Skip to content

Release 0.41.0

Choose a tag to compare

View all tags
@LuisAlfredoNu LuisAlfredoNu released this 14 Apr 15:22
· 1 commit to v0.41.0_release since this release
v0.41.0
0c97e03

New features since last release

  • Add native support for qml.PCPhase for lightning.qubit and lightning.gpu. (#1107)

  • Integrate sparse kernels into the Lightning-Qubit dynamic dispatcher, enhance the Python layers, and expand the testing process for dense and sparse operations. (#1094)

  • Add qml.PSWAP gate native implementation to PennyLane-Lightning devices. (#1088)

  • Implement new C++ kernels for efficient in-place multiplication of sparse matrices to state vectors, supporting both controlled and non-controlled gates, and add comprehensive tests for this new functionality. (#1085)

  • Add support for sparse qml.QubitUnitary gates for Lightning state-vector simulators. (#1068)

  • Lightning devices support dynamically allocated wires (e.g. qml.device("lightning.qubit")). (#1043)

  • Add support for Python 3.13 Lightning wheel builds. (#1001)

  • Add native qml.SX and qml.ctrl(qml.SX) gates to all lightning devices. (#731)

Breaking changes

  • Update the minimum required version of PennyLane to v0.40.0. (#1033)

Improvements

  • Install Pytorch 2.5.1 CPU in CIs and update make python command. (#1118)

  • Update TF, Keras and Torch versions in the CIs based on PennyLane's pinned versions. (#1112)

  • Build Lightning-GPU and Lightning-Tensor wheels against Python 3.10 and 3.13 on non-release PRs. (#1112)

  • Use cuQuantum API for calculating expectation value of Pauli sentences in Lightning-GPU. (#1104)

  • Update CIs to use requirements-tests.txt instead of requirements-dev.txt. (#1105)

  • Hide internal C++ APIs in Lightning docs. (#1096)

  • Implement the jaxpr_jvp method to compute the jvp of a jaxpr using lightning.qubit.
    This method currently only support the adjoint differentiation method. (#1087) (#1106)

  • Modify expval of named operators in Lightning-Qubit for in-place computation of expectation value, to avoid creating an intermediate statevector (#1079) (#565)

  • Device ("lightning.qubit", "lightning.gpu", "lightning.kokkos") pre-processing is now included in the execution pipeline when program capture is enabled. (#1084)

  • Add semgrep/bandit security scanning for pull requests. (#1100)

  • Hide anonymous namespaces in Lightning docs. (#1097)

  • Expand test structure to efficiently handle sparse data. (#1085)

  • Remove redundant reset_state calls for circuit execution when state vector is freshly initialized. (#1076)

  • Introduce a generalized sparse gate selection system via the _observable_is_sparse method in the base measurement class, enabling future expansion for any number of sparse observables. (#1068)

  • Optimize the copy of a input state-vector into Lightning-GPU. (#1071)

  • Add new GitHub workflow to enable benchmarking within pull request. (#1073)

  • Fix wheel naming in Docker builds for setuptools v75.8.1 compatibility. (#1075)

  • Use native C++ kernels for controlled-adjoint and adjoint-controlled of supported operations. (#1063)

  • In Lightning-Tensor, allow qml.MPSPrep to accept an MPS with len(MPS) = n_wires-1. (#1064)

  • Capture execution via dev.eval_jaxpr can now be used with jax.jit and jax.vmap. (#1055)

  • Add an execution_config keyword argument to LightningBase.eval_jaxpr to accommodate a Device API change. (#1067)

  • Remove the old device API references in the Lightning repo and test suite. (#1057)

  • Update GPU workflow to use the new self-hosted GPU runner infrastructure. (#1061)

  • Optimize MPO operation in lightning.tensor. (#1054)

  • Update qml.ControlledQubitUnitary tests following the latest updates in PennyLane. (#1047)

  • Remove unnecessary adjoint pytest skip for Lightning-Kokkos. (#1048)

  • Add isort to the precommit hook. (#1052)

  • Update source code to use black formatter 25.1.0. (#1059)

  • Replace the type checking using the property return_type of MeasurementProcess with direct isinstance checks. (#1044)

  • Update Lightning integration tests following the gradient_kwargs deprecation in Pennylane. (#1045)

  • Update qml.MultiControlledX tests following the latest updates in PennyLane. (#1040)

  • Merge the v0.40.0-rc branch to the master and bump version. (#1038)

  • Reduce the number of shots in the PennyLane Python tests on CIs, from 20k to 10k. (#1046)

  • Program transformed by qml.defer_measurements can be executed on lightning.qubit. Supports ctrl and adjoint with program capture in lightning.qubit. (#1069)

Documentation

  • Modify CHANGELOG.md to prepare it for release v0.41.0 (#1117)

  • Update install instructions for lightning.gpu. (#1037)

Bug fixes

  • Fix the issue with decomposing controlled qml.SProd and qml.Exp operations. (#1120)

  • Fix the validation for all wires present after adding the extra wires from apply mid_circuit_measurements. (#1119)

  • Fix the issue with pip installing PennyLane (and Lightning-Qubit) on Windows. (#1116)

  • Fix the stable/stable issue with missing pytest-split. (#1112)

  • Add missing GH workflow step id, python setup, and virtual environment for aarch64 cuda. (#1113)

  • Fix the development wheel upload step for Python 3.13 by following the same syntax as for the other Python wheels. (#1111)

  • Add a workflow to run label-external-pull-request to label pull requests from external forks. (#1110)

  • Fix the test_preprocess test skip condition for lightning.tensor. (#1092)

  • Fix measurements with empty wires and operators for statevectors with dynamically allocated wires. (#1081)

  • Fix unit tests that were being skipped in testApplyControlledPhaseShift. (#1083)

  • Fix Github CI for aarch64 cuda to clean up after runs. (#1074)

  • Increase maximum time for aarch64-CUDA Github CI action. (#1072)

  • Fix SyntaxWarning from is with a literal in Python tests. (#1070)

  • Fix CI to collect Python code coverage for Lightning-Qubit and Lightning-Kokkos CPU. (#1053)

  • Upgrade the version of QEMU image to fix AARCH64 wheel action. (#1056)

  • Patch MultiControlledX tests to fix stable/latest failures after the v0.40.0 release. (#1046)

  • Update Github CI to use Ubuntu 24 and remove libopenblas-base package. (#1041)

  • Update the eval_jaxpr method to handle the new signatures for the cond, while, and for primitives. (#1051)

Contributors

This release contains contributions from (in alphabetical order):

Runor Agbaire,
Catalina Albornoz,
Ali Asadi,
Saeed Bohloul,
Astral Cai,
Yushao Chen,
Amintor Dusko,
Pietropaolo Frisoni,
Christina Lee,
Joseph Lee,
Rashid N H M,
Luis Alfredo Nuñez Meneses,
Lee J. O'Riordan,
Mudit Pandey,
Andrija Paurevic,
Alex Preciado,
Shuli Shu,
Jake Zaia

Assets 59
Loading