Automating quantum control with machine learning

Información del proyecto

qx

Identificador del acuerdo de subvención: 190129990

DOI

10.3030/190129990

Proyecto cerrado

Fecha de la firma de la CE 2 Febrero 2023

Fecha de inicio 1 Abril 2023

Fecha de finalización 30 Septiembre 2025

Financiado con arreglo a

• The European Innovation Council (EIC)

Coste total

€ 3 565 000,00

Aportación de la UE

€ 2 495 500,00

Coordinado por

QUANTROLOX FINLAND OY
Finland

Periodic Reporting for period 2 - qx (Automating quantum control with machine learning)

Período documentado: 2024-04-01 hasta 2025-09-30

Resumen del contexto y de los objetivos generales del proyecto

Quantum processing units (QPUs) are at the heart of quantum computers that utilize quantum mechanical properties of qubits, such as entanglement and interference, to perform calculations that classical computers cannot. The slowness of bringing up qubits, characterizing and testing them and their coupling and entanglements has substantially hampered the development of QPUs. For QPUs with just 10s of qubits the process is known to take anywhere from several weeks to a few months resulting in ponderous iterative development. Even after 25 years since the invention of the first solid state qubit in 1999 we are still struggling in the 100 qubit era.

The difficulty in the characterisation process arises due to variations in the fabrication process and inherent defects in the material, resulting in qubits that vary significantly across a quantum processing unit (QPU). This means that no two qubits are the same and devices vary significantly in their behavior: Each qubit in a QPU has to be painstakingly characterized and tested using tools that require significant manual operations by quantum experts.

The overall objective of the project is to develop software that learns the variability of each qubit and decides the appropriate microwave pulse sequences to ensure automated qubit bring-up, characterization and testing with just a press of a button, thereby speeding up the development to QPUs which is turn accelerates the path to practical quantum computers.

The software should also seamlessly integrate the workflow of the QPU development and production processes. This is crucial for scaling quantum computers.

Trabajo realizado desde el comienzo del proyecto hasta el final del período abarcado por el informe y los principales resultados hasta la fecha

The main achievements during the reporting period where:

-Development of core algorithms for superconducting qubits where we implemented device characterisation and single qubit tuning.
-We launched Quantum EDGE for automation of qubit bring up, characterization, testing and tune up.
-We developed an easy to use user interface enabling QPU monitoring that gathers and displays key qubit metrics; design and execution of qubit characterization workflows, as well as the data visualization of all measurements performed in the workflows.
-We realized an automated device bring-up and single qubit tuning workflow with ongoing work on automating 2 qubit gates to enable entanglement of qubit pairs on a QPU.
-We support three superconducting qubit architectures: flux tunable and fixed frequency transmons, as well as fluxonium qubits.
-Integration interfaces with major control electronics manufacturers namely: Qblox, Quantum Machines, Zurich Instruments

Avances que van más allá del estado de la técnica e impacto potencial esperado (incluida la repercusión socioeconómica y las implicaciones sociales más amplias del proyecto hasta la fecha)

With only it’s second release, QuantrolOx’s Quantum EDGE qubit automation software has set a new benchmark for speed and ease of use for qubit bring up, characterisation and testing (characterisation for short) with over 100x speed up compared to current industry benchmarks, thereby removing a major bottleneck in the rate of development quantum processing units (QPUs) and accelerating the path to developing practical quantum computers.

The benefit of Quantum EDGE over current solution is demonstrated through two case studies:

Case Study 1: Speed up qubit device bring-up, characterisation and testing: With current tools the industry norm is 1-2 days to bring up, characterize and test a single qubit, with a need for frequent manual intervention in the process. QuantroloOx showed that Quantum EDGE is able to complete the entire process in less than 10 minutes with a single click. This represents a speed up of two orders of magnitude.

Case Study 2: Speed up shipments of superconducting quantum chips: A company producing 〜20 qubit QPU currently takes 14 days to it bring up, characterize and test it. As each QPU needs to be characterized and tested before shipping to customers, the 14 days period adds a significant cost to the production process. With Quantum EDGE the entire bring up, characterisation and testing process takes less than a day.

These case studies showcase that Quantum EDGE dramatically increases the speed of qubit and QPU architecture development, reducing the need for highly trained quantum experts in the process, and accelerating innovation and increasing the production throughput of quantum devices.

Quantum computing is in a nascent stage with the most advanced systems limits to around 100 physical qubits. To build useful quantum computers we will need to have quantum computers with millions of physical qubits. We need to further develop our software in close partnership with the community so that we can develop the technology stack step by step to overcome the myriad of upcoming technological challenges.

Quantum EDGE Automation use case summary

Quantum EDGE Automation Workflow screen with device bring-up and single qubit tuning workflows.

Quantum EDGE QPU monitoring screen

Quantum EDGE Data visualization screen showing a 2D plot of Qubit power spectroscopy.