Pennylane

Bug fixes

* Fixed a bug where the interfaces submodule was not correctly being packaged via setup.py.

New features since last release

* PennyLane now includes a new `interfaces` submodule, which enables QNode integration with additional machine learning libraries (165).
* Adds support for an experimental PyTorch interface for QNodes
* Adds support for an experimental TensorFlow eager execution interface for QNodes
* Adds a PyTorch+GPU+QPU tutorial to the documentation
* Documentation now includes links and tutorials including the new [PennyLane-Forest](https://github.com/rigetti/pennylane-forest) plugin.

Improvements

* Printing a QNode object, via `print(qnode)` or in an interactive terminal, now displays more useful information regarding the QNode, including the device it runs on, the number of wires, it's interface, and the quantum function it uses:
python
>>> print(qnode)
<QNode: device='default.qubit', func=circuit, wires=2, interface=PyTorch>


Contributors

This release contains contributions from:

Josh Izaac and Nathan Killoran.

<h3>Bug fixes</h3>

* Add missing OpenQASM backend in binary distribution, which relies on the latest version of the AWS Braket plugin for PennyLane to resolve dependency issues between the plugin, Catalyst, and PennyLane. The Lightning-Kokkos backend with Serial and OpenMP modes is also added to the binary distribution. [198](https://github.com/PennyLaneAI/catalyst/pull/198)

<h3>Improvements</h3>

* When using OpenQASM-based devices the string representation of the circuit is printed on exception. [199](https://github.com/PennyLaneAI/catalyst/pull/199)

* Use ``pybind11::module`` interface library instead of ``pybind11::embed`` in the runtime for OpenQasm backend to avoid linking to the python library at compile time. [200](https://github.com/PennyLaneAI/catalyst/pull/200)

<h3>Contributors</h3>

This release contains contributions from (in alphabetical order):

Ali Asadi, David Ittah.

New features since last release

* Added the `Identity` expectation value for both CV and qubit models (135)
* Added the `templates.py` submodule, containing some commonly used QML models to be used as ansatz in QNodes (133)
* Added the `qml.Interferometer` CV operation (152)
* Wires are now supported as free QNode parameters (151)
* Added ability to update stepsizes of the optimizers (159)

Improvements

* Removed use of hardcoded values in the optimizers, made them parameters (see 131 and 132)
* Created the new `PlaceholderExpectation`, to be used when both CV and qubit expval modules contain expectations with the same name
* Provide the plugins a way to view the operation queue _before_ applying operations. This allows for on-the-fly modifications of the queue, allowing hardware-based plugins to support the full range of qubit expectation values. (143)
* QNode return values now support _any_ form of sequence, such as lists, sets, etc. (144)
* CV analytic gradient calculation is now more robust, allowing for operations which may not themselves be differentiated, but have a well defined `_heisenberg_rep` method, and so may succeed operations that are analytically differentiable (152)

Bug fixes

* Fixed a bug where the variational classifier example was not batching when learning parity (see 128 and 129)
* Fixed an inconsistency where some initial state operations were documented as accepting complex parameters - all operations now accept real values (146)

Contributors

This release contains contributions from:

Christian Gogolin, Josh Izaac, Nathan Killoran, and Maria Schuld.

<h3>New features</h3>

* Add an option to print verbose messages explaining the compilation process
[68](https://github.com/PennyLaneAI/catalyst/pull/68)

* Allow `catalyst.grad` to be used on any traceable function (within a qjit context).
This means the operation is no longer resticted to acting on `qml.qnode`s only.
[75](https://github.com/PennyLaneAI/catalyst/pull/75)

<h3>Improvements</h3>

* Work in progress on a Lightning-Kokkos backend:

Bring feature parity to the Lightning-Kokkos backend simulator.
[55](https://github.com/PennyLaneAI/catalyst/pull/55)

Add support for variance measurements for all observables.
[70](https://github.com/PennyLaneAI/catalyst/pull/70)

* Build the runtime against qir-stdlib v0.1.0.
[58](https://github.com/PennyLaneAI/catalyst/pull/58)

* Replace input-checking assertions with exceptions.
[67](https://github.com/PennyLaneAI/catalyst/pull/67)

* Perform function inlining to improve optimizations and memory management within the compiler.
[72](https://github.com/PennyLaneAI/catalyst/pull/72)

<h3>Breaking changes</h3>

<h3>Bug fixes</h3>

* Several fixes to address memory leaks in the compiled program:

Fix memory leaks from data that flows back into the Python environment.
[54](https://github.com/PennyLaneAI/catalyst/pull/54)

Fix memory leaks resulting from partial bufferization at the MLIR level. This fix makes the
necessary changes to reintroduce the ``-buffer-deallocation`` pass into the MLIR pass pipeline.
The pass guarantees that all allocations contained within a function (that is allocations that are
not returned from a function) are also deallocated.
[61](https://github.com/PennyLaneAI/catalyst/pull/61)

Lift heap allocations for quantum op results from the runtime into the MLIR compiler core. This
allows all memref buffers to be memory managed in MLIR using the
[MLIR bufferization infrastructure](https://mlir.llvm.org/docs/Bufferization/).
[63](https://github.com/PennyLaneAI/catalyst/pull/63)

Eliminate all memory leaks by tracking memory allocations at runtime. The memory allocations
which are still alive when the compiled function terminates, will be freed in the
finalization / teardown function.
[78](https://github.com/PennyLaneAI/catalyst/pull/78)

* Fix returning complex scalars from the compiled function.
[77](https://github.com/PennyLaneAI/catalyst/pull/77)

<h3>Contributors</h3>

This release contains contributions from (in alphabetical order):

Ali Asadi,
David Ittah,
Erick Ochoa Lopez,
Sergei Mironov.

<h3>New features</h3>

* Adds support for interpreting control flow operations.
[31](https://github.com/PennyLaneAI/catalyst/pull/31)

<h3>Improvements</h3>

* Adds fallback compiler drivers to increase reliability during linking phase. Also adds support for a
CATALYST_CC environment variable for manual specification of the compiler driver used for linking.
[30](https://github.com/PennyLaneAI/catalyst/pull/30)

<h3>Breaking changes</h3>

<h3>Bug fixes</h3>

* Fixes to codecov, build-lightning, and GH actions.
[34](https://github.com/PennyLaneAI/catalyst/pull/34)

* Fixes the Catalyst image path in the readme to properly render on PyPI.

<h3>Contributors</h3>

This release contains contributions from (in alphabetical order):

Ali Asadi,
Erick Ochoa Lopez.