Release 0.38.0

New features since last release

• Add qml.StatePrep() and qml.QubitStateVector() support to lightning.tensor. (#849)

• Add analytic qml.probs() measurement support to lightning.tensor. (#830)

• Add qml.state() measurement support to lightning.tensor. (#827)

• Add Lightning-GPU Linux (AArch64 + GraceHopper) wheels to PyPI. (#815)

• Add var support to lightning.tensor. Note that var support is added via obs**2 and this implementation scales as O(num_obs**2). (#804)

Breaking changes

• Update python packaging to follow PEP 517/518/621/660 standards. (#832)

• Add getData() in the lightning.tensor C++ backend. Users are responsible for ensuring sufficient host memory is allocated for the full state vector. (#827)

• Remove NDpermuter.hpp which is no longer required. (#795)

• Remove temporary steps from the CI, such as downgrading Scipy to <1.14 and installing Kokkos v4.2 for lightning-version == 'stable'. (#792)

• Do not run GPU tests and Docker workflows on release. (#788)

Improvements

• Update Lightning tests to support the generalization of basis state preparation. (#864)

• Add SetState and SetBasisState to LightningKokkosSimulator. (#861)

• Remove use of the deprecated Operator.expand in favour of Operator.decomposition. (#846)

• The setBasisState and setStateVector methods of StateVectorLQubit and StateVectorKokkos are overloaded to support PennyLane-like parameters. (#843)

• Move setBasisState, setStateVector and resetStateVector from StateVectorLQubitManaged to StateVectorLQubit. (#841)

• Update generate_samples in LightningKokkos and LightningGPU to support qml.measurements.Shots type instances. (#839)

• Add a Catalyst-specific wrapping class for Lightning Kokkos. (#837) (#770)

• Lightning-Qubit natively supports the PauliRot gate. (#834)

• Multiple calls to the append_mps_final_state() API is allowed in lightning.tensor. (#830)

• Add initial_state_prep option to Catalyst TOML file. (#826)

• ENABLE_LAPACK is OFF by default for all Lightning backends. (#825)

• Update ctrl_decomp_zyz tests with len(control_wires) > 1. (#821)

• Update the Catalyst-specific wrapping class for Lightning Kokkos to track Catalyst's new support for MCM seeding. (#819)

• Replace size_t by std::size_t everywhere. (#816)

• Shot batching is made more efficient by executing all the shots in one go on Lightning-Qubit. (#814)

• Lightning-Qubit calls generate_samples(wires) on a minimal subset of wires when executing in finite-shot mode. (#813)

• Update LightingQubit.preprocess to work with changes to preprocessing for mid-circuit measurements. (#812)

• Avoid unnecessary memory reset in Lightning-Qubit's state vector class constructor. (#811)

• Add generate_samples(wires) support in Lightning-Qubit, which samples faster for a subset of wires. (#809)

• Optimize the OpenMP parallelization of Lightning-Qubit's probs for all number of targets. (#807)

• Optimize probs(wires) of Lightning-Kokkos using various kernels. Which kernel is to be used depends on the device, number of qubits and number of target wires. (#802)

• Add GPU device compute capability check for Lightning-Tensor. (#803)

• Refactor CUDA utils Python bindings to a separate module. (#801)

• Parallelize Lightning-Qubit probs with OpenMP when using the -DLQ_ENABLE_KERNEL_OMP=1 CMake argument. (#800)

• Implement probs(wires) using a bit-shift implementation akin to the gate kernels in Lightning-Qubit. (#795)

• Enable setting the PennyLane version when invoking, for example, make docker-build version=master pl_version=master. (#791)

Documentation

• The installation instructions for all lightning plugins have been improved. (#858) (#851)

• Updated the README and added citation format for Lightning arXiv preprint. (#818)

Bug fixes

• Point to the right Lightning root folder independently from the invocation location, when configuring the project. (#874)

• Update dependencies and build command options following changes in the build system. (#863)

• Replace structured bindings by variables in GateImplementationsLM.hpp. (#856)

• Remove wrong -m when calling setup.py. (#854)

• Fix plugin-test-matrix CI/CD workflows. (#850)

• Set the immutable parameter value as false for the cutensornetStateApplyTensorOperator to allow the following cutensornetStateUpdateTensorOperator call. (#845)

• Fix cuQuantum SDK path pass-thought in CMake. (#831)

