Chinese team just pulled off something wild: they crammed 16 qubits into 4 photons on a single silicon chip using path encoding. Instead of treating each photon as 1 qubit, they're routing it through 16 waveguide paths, turning each photon into a 4-level qudit that carries 4 qubits worth of info.
The killer move here is dodging the exponential photon-generation bottleneck that's been strangling photonic quantum computing. Normally you need more photons for bigger entangled states, but coincidence rates tank hard as photon count climbs. Their solution: extract more data per photon instead of generating more photons.
They built this on standard silicon-on-insulator using Mach-Zehnder interferometers and thermo-optic phase shifters. No cryogenic cooling needed, runs at room temp, and it's compatible with existing fiber-optic infrastructure.
Results are legitimately impressive:
- Generated a 16-qubit GHZ state (largest verified entangled state on an optical chip)
- Verified genuine multipartite entanglement across 10 qubits
- Ran Grover's search on a 4-qubit cluster state with 98.7% accuracy (vs Stuttgart's previous 80.8% benchmark)
The architecture converts multi-photon state prep into single-photon operations, which is way more practical for scaling. Downside: losing one photon still nukes multiple qubits, so photon-loss management is still critical. But overall coincidence-rate demands drop massively compared to traditional multi-photon schemes.
This is measurement-based quantum computing (MBQC) done right. The high-dimensional encoding approach could be the blueprint for scaling photonic quantum processors without waiting for new hardware breakthroughs.
The killer move here is dodging the exponential photon-generation bottleneck that's been strangling photonic quantum computing. Normally you need more photons for bigger entangled states, but coincidence rates tank hard as photon count climbs. Their solution: extract more data per photon instead of generating more photons.
They built this on standard silicon-on-insulator using Mach-Zehnder interferometers and thermo-optic phase shifters. No cryogenic cooling needed, runs at room temp, and it's compatible with existing fiber-optic infrastructure.
Results are legitimately impressive:
- Generated a 16-qubit GHZ state (largest verified entangled state on an optical chip)
- Verified genuine multipartite entanglement across 10 qubits
- Ran Grover's search on a 4-qubit cluster state with 98.7% accuracy (vs Stuttgart's previous 80.8% benchmark)
The architecture converts multi-photon state prep into single-photon operations, which is way more practical for scaling. Downside: losing one photon still nukes multiple qubits, so photon-loss management is still critical. But overall coincidence-rate demands drop massively compared to traditional multi-photon schemes.
This is measurement-based quantum computing (MBQC) done right. The high-dimensional encoding approach could be the blueprint for scaling photonic quantum processors without waiting for new hardware breakthroughs.