Releases: PennyLaneAI/pennylane-qiskit
Release 0.11.0
New features since last release
-
Qiskit devices now support custom wire labels. (#99) (#100)
One can now specify any string or number as a custom wire label, and use these labels to address subsystems on the device:
dev = qml.device('qiskit.ibmq', wires=['q1', 'ancilla', 0, 1]) def circuit(): qml.Hadamard(wires='q1') qml.CNOT(wires=[1, 'ancilla'])
Improvements
- Adds support for Qiskit v0.20. (#101)
Bug fixes
- When converting QASM or Qiskit circuit to PennyLane templates, the
CU1
gate is now natively supported and converted to aQubitUnitary
. (#101)
Contributors
This release contains contributions from (in alphabetical order):
Maria Schuld, Antal Száva
Release 0.8.2
Bug fixes
- Fixed a bug where users with
IBMQ
tokens linked to multiple
providers would experience an error.
(#74)
Contributors
This release contains contributions from (in alphabetical order):
Antal Száva
Release 0.8.1
Bug fixes
- Fixed a bug where gradient computations always returned 0 when
loading a parametrized Qiskit circuit as a PennyLane template.
(#71)
Contributors
This release contains contributions from (in alphabetical order):
Josh Izaac
Release 0.8
Bug fixes
- Removed v1
IBMQ
credentials, disallowedunitary_simulator
backend to
havememory=True
and discontinuing support forQubitStateVector
on
theunitary_simulator
backend due to Qiskit's 0.14.0 version.
#65
Contributors
This release contains contributions from (in alphabetical order):
Antal Száva
Release 0.6
New features since last release
- All Qiskit devices now support tensor observables using the
return expval(qml.PauliZ(0) @ qml.Hermitian(A, [1, 2]))
syntax introduced in PennyLane v0.6.
Contributors
This release contains contributions from (in alphabetical order):
Josh Izaac
Release 0.5.1
This is a significant release, bringing the plugin up to date with the latest
PennyLane and Qiskit features.
New features since last release
-
The plugin and the tests have been completely re-written from scratch, to ensure high quality and remove technical debt. #44 #43
-
Samples and variance support have been added to all devices. #46
-
Multi-qubit hermitian observables are now supported, due to support being added in Qiskit version 0.12.
-
Support has been added for IBM Q Experience 2.0. #44
Improvements
-
Hardware and software devices are now treated identically, with expectations, variance, and samples computed via the probability, not the amplitudes. This has several consequences:
-
It makes the code cleaner and simpler, as there is now one defined way of computing statistics.
-
It is faster in most cases, as this does not require computing large matrix-vector products, or Kronecker products. Instead, eigenvalues of single- and multi-qubit observables are computed using dyanamic programming.
-
It reduces the number of tests required.
-
-
Test suite now includes tests against the IBM Q Experience, in addition to the local simulators.
Bug fixes
-
Due to the move to IBM Q 2.0 credentials, users remaining with IBM Q v1.0 now must pass an additional URL argument to the service. The plugin has been modified to allow the user to pass this argument if required. #44
-
Fixed a bug where backend keyword arguments, such as
backend_options
andnoise_model
, were being passed to backends that did not support it. #51
Contributors
This release contains contributions from:
Shahnawaz Ahmed, Josh Izaac
Compatible with Pennylane v0.4 and QISKIT 0.10.1
This release fixes the issues with the latest QISKIT and Pennylane. There were some breaking changes for QISKIT as well as Pennylane due to changes in the API for both which is fixed in this version of the plugin.