In an article recently published in the journal Astronomy and Computing, researchers investigated the feasibility of emerging quantum computers for applications in radio astronomy, specifically radio astronomy calibration.
Computational Challenges of Radio Telescopes
Large-scale radio telescopes’ computational requirements are anticipated to scale beyond the conventional digital resources’ capabilities in the future. The planned and current telescopes cannot process substantial amounts of generated data efficiently.
Large-scale radio telescope calibration, specifically the calibration of phased array telescopes, is computationally expensive. For instance, the computational cost of an 8-hour LOFAR two-meter sky survey (LoTSS) observation is approximately 52,000 core hours. Thus, methods must be developed to reduce or meet this computational demand efficiently and decrease energy requirements.
Redundant Baseline Calibration
The regular array-introduced redundancy is exploited by one class of calibration for self-calibrating the array in a statistically efficient way. This study explored the quantum-accelerated version of this calibration form owing to its real-world applicability and simple structure. Additionally, redundancy calibration primarily depends on solving linear equation sets, for which effective quantum algorithms currently exist.
The hydrogen epoch of reionization array (HERA) is one radio telescope that uses an extremely regular array structure. This telescope consists of several antennas in a regular hexagonal configuration, which results in a substantial redundancy between baselines, making HERA suitable for redundancy calibration.
The Study
In this study, researchers investigated the potential use of combinatorial solvers in quantum annealers (QAs) and variational quantum linear solvers (VQLSs) in noisy intermediate-scale quantum (NISQ) computers for a radio astronomy calibration pipeline. Specifically, two distinct quantum computing approaches, including QAs developed by D-Wave and gate-based quantum computers provided by IBM, were explored for calibration, a computationally expensive component of typical radio astronomy processing pipelines requiring solving of linear equations’ sets.
The objective of this study was to display the effectiveness of these approaches in realizing satisfying results when they are integrated into calibration pipelines to minimize the computational cost. Although the Harrow-Hassid-Im-Lloyd (HHL) method is one of the promising quantum algorithms that offer a substantial speedup compared to classical methods, this method has several limitations, including hardware limitations and boundary conditions on input data.
Thus, a variational approach, designated as VQLS, was investigated as this approach is more suitable for current hardware. Variational quantum algorithms have gained significant attention as an effective approach for harnessing quantum computing power in the NISQ era. Newer variations of the original VQLS method have also been proposed to mitigate its limitations. Many studies have exploited this method to solve finite-element problems.
QAs serve as a good alternative to gate-based quantum computers that have been evaluated extensively for real-world applications. They are a specialized hardware type that is very effective for solving minimization problems and has been utilized to solve power grid management and scheduling problems.
They were also used in fundamental studies in structural biology and acoustics. These studies employed D-Wave QAs accessible through the cloud and containing over 2,000 qubits. In addition to binary problems, QAs are also suitable for solving linear systems and floating-point calculations.
Researchers integrated a variational and a combinatorial quantum linear solver (VQLS) and quadratic unconstrained binary optimization (QUBO) solvers into the redundancy calibration pipeline of the HERA telescope using the Hera linsolve package’s dedicated fork.
The robust integration of quantum solvers within this software suite enabled seamless switching between quantum and classical resources for calibration. Researchers performed experiments in both ideal and realistic settings. In the ideal case, the quantum computers had no restrictions on noise, coherence time, and qubit connectivity.
They compared the results obtained using the quantum and classical solvers based on accuracy. Additionally, the VQLS solver employed a full correlation between qubits and a real-amplitude variational ansatz with three repetitions, while 11 qubits were used for the QUBO solver to encode the solution vector’s every floating-point number. However, the existing significant limitations of current quantum computers were considered for their evaluation in realistic settings. For instance, QAs like the D-Wave chips possess limited connectivity graphs for every qubit.
Significance of the Study
The results demonstrated that quantum linear solvers could be a viable tool for obtaining the initial estimation of antennas’ gains in the ideal case where various quantum hardware were not limited by the coherence time or qubit connectivity. However, the existing limitations on coherence time and qubit connectivity significantly hampered the performance of the evaluated quantum linear solvers in realistic settings.
Although the variational method implemented on gate-based quantum computers required a small number of qubits for large arrays, this approach needed an extremely efficient matrix decomposition scheme to compete with classical approaches. The computational cost was prohibitive without such a decomposition scheme.
Similarly, the combinatorial approach that relied on QAs yielded very accurate results but required a substantial number of physical qubits to address the limited-scale inter-qubit connectivity on real devices. Thus, existing QAs could accommodate only small antenna arrays where classical methods are sufficiently fast.
Overall, no concrete quantum advantage could be identified for radio astronomy calibration with the currently available hardware due to the limitations of the existing state-of-the-art quantum computers. Thus, more research is required to develop new quantum solvers with improved performance and new hardware with less demanding limitations to realize computational advantages.
Journal Reference
Renaud, N., Rodríguez-Sánchez, P., Hidding, J., Broekema, P. C. (2024). Quantum radio astronomy: Quantum linear solvers for redundant baseline calibration. Astronomy and Computing, 47, 100803. https://doi.org/10.1016/j.ascom.2024.100803, https://www.sciencedirect.com/science/article/pii/S2213133724000180
Wanda Parisien is a computing expert who navigates the vast landscape of hardware and software. With a focus on computer technology, software development, and industry trends, Wanda delivers informative content, tutorials, and analyses to keep readers updated on the latest in the world of computing.
