Quantum reservoir computing for efficient signal processing

Project Information

QRC-4-ESP

Grant agreement ID: 101129663

DOI

10.3030/101129663

EC signature date 27 October 2023

Start date 1 January 2024

End date 31 December 2026

Funded under

• The European Innovation Council (EIC)

Total cost

€ 2 522 411,25

EU contribution

€ 2 522 411,25

Coordinated by

LEIBNIZ-INSTITUT FUER PHOTONISCHE TECHNOLOGIEN E.V.
Germany

Periodic Reporting for period 1 - QRC-4-ESP (Quantum reservoir computing for efficient signal processing)

Reporting period: 2024-01-01 to 2024-12-31

Summary of the context and overall objectives of the project

The long-term vision of the QRC-4-ESP project is to produce the World’s first quantum reservoir computing (QRC) systems based on superconducting qubits and silicon carbide (SiC) defect qubits. This new disruptive technology will create drastic improvements in speed and reduction in power consumption – two or more orders of magnitude (>100X)
- compared to classical machine learning systems. Ultimately, the technology will enable ground-breaking microwave range, open air, quantum communication and new optical range, fibre network, quantum sensors. Due to their structure, superconducting qubits naturally operate in the microwave range (hundreds of MHz to tens of GHz), which means they are well-matched to the required frequency range for satellite communications, because signals in this frequency band are minimally disturbed by fog and clouds. Currently, operating open-air quantum communications in this range is difficult due to strong background thermal noise. However, the development of superconducting quantum sensors and their integration with superconducting QRC could resolve this issue by enabling the routine use of well-developed quantum key distribution protocols. Thereby, quantum satellite communication could be successfully deployed - once and for all - without the risk of interception and decryption. Defect-based qubits in SiC can operate in several frequency bands, including the optical band. Here we are especially interested in their operation in the near-infrared, which would make them a natural match for fibre-optical networks. Long-range open-air quantum communication in the optical range is impractical due to atmospheric interference. However, the inclusion of prospective QRC devices - with quantum inputs and outputs - as quantum repeaters could significantly increase performance and reduce costs. Another application would be to integrate an optical-range quantum sensor with an image-processing QRC, which would be very useful for medical diagnostics.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

By the end of Period 1 (Year 1), the QRC-4-ESP project had made the following progress

WP1 – QR Theory and Models
• A first-principles quantitative model of a set of transmon qubits in a waveguide is developed and successfully tested on the experimental data from WP2.
• A generic model of a qubit-based quantum reservoir developed and used to simulate different regimes of quantum reservoir operation.
• A new approach to quantum reservoir operation (“measurevoir”) proposed.
• A quantitative model of a defect-based qubit developed.

WP2 – Superconducting QR
• Designed, fabricated and tested a sample containing 5 superconducting qubits coupled to coplanar waveguide resonator.
• Determined the dependence transmission coefficient on the applied power and magnetic field.
• Observed bistability, explained by the theory of WP1.

WP3 – Defect-based QR
• Designed a quantum system consisting of one to several Si vacancies in 4H-SiC with electron spin S=3/2. Each Si vacancy has a quartet spin state in both the ground and excited state, |±1/2> and |±3/2>, which can be optically readout by photoluminescence excitation (PLE).
• Realised isolated Si vacancy emitters and clusters of several Si vacancies by electron irradiation in 15 pure 4H-SiC samples. Photoluminescence (PL) mapping using confocal microscope allows to determine the density of emitters for choosing and optimising quantum systems for optically detected magnetic resonance (ODMR) and PLE studies.
• Used isotope-pure layers to reduce the magnetic noises from the nuclear spin bath (13C and 29Si isotopes), the pure layers help to reduce both the decoherence caused by dipole interaction of other defects and impurities and the electric noise from local environment induced by ionisation of other defects under optical excitation. Made hundreds of PLE scans continuously under resonant excitation without repump laser that showed stable single Si emitters without signs of ionisation.
• Demonstrated that single Si vacancies within quantum system nonlinearly interact with each other, with spins of other defects and the nuclear spin bath via dipole-dipole interaction and with charges in the local environment, which induce changes in PLE spectra. Potential inputs can be the laser power, the external magnetic field, and the microwave (MW) power, and outputs are the energy shift of the zero-phonon lines (ZPL) corresponding to optical transitions between |±1/2> (A1 line) states and between |±3/2> (A2 line), and their intensities. These parameters have been accounted for in theoretical models as perturbations.
• Collected PLE data for some clusters with different sizes. The obtained PLE data confirm that the model of Si vacancy quantum system is working in principle.

WP4 – QRC Integration, Demonstration and Benchmarking
• Implemented effective input/output strategies for Quantum Reservoir Computing (QRC) systems integrated with neural networks. Successfully tested a Python-based simulator for data encoding and extraction, validating the linear readout mechanism for regression and classification tasks.
• Designed a benchmark suite to evaluate QRC systems' performance on tasks like regression, classification, time-series forecasting, clustering, and anomaly detection using synthetic datasets. Performance metrics (MSE, NMSE) are used to systematically assess QR-enhanced and traditional networks.
• Initiated benchmarking of experimental QRC systems for industrial tasks like signal processing and anomaly detection, with plans for future integration of refined prototypes and real-world applications.

WP5 – Dissemination, Communication and Exploitation
• Presented scientific results via 10 papers and posters at international conferences.
• Published a short-animation explanatory video about the QRC-4-ESP project.
• Published 20 separate news announcements via the project website and social media accounts.
• Participated in numerous outreach events (Pint of Science, online brokerage and showcase events).

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

Theoretical models for superconducting and defect-based Quantum Reservoir Computing (QRC) have been developed, achieving accurate fits to experimental data and enabling simulations of QRC operations. These models guide ongoing experiments and further model improvements, with the objective on track for completion.
A quantum reservoir using superconducting qubits was fabricated and tested, featuring a coplanar resonator with flux-tunable transmon qubits. Initial experiments showed promising results, but further measurements and design improvements are needed for deeper insights.
A SiC-based quantum reservoir has been designed, leveraging silicon vacancies with spin states readable via photoluminescence excitation (PLE). Preliminary studies identified optimal material conditions to minimize decoherence. Electron irradiation and PLE experiments have yielded promising samples, setting the stage for optimizing defect ensembles and input-output methods.
Progress includes integrating a neural network with a QRC as an untrained layer, enhancing computational efficiency for tasks like signal processing. Benchmarking with synthetic data demonstrated the potential of the QRC layer. Future work involves validating strategies with experimental data and scaling for industrial applications.

Project logo