spin lock sequence by jevillegasd · Pull Request #1540 · qiboteam/qibocal

jevillegasd · 2026-06-15T14:16:33Z

This is a first trial to address #1525

The experiment runs, but I run out of memory surprisingly fast, even though I use sweepers for both amplitude and time iterators. Maybe this works better if I use of of the iterations to be a for loop outside (at the cost of the experiment being longer than necessary).

for more information, see https://pre-commit.ci

…cation. This is still however unable to run for long pulse durations in the spin lock pulse.

for more information, see https://pre-commit.ci

codecov · 2026-06-15T15:24:15Z

Codecov Report

❌ Patch coverage is 36.57143% with 111 lines in your changes missing coverage. Please review.
✅ Project coverage is 93.40%. Comparing base (e0d5dcc) to head (b00f190).

Files with missing lines	Patch %	Lines
src/qibocal/protocols/coherence/spin_lock.py	36.20%	111 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1540      +/-   ##
==========================================
- Coverage   94.36%   93.40%   -0.97%     
==========================================
  Files         136      137       +1     
  Lines       10634    10809     +175     
==========================================
+ Hits        10035    10096      +61     
- Misses        599      713     +114

Flag	Coverage Δ
unittests	`93.40% <36.57%> (-0.97%)`	⬇️

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

Files with missing lines	Coverage Δ
src/qibocal/protocols/coherence/__init__.py	`100.00% <100.00%> (ø)`
src/qibocal/protocols/coherence/spin_lock.py	`36.20% <36.20%> (ø)`

... and 5 files with indirect coverage changes

🚀 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.

jevillegasd · 2026-06-15T15:52:40Z

Im kind of stuck trying to make the protocol run for long durations of the spin lock pulse. I see that the driver supports batching , would this be a good way to allow longer sequences.

for more information, see https://pre-commit.ci

alecandido · 2026-06-19T06:15:11Z

Im kind of stuck trying to make the protocol run for long durations of the spin lock pulse. I see that the driver supports batching , would this be a good way to allow longer sequences.

Just for the sake of future reference: batching is quite the opposite. It is a way to accelerate the execution of short sequences, but quite ineffective on long ones.

For very long sequences, we just need to identify the limiting factor (instruction memory, waveform memory, acquisition memory) and check whether there is a specific implementation lifting the bottleneck.

In this case, the sequence is in a sense small, since it is made by few operations. The reason it is failing is because of waveform memory limitations, which can only be lifted by not realizing the spin-locking pulse as a single arbitrary waveform. Which can be realized by either decomposing it in parts (e.g. repeating many back-to-back identical rectangular units) or with a compressed pulse implementation (which is possible for the flux pulse).

Use LongPulse instead of rectangular pulses

lballerio

Overall I think is good, just small suggestions.
Btw @jevillegasd can you also paste on this PR some data plot you generated while running this experiment?

lballerio · 2026-06-23T11:53:20Z

+    duration_min: int
+    """Minimum spin-lock pulse duration [ns]."""
+    duration_max: int
+    """Maximum spin-lock pulse duration [ns]."""
+    duration_step: int
+    """Step spin-lock pulse duration [ns]."""


we decided to move into ranges as input, i.e. tuples of the form (init_value, final_value, step).

Suggested change

duration_min: int

"""Minimum spin-lock pulse duration [ns]."""

duration_max: int

"""Maximum spin-lock pulse duration [ns]."""

duration_step: int

"""Step spin-lock pulse duration [ns]."""

duration_range: tuple[float, float, float]

"""Spin-lock pulse duration range [ns]."""

Also in theory we can admit times to be float for sub-ns sampling rate; then drivers will convert it accordingly.

lballerio · 2026-06-23T11:53:58Z

+    @property
+    def duration_range(self) -> tuple[int, int, int]:
+        """Return a tuple with the spin-lock pulse duration range."""
+        return self.duration_min, self.duration_max, self.duration_step


noy necessary anymore since we are taking range as input.

Suggested change

@property

def duration_range(self) -> tuple[int, int, int]:

"""Return a tuple with the spin-lock pulse duration range."""

return self.duration_min, self.duration_max, self.duration_step

lballerio · 2026-06-23T11:55:48Z

+    amplitude_min: float
+    """Minimum spin-lock pulse amplitude [a.u.]."""
+    amplitude_max: float
+    """Maximum spin-lock pulse amplitude [a.u.]."""
+    n_amplitudes: int
+    """Number of amplitude points, log-spaced between amplitude_min and amplitude_max."""


Honestly here I don't know whether is better to ask for a tuple or not. However I will write amplitude_values as a @property of the SpinLockParameters class.

lballerio · 2026-06-23T11:57:18Z

+    gamma: dict[QubitId, list[float]] = field(default_factory=dict)
+    """Relaxation rate :math:`\\Gamma_{1\\rho} = S(\\nu_R)` for each amplitude [Hz]."""
+    gamma_error: dict[QubitId, list[float]] = field(default_factory=dict)
+    """Error on :math:`\\Gamma_{1\\rho}` [Hz]."""


