Processing math: 100%

Using PennyLane

Release news

Development

API

Internals

  1. Docs
  2. qml.io
  3. qml.io.FromBloq

qml.io.FromBloq

class FromBloq(bloq, wires)[source]

Bases: pennylane.operation.Operation

An adapter for using a Qualtran Bloq as a PennyLane Operation.

Note

This class requires the latest version of Qualtran. We recommend installing the main branch via pip:

pip install qualtran
Parameters

  • bloq (qualtran.Bloq) – an initialized Qualtran Bloq to be wrapped as a PennyLane operator

  • wires (WiresLike) – The wires the operator acts on. The number of wires can be determined by using the signature of the Bloq using bloq.signature.n_qubits().

Raises

TypeError – bloq must be an instance of Bloq.

Example

This example shows how to use qml.FromBloq:

>>> from qualtran.bloqs.basic_gates import CNOT
>>> qualtran_cnot = qml.FromBloq(CNOT(), wires=[0, 1])
>>> qualtran_cnot.matrix()
array([[1.+0.j, 0.+0.j, 0.+0.j, 0.+0.j],
   [0.+0.j, 1.+0.j, 0.+0.j, 0.+0.j],
   [0.+0.j, 0.+0.j, 0.+0.j, 1.+0.j],
   [0.+0.j, 0.+0.j, 1.+0.j, 0.+0.j]])

This example shows how to use qml.FromBloq inside a device:

>>> from qualtran.bloqs.basic_gates import CNOT
>>> dev = qml.device("default.qubit") # Execute on device
>>> @qml.qnode(dev)
... def circuit():
...     qml.FromBloq(CNOT(), wires=[0, 1])
...     return qml.state()
>>> circuit()
array([1.+0.j, 0.+0.j, 0.+0.j, 0.+0.j])

This example shows how to use qml.FromBloq to implement a textbook Quantum Phase Estimation Bloq inside a device:

from qualtran.bloqs.phase_estimation import RectangularWindowState, TextbookQPE
from qualtran.bloqs.chemistry.trotter.ising import IsingXUnitary, IsingZZUnitary
from qualtran.bloqs.chemistry.trotter.trotterized_unitary import TrotterizedUnitary

# Parameters for the TrotterizedUnitary
nsites = 5
j_zz, gamma_x = 2, 0.1
zz_bloq = IsingZZUnitary(nsites=nsites, angle=0.02 * j_zz)
x_bloq = IsingXUnitary(nsites=nsites, angle=0.01 * gamma_x)
trott_unitary = TrotterizedUnitary(
    bloqs=(x_bloq, zz_bloq),  timestep=0.01,
    indices=(0, 1, 0), coeffs=(0.5 * gamma_x, j_zz, 0.5 * gamma_x)
)

# Instantiate the TextbookQPE and pass in the unitary
textbook_qpe = TextbookQPE(trott_unitary, RectangularWindowState(3))

# Execute on device
dev = qml.device("default.qubit")
@qml.qnode(dev)
def circuit():
    qml.FromBloq(textbook_qpe, wires=range(textbook_qpe.signature.n_qubits()))
    return qml.probs(wires=[5, 6, 7])

circuit()

The decomposition of a Bloq wrapped in qml.FromBloq may use more wires than expected. For example, when we wrap Qualtran’s CZPowGate, we get

>>> from qualtran.bloqs.basic_gates import CZPowGate
>>> qml.FromBloq(CZPowGate(0.468, eps=1e-11), wires=[0, 1]).decomposition()
[FromBloq(And, wires=Wires([0, 1, 'alloc_free_2'])),
FromBloq(Z**0.468, wires=Wires(['alloc_free_2'])),
FromBloq(And†, wires=Wires([0, 1, 'alloc_free_2']))]

This behaviour results from the decomposition of CZPowGate as defined in Qualtran, which allocates and frees a wire in the same bloq. In this situation, PennyLane automatically allocates this wire under the hood, and that additional wire is named alloc_free_{idx}. The indexing starts at the length of the wires defined in the signature, which in the case of CZPowGate is 2. Due to the current limitations of PennyLane, these wires cannot be accessed manually or mapped.

Previous
Next