Pulse (qiskit.pulse)

Qiskit-Pulse is a pulse-level quantum programming kit. This lower level of programming offers the user more control than programming with QuantumCircuit s.

Extracting the greatest performance from quantum hardware requires real-time pulse-level instructions. Pulse answers that need: it enables the quantum physicist user to specify the exact time dynamics of an experiment. It is especially powerful for error mitigation techniques.

The input is given as arbitrary, time-ordered signals (see: Instructions (qiskit.pulse.instructions)) scheduled in parallel over multiple virtual hardware or simulator resources (see: Channels (qiskit.pulse.channels)). The system also allows the user to recover the time dynamics of the measured output.

This is sufficient to allow the quantum physicist to explore and correct for noise in a quantum system.

Instructions (qiskit.pulse.instructions)

instructions	The instruction module holds the various Instruction s which are supported by Qiskit Pulse.
Acquire(duration, channel[, mem_slot, …])	The Acquire instruction is used to trigger the ADC associated with a particular qubit; e.g.
Call(subroutine[, value_dict, name])	Pulse Call instruction.
Delay(duration, channel[, name])	A blocking instruction with no other effect.
Play(pulse, channel[, name])	This instruction is responsible for applying a pulse on a channel.
SetFrequency(frequency, channel[, name])	Set the channel frequency.
ShiftFrequency(frequency, channel[, name])	Shift the channel frequency away from the current frequency.
SetPhase(phase, channel[, name])	The set phase instruction sets the phase of the proceeding pulses on that channel to phase radians.
ShiftPhase(phase, channel[, name])	The shift phase instruction updates the modulation phase of proceeding pulses played on the same Channel.
Snapshot(label[, snapshot_type, name])	An instruction targeted for simulators, to capture a moment in the simulation.

Pulse Library (waveforms qiskit.pulse.library)

library	This library provides Pulse users with convenient methods to build Pulse waveforms.
library.discrete	Module for builtin discrete pulses.
Waveform(samples[, name, epsilon, …])	A pulse specified completely by complex-valued samples; each sample is played for the duration of the backend cycle-time, dt.
Constant(duration, amp[, name])	A simple constant pulse, with an amplitude value and a duration:
Drag(duration, amp, sigma, beta[, name])	The Derivative Removal by Adiabatic Gate (DRAG) pulse is a standard Gaussian pulse with an additional Gaussian derivative component.
Gaussian(duration, amp, sigma[, name])	A truncated pulse envelope shaped according to the Gaussian function whose mean is centered at the center of the pulse (duration / 2):
GaussianSquare(duration, amp, sigma[, …])	A square pulse with a Gaussian shaped risefall on both sides. Either risefall_sigma_ratio

Channels (qiskit.pulse.channels)

Pulse is meant to be agnostic to the underlying hardware implementation, while still allowing low-level control. Therefore, our signal channels are virtual hardware channels. The backend which executes our programs is responsible for mapping these virtual channels to the proper physical channel within the quantum control hardware.

Channels are characterized by their type and their index. See each channel type below to learn more.

channels	This module defines Pulse Channels.
DriveChannel(index)	Drive channels transmit signals to qubits which enact gate operations.
MeasureChannel(index)	Measure channels transmit measurement stimulus pulses for readout.
AcquireChannel(index)	Acquire channels are used to collect data.
ControlChannel(index)	Control channels provide supplementary control over the qubit to the drive channel.
RegisterSlot(index)	Classical resister slot channels represent classical registers (low-latency classical memory).
MemorySlot(index)	Memory slot channels represent classical memory storage.

Schedules

Schedules are Pulse programs. They describe instruction sequences for the control hardware.

Schedule(*schedules[, name, metadata])	A quantum program schedule with exact time constraints for its instructions, operating over all input signal channels and supporting special syntaxes for building.
ScheduleBlock([name, metadata, …])	A ScheduleBlock is a time-ordered sequence of instructions and transform macro to manage their relative timing.
Instruction(operands[, duration, channels, name])	The smallest schedulable unit: a single instruction.