Maybe better to save them together?
Something like:

Suggested change

gamma: dict[QubitId, list[float]] = field(default_factory=dict)

"""Relaxation rate :math:`\\Gamma_{1\\rho} = S(\\nu_R)` for each amplitude [Hz]."""

gamma_error: dict[QubitId, list[float]] = field(default_factory=dict)

"""Error on :math:`\\Gamma_{1\\rho}` [Hz]."""

gamma: dict[QubitId, list[float, float]] = field(default_factory=dict)

"""Relaxation rate :math:`\\Gamma_{1\\rho} = S(\\nu_R)` and its error for each amplitude [Hz]."""

lballerio · 2026-06-23T12:00:41Z

+) -> SpinLockData:
+    """Data acquisition for the spin-lock (T1rho) experiment."""
+
+    duration_range = np.arange(*params.duration_range)


I would call in a different way since it is not a range and then might be confused with params.duration_range.

lballerio · 2026-06-23T12:05:03Z

+        qd_channel = platform.qubits[q].drive
+        ro_channel, ro_pulse = natives.MZ()[0]
+        rx_pulse = natives.RX()[0][1]


Is there a reason why not:

ro_channel, ro_pulse = natives.MZ()[0] rx_pulse, qd_channel = natives.RX()[0][1]

that I cannot see?

lballerio · 2026-06-23T12:11:33Z

+            nshots=params.nshots,
+            relaxation_time=params.relaxation_time,
+            acquisition_type=AcquisitionType.DISCRIMINATION,
+            averaging_mode=AveragingMode.SINGLESHOT,


Is there a reason why you set SINGLESHOT as acquisition mode and not CYCLIC?

lballerio · 2026-06-23T12:12:29Z

+        )
+
+        for q in targets:
+            prob = probability(results[ro_pulses[q].id], state=1)


if you use CYCLIC:

Suggested change

prob = probability(results[ro_pulses[q].id], state=1)

prob = results[ro_pulses[q].id]

lballerio · 2026-06-23T12:18:33Z

+    gamma = {}
+    gamma_error = {}


as before, maybe error and measurement can be incorporated

lballerio · 2026-06-23T12:23:27Z

+    return SpinLockResults(t1rho, gamma, gamma_error, rabi_frequency, fitted_parameters)
+
+
+def _plot(data: SpinLockData, target: QubitId, fit: SpinLockResults = None):


Suggested change

def _plot(data: SpinLockData, target: QubitId, fit: SpinLockResults = None):

def _plot(data: SpinLockData, target: QubitId, fit: SpinLockResults | None = None):

lballerio · 2026-06-23T12:26:56Z

Btw also rebase (in this case is straightforward) and try to solve the failing tests

jevillegasd and others added 4 commits June 15, 2026 18:12

base implementation of spin lock sequence

4b29d21

[pre-commit.ci] auto fixes from pre-commit.com hooks

e1adea5

for more information, see https://pre-commit.ci

Use a pulse library like in #1508 to try a more efficient memory allo…

2ab0cd7

…cation. This is still however unable to run for long pulse durations in the spin lock pulse.

[pre-commit.ci] auto fixes from pre-commit.com hooks

b00f190

for more information, see https://pre-commit.ci

jevillegasd and others added 6 commits June 16, 2026 13:02

Use LongePulse instead of rectangular

be60d79

[pre-commit.ci] auto fixes from pre-commit.com hooks

2434eea

for more information, see https://pre-commit.ci

change amplitude range to be logspaced (run in sequences)

bd44988

[pre-commit.ci] auto fixes from pre-commit.com hooks

c6cf9e0

for more information, see https://pre-commit.ci

format fixes

5e0ba8a

[pre-commit.ci] auto fixes from pre-commit.com hooks

9128150

for more information, see https://pre-commit.ci

jevillegasd added 2 commits June 22, 2026 17:02

Merge pull request #1543 from qiboteam/sls_long_pulses

27e9626

Use LongPulse instead of rectangular pulses

Merge branch 'main' into spin_lock_spectroscopy

a9f4480

lballerio requested changes Jun 23, 2026

View reviewed changes

	prob = probability(results[ro_pulses[q].id], state=1)
	prob = results[ro_pulses[q].id]

		return SpinLockResults(t1rho, gamma, gamma_error, rabi_frequency, fitted_parameters)


		def _plot(data: SpinLockData, target: QubitId, fit: SpinLockResults = None):

Uh oh!

Conversation

jevillegasd commented Jun 15, 2026

Uh oh!

codecov Bot commented Jun 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

jevillegasd commented Jun 15, 2026

Uh oh!

alecandido commented Jun 19, 2026

Uh oh!

lballerio 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!

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!

lballerio commented Jun 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

codecov Bot commented Jun 15, 2026 •

edited

Loading