Chinese scientists claim to have built the world’s fastest programmable quantum computers, which appear to crack problems that are currently not feasible for ”classical” non-quantum computers.
The researchers led by Pan Jianwei from the University of Science and Technology of China (USTC), said one of the quantum computing systems — Zuchongzhi 2.1 — is a million times more powerful than its nearest competitor, Google’s Sycamore.
Besides, their photonic quantum computer based on light — Jiuzhang 2 — can carry out calculations 100 trillion times faster than the world’s fastest existing supercomputer, the physicists noted in another study, published in the journal Physical Review Letters on Monday.
In conventional computers, the most basic unit of information is a bit, and data is fundamentally stored in binary codes of 1s and 0s. On the other hand, quantum computers make use of the special properties of the smallest particles in the universe which can exist in multiple states — as zeros and ones at the same time, or in any position between.
This flexibility of quantum particles allows for quantum bits, or qubits, using which many different calculations can be performed simultaneously, scientists said.
While there are many approaches to achieve quantum computing, the Chinese team has built two different systems — one is the light-based photonic quantum computers, and the other is a superconducting quantum computer that needs to be kept at very low temperatures to work efficiently.
In photonic quantum computers, light’s energy units, the photons, are manipulated with mirrors, beam splitters, and phase shifters, while in the latter the state of the qubits is manipulated using an electromagnetic field.
These manipulations execute operations on the photons similar to adding ones and zeroes in classical computers, and single-photon detectors help read what changes the photons have undergone.
The common theme in both these types of quantum computers is that they accept multiple quantum states as inputs, have the states travel through a circuit, and deliver multiple states as output.
For instance, in photonic quantum computers, the scientists said, single photons arrive as input in parallel to an optical circuit in which components like beam splitters cause the photons to interfere, causing their states to change, and they emerge from multiple output ports.
In experiments, the scientists used the two quantum computers to calculate the probability that a certain input configuration may lead to a particular output configuration.
Since these circuits have tens of inputs and outputs, such probability calculations, scientists say, are infeasible for classical computers.
But in both photonic and superconducting quantum computers, they say the quantum nature of these systems helps increase the number of parallel computations that can happen, making such probability calculations feasible.
While these machines are not expected to completely replace classical computers, they can carry out specific complex calculations for short periods of time.
Pan and his team showed that for their system which has 1043 possible outcomes, their photonic quantum computer can sample the output 1024 times faster than a classical supercomputer – an upgrade from the team’s December result of 1014 times faster operation.
The scientists also claim that their sampling calculation using their superconducting quantum computer is about 1,000 times more difficult to do on a classical computer.
“We estimate that the sampling task finished by Zuchongzhi in about 1.2 h [hours] will take the most powerful supercomputer at least 8 yr [years],” the scientists wrote in the study.
Barry Sanders, director of the Institute for Quantum Science and Technology at the University of Calgary in Canada, who was unrelated to the study, said in a linked commentary that the two experimental quantum computers “tackle the most complex problems yet”.
“This indicates that our research has entered its second stage to start realising fault-tolerating quantum computing and near-term applications such as quantum machine learning and quantum chemistry,” the study’s co-author Zhu Xiaobo told Chinese state media.