Hefei and USTC