Calibration experiments#
Let’s see some examples of the typical experiments needed to calibrate and characterize a qubit.
Note
This is just for demonstration purposes! In the Qibo framework these experiments are already coded and available in the Qibocal API.
Let’s consider a platform called single_qubit with, as expected, a single qubit.
Resonator spectroscopy#
The first experiment we conduct is a resonator spectroscopy. The experiment is as follows:
- A measurement pulse (pulse on the readout line, followed by an acquisition)
is fired at a specific frequency.
We repeat point 1 for different frequencies.
We plot the acquired amplitudes, identifying the peak/deep value as the resonator frequency.
We start by initializing the platform, creating a sequence composed of only a measurement and a sweeper around the pre-defined frequency. We then define the execution parameters and launch the experiment. In few seconds, the experiment will be finished and we can proceed to plot it. This is done in the following script:
import numpy as np
import matplotlib.pyplot as plt
from qibolab import (
AcquisitionType,
AveragingMode,
Parameter,
PulseSequence,
Sweeper,
create_platform,
)
# allocate platform
platform = create_platform("dummy")
qubit = platform.qubits[0]
natives = platform.natives.single_qubit[0]
sequence = natives.MZ.create_sequence()
# allocate frequency sweeper
f0 = platform.config(qubit.probe).frequency
sweeper = Sweeper(
parameter=Parameter.frequency,
range=(f0 - 2e8, f0 + 2e8, 1e6),
channels=[qubit.probe],
)
results = platform.execute(
[sequence],
[[sweeper]],
nshots=1000,
relaxation_time=50,
averaging_mode=AveragingMode.CYCLIC,
acquisition_type=AcquisitionType.INTEGRATION,
)
acq = sequence.acquisitions[0][1]
signal = results[acq.id]
amplitudes = np.abs(signal[..., 0] + 1j * signal[..., 1])
frequencies = sweeper.values
plt.title("Resonator Spectroscopy")
plt.xlabel("Frequencies [Hz]")
plt.ylabel("Amplitudes [a.u.]")
plt.plot(frequencies, amplitudes)
plt.show()
Qubit spectroscopy#
For a qubit spectroscopy experiment, the procedure is almost identical. A typical qubit spectroscopy experiment is as follows:
A first pulse is sent to the drive line, in order to excite the qubit. Since the qubit parameters are not known, this is typically a very long pulse (2 microseconds) at low amplitude.
A measurement, tuned with resonator spectroscopy, is performed.
We repeat point 1 for different frequencies of the drive pulse.
We plot the acquired amplitudes, identifying the deep/peak value as the qubit frequency.
The main difference introduced by this experiment is a slightly more complex pulse sequence. Therefore with start with that:
import numpy as np
import matplotlib.pyplot as plt
from qibolab import (
AcquisitionType,
AveragingMode,
Parameter,
PulseSequence,
Sweeper,
create_platform,
)
# allocate platform
platform = create_platform("dummy")
qubit = platform.qubits[0]
natives = platform.natives.single_qubit[0]
# create pulse sequence and add pulses
sequence = natives.RX() | natives.MZ()
# allocate frequency sweeper
f0 = platform.config(qubit.drive).frequency
sweeper = Sweeper(
parameter=Parameter.frequency,
range=(f0 - 2e8, f0 + 2e8, 1e6),
channels=[qubit.drive],
)
results = platform.execute(
[sequence],
[[sweeper]],
nshots=1000,
relaxation_time=50,
averaging_mode=AveragingMode.CYCLIC,
acquisition_type=AcquisitionType.INTEGRATION,
)
acq = sequence.acquisitions[0][1]
signal = results[acq.id]
amplitudes = np.abs(signal[..., 0] + 1j * signal[..., 1])
frequencies = sweeper.values
plt.title("Qubit Spectroscopy")
plt.xlabel("Frequencies [Hz]")
plt.ylabel("Amplitudes [a.u.]")
plt.plot(frequencies, amplitudes)
plt.show()
Note that the drive pulse has been changed to match the characteristics required for the experiment.
Single shot classification#
To avoid seeing other very similar experiment, let’s jump to the single shot classification experiment. The single-shot classification experiment is conducted towards the end of the single-qubit calibration process and assumes the availability of already calibrated pulses.
Two distinct pulse sequences are prepared for the experiment:
Sequence with only a measurement pulse.
Sequence comprising an RX pulse (X gate) followed by a measurement pulse.
For each sequence, the qubit is initialized in state 0 (no operation applied), and a measurement is executed. This process is repeated multiple times. Unlike previous experiments, the results of each individual measurement are saved separately, avoiding averaging. Both measurements are repeated: first with the single-pulse sequence and then with the two-pulse sequence. The goal is to compare the outcomes and visualize the differences in the IQ plane between the two states.
Prepare the single-pulse sequence: Measure the qubit multiple times in state 0.
Prepare the two-pulse sequence: Apply an RX pulse followed by measurement, and perform the same measurement multiple times.
Plotting the Results: Plot the single-shot results for both sequences, highlighting the differences in the IQ plane between the two states.
This experiment serves to assess the effectiveness of single-qubit calibration and its impact on qubit states in the IQ plane.
import numpy as np
import matplotlib.pyplot as plt
from qibolab import (
AcquisitionType,
AveragingMode,
Parameter,
Sweeper,
create_platform,
)
# allocate platform
platform = create_platform("dummy")
qubit = platform.qubits[0]
natives = platform.natives.single_qubit[0]
# create pulse sequence 1
zero_sequence = natives.MZ()
# create pulse sequence 2
one_sequence = natives.RX() | natives.MZ()
results = platform.execute(
[zero_sequence, one_sequence],
nshots=1000,
relaxation_time=50_000,
averaging_mode=AveragingMode.SINGLESHOT,
acquisition_type=AcquisitionType.INTEGRATION,
)
acq0 = zero_sequence.acquisitions[0][1]
acq1 = one_sequence.acquisitions[0][1]
plt.title("Single shot classification")
plt.xlabel("I [a.u.]")
plt.ylabel("Q [a.u.]")
plt.scatter(
results[acq1.id][..., 0],
results[acq1.id][..., 1],
label="One state",
)
plt.scatter(
results[acq0.id][..., 0],
results[acq0.id][..., 1],
label="Zero state",
)
plt.show()
Note that in this experiment we passed both sequences in the same platform.execute
command.
In this case the sequences will be unrolled to a single sequence automatically, which is
then deployed with a single communication with the instruments, to reduce communication bottleneck.