Configuration

InstructionScheduleMap()

Mapping from QuantumCircuit qiskit.circuit.Instruction names and qubits to Schedule s. In particular, the mapping is formatted as type::.

Schedule Transforms (qiskit.pulse.transforms)

Schedule transforms take Schedule s as input and return modified Schedule s.

transforms.align_measures(schedules[, …])	Return new schedules where measurements occur at the same physical time.
transforms.add_implicit_acquires(schedule, …)	Return a new schedule with implicit acquires from the measurement mapping replaced by explicit ones.
transforms.pad(schedule[, channels, until, …])	Pad the input Schedule with Delay``s on all unoccupied timeslots until ``schedule.duration or until if not None.

Exceptions

PulseError(*message)

Errors raised by the pulse module.

Pulse Builder (builder)

Warning

The pulse builder interface is still in active development. It may have breaking API changes without deprecation warnings in future releases until otherwise indicated.

The pulse builder provides an imperative API for writing pulse programs with less difficulty than the Schedule API. It contextually constructs a pulse schedule and then emits the schedule for execution. For example to play a series of pulses on channels is as simple as:

from qiskit import pulse

dc = pulse.DriveChannel
d0, d1, d2, d3, d4 = dc(0), dc(1), dc(2), dc(3), dc(4)

with pulse.build(name='pulse_programming_in') as pulse_prog:
    pulse.play([1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1], d0)
    pulse.play([1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0], d1)
    pulse.play([1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0], d2)
    pulse.play([1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0], d3)
    pulse.play([1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 0, 1, 0], d4)

pulse_prog.draw()

In the future the pulse builder will be coupled to the QuantumCircuit with an equivalent circuit builder interface.

build([backend, schedule, name, …])

Create a context manager for launching the imperative pulse builder DSL.

Channels

Methods to return the correct channels for the respective qubit indices.

from qiskit import pulse
from qiskit.test.mock import FakeArmonk

backend = FakeArmonk()

with pulse.build(backend) as drive_sched:
    d0 = pulse.drive_channel(0)
    print(d0)

DriveChannel(0)

acquire_channel(qubit)	Return AcquireChannel for qubit on the active builder backend.
control_channels(*qubits)	Return ControlChannel for qubit on the active builder backend.
drive_channel(qubit)	Return DriveChannel for qubit on the active builder backend.
measure_channel(qubit)	Return MeasureChannel for qubit on the active builder backend.

Instructions

Pulse instructions are available within the builder interface. Here’s an example:

from qiskit import pulse
from qiskit.test.mock import FakeArmonk

backend = FakeArmonk()

with pulse.build(backend) as drive_sched:
    d0 = pulse.drive_channel(0)
    a0 = pulse.acquire_channel(0)

    pulse.play(pulse.library.Constant(10, 1.0), d0)
    pulse.delay(20, d0)
    pulse.shift_phase(3.14/2, d0)
    pulse.set_phase(3.14, d0)
    pulse.shift_frequency(1e7, d0)
    pulse.set_frequency(5e9, d0)

    with pulse.build() as temp_sched:
        pulse.play(pulse.library.Gaussian(20, 1.0, 3.0), d0)
        pulse.play(pulse.library.Gaussian(20, -1.0, 3.0), d0)

    pulse.call(temp_sched)
    pulse.acquire(30, a0, pulse.MemorySlot(0))

drive_sched.draw()

acquire(duration, qubit_or_channel, …)	Acquire for a duration on a channel and store the result in a register.
barrier(*channels_or_qubits[, name])	Barrier directive for a set of channels and qubits.
call(target[, name, value_dict])	Call the target within the currently active builder context with arbitrary parameters which will be assigned to the target program.
delay(duration, channel[, name])	Delay on a channel for a duration.
play(pulse, channel[, name])	Play a pulse on a channel.
set_frequency(frequency, channel[, name])	Set the frequency of a pulse channel.
set_phase(phase, channel[, name])	Set the phase of a pulse channel.
shift_frequency(frequency, channel[, name])	Shift the frequency of a pulse channel.
shift_phase(phase, channel[, name])	Shift the phase of a pulse channel.
snapshot(label[, snapshot_type])	Simulator snapshot.

