In the lab, Guizhen Chip demonstrates a sixteen-qubit state
Inside a custom programmable photonic chip, a team from Hefei-based Guizhen Chip Technology and Professor Ren Xifeng’s group at the University of Science and Technology of China assembled a four-photon sixteen-qubit GHZ state, a shared entangled state, alongside a four-qubit cluster state. Ten of the sixteen qubits were verified as genuinely entangled. The work is reported as the largest on-chip multi-qubit entangled state to date.
The obstacle is scale. Photonic quantum computers need many entangled photons, but multi-photon sources scale exponentially poorly, while deterministic two-photon gates are difficult to build. The team instead uses MBQC — measurement-based quantum computing — where computation is driven by single-qubit measurements on an entangled resource prepared in advance.
The chip gives each photon several qubits through its path, then routes those paths through a four-layer programmable measurement module. In effect, the architecture trades a hard multi-photon preparation problem for a high-dimensional single-photon design and layered measurement, keeping the optical hardware simpler while making resource-state preparation more deterministic.
The team also used its four-qubit cluster state to run Grover’s search algorithm, reporting an average identification probability of 0.987. The result is published as an arXiv preprint, so it is a laboratory milestone rather than a deployed quantum computer or a finished million-qubit system.
So what changes, concretely? Guizhen Chip has shown an engineering route for putting larger entangled resource states directly on a photonic chip, a building block for measurement-based and fusion-based quantum computing. The next test is harder: a measured demonstration of fault-tolerant logical qubits under realistic noise. Until then, the result marks a building block toward scalable optical quantum hardware, not proof that million-qubit machines are ready.
Comments
Loading the thread…
Sign in to leave a comment. Sign in