feat: qudit noise model (2/5) (#1854) by bramathon · Pull Request #1859 · rigetti/pyquil

bramathon · 2026-07-04T12:40:29Z

Part of the effort to land the experimental noise model and simulation module (originally PR #1848, branch nonbreaking-noise-model) as a reviewable stack. This is PR 2 of 5, stacked on #1858 — review/merge in order. Closes #1854.

This PR introduces the new quax-backed qudit noise model under pyquil/noise/:

Channel, MeasurementChannel, ResetChannel, and CycleChannel (_channels.py), and the new NoiseModel container plus DepolarizingNoiseModel / CompositeNoiseModel and fidelity estimators (_noise_model.py).
The historical pyquil.noise public API is preserved: pyquil/noise.py is moved to pyquil/noise/_legacy_noise.py and re-exported from pyquil/noise/__init__.py, with the legacy NoiseModel/KrausModel now marked deprecated in favor of the quax-based model.

It also adds qudit DefGate support in quilbase.py (matrix dimensions may now be any perfect power, e.g. 3, 9, …, not just powers of two), which the qutrit noise channels and later simulators rely on, and enables float64 (jax_enable_x64) for the test suite.

A handful of ResetChannel tests verify behavior end-to-end through the density-matrix simulator; since that simulator lands in the next PR, those specific assertions are deferred to #1855 (they are restored there). Everything else in the noise module is fully tested here (test_noise_model.py).

Stack

Drop Python 3.9 #1852 — Drop Python 3.9
Qudit Noise model #1854 — Qudit noise model (this PR)
Experimental exact simulators #1855 — Experimental exact simulators
Trajectory simulator #1856 — Trajectory simulator
Dynamic shape trajectory simulator #1857 — Dynamic-shape trajectory simulator

Tracked in #1863.

github-actions · 2026-07-04T12:43:20Z

Bencher Report

Project	pyquil
Branch	1854-qudit-noise-model
Testbed	ci-runner-linux

⚠️ WARNING: No Threshold found!
Without a Threshold, no Alerts will ever be generated.
Latency (nanoseconds (ns))
Click here to create a new Threshold
For more information, see the Threshold documentation.
To only post results if a Threshold exists, set the --ci-only-thresholds flag.

Click to view all benchmark results

Benchmark	Latency	seconds (s)
test/benchmarks/test_program.py::test_copy_everything_except_instructions	📈 view plot ⚠️ NO THRESHOLD	11.08 s
test/benchmarks/test_program.py::test_instructions	📈 view plot ⚠️ NO THRESHOLD	3.74 s
test/benchmarks/test_program.py::test_iteration	📈 view plot ⚠️ NO THRESHOLD	3.73 s

🐰 View full continuous benchmarking report in Bencher

asaites

I've made several comments, which I hope will be useful. I might spend a little more time looking through _noise_model.py, but for the most part, I think I've gone through this pretty thoroughly (and am doing my best to keep up with the physics).

Most of my comments are more style than anything, but I have a few concerns. Overall, though, this looks pretty solid!

Introduce the new noise model under pyquil/noise/ (Channel, MeasurementChannel, ResetChannel, CycleChannel and the quax-based NoiseModel), preserving the legacy pyquil.noise API via _legacy_noise.py. Add qudit DefGate support in quilbase and enable float64 for tests. Part of splitting PR #1848 into a reviewable stack. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

github-actions · 2026-07-17T09:57:29Z

☂️ Python Coverage

current status: ❌

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
8296	7246	87%	87%	🟢

New Files

File	Coverage	Status
pyquil/noise/init.py	100%	🟢
pyquil/noise/_channels.py	85%	🔴
pyquil/noise/_noise_model.py	76%	🔴
TOTAL	87%	🔴

Modified Files

File	Coverage	Status
pyquil/quilbase.py	94%	🟢
TOTAL	94%	🟢

updated for commit: 69edd32 by action🐍

bramathon · 2026-07-20T20:09:13Z

I've done a substantial refactor to add support for Lindbladian channels. The way this works is as follows:

ChannelProtocol defines the protocol for a channel which provides a superoperator associated with a gate. There are two concrete classes for this: Channel and SuperopChannel. The Channel has the _LindbladianBacked mixin, which means that it not only has a superoperator, but a lindbladian which generates it. This is now our canonical Channel while SuperopChannel is relegated to special use cases where we don't know the Lindbladian.

ResetChannel receives the same treatment, becoming _LindbladianBacked and with two concrete implementations.

CycleChannel and MeasureChannel are unchanged.

asaites · 2026-07-21T07:12:14Z

I think there's a problem with the jax or quax dependency version.

