Amazon Braket launches new superconducting quantum processor from IQM

This post was contributed by Daniela Becker, Tim (Yi-Ting) Chen at AWS, and Max Haeberlein at IQM Quantum Computers.

Amazon Braket enables customers to design and run quantum algorithms and explore applications of quantum computing, by providing access to a choice of quantum hardware through a unified interface. Today, we further expand on that choice and announce the general availability of the first publicly available device from IQM Quantum Computers. The device, named Garnet, is a 20-qubit quantum processing unit (QPU), based on superconducting technology. Adding access to IQM’s Garnet processor, which is located in Finland, also expands Braket to the Europe (Stockholm) Region. This launch now enables customers in the European Union to conveniently experiment with this QPU during their work hours, while helping them meet their data sovereignty needs.

Figure 1: The Garnet processor in IQM’s facilities.

In this early phase of quantum computing, experimentation across a variety of available devices is crucial for developing quantum algorithms to eventually address complex problems across domains like finance, energy, pharmaceuticals, and logistics. Customers all over the world now have the ability to run experiments on IQM’s quantum hardware, using on-demand access for designing and executing programs, and priority access via  Hybrid Jobs for running algorithms, all with simple pay-as-you-go pricing. Furthermore, for their most latency or time-sensitive workloads, customers can reserve dedicated capacity on the Garnet QPU, and engage directly with AWS and IQM’s scientific expertise through Braket Direct, with hourly pricing and no upfront commitments.

Garnet is based on superconducting transmon qubits, where the qubits are arranged in a square lattice and connected by tunable couplers as shown in Figure 2. The device has achieved median 1-qubit gate fidelity of 99.92%, median 2-qubit gate fidelity of 99.51%, and a  Quantum Volume  of 2⁵=32. Up-to-date characterization data can be found on Garnet’s device details page in the Braket Console. More information on the latest benchmarking results can also be found here.

Figure 2: The Garnet QPU topology showing a square lattice with tunable couplers.

In addition to the new hardware offering, any European customers with data residency needs can now more easily demonstrate compliance when using this EU-based QPU in the Europe (Stockholm) Region. For added convenience to customers in Europe, the Middle-East, and East Asia, the Garnet processor will be available for use Monday to Wednesday from 8:00 to 10:00 UTC (9 am-11 am CET), and on Thursday and Friday from 13:00 to 15:00 UTC (2-4pm CET).

Getting Started with Garnet on Braket

One of the core value propositions of Braket is to provide customers a unified, easy-to-use programming interface to access a choice of hardware. To execute programs on Garnet, simply specify the Amazon Resource Name (ARN) in the device definition before running your circuit, as shown in the snippet below.

device = AwsDevice("arn:aws:braket:eu-north-1::device/qpu/iqm/Garnet")

You can run a circuit with as little as two lines of code as shown in the Bell State example below:

# run circuit

bell = Circuit().h(0).cnot(control=0, target=1)

result = device.run(bell, shots=1000).result()

Experiment with large-scale entanglement on Garnet

A critical requirement for harnessing the full potential of quantum computing technology is achieving a highly coherent quantum state involving many qubits. The Greenberger-Horne-Zeilinger (GHZ) state is such an entangled state of many qubits. If the fidelity of the GHZ state exceeds 50%, it serves as an indicator that genuine multi-qubit entanglement has been achieved within the quantum system.

Below we show how to prepare a 20-qubit GHZ state using the Garnet device:

from braket.aws import AwsDevice
from braket.circuits import Circuit

device = AwsDevice("arn:aws:braket:eu-north-1::device/qpu/iqm/Garnet")

N = 20

edges = [
    (9, 8), (8, 7), (9, 10), (7, 2), (8, 13), (9, 4), 
    (10, 15), (7, 12),(13, 17), (2, 3), (9, 14), (10, 11),
    (15, 19), (4, 1), (1, 0), (14, 18), (15, 16), (11, 6), (10, 5)
] 
circuit = Circuit().h(9)
for i in range(N-1):
    q1, q2 = edges[i]
    circuit.cnot(q1,q2)
task = device.run(circuit, shots=5000)

Figure 3 shows the fidelities of 10-qubit to 20-qubit GHZ states prepared on Garnet. All GHZ states achieve fidelities above 50%. For more system-level benchmark results, check out IQM’s blog [2].

Figure 3: Fidelity of 10-qubit to 20-qubit GHZ states on IQM’s Garnet.

Visit the Amazon Braket console to view the device topology, availability, and get up-to-date information about the single and two-qubit gate fidelities, and readout fidelities. To get started with running your own quantum programs on the Garnet processor, refer to our GitHub repository for example notebooks and tutorials. You can run these programs using one of our managed Jupyter notebooks, or from your local development environment. To reserve dedicated capacity on the Garnet QPU through Braket Direct, navigate to the Braket Direct Console page and click Reserve Device to get started.

References

[1] Gary J Mooney et. al., Generation and verification of 27-qubit Greenberger-Horne-Zeilinger states in a superconducting quantum computer, J. Phys. Commun. 5 095004 (2021)

[2] IQM Quantum Computers achieves a new benchmark result on 20-qubit quantum computer