Release v0.30.0

New features since last release

• Add MCMC sampler.
  (#384)

• Serialize PennyLane's arithmetic operators when they are used as observables
  that are expressed in the Pauli basis.
  (#424)

Breaking changes

• Lightning now works with the new return types specification that is now default in PennyLane.
  See the PennyLane qml.enable_return documentation for more information on this change.
  (#427)

Instead of creating potentially ragged numpy array, devices and QNode's now return an object of the same type as that
returned by the quantum function.

>>> dev = qml.device('lightning.qubit', wires=1)
>>> @qml.qnode(dev, diff_method="adjoint")
... def circuit(x):
...     qml.RX(x, wires=0)
...     return qml.expval(qml.PauliY(0)), qml.expval(qml.PauliZ(0))
>>> x = qml.numpy.array(0.5)
>>> circuit(qml.numpy.array(0.5))
(array(-0.47942554), array(0.87758256))

Interfaces like Jax or Torch handle tuple outputs without issues:

>>> jax.jacobian(circuit)(jax.numpy.array(0.5))
(Array(-0.87758255, dtype=float32, weak_type=True),
Array(-0.47942555, dtype=float32, weak_type=True))

Autograd cannot differentiate an output tuple, so results must be converted to an array before
use with qml.jacobian:

>>> qml.jacobian(lambda y: qml.numpy.array(circuit(y)))(x)
array([-0.87758256, -0.47942554])

Alternatively, the quantum function itself can return a numpy array of measurements:

>>> dev = qml.device('lightning.qubit', wires=1)
>>> @qml.qnode(dev, diff_method="adjoint")
>>> def circuit2(x):
...     qml.RX(x, wires=0)
...     return np.array([qml.expval(qml.PauliY(0)), qml.expval(qml.PauliZ(0))])
>>> qml.jacobian(circuit2)(np.array(0.5))
array([-0.87758256, -0.47942554])

Improvements

• Remove deprecated set-output commands from workflow files.
  (#437)

• Lightning wheels are now checked with twine check post-creation for PyPI compatibility.
  (#430)

• Lightning has been made compatible with the change in return types specification.
  (#427)

• Lightning is compatible with clang-tidy version 16.
  (#429)

Contributors

This release contains contributions from (in alphabetical order):

Christina Lee, Vincent Michaud-Rioux, Lee James O'Riordan, Chae-Yeun Park, Matthew Silverman

Release v0.29.0

Improvements

• Remove runtime dependency on ninja build system. (#414)

• Allow better integration and installation support with CMake targeted binary builds. (#403)

• Remove explicit Numpy and Scipy requirements. (#412)

• Get llvm installation root from the environment variable LLVM_ROOT_DIR (or fallback to brew). (#413)

• Update AVX2/512 kernel infrastructure for additional gate/generator operations. (#404)

• Remove unnecessary lines for resolving CodeCov issue. (#415)

• Add more AVX2/512 gate operations. (#393)

Bug fixes

• Ensure error raised when asking for out of order marginal probabilities. Prevents the return of incorrect results. (#416)

• Fix Github shields in README. (#402)

Contributors

Amintor Dusko, Vincent Michaud-Rioux, Lee James O'Riordan, Chae-Yeun Park

Release 0.28.2

Bug fixes

• Fix Python module versioning for Linux wheels. (#408)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko

Release 0.28.1

Bug fixes

• Fix Pybind11 module versioning and locations for Windows wheels. (#400)

Contributors

This release contains contributions from (in alphabetical order):

Lee J. O'Riordan

Release 0.28.0

Breaking changes

• Deprecate support for Python 3.7. (#391)

Improvements

• Improve Lightning package structure for external use as a C++ library. (#369)

• Improve the stopping condition method. (#386)

Bug fixes

• Pin CMake to 3.24.x in wheel-builder to avoid Python not found error in CMake 3.25, when building wheels for PennyLane-Lightning-GPU. (#387)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko, Lee J. O'Riordan

Release v0.27.0

Release 0.27.0

New features since last release

• Enable building of python 3.11 wheels and upgrade python on CI/CD workflows to 3.8. (#381)

Improvements

• Update clang-tools version in Github workflows. (#351)

• Improve tests and checks CI/CD pipelines. (#353)

• Implement 3 Qubit gates (CSWAP & Toffoli) & 4 Qubit gates (DoubleExcitation, DoubleExcitationMinus, DoubleExcitationPlus) using (low-memory) LM architecture. (#362)

• Upgrade Kokkos and Kokkos Kernels to 3.7.00, and improve sparse matrix-vector multiplication performance and memory usage. (#361)

• Update Linux (ubuntu-latest) architecture x86_64 wheel-builder from GCC 10.x to GCC 11.x. (#373)

• Update gcc and g++ 10.x to 11.x in CI tests. This update brings improved support for newer C++ features. (#370)

• Change Lightning to inherit from QubitDevice instead of DefaultQubit. (#365)

Bug fixes

• Use mutex when accessing cache in KernelMap. (#382)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko, Chae-Yeun Park, Monit Sharma, Shuli Shu

Release v0.26.1

Release 0.26.1

Bug fixes

• Fixes the transposition method used in the probability calculation.
  (#377)

Contributor

Amintor Dusko

Release v0.26.0

Improvements

• Introduces requirements-dev.txt and improves dockerfile. (#330)

• Support expval for a Hamiltonian. (#333)

• Implements caching for Kokkos installation. (#316)

• Supports measurements of operator arithmetic classes such as Sum, Prod,
  and SProd by deferring handling of them to DefaultQubit. (#349)

@qml.qnode(qml.device('lightning.qubit', wires=2))
def circuit():
    obs = qml.s_prod(2.1, qml.PauliZ(0)) + qml.op_sum(qml.PauliX(0), qml.PauliZ(1))
    return qml.expval(obs)

Bug fixes

• Test updates to reflect new measurement error messages. (#334)

• Updates to the release tagger to fix incompatibilities with RTD. (#344)

• Update cancel-workflow-action and bot credentials. (#345)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko, Christina Lee, Lee J. O'Riordan, Chae-Yeun Park

v0.25.1

Release 0.25.1

Bug fixes

• Ensure build type for CMake in all compilers and operational systems.
  (#341)

Contributor

Amintor Dusko

Release v0.25.0

New features since last release

Breaking changes

• We explicitly disable support for PennyLane's parameter broadcasting. #317

• We explicitly remove support for PennyLane's Sum, SProd and Prod
  as observables. (#326)

Improvements

• CI builders use a reduced set of resources and redundant tests for PRs. (#319)

• Parallelize wheel-builds where applicable. (#314)

• AVX2/512 kernels are now available on Linux/MacOS with x86-64 architecture. (#313)

Documentation

• Updated ReadTheDocs runner version from Ubuntu 20.04 to 22.04 (#327)

Bug fixes

• Test updates to reflect new additions to PennyLane. (#318)

Contributors

This release contains contributions from (in alphabetical order):

Amintor Dusko, Christina Lee, Rashid N H M, Lee J. O'Riordan, Chae-Yeun Park