ContextAwareDynamicalDecoupling

class qiskit.transpiler.passes.ContextAwareDynamicalDecoupling(*args, **kwargs)

Bases: TransformationPass

Implement an X-sequence dynamical decoupling considering the gate- and qubit-context.

This pass implements a context-aware dynamical decoupling (DD) [1], which ensures that

1. simultaneously occurring DD sequences on device-adjacent qubits are mutually orthogonal, and
2. DD sequences on spectator qubits of ECR and CX gates are orthogonal to the echo pulses on the neighboring target/control qubits.

The mutually orthogonal DD sequences are currently Walsh-Hadamard sequences, consisting of only X gates. In some cases it might therefore be beneficial to use PadDynamicalDecoupling with more generic sequences, such as XY4.

This pass performs best if the two-qubit interactions have the same durations on the device, as it allows to align delay sequences and take into account potential control and target operations on neighboring qubits. However, it is still valid if this is not the case.

Example:

from qiskit.circuit import QuantumCircuit
from qiskit.circuit.library import QFTGate
from qiskit.transpiler import PassManager
from qiskit.transpiler.passes import ALAPScheduleAnalysis, ContextAwareDynamicalDecoupling
from qiskit.transpiler.preset_passmanagers import generate_preset_pass_manager
from qiskit_ibm_runtime.fake_provider import FakeSherbrooke
 
num_qubits = 10
circuit = QuantumCircuit(num_qubits)
circuit.append(QFTGate(num_qubits), circuit.qubits)
circuit.measure_all()
 
target = FakeSherbrooke().target
 
pm = generate_preset_pass_manager(optimization_level=2, target=target)
dd = PassManager([
    ALAPScheduleAnalysis(target=target),
    ContextAwareDynamicalDecoupling(target=target),
])
 
transpiled = pm.run(circuit)
with_dd = dd.run(transpiled)
 
print(with_dd.draw(idle_wires=False))

References

[1] A. Seif et al. (2024). Suppressing Correlated Noise in Quantum Computers via

Context-Aware Compiling, arXiv:2403.06852.

Parameters

• target – The Target of the device to run the circuit.
• min_duration – Minimal delay duration (in dt) to insert a DD sequence. This can be useful, e.g. if a big delay block would be interrupted and split into smaller blocks due to a very short, adjacent delay. If None, this is set to be at least twice the difference of the longest/shortest CX or ECR gate.
• skip_reset_qubits – Skip initial delays and delays after a reset.
• skip_dd_threshold – Skip dynamical decoupling on an idle qubit, if the duration of the decoupling sequence exceeds this fraction of the idle window. For example, to skip a DD sequence if it would take up more than 95% of the idle time, set this value to 0.95. A value of 1. means that the DD sequence is applied if it fits into the window.
• pulse_alignment – The hardware constraints (in dt) for gate timing allocation. If provided, the inserted gates will only be executed on integer multiples of this value. This is usually provided on backend.configuration().timing_constraints. If None, this is extracted from the target.
• coloring_strategy – The coloring strategy used for rx.greedy_graph_color. Defaults to a saturation strategy, which is optimal on bipartite graphs, see Section 1.2.2.8 of [2].

References

[2] A. Kosowski, and K. Manuszewski, Classical Coloring of Graphs, Graph Colorings,

2-19, 2004. ISBN 0-8218-3458-4.

Attributes

is_analysis_pass

is_transformation_pass

Methods

execute

get_orthogonal_sequence

name

run

update_status