• Fix CUDA sync issues on AArch64 + GraceHopper. (#823)

• Check for the number of wires for Hermitian observables in Lightning-Tensor. Only 1-wire Hermitian observables are supported as of cuTensorNet-v24.03.0. (#806)

• Set PL_BACKEND for the entire build-wheel-lightning-gpu Docker-build stage to properly build the Lightning-GPU wheel. (#791)

• Fix conditions for skipping build & push steps in the Docker build workflows. (#790)

• Downgrade Scipy on Lightning stable version tests. (#783)

• Fix checkout command in test workflows for rc branches. (#777)

Contributors

This release contains contributions from (in alphabetical order):

Ali Asadi, Astral Cai, Ahmed Darwish, Amintor Dusko, Vincent Michaud-Rioux, Erick Ochoa Lopez, Lee J. O'Riordan, Mudit Pandey, Shuli Shu, Raul Torres, Paul Haochen Wang

Release 0.37.0

New features since last release

• Implement Python interface to the lightning.tensor device. (#748)

• Add inverse support for gate operations in lightning.tensor in the C++ layer. (#753)

• Add observable and expval support to the cutensornet-backed lightning.tensor C++ layer. (#728)

• Add gate support to cutensornet-backed lightning.tensor C++ layer. (#718)

• Add cutensornet-backed MPS C++ layer to lightning.tensor. (#704)

• Add support for C(BlockEncode) to Lightning devices. (#743)

Breaking changes

• Removed the QuimbMPS class and the corresponding backend from lightning.tensor. (#737)

• Changed the name of default.tensor to lightning.tensor with the quimb backend. (#730)

• dynamic_one_shot uses shot-vectors in the auxiliary tape to tell the device how many times to repeat the tape. Lightning-Qubit is updated accordingly. (#724)

• dynamic_one_shot deals with post-selection during the post-processing phase, so Lightning-Qubit does not return None-valued measurements for mismatching samples anymore. (#720)

Improvements

• Release candidate branches automatically use the new large GitHub runner pool. (#769)

• Lightning-Kokkos dev wheels for MacOS (x86_64, ARM64) and Linux (aarch64) are uploaded to TestPyPI upon merging a pull request. (#765)

• Lightning-Kokkos Linux (x86_64) dev wheels are pushed to Test PyPI upon merging a pull request. (#763)

• Change the type of tensor network objects passed to ObservablesTNCuda and MeasurementsTNCuda classes from StateTensorT to TensorNetT. (#759)

• Silence NDPermuter linting warnings. (#750)

• Rationalize MCM tests, removing most end-to-end tests from the native MCM test file, but keeping one that validates multiple mid-circuit measurements with any allowed return. (#754)

• Rename lightning.tensor C++ libraries. (#755)

• Set state_tensor as const for the MeasurementTNCuda class. (#753)

• Updated Kokkos version and support to 4.3.01. (#725)

• Lightning-Kokkos' functors are rewritten as functions wrapping around generic gate and generator functors templated over a coefficient interaction function. This reduces boilerplate while clarifying how the various kernels differ from one another. (#640)

• Update C++ and Python GitHub actions names to include the matrix info. (#717)

• Remove CPhase in favour of CPhaseShift in Lightning devices. (#717)

• The various OpenMP configurations of Lightning-Qubit are tested in parallel on different Github Actions runners. (#712)

• Update Linux wheels to use manylinux_2_28 images. (#667)

• Add support for qml.expval and qml.var in the lightning.tensor device for the quimb interface and the MPS method. (#686)

• Changed the name of lightning.tensor to default.tensor with the quimb backend. (#719)

• lightning.qubit and lightning.kokkos adhere to user-specified mid-circuit measurement configuration options. (#736)

• Patch the C++ Measurements.probs(wires) method in Lightning-Qubit and Lightning-Kokkos to Measurements.probs() when called with all wires.
  This will trigger a more optimized implementation for calculating the probabilities of the entire system. (#744)

• Remove the daily schedule from the "Compat Check w/PL - release/release" GitHub action. (#746)

• Remove the required scipy config file for LightningQubit. The support is now maintained by passing SCIPY_LIBS_PATH to the compiler. (#775)

Documentation

• Add installation instructions and documentation for lightning.tensor. (#756)

Bug fixes

• Don't route probs(wires=[]) to probs(all_wires) in Lightning-Kokkos. (#762)

