Using PennyLane

Release news

Development

API

Internals

  1. Docs
  2. qml
  3. qml.from_qiskit

qml.from_qiskit

from_qiskit(quantum_circuit, measurements=None)[source]

Converts a Qiskit QuantumCircuit into a PennyLane quantum function.

Note

This function depends upon the PennyLane-Qiskit plugin. Follow the installation instructions to get up and running. You may need to restart your kernel if you are running in a notebook environment.

Parameters:

  • quantum_circuit (qiskit.QuantumCircuit) – a quantum circuit created in Qiskit

  • measurements (None | MeasurementProcess | list[MeasurementProcess]) – an optional PennyLane measurement or list of PennyLane measurements that overrides any terminal measurements that may be present in the input circuit

Returns:

The PennyLane quantum function, created based on the input Qiskit QuantumCircuit object.

Return type:

function

Example:

import pennylane as qml
from qiskit import QuantumCircuit

qc = QuantumCircuit(2, 2)
qc.rx(0.785, 0)
qc.ry(1.57, 1)

my_qfunc = qml.from_qiskit(qc)

The my_qfunc function can now be used within QNodes, as a two-wire quantum template. We can also pass wires when calling the returned template to define which wires it should operate on. If no wires are passed, it will default to sequential wire labels starting at 0.

dev = qml.device("default.qubit")

@qml.qnode(dev)
def circuit():
    my_qfunc(wires=["a", "b"])
    return qml.expval(qml.Z("a")), qml.var(qml.Z("b"))

>>> circuit()
(tensor(0.70738827, requires_grad=True),
tensor(0.99999937, requires_grad=True))

The measurements can also be passed directly to the function when creating the quantum function, making it possible to create a PennyLane circuit with qml.QNode:

>>> measurements = [qml.expval(qml.Z(0)), qml.var(qml.Z(1))]
>>> circuit = qml.QNode(qml.from_qiskit(qc, measurements), dev)
>>> circuit()
(tensor(0.70738827, requires_grad=True),
tensor(0.99999937, requires_grad=True))

Note

The measurements keyword allows one to add a list of PennyLane measurements that will override any terminal measurements present in the QuantumCircuit, so that they are not performed before the operations specified in measurements. measurements=None.

If an existing QuantumCircuit already contains measurements, from_qiskit will return those measurements, provided that they are not overridden as shown above. These measurements can be used, e.g., for conditioning with qml.cond(), or simply included directly within the QNode’s return:

qc = QuantumCircuit(2, 2)
qc.rx(np.pi, 0)
qc.measure_all()

@qml.qnode(dev)
def circuit():
    # Since measurements=None, the measurements present in the QuantumCircuit are returned.
    measurements = qml.from_qiskit(qc)()
    return [qml.expval(m) for m in measurements]

>>> circuit()
[tensor(1., requires_grad=True), tensor(0., requires_grad=True)]

Note

The measurements returned from a QuantumCircuit are in the computational basis with 0 corresponding to \(|0\rangle\) and 1 corresponding to \(|1 \rangle\). This corresponds to the \(|1 \rangle \langle 1|\) observable rather than the \(Z\) Pauli operator.

See below for more information regarding how to translate more complex circuits from Qiskit to PennyLane, including handling parametrized Qiskit circuits, mid-circuit measurements, and classical control flows.

A Qiskit QuantumCircuit is parametrized if it contains Parameter or ParameterVector references that need to be given defined values to evaluate the circuit. These can be passed to the generated quantum function as keyword or positional arguments. If we define a parametrized circuit:

from qiskit.circuit import QuantumCircuit, Parameter

angle0 = Parameter("x")
angle1 = Parameter("y")

qc = QuantumCircuit(2, 2)
qc.rx(angle0, 0)
qc.ry(angle1, 1)
qc.cx(1, 0)

Then this circuit can be converted into a differentiable circuit in PennyLane and executed:

import pennylane as qml
from pennylane import numpy as np

dev = qml.device("default.qubit")

qfunc = qml.from_qiskit(qc, measurements=qml.expval(qml.Z(0)))
circuit = qml.QNode(qfunc, dev)

Now, circuit has a signature of (x, y). The parameters are ordered alphabetically.

