A team of researchers from various research institutes across China have recently demonstrated quantum supremacy thanks to a photonic quantum computer. A paper recently published by the journal Science describes the quantum computer as “Jiuzhang”.
As reported by LiveScience, the quantum computer, designed mainly by researchers at the University of Science and Technology, is reportedly substantially more powerful than the quantum computer designed by Google in 2019. In 2019, Google claimed that it had designed the first computer ever to achieve “quantum supremacy”, which references the use of quantum-based computers to outperform current, traditional supercomputers. Reportedly, Jiuzhang is around 10 billion times faster than the quantum computer designed by Google.
In the past few years, China has made massive investments in the area of quantum computing, funding the research at the nation’s National Laboratory for Quantum Information Sciences for approximately $10 billion dollars. In addition, China is currently one of the world leaders in quantum networking. Quantum networking makes use of quantum mechanics to encode data as it is transmitted over long distances.
Quantum computers take advantage of the unique properties of quantum particles in order to obtain better performance than traditional computers. Classical computers can only process data that exists in one of two different states. Bits in this binary system uses ones and zeroes to represent data, and it is inherently limited compared to quantum bits (qubits), which can exist in more than two states at the same time. This property enables quantum computers to handle more complex problems and process tasks far more quickly than even the best supercomputers today.
It has long been theorized that quantum computers could dramatically beat out modern computers, but producing a reliable quantum computer is an engineering challenge that is still ongoing. Quantum computers often need to be located in controlled environments that prevent fluctuations in temperature or other environmental variables that could throw off a quantum computer’s calculations. Research groups around the globe have experimented with different ways of building quantum computers. While Google’s won quantum computer relied on superconducting materials integrated with chips, Jiuzhang relies on optical circuits.
In order to test Jizhang, the research team had it calculate the output of a circuit that uses light and returns a list of numbers. This process is known as Gaussian Boson Sampling. The goal was to detect as many photons as possible. Jiuzhang is an optical circuit itself and it managed to detect an average of 43 photons, hitting a record of 76 photons.
According to the paper published in Science, it took approximately 200 seconds to generate the list of numbers for every trial run of the quantum computer. Traditional supercomputers would take around 2.5 billion years to generate the same list of numbers. If the same rate of computation holds true for other tasks, quantum computers may be able to carry out computations around 100 trillion times more quickly than traditional supercomputers.
It’s important to note that Jiuzhang can only carry out the narrow range of tasks that it was developed for, those which center around Gaussian Boson Sampling. Jiuzhang is not a general quantum computer. However, it is a step towards the creation of practical quantum computers.
As TechXplore reported, the Jiuzhang computer isn’t the only recent example of advances in light-based computing technology to have potential impacts on artificial intelligence. A team of researchers has recently reviewed recent advances regarding the application of optical computing to visual-computing technologies, finding that optical computing platforms can potentially mesh with deep neural networks.
The research team studied several examples of optical computing alongside AI to find that AI inference based on light moving across optical devices could be used to create new forms of visual-computing technologies. These include optical neural networks that can quickly process and classify objects without the need for an external power supply, relying on incoming light to power the computations.
AI devices operating in systems like smart homes, remote sensors and autonomous vehicles could enhance the power of a regular electronic computer by using light to quickly analyze objects and the environment surrounding that object. Hybrid optical computer systems could leverage both the flexibility of traditional computers with the parallelism and speed of optical computers.