• ControlledQubitUnitary is present in the Python device but not the TOML files. It is added to the decomposition gates since it can be implemented in its alternate form of C(QubitUnitary). (#767)

• Update the Lightning TOML files to indicate that non-commuting observables are supported. (#764)

• Fix regex matching issue with auto on-boarding of release candidate branch to using the large runner queue. (#774)

• Fix random CI failures for lightning.tensor Python unit tests and ignore lightning_tensor paths. (#761)

• lightning.qubit and lightning.kokkos use qml.ops.Conditional.base instead of qml.ops.Conditional.then_op. (#752)

• The preprocessing step in lightning.qubit now uses interface information to properly support the hardware-like postselection for mid-circuit measurements. (#760)

• Fix AVX streaming operation support with newer GCC. (#729)

• Revert changes calling the templated IMAG, ONE, ZERO functions in Kokkos kernels since they are incompatible with device execution. (#733)

• The tests_lkcpu_python.yml workflow properly checks out the release or stable version of Lightning-Qubit during the test job. (#723)

• Fix PennyLane Lightning-Kokkos and LightningQubit tests for stable/stable configuration. (#734)

• Remove the Autoray dependency from requirement files. (#736)

• Fix the cuda-runtime-12-0 dependency issue on RHEL8. (#739)

• Fix the memory segmentation fault when initializing zero-wire Lightning-Kokkos. (#757)

• Remove pennylane.ops.op_math.controlled_decompositions.ctrl_decomp_zyz tests with len(control_wires) > 1. (#757)

• Add support for Scipy v1.14. (#776)

• Add pickle support for the DevPool object in lightning.gpu. (#772)

Contributors

This release contains contributions from (in alphabetical order):

Ali Asadi, Amintor Dusko, Lillian Frederiksen, Pietropaolo Frisoni, David Ittah, Vincent Michaud-Rioux, Lee James O'Riordan, Mudit Pandey, Shuli Shu, Jay Soni

Release 0.36.0

New features since last release

• Add compile-time support for AVX2/512 streaming operations in lightning.qubit. (#664)

• lightning.kokkos supports mid-circuit measurements. (#672)

• Add dynamic linking to LAPACK/OpenBlas shared objects in scipy.libs for both C++ and Python layer. (#653)

• lightning.qubit supports mid-circuit measurements. (#650)

• Add finite shots support in lightning.qubit2. (#630)

• Add collapse and normalize methods to the StateVectorLQubit classes, enabling "branching" of the wavefunction. Add methods to create and seed an RNG in the Measurements modules. (#645)

• Add two new python classes (LightningStateVector and LightningMeasurements) to support lightning.qubit2. (#613)

• Add analytic-mode qml.probs and qml.var support in lightning.qubit2. (#627)

• Add LightningAdjointJacobian to support lightning.qubit2. (#631)

• Add lightning.qubit2 device which uses the new device API. (#607) (#628)

• Add Vector-Jacobian Product calculation support to lightning.qubit. (#644)

• Add support for using new operator arithmetic as the default. (#649)

Breaking changes

• Split Lightning-Qubit and Lightning-Kokkos CPU Python tests with pytest-split. Remove SERIAL from Kokkos' exec_model matrix. Remove all from Lightning-Kokkos' pl_backend matrix. Move clang-tidy checks to tidy.yml. Avoid editable pip installations. (#696)

• Update lightning.gpu and lightning.kokkos to raise an error instead of falling back to default.qubit. (#689)

• Add paths directives to test workflows to avoid running tests that cannot be impacted by changes. (#699) (#695)

• Move common components of /src/simulator/lightning_gpu/utils/ to /src/utils/cuda_utils/. (#676)

• Deprecate static LAPACK linking support. (#653)

• Migrate lightning.qubit to the new device API. (#646)

• Introduce ci:build_wheels label, which controls wheel building on pull_request and other triggers. (#648)

• Remove building wheels for Lightning Kokkos on Windows. (#693)

Improvements

• Add tests for Windows Wheels, fix ill-defined caching, and set the proper backend for lightning.kokkos wheels. (#693)

• Replace string comparisons by isinstance checks where possible. (#691)

• Refactor cuda_utils to remove its dependency on custatevec.h. (#681)

• Add test_templates.py module where Grover and QSVT are tested. (#684)