- Remove the awkward `SupportsReal` protocol; type `_resolve_params` against the canonical `Sequence[ParameterDesignator]` (drops the `ResolvableParam` alias and the unused `Protocol`/`SupportsFloat`/ `runtime_checkable` imports). - Inline the one-line noise combination into `Channel.__add__` and have `__matmul__` delegate to it (drop `_combine_noise_keeping_gate`). - Fix D401 (imperative mood) on the two `as_post_gate_noise` docstrings. - Drop `# type: ignore[attr-defined]` in the channel `__eq__` methods: mypy narrows `other` after the `type(self) is not type(other)` guard, so the comments were flagged unused. - Sort the `quilbase` import block (ruff I001). Makes `ruff check`, `ruff format --check`, and `mypy pyquil` pass. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ViuQ59tEfPw6z36i1fkFhg

The locked jax 0.10.0 predates the `enable_eigvec_derivs` kwarg that rigetti-quax 0.7.0 passes to `jax.lax.linalg.eig`, so every quax-backed Channel construction raised `TypeError` in CI. Update the lock to the newest jax each supported interpreter allows — jax 0.10.2 on Python 3.11 and jax 0.11.0 on Python 3.12 (0.11.0 dropped 3.11) — both of which include the kwarg. Proper version pinning to follow upstream in quax. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ViuQ59tEfPw6z36i1fkFhg

The pinned `pandoc` 2.4b0 (a beta) imports the removed `pkg_resources` module in `pandoc/utils.py`, which fails under the project's setuptools >=83 (setuptools dropped `pkg_resources` in 81). The docs `conf.py` imports `pandoc` in its `builder-inited` handler to render the changelog, so `make doctest` aborted with `No module named 'pkg_resources'`. The stable 2.4 release switched to `importlib.resources` and imports cleanly. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ViuQ59tEfPw6z36i1fkFhg

windsurf-bot

1 file skipped due to size limits:

pyquil/noise/_channels.py

Looks good to me 🤙

💡 To request another review, post a new comment with "/windsurf-review".

bramathon · 2026-07-21T20:24:30Z

I think there's a problem with the jax or quax dependency version.

Confirmed, we need to bump the minimum version of quax

erichulburd

I've reviewed everything up to quilbase and test_noise_model, which I'll shoot to cover tomorrow. This round's comments are all discretionary.

erichulburd

Largely looks good to me. I'm going to go through and react to my comments that I care more about with ❗ . Close everything out at your discretion and then I'll approve.

@deprecated

Correctness fixes, each with a regression test: - from_coherence_times: the T2 -> Tphi conversion omitted the factor of two on T1 (1/T2 = 1/(2*T1) + 1/Tphi), so the default t2 = 2*t1 produced a negative pure-dephasing time, NaN fidelities and a LinAlgError. Every physical T1 < T2 <= 2*T1 was affected. Fixed in Channel and ResetChannel, with a T2 <= 2*T1 guard. - to_stochastic_channel: hand-rolled Kronecker products used the opposite superoperator convention to quax (kron(conj(U), U)) and an inverse that is not the superoperator of any unitary, yielding a non-CP, non-TP channel with process fidelity 0.003 instead of 0.996. Now composed through quax. - pauli_twirl: twirled the full process, gate included, which deletes any gate whose transfer matrix is not already diagonal (RX(pi/2): 0.9925 -> 0.4975). Now twirls error_process and recomposes the gate, which for Clifford gates is exactly the standard twirl (1/4^n) sum_k P'_k^dag E P_k with P'_k = U P_k U^dag. - SuperopChannel.from_pauli_noise: an explicit identity key overrode the implicit normalization, breaking trace preservation. Now rejected. - Channel.__add__: raise on a gate_time mismatch instead of silently reinterpreting the other operand's per-unit-time rates. - SuperopResetChannel.from_reset_fidelity: the argument was applied as a depolarizing shrink factor, so requesting 0.9 gave 0.95. Now inverted through F = (1 + (d-1)p)/d so the channel reports back the requested fidelity. - CycleChannel.average_gate_fidelity: average gate fidelity does not factorize over a tensor product; convert once from the exact process-fidelity product. - Channel.__matmul__: no longer aliased to __add__. `@` is the exact superoperator composition (returning SuperopChannel); `+` is the generator-level noise sum. The two agree only for commuting generators. - from_random_coherent_error: solve the rotation angle numerically so the requested process fidelity is hit for multi-qubit gates too. - Raise instead of silently clipping out-of-range depolarizing constants, readout fidelities and asymmetries, and imaginary gate parameters. - Guard the qubit-only Pauli constructors and analyses against qudits with a message naming the operation, rather than a raw JAX shape error. - DefPermutationGate.num_args: read prime-power lengths like DefGate does. - from_isa: crashed on a complex ISA parameter; skipped perfect gate fidelities but not perfect readout fidelities. Review comments addressed: - Rename the fidelity properties to match quax: fidelity -> average_gate_fidelity, pauli_fidelity -> process_fidelity (and the infidelities). `fidelity` previously meant an average gate fidelity on gate channels but a process fidelity on reset channels. - Rename unitary -> ideal_unitary, and as_post_gate_noise() -> the error_process property, so the error-only object carries the error name. - CycleChannel: serialize the cycle instruction and DEFCIRCUIT so an externally-built cycle survives a round trip; add __or__ so `a | b | c` builds one flat cycle; validate by instruction equality rather than str(); return expanded_instructions as a tuple. - NoiseModel.from_isa now takes a qcs_sdk InstructionSetArchitecture; the rpcq CompilerISA path is from_compiler_isa, deprecated for pyQuil v5. - Replace NoiseModel's hand-written __init__ with __post_init__ validation. - Drop the @deprecated decorator from the legacy Kraus NoiseModel: it fired from pyQuil's own internals (get_qc(noisy=True)) while the replacement is still private. - Re-export the remaining private names from the old pyquil/noise.py. - TypeAlias on the union aliases, TypedDict for the packed JSON payloads, @Final on the concrete channels, _DEFAULT_GATE_TIME, _resolve_params -> _evaluate_parameter_designators, _as_channel -> _to_superop_channel, _qudit_dims -> _operator_dims_from_dimension. - Say "prime power" consistently in the DefGate dimension errors, and add qudit table tests for num_args and _integer_base_and_exponent. - Memoize DepolarizingNoiseModel's synthesized channels; resolve each gate unitary once per construction; use math.prod in estimate_program_fidelity; drop an unreachable serialization branch. - Fix the quax keyword in test_from_depolarizing_constant that was failing CI. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013QB5EGDshZbFtBvyfPPQqU

