Photo: IBM Q System One Quantum Computer on display at the Consumer Electronics Show 2020
Noisy intermediate-scale quantum (NISQ) computers are shifting the paradigm of information processing from the classical bit to the quantum bit (qubit) thanks to the intrinsic massive parallelisation they can offer. Quantum computers therefore hold the potential to be amongst the most disruptive technologies of the 21st century. The power of quantum computing is made possible by the complexity of quantum states that can be created, manipulated, and measured by a quantum computer. NISQ computers operate by applying a set of quantum gates to an initial state to achieve a final state that encodes the solution to a computationally hard problem, e.g. evaluating the energy of a molecule or the shortest path in a travelling salesman problem.
Implementing multi-qubit gates efficiently with high fidelity is essential for achieving universal fault tolerant computing. Quantum optimal control enables the realisation of accurate operations, such as quantum gates, and supports the development of quantum technologies. A key figure of merit of the performance of a quantum gate is the fidelity, which is a measure of the distance between the quantum state after the application of the quantum gate and a desired target state. Unfortunately, state-of-the-art quantum gate fidelities are too low for the application of a sufficient number of gates in order to attain the solution of such complex problems. Quantum optimal control theory is deemed to be the cornerstone for enabling quantum technologies by devising and implementing the shapes of the electromagnetic pulses that drive the evolution from the initial to the final quantum state.
The RLQuantOpt Project proposes to improve the fidelity of quantum gates and circuits using deep reinforcement learning (RL) techniques to optimise the pulses which generate these gates. After training a number of RL agents in a simulated environment, the best agents will be calibrated and evaluated on a physical quantum computer. The RLQuantOpt project will have a significant impact on improving the operational availability of quantum computers in a hardware independent manner, as less time will be required for calibration and running benchmarks, as well as other aspects such as gate synthesis.
The RLQuantOpt Project is an interdisciplinary one, relying on the collaboration and expertise of two departments within the University of Malta. The research team is led by Prof. In摹. Gianluca Valentino from the Department of Communications and Computer Engineering and Dr Tony Apollaro from the Department of Physics, and includes Dr Leander Grech and Mr Mirko Consiglio.
Project RLQuantOpt financed by the Malta Council for Science & Technology, for and on behalf of the Foundation for Science and Technology, through the FUSION: R&I Research Excellence Programme.