• Create cuda_utils for common usage of CUDA related backends. (#676)

• Refactor lightning_gpu_utils unit tests to remove the dependency on statevector class. (#675)

• Upgrade GitHub actions versions from v3 to v4. (#669)

• Initialize the private attributes gates_indices_ and generators_indices_ of StateVectorKokkos using the definitions of the Pennylane::Gates::Constant namespace. (#641)

• Add isort to requirements-dev.txt and run before black upon make format to sort Python imports. (#623)

• Improve support for new operator arithmetic with QuantumScriptSerializer.serialize_observables. (#670)

• Add workflow_dispatch to wheels recipes; allowing developers to build wheels manually on a branch instead of temporarily changing the headers. (#679)

• Add the ENABLE_LAPACK compilation flag to toggle dynamic linking to LAPACK library. (#678)

Documentation

Bug fixes

• Fix wire order permutations when using qml.probs with out-of-order wires. (#707)

• Lightning Qubit once again respects the wire order specified on device instantiation. (#705)

• dynamic_one_shot was refactored to use SampleMP measurements as a way to return the mid-circuit measurement samples. LightningQubit's simulate is modified accordingly. (#694)

• LightningQubit correctly decomposes state prep operations when used in the middle of a circuit. (#687)

• LightningQubit correctly decomposes qml.QFT and qml.GroverOperator if len(wires) is greater than 9 and 12 respectively. (#687)

• Specify isort --py (Python version) and -l (max line length) to stabilize isort across Python versions and environments. (#647)

• Fix random coverage xml CI issues. (#635)

• lightning.qubit correctly decomposed state preparation operations with adjoint differentiation. (#661)

• Fix the failed observable serialization unit tests. (#683)

• Update the LightningQubit new device API to work with Catalyst. (#665)

• Update the version of codecov-action to v4 and fix the CodeCov issue with the PL-Lightning check-compatibility actions. (#682)

• Refactor of dev prerelease auto-update-version workflow. (#685)

• Remove gates unsupported by catalyst from toml file. (#698)

• Increase tolerance for a flaky test. (#703)

Contributors

This release contains contributions from (in alphabetical order):

Ali Asadi, Amintor Dusko, Thomas Germain, Christina Lee, Erick Ochoa Lopez, Vincent Michaud-Rioux, Rashid N H M, Lee James O'Riordan, Mudit Pandey, Shuli Shu

Release 0.35.1

Improvements

• Use the adjoint gate parameter to apply qml.Adjoint operations instead of matrix methods in lightning.qubit.
  (#632)

Bug fixes

• Fix qml.Adjoint support in lightning.gpu and lightning.kokkos.
  (#632)

• Fix finite shots support in lightning.qubit, lightning.gpu and lightning.kokkos. The bug would impact calculations with measurements on observables with non-trivial diagonalizing gates and calculations with shot vectors.
  (#632)

Contributors

This release contains contributions from (in alphabetical order):

Vincent Michaud-Rioux

Release 0.35.0

New features since last release

• All backends now support GlobalPhase and C(GlobalPhase) in forward pass.
  (#579)

• Add Hermitian observable support for shot-noise measurement and Lapack support.
  (#569)

Breaking changes

• Migrate lightning.gpu to CUDA 12.
  (#606)

Improvements

• Expand error values and strings returned from CUDA libraries.
  (#617)

• C(MultiRZ) and C(Rot) gates are natively supported (with LM kernels).
  (#614)

• Add adjoint support for GlobalPhase in Lightning-GPU and Lightning-Kokkos.
  (#615)

• Lower the overheads of Windows CI tests.
  (#610)

• Decouple LightningQubit memory ownership from numpy and migrate it to Lightning-Qubit managed state-vector class.
  (#601)

• Expand support for Projector observables on Lightning-Kokkos.
  (#601)

• Split Docker build cron job into two jobs: master and latest. This is mainly for reporting in the plugin-test-matrix repo.
  (#600)

• The BlockEncode operation from PennyLane is now supported on all Lightning devices.
  (#599)

• OpenMP acceleration can now be enabled at compile time for all lightning.qubit gate kernels using the "-DLQ_ENABLE_KERNEL_OMP=1" CMake argument.
  (#510)

• Enable building Docker images for any branch or tag. Set the Docker build cron job to build images for the latest release and master.
  (#598)