This was referenced Jul 4, 2026

feat: experimental exact simulators (3/5) (#1855) #1860

Merged

Experimental noise model & simulation module (tracking) #1863

Open

asaites self-assigned this Jul 13, 2026

Base automatically changed from 1852-drop-python-39 to master July 16, 2026 19:06

asaites reviewed Jul 16, 2026

View reviewed changes

bramathon and others added 2 commits July 17, 2026 08:46

Address review comments

65aa919

bramathon force-pushed the 1854-qudit-noise-model branch from a2e40e8 to 65aa919 Compare July 17, 2026 09:52

Add Lindbladian channels

948937d

asaites reviewed Jul 20, 2026

View reviewed changes

Comment thread pyquil/quilbase.py Outdated

Comment thread pyquil/quilbase.py Outdated

Comment thread pyquil/quilbase.py Outdated

Comment thread pyquil/noise/_channels.py Outdated

Renames

f7fdf61

asaites reviewed Jul 21, 2026

View reviewed changes

Comment thread pyquil/noise/_noise_model.py Outdated

Comment thread pyquil/noise/_channels.py Outdated

bramathon and others added 4 commits July 21, 2026 17:26

Refactor channels

2055f93

bramathon marked this pull request as ready for review July 21, 2026 20:19

bramathon requested a review from a team as a code owner July 21, 2026 20:19

windsurf-bot Bot reviewed Jul 21, 2026

View reviewed changes

Make qubits a plain property

22a7e41

Fix isinstance checks

f45823c

erichulburd reviewed Aug 3, 2026

View reviewed changes

erichulburd reviewed Aug 4, 2026

View reviewed changes

erichulburd approved these changes Aug 18, 2026

View reviewed changes

erichulburd merged commit 1dbd8e9 into master Aug 18, 2026
14 checks passed

erichulburd deleted the 1854-qudit-noise-model branch August 18, 2026 23:40

bramathon mentioned this pull request Sep 18, 2026

feat: trajectory simulator (4/5) (#1856) #1861

Merged

Conversation

bramathon commented Jul 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Stack

Uh oh!

github-actions Bot commented Jul 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Bencher Report

⚠️ WARNING: No Threshold found!

Uh oh!

asaites 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!

Uh oh!

github-actions Bot commented Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

☂️ Python Coverage

Overall Coverage

New Files

Modified Files

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

bramathon commented Jul 20, 2026

Uh oh!

Uh oh!

Uh oh!

asaites commented Jul 21, 2026

Uh oh!

windsurf-bot Bot left a comment

Choose a reason for hiding this comment

Uh oh!

bramathon commented Jul 21, 2026

Uh oh!

erichulburd 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!

Uh oh!

erichulburd 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!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

bramathon commented Jul 4, 2026 •

edited

Loading

github-actions Bot commented Jul 4, 2026 •

edited

Loading

github-actions Bot commented Jul 17, 2026 •

edited

Loading