, _fit, _plot, _update)


Yes, I was thinking about this. Perhaps at this point it is easier to just remove the update for this protocol. What do you think?

andrea-pasquale · 2024-12-12T13:25:20Z

src/qibocal/protocols/coherence/spin_echo.py

Actually, the spin-echo protocol doesn't implement the CPMG sequence but the CP sequence, which is slightly different.

qibocal/src/qibocal/protocols/coherence/spin_echo.py

Line 44 in db00d75

sequence, delays = dynamical_decoupling_sequence(platform, targets, kind="CP")

But yes, generally I agree the two protocols are exactly the same apart from the sequence type.

andrea-pasquale · 2024-12-12T13:36:47Z

src/qibocal/protocols/coherence/utils.py

+        sequence.append((qd_channel, rx90_pulse))
+
+        for i in range(n):
+            sequence.append((qd_channel, drive_delays[2 * i]))
+            sequence.append((ro_channel, ro_delays[2 * i]))
+            sequence.append((qd_channel, pulse))
+            sequence.append((qd_channel, drive_delays[2 * i + 1]))
+            sequence.append((ro_channel, ro_delays[2 * i + 1]))
+
+        sequence.append((qd_channel, rx90_pulse))
+
+        sequence.append(
+            (
+                ro_channel,
+                Delay(duration=2 * rx90_pulse.duration + n * pulse.duration),
+            )


The sequence can be re-written as
RX90 - (wait(t/2N) RY wait(t/2N))^N RX90
which is exactly what I am doing in the for loop.

I've specified that the delay should have also a factor 1/2N in the acquisition

qibocal/src/qibocal/protocols/coherence/cpmg.py

Line 61 in db00d75

values=wait_range / 2 / params.n,

andrea-pasquale · 2024-12-12T13:42:13Z

doc/source/protocols/cpmg/cpmg.rst

+
+The CPMG sequence consists in applying N equally spaced :math:`\pi` pulses
+within two :math:`\pi` pulses. By increasing the number of :math:`\pi` pulses :math:`T_2`
+should increase until reaching the :math:`2 T_1` limit.


I am not sure about this. Generally by increasing the number of pi pulses according to here I believe that we should be able to mitigate noises of the type 1/f. Then I don't know whether we could have other noises which are not 1/f but I don't know if we can detect them through a CPMG sequence.
Did you have something else in mind?

andrea-pasquale added 2 commits December 9, 2024 18:35

feat: Adding CPMG protocol

0b625f3

feat: Adding documentation for CPMG

db00d75

andrea-pasquale added the calibration label Dec 11, 2024

andrea-pasquale self-assigned this Dec 11, 2024

andrea-pasquale requested review from Edoardo-Pedicillo and ElStabilini December 11, 2024 08:15

ElStabilini approved these changes Dec 11, 2024

View reviewed changes

Edoardo-Pedicillo reviewed Dec 12, 2024

View reviewed changes

Edoardo-Pedicillo added this to the Qibocal - Qibolab 0.2 milestone Dec 12, 2024

andrea-pasquale commented Dec 12, 2024

View reviewed changes

andrea-pasquale added 2 commits December 12, 2024 22:47

doc: Improve documentation

6fcf9d6

refactor: Reduce code duplication

1a8f9f7

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding CPMG protocol #1057

Adding CPMG protocol #1057

andrea-pasquale commented Dec 11, 2024

codecov bot commented Dec 11, 2024 •

edited

Loading

Edoardo-Pedicillo Dec 12, 2024

Edoardo-Pedicillo Dec 12, 2024

andrea-pasquale Dec 12, 2024

Edoardo-Pedicillo Dec 12, 2024

andrea-pasquale Dec 12, 2024

Edoardo-Pedicillo Dec 12, 2024 •

edited

Loading

andrea-pasquale Dec 12, 2024

Edoardo-Pedicillo Dec 12, 2024

Edoardo-Pedicillo Dec 12, 2024

andrea-pasquale Dec 12, 2024

Edoardo-Pedicillo Dec 12, 2024

andrea-pasquale left a comment

andrea-pasquale Dec 12, 2024

andrea-pasquale Dec 12, 2024

andrea-pasquale Dec 12, 2024

andrea-pasquale Dec 12, 2024

		return figures, fitting_report


		cpmg = Routine(_acquisition, _fit, _plot, _update)

Adding CPMG protocol #1057

Are you sure you want to change the base?

Adding CPMG protocol #1057

Conversation

andrea-pasquale commented Dec 11, 2024

codecov bot commented Dec 11, 2024 • edited Loading

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Edoardo-Pedicillo Dec 12, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

andrea-pasquale left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov bot commented Dec 11, 2024 •

edited

Loading

Edoardo-Pedicillo Dec 12, 2024 •

edited

Loading