• Enable choosing the PennyLane-Lightning version and disabling push to Docker Hub in the Docker build workflow. Add a cron job calling the Docker build workflow.
  (#597)

• Pull Kokkos v4.2.00 from the official Kokkos repository to test Lightning-Kokkos with the CUDA backend.
  (#596)

• Remove deprecated MeasurementProcess.name.
  (#605)

Documentation

• Update requirements to build the documentation.
  (#594)

Bug fixes

• Downgrade auditwheel due to changes with library exclusion list.
  (#620)

• List GlobalPhase gate in each device's TOML file.
  (#615)

• Lightning-GPU's gate cache failed to distinguish between certain gates.
  For example, MultiControlledX([0, 1, 2], "111") and MultiControlledX([0, 2], "00") were applied as the same operation.
  This could happen with (at least) the following gates: QubitUnitary,ControlledQubitUnitary,MultiControlledX,DiagonalQubitUnitary,PSWAP,OrbitalRotation.
  (#579)

• Ensure the stopping condition decompositions are respected for larger templated QFT and Grover operators.
  (#609)

• Move concurrency group specifications from reusable Docker build workflow to the root workflows.
  (#604)

• Fix lightning-kokkos-cuda Docker build and add CI workflow to build images and push to Docker Hub.
  (#593)

• Update jax.config imports.
  (#619)

• Fix apply state vector when using a Lightning handle.
  (#622)

• Pinning Pytest to a version compatible with Flaky.
  (#624)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko, David Ittah, Vincent Michaud-Rioux, Lee J. O'Riordan, Shuli Shu, Matthew Silverman

Release v0.34.0

New features since last release

• Support added for Python 3.12 wheel builds. (#541)

• Lightning-Qubit support arbitrary controlled gates (any wires and any control values). The kernels are implemented in the LM module. (#576)

• Shot-noise related methods now accommodate observable objects with arbitrary eigenvalues. Add a Kronecker product method for two diagonal matrices. (#570)

• Add shot-noise support for probs in the C++ layer. Probabilities are calculated from generated samples. All Lightning backends support this feature. Please note that target wires should be sorted in ascending manner. (#568)

• Add LM kernels to apply arbitrary controlled operations efficiently. (#516)

• Add shots support for variance value, probs, sample, counts calculation for given observables (NamedObs, TensorProd and Hamiltonian) based on Pauli words, Identity and Hadamard in the C++ layer. All Lightning backends support this support feature. (#561)

• Add shots support for expectation value calculation for given observables (NamedObs, TensorProd and Hamiltonian) based on Pauli words, Identity and Hadamard in the C++ layer by adding measure_with_samples to the measurement interface. All Lightning backends support this support feature. (#556)

• qml.QubitUnitary operators can be included in a circuit differentiated with the adjoint method. Lightning handles circuits with arbitrary non-differentiable qml.QubitUnitary operators. 1,2-qubit qml.QubitUnitary operators with differentiable parameters can be differentiated using decomposition. (#540)

Breaking changes

• Set the default version of Kokkos to 4.2.00 throughout the project (CMake, CI, etc.) (#578)

• Overload applyOperation with a fifth matrix argument to all state vector classes to support arbitrary operations in AdjointJacobianBase. (#540)

Improvements

• Ensure aligned memory used for numpy arrays with state-vector without reallocations. (#572)

• Unify error messages of shot measurement related unsupported observables to better Catalyst. (#577)

• Add configuration files to improve compatibility with Catalyst. (#566)

• Refactor shot-noise related methods of MeasurementsBase class in the C++ layer and eigenvalues are not limited to 1 and -1. Add getObs() method to Observables class. Refactor applyInPlaceShots to allow users to get eigenvalues of Observables object. Deprecated _preprocess_state method in MeasurementsBase class for safer use of the LightningQubitRaw backend. (#570)

• Modify setup.py to use backend-specific build directory (f"build_{backend}") to accelerate rebuilding backends in alternation. (#540)

• Update Dockerfile and rewrite the build-wheel-lightning-gpu stage to build Lightning-GPU from the pennylane-lightning monorepo. (#539)

• Add the MPI test CI workflows of Lightning-GPU in compatibility cron jobs. (#536)

• Add MPI synchronization in places to safely handle communicated data. (#538)

• Add release option in compatibility cron jobs to test the release candidates of PennyLane and the Lightning plugins against one another. (#531)

