We’re at an exhilarating point in the Silicon Age, where computers seem to have the answers for everything from the mysteries of the human genome to starting colonies on faraway planets. However, the evolution of computing hardware still lags behind the evolution of our data needs; while researchers have crafted equations that could answer more complex questions, today’s computers may struggle to process them.
For computer scientists, quantum computers may be the answer. Where classical computing uses a binary system of ones and zeroes to encode information in units called bits, quantum computing uses subatomic particles such as electrons or photons, which can exist in multiple states at the same time. The result? Computing power in quantum units, or qubits, that (at least in theory) can effortlessly solve large-scale calculations that would take classical computers millions of years.
Currently available quantum computers are still difficult to use for real-world engineering problems such as simulating electromagnetism, explained Wei-Bin Ewe, a Senior Researcher from A*STAR’s Institute of High Performance Computing (IHPC). For one, slight environmental changes can affect the hardware, causing ‘noise’ or calculation errors; for another, today’s most advanced machines still only contain a relatively low number of qubits.
One solution may be cloud-based variational quantum algorithms (VQAs); hybrid techniques that pair up classical and quantum computers, splitting the computing work to either machine based on what either does best.
To investigate their potential, Ewe and his team designed a VQA based on a quantum linear system algorithm (QLSA) to simulate electromagnetic waves travelling through a metallic waveguide. Such wave problems are regularly tackled by engineers as an integral part of designing microwave circuits—key components in telecommunications devices ranging from radars to radios.
“QLSAs can solve linear equations faster and represent N data using log(N) qubits, both of which speed up the solving of large electromagnetic problems,” commented Ewe.
In their study, the team first broke down a waveguide into grids of tiny cells before they computed the possible frequencies of electric and magnetic waves across all cells with VQAs. To reduce the amount of quantum processing power required, the grid was decomposed further into smaller tables.
With this new approach, the researchers were able to reproduce results derived from hybrid quantum-classical algorithms more efficiently, with minimal quantum resources required. Ewe said that this demonstrates the almost limitless potential VQAs could have in the industry. “It is the first variational quantum algorithm to solve such problems and could be adapted to other problems in engineering,” explained Ewe.
However, technical barriers remain to its commercial applications. The team is working on noise mitigation schemes to enhance VQA accuracy while searching for other problems in electronics and photonics that could benefit from VQAs. "I hope that the ideas from our work will inspire more research in the use of VQAs to solve real-world problems,” Ewe said.
The A*STAR-affiliated researchers contributing to this research are from the Institute of High Performance Computing (IHPC).