D-Wave Ocean plugin for IBM Qiskit

Enables Qiskit users to work with D-Wave's quantum resources, available via Leap.

DWaveMinimumEigensolver

The package provides an implementation of Qiskit Optimization's SamplingMinimumEigensolver interface (available as DWaveMinimumEigensolver) which can be used directly on qubit operators, or via qiskit_optimization's MinimumEigenOptimizer.

Examples

Solve a QuadraticProgram with MinimumEigenOptimizer using DWaveMinimumEigensolver:

>>> from qiskit_optimization import QuadraticProgram
>>> from qiskit_optimization.algorithms import MinimumEigenOptimizer
>>> from dwave.plugins.qiskit import DWaveMinimumEigensolver
...
>>> # Construct a simple quadratic program
>>> qp = QuadraticProgram()
>>> qp.binary_var('x')
>>> qp.binary_var('y')
>>> qp.minimize(quadratic={'xy': 1})
...
>>> # Solve using Qiskit's MinimumEigenOptimizer on D-Wave QPU as a minimum eigen solver
>>> dwave_mes = DWaveMinimumEigensolver()
>>> optimizer = MinimumEigenOptimizer(dwave_mes)
>>> result = optimizer.solve(qp)
...
>>> print(result)
fval=0.0, x=0.0, y=0.0, status=SUCCESS
>>> [(''.join(str(int(v)) for v in s.x), s.fval, s.probability) for s in result.samples]
[('00', 0.0, 0.33), ('10', 0.0, 0.33), ('01', 0.0, 0.33)]

Solve a 6-city TSP (or some other optimization application), a 36-qubit Ising Hamiltonian:

>>> from qiskit_optimization.applications import Tsp
>>> from qiskit_optimization.algorithms import MinimumEigenOptimizer
>>> from dwave.plugins.qiskit import DWaveMinimumEigensolver
...
>>> tsp = Tsp.create_random_instance(6, seed=123)
>>> qp = tsp.to_quadratic_program()
...
>>> dwave_mes = DWaveMinimumEigensolver(num_reads=1000)
>>> result = MinimumEigenOptimizer(dwave_mes).solve(qp)
...
>>> tsp.interpret(result)
[3, 4, 2, 1, 5, 0]

For comparison, trying this on NumPyMinimumEigensolver (which constructs the full 2^36 state space) produces:

>>> from qiskit_optimization.minimum_eigensolvers import NumPyMinimumEigensolver
>>> result = MinimumEigenOptimizer(NumPyMinimumEigensolver()).solve(qp)
# snipped for brevity
memory allocation of 1818775484491218187754844912 bytes failed
Aborted (core dumped)

and trying with QAOA backed by the reference StatevectorSampler primitive produces:

>>> import numpy as np
>>> from qiskit.primitives import StatevectorSampler
>>> from qiskit_optimization.minimum_eigensolvers import QAOA
>>> from qiskit_optimization.optimizers import COBYLA
...
>>> qaoa_mes = QAOA(sampler=StatevectorSampler(), optimizer=COBYLA(),
...                 initial_point=np.array([0.0, 0.0]))
>>> result = MinimumEigenOptimizer(qaoa_mes).solve(qp)
# snipped for brevity
MemoryError: Unable to allocate 1.00 TiB for an array with shape (68719476736,) and data type complex128

QCDL Translators

dwave.plugins.qiskit.qcdl.translators converts Qiskit QuantumCircuit objects into D-Wave's QCDL program format, for running gate-model circuits on D-Wave's gate-model hardware/simulator via dwave-gate.

Examples

Translate a Bell state Qiskit circuit into a QCDL program:

>>> from qiskit import QuantumCircuit
>>> from dwave.gate.qcdl import print_qcdl
>>> from dwave.plugins.qiskit.qcdl.translators import circuit_to_qcdl
...
>>> qc = QuantumCircuit(2, 2, name="bell")
>>> qc.h(0)
>>> qc.cx(0, 1)
>>> qc.measure([0, 1], [0, 1])
...
>>> result = circuit_to_qcdl(qc)
...
>>> print_qcdl(result.qcdl)
begin quantum
   q0.initialize(q1)
   h([q0], q0)
   cx([q0, q1], q0, q1)
   measure([q0], q0, log=True, tag="0")
   measure([q1], q1, log=True, tag="1")
end quantum

DWaveProvider

DWaveProvider exposes D-Wave's Leap QCDL simulator solvers through the standard Qiskit provider/backend interface: circuits passed to QCDLSimulatorBackend.run() are translated to QCDL, submitted to a Leap solver, and the answers are returned in QCDLResult, a qiskit.result.Result subclass. Leap credentials are picked up from the standard dwave-cloud-client configuration (configuration file or environment variables), or can be passed to the provider directly.

Examples

Run a Bell state circuit on a Leap QCDL solver:

>>> from qiskit import QuantumCircuit
>>> from dwave.plugins.qiskit import DWaveProvider
...
>>> qc = QuantumCircuit(2, 2, name="bell")
>>> qc.h(0)
>>> qc.cx(0, 1)
>>> qc.measure([0, 1], [0, 1])
...
>>> with DWaveProvider() as provider:
...     backend = provider.get_backend()
...     job = backend.run(qc, shots=1000)
...     counts = job.result().get_counts()
>>> counts                                      # doctest: +SKIP
{'00': 512, '11': 488}

run() also accepts a list of circuits; by default they are packed into as few QCDL programs as estimated to fit (disable with pack_qcdls=False to submit one QCDL program per circuit).

Installation

Compatible with Python 3.11+, Qiskit 2.0+, qiskit-optimization 0.7+, and Ocean's dwave-system 1.20+.

pip install dwave-qiskit-plugin

To install from source:

pip install --group dev
pip install --editable .

Test dependencies are defined in the test dependency group in pyproject.toml, and can be installed with:

pip install --group test .
python -m pytest

License

Released under the Apache License 2.0. See LICENSE file.

Contributing

Ocean's contributing guide has guidelines for contributing to Ocean packages.

Release Notes

We use reno to manage release notes.

See reno's user guide for details.