• Add GPU workflows in compatibility cron jobs to test Lightning-GPU and Lightning-Kokkos with the Kokkos CUDA backend. (#528)

Documentation

• Fixed a small typo in the documentation page for the PennyLane-Lightning GPU device. (#563)

• Add OpenGraph social preview for Lightning docs. (#574)

Bug fixes

• Fix CodeCov file contention issue when uploading data from many workloads. (#584)

• Ensure the lightning.gpu intermediate wheel builds are uploaded to TestPyPI. (#575)

• Allow support for newer clang-tidy versions on non-x86_64 platforms. (#567)

• Do not run C++ tests when testing for compatibility with PennyLane, hence fixing plugin-matrix failures. Fix Lightning-GPU workflow trigger. (#571)

• Revert single-node multi-GPU batching behaviour to match PennyLaneAI/pennylane-lightning-gpu#27. (#564)

• Move deprecated stateprep QuantumScript argument into the operation list in mpitests/test_adjoint_jacobian.py. (#540)

• Fix MPI Python unit tests for the adjoint method. (#538)

• Fix the issue with assigning kernels to ops before registering kernels on macOS. (#582)

• Update MANIFEST.in to include device config files and CHANGELOG.md. (#585)

Contributors

This release contains contributions from (in alphabetical order):

Ali Asadi, Isaac De Vlugt, Amintor Dusko, Vincent Michaud-Rioux, Erick Ochoa Lopez, Lee James O'Riordan, Shuli Shu

Release 0.33.1

New features since last release

• pip-installed CUDA runtime libraries can now be accessed from a virtualenv. (#543 )

Bug fixes

• The pybind11 compiled module RPATH linkage has been restored to pre-0.33 behaviour. (#543 )

Contributors

This release contains contributions from (in alphabetical order):

Lee J. O'Riordan

Release 0.33.0

New features since last release

• Add documentation updates for the lightning_gpu backend. (#525)

• Add SparseHamiltonian support for Lightning-Qubit and Lightning-GPU. (#526)

• Add SparseHamiltonian support for Lightning-Kokkos. (#527)

• Integrate python/pybind layer of distributed Lightning-GPU into the Lightning monorepo with python unit tests. (#518)

• Integrate the distributed C++ backend of Lightning-GPU into the Lightning monorepo. (#514)

• Integrate Lightning-GPU into the Lightning monorepo. The new backend is named lightning_gpu and includes all single-GPU features. (#499)

• Build Linux wheels for Lightning-GPU (CUDA-11). (#517)

• Add Dockerfile in docker and make docker workflow in Makefile. The Docker images and documentation are available on DockerHub. (#496)

• Add mid-circuit state preparation operation tests. (#495)

Improvements

• Improve Python testing for Lightning-GPU (+MPI) by adding jobs in Actions files and adding Python tests to increase code coverage. (#522)

• Add support for pip install pennylane-lightning[kokkos] for the OpenMP backend. (#515)

• Update setup.py to allow for multi-package co-existence. The PennyLane_Lightning package now is the responsible for the core functionality, and will be depended upon by all other extensions. (#504)

• Redesign Lightning-Kokkos StateVectorKokkos class to use Kokkos RangePolicy together with special functors in applyMultiQubitOp to apply 1- to 4-wire generic unitary gates. For more than 4 wires, the general implementation using Kokkos TeamPolicy is employed to yield the best all-around performance. (#490)

• Redesign Lightning-Kokkos Measurements class to use Kokkos RangePolicy together with special functors to obtain the expectation value of 1- to 4-wire generic unitary gates. For more than 4 wires, the general implementation using Kokkos TeamPolicy is employed to yield the best all-around performance. (#489)

• Add tests to increase Lightning-Kokkos coverage. (#485)

• Add memory locality tag reporting and adjoint diff dispatch for lightning.qubit statevector classes. (#492)

• Add support for dependent external packages to C++ core. (#482)

• Add support for building multiple backend simulators. (#497)

Breaking changes

• Add tests_gpu.yml workflow to test the Lightning-Kokkos backend with CUDA-12. (#494)

• Implement LM::GeneratorDoubleExcitation, LM::GeneratorDoubleExcitationMinus, LM::GeneratorDoubleExcitationPlus kernels. Lightning-Qubit default kernels are now strictly from the LM implementation, which requires less memory and is faster for large state vectors. (#512)