Contexts

Builder aware contexts that modify the construction of a pulse program. For example an alignment context like align_right() may be used to align all pulses as late as possible in a pulse program.

from qiskit import pulse

d0 = pulse.DriveChannel(0)
d1 = pulse.DriveChannel(1)

with pulse.build() as pulse_prog:
    with pulse.align_right():
        # this pulse will start at t=0
        pulse.play(pulse.Constant(100, 1.0), d0)
        # this pulse will start at t=80
        pulse.play(pulse.Constant(20, 1.0), d1)

pulse_prog.draw()

align_equispaced(duration)	Equispaced alignment pulse scheduling context.
align_func(duration, func)	Callback defined alignment pulse scheduling context.
align_left()	Left alignment pulse scheduling context.
align_right()	Right alignment pulse scheduling context.
align_sequential()	Sequential alignment pulse scheduling context.
circuit_scheduler_settings(**settings)	Set the currently active circuit scheduler settings for this context.
frequency_offset(frequency, *channels[, …])	Shift the frequency of inputs channels on entry into context and undo on exit.
inline()	Deprecated.
pad(*chs)	Deprecated.
phase_offset(phase, *channels)	Shift the phase of input channels on entry into context and undo on exit.
transpiler_settings(**settings)	Set the currently active transpiler settings for this context.

Macros

Macros help you add more complex functionality to your pulse program.

from qiskit import pulse
from qiskit.test.mock import FakeArmonk

backend = FakeArmonk()

with pulse.build(backend) as measure_sched:
    mem_slot = pulse.measure(0)
    print(mem_slot)

MemorySlot(0)

measure(qubits[, registers])	Measure a qubit within the currently active builder context.
measure_all()	Measure all qubits within the currently active builder context.
delay_qubits(duration, *qubits)	Insert delays on all of the channels.Channels that correspond to the input qubits at the same time.

Circuit Gates

To use circuit level gates within your pulse program call a circuit with qiskit.pulse.builder.call().

Warning

These will be removed in future versions with the release of a circuit builder interface in which it will be possible to calibrate a gate in terms of pulses and use that gate in a circuit.

import math

from qiskit import pulse
from qiskit.test.mock import FakeArmonk

backend = FakeArmonk()

with pulse.build(backend) as u3_sched:
    pulse.u3(math.pi, 0, math.pi, 0)

cx(control, target)	Call a CXGate on the input physical qubits.
u1(theta, qubit)	Call a U1Gate on the input physical qubit.
u2(phi, lam, qubit)	Call a U2Gate on the input physical qubit.
u3(theta, phi, lam, qubit)	Call a U3Gate on the input physical qubit.
x(qubit)	Call a XGate on the input physical qubit.

Utilities

The utility functions can be used to gather attributes about the backend and modify how the program is built.

from qiskit import pulse

from qiskit.test.mock import FakeArmonk

backend = FakeArmonk()

with pulse.build(backend) as u3_sched:
    print('Number of qubits in backend: {}'.format(pulse.num_qubits()))

    samples = 160
    print('There are {} samples in {} seconds'.format(
        samples, pulse.samples_to_seconds(160)))

    seconds = 1e-6
    print('There are {} seconds in {} samples.'.format(
        seconds, pulse.seconds_to_samples(1e-6)))

Number of qubits in backend: 1
There are 160 samples in 3.5555555555555554e-08 seconds
There are 1e-06 seconds in 4500 samples.

active_backend()	Get the backend of the currently active builder context.
active_transpiler_settings()	Return the current active builder context’s transpiler settings.
active_circuit_scheduler_settings()	Return the current active builder context’s circuit scheduler settings.
num_qubits()	Return number of qubits in the currently active backend.
qubit_channels(qubit)	Returns the set of channels associated with a qubit.
samples_to_seconds(samples)	Obtain the time in seconds that will elapse for the input number of samples on the active backend.
seconds_to_samples(seconds)	Obtain the number of samples that will elapse in seconds on the active backend.