>>> x = np.pi / 4
>>> y = 0
>>> circuit(x, y)
tensor(0.70710678, requires_grad=True)

>>> qml.grad(circuit, argnum=[0, 1])(np.pi/4, np.pi/6)
(array(-0.61237244), array(-0.35355339))

The QuantumCircuit may also be parametrized with a ParameterVector. These can be similarly converted:

from qiskit.circuit import ParameterVector

angles = ParameterVector("angles", 2)

qc = QuantumCircuit(2, 2)
qc.rx(angles[0], 0)
qc.ry(angles[1], 1)
qc.cx(1, 0)

@qml.qnode(dev)
def circuit(angles):
    qml.from_qiskit(qc)(angles)
    return qml.expval(qml.Z(0))

>>> angles = [3.1, 0.45]
>>> circuit(angles)
tensor(-0.89966835, requires_grad=True)

When measurement=None, all of the measurements performed in the QuantumCircuit will be returned by the quantum function in the form of a mid-circuit measurement. For example, if we define a QuantumCircuit with measurements:

import pennylane as qml
from qiskit import QuantumCircuit

qc = QuantumCircuit(2, 2)
qc.h(0)
qc.measure(0, 0)
qc.rz(0.24, [0])
qc.cx(0, 1)
qc.measure_all()

Then we can create a PennyLane circuit that uses this as a sub-circuit, and performs additional operations conditional on the results. We can also calculate standard mid-circuit measurement statistics, like expectation value, on the returned measurements:

@qml.qnode(qml.device("default.qubit"))
def circuit():
    # apply the QuantumCircuit and retrieve the measurements
    mid_measure0, m0, m1 = qml.from_qiskit(qc)()

    # conditionally apply an additional operation based on the results
    qml.cond(mid_measure0==0, qml.RX)(np.pi/2, 0)

    # return the expectation value of one of the mid-circuit measurements, and a terminal measurement
    return qml.expval(mid_measure0), qml.expval(m1)

>>> circuit()
(tensor(0.5, requires_grad=True), tensor(0.5, requires_grad=True))

Note

The order of mid-circuit measurements returned by qml.from_qiskit() in the example above is determined by the order in which measurements appear in the input Qiskit QuantumCircuit.

Furthermore, the Qiskit IfElseOp, SwitchCaseOp and c_if conditional workflows are automatically translated into their PennyLane counterparts during conversion. For example, if we construct a QuantumCircuit with these workflows:

qc = QuantumCircuit(4, 1)
qc.h(0)
qc.measure(0, 0)

# Use an `IfElseOp` operation.
noop = QuantumCircuit(1)
flip_x = QuantumCircuit(1)
flip_x.x(0)
qc.if_else((qc.clbits[0], True), flip_x, noop, [1], [])

# Use a `SwitchCaseOp` operation.
with qc.switch(qc.clbits[0]) as case:
    with case(0):
        qc.y(2)

# Use the `c_if()` function.
qc.z(3).c_if(qc.clbits[0], True)

qc.measure_all()

We can convert the QuantumCircuit into a PennyLane quantum function using:

dev = qml.device("default.qubit")

measurements = [qml.expval(qml.Z(i)) for i in range(qc.num_qubits)]
cond_circuit = qml.QNode(qml.from_qiskit(qc, measurements=measurements), dev)

The result is:

>>> print(qml.draw(cond_circuit)())
0: ──H──┤↗├──────────╭||─┤  <Z>
1: ──────║───X───────├||─┤  <Z>
2: ──────║───║──Y────├||─┤  <Z>
3: ──────║───║──║──Z─╰||─┤  <Z>
         ╚═══╩══╩══╝
Previous
Next