• Add workflows validating compatibility between PennyLane and Lightning's most recent stable releases and development (latest) versions. (#507) (#498)

• Introduce timeout-minutes in various workflows, mainly to avoid Windows builds hanging for several hours. (#503)

• Cast integral-valued arrays to the device's complex type on entry in _preprocess_state_vector to ensure the state is correctly represented with floating-point numbers. (#501)

• Update DefaultQubit to DefaultQubitLegacy on Lightning fallback. (#500)

• Enums defined in GateOperation.hpp start at 1 (previously 0). ::BEGIN is introduced in a few places where it was assumed 0 accordingly. (#485) ::BEGIN was later reverted to 0, keeping the rest of the changes, to fix a compatibility issue with Catalyst. (#521)

• Enable pre-commit hooks to format all Python files and linting of all Python source files. (#485)

Bug fixes

• Re-add support for pip install pennylane-lightning[gpu]. (#515)

• Switch most Lightning-Qubit default kernels to LM. Add LM::multiQubitOp tests, failing when targeting out-of-order wires clustered close to num_qubits-1. Fix the LM::multiQubitOp kernel implementation by introducing a generic revWireParity routine and replacing the bitswap-based implementation. Mimic the changes fixing the corresponding multiQubitOp and expval functors in Lightning-Kokkos. (#511)

• Fix RTD builds by removing unsupported system_packages configuration option. (#491)

Contributors

This release contains contributions from (in alphabetical order):

Ali Asadi, Amintor Dusko, Vincent Michaud-Rioux, Lee J. O'Riordan, Shuli Shu

Release 0.32.0

Improvements

• Refactor LKokkos Measurements class to use (fast) specialized functors whenever possible. (#481)

• Merge Lightning Qubit and Lightning Kokkos backends in the new repository. (#472)

New features since last release

• The lightning_kokkos backend supports Nvidia GPU execution (with Kokkos v4 and CUDA v12). (#477)

• Complete overhaul of repository structure to facilitate integration of multiple backends. Refactoring efforts we directed to improve development performance, and code reuse and decrease overall overhead to propagate changes through backends. The new C++ modular build strategy allows faster test builds restricted to a module. Update CI/CD actions concurrency strategy. Change the minimal Python version to 3.9. (#472)

• Wheels are built with native support for sparse Hamiltonians. (#470)

• Add native support to sparse Hamiltonians in the absence of Kokkos & Kokkos-kernels. (#465)

Breaking changes

• Rename QubitStateVector to StatePrep in the LightningQubit and LightningKokkos classes. (#486)

• Modify adjointJacobian methods to accept a (maybe unused) reference StateVectorT, allowing device-backed simulators to directly access state vector data for adjoint differentiation instead of copying it back-and-forth into JacobianData (host memory). (#477)

Documentation

• Integrated new unified docs for Lightning Kokkos and Lightning Qubit packages. (#473)

Bug fixes

• Ensure PennyLane has an active_return attribute before calling it. (#483)

• Do no import sqrt2_v from <numbers> in Util.hpp to resolve issue with Lightning-GPU builds. (#479)

• Update the CMake internal references to enable sub-project compilation with affecting the parent package. (#478)

• apply no longer mutates the inputted list of operations. (#474)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko, Christina Lee, Vincent Michaud-Rioux, Lee J. O'Riordan

Release v0.31.0

New features since last release

• Update Kokkos support to 4.0.01. [(#439)] (#439)

Breaking changes

• Update tests to be compliant with PennyLane v0.31.0 development changes and deprecations. (#448)

Improvements

• Remove logic from setup.py and transfer paths and env variable definitions into workflow files. (#450)

• Detect MKL or CBLAS if ENABLE_BLAS=ON making sure that BLAS is linked as expected. (#449)

Documentation

• Fix LightningQubit class parameter documentation. (#456)

Bug fixes

• Ensure cross-platform wheels continue to build with updates in git safety checks. (#452)

• Fixing Python version bug introduce in (#450)
  when Python_EXECUTABLE was removed from setup.py. (#461)

• Ensure aligned allocator definition works with C++20 compilers. (#438)

• Prevent multiple threads from calling Kokkos::initialize or Kokkos::finalize. (#439)

Contributors

This release contains contributions from (in alphabetical order):

Vincent Michaud-Rioux, Lee J. O'Riordan, Chae-Yeun Park