Scientists have made significant progress with quantum technologies that could transform the modeling of complex systems with an accurate and efficient approach requiring significantly reduced memory.
Complex systems play a vital role in our daily lives, whether predicting traffic patterns, weather forecasts, or understanding financial markets. However, to accurately predict these behaviors and make informed decisions requires storing and tracking vast information about events from the distant past – a process that presents enormous challenges.
Current models using artificial intelligence see their memory requirements more than a hundredfold every two years and can often involve optimization on billions, even trillions, of parameters. These immense amounts of information lead to a bottleneck where we have to trade off memory cost against predictive accuracy.
A collaborative team of researchers from the University of Manchester, University of Science and Technology of China (USTC), Center for Quantum Technologies (CQT), National University of Singapore and Nanyang Technological University (NTU) proposes that quantum technologies could provide a way to mitigate this trade-off.
The team successfully implemented quantum models capable of simulating a family of complex processes with a single qubit of memory – the basic unit of quantum information – offering dramatically reduced memory requirements.
Unlike classical models that rely on increasing memory capacity as new data from past events is added, these quantum models will only ever need one qubit of memory.
The development, published in the journal Nature Communications, represents a significant advance in the application of quantum technologies to modeling complex systems.
Dr Thomas Elliott, Project Leader and Dame Kathleen Ollerenshaw Fellow at the University of Manchester, said: “Many quantum advantage proposals focus on using quantum computers to calculate things faster. We take a complementary approach and instead look at how quantum computers can help us. reduce the size of the memory we need for our calculations.
“One of the benefits of this approach is that by using as few qubits as possible for memory, we get closer to what is practical with near-future quantum technologies. Additionally, we can use all the extra qubits that we release to help mitigate errors in our quantum simulators.”
The project builds on an earlier theoretical proposal by Dr Elliott and the Singapore team. To test the feasibility of the approach, they partnered with the USTC, which used a photon-based quantum simulator to implement the proposed quantum models.
The team achieved greater accuracy than is possible with any conventional simulator equipped with the same amount of memory. The approach can be adapted to simulate other complex processes with different behaviors.
Dr. Wu Kang-Da, postdoctoral researcher at USTC and first co-author of the research, said: “Quantum photonics represents one of the least error-prone architectures that have been proposed for quantum computing, especially on a smaller scale. Additionally, because we configure our quantum simulator to model a particular process, we are able to fine-tune our optical components and achieve smaller errors than are typical of today’s universal quantum computers.”
Dr. Chengran Yang, researcher at CQT and also co-first author of the research, added: “This is the first realization of a stochastic quantum simulator where the propagation of information through memory over the time is conclusively demonstrated, as well as evidence of greater accuracy than possible with any conventional simulator of the same memory size.”
Beyond the immediate results, the scientists say the research presents opportunities for further investigation, such as exploring the benefits of reduced heat dissipation in quantum modeling compared to classical models. Their work could also find potential applications in financial modeling, signal analysis and quantum enhanced neural networks.
Next steps include plans to explore these connections and to adapt their work to higher-dimensional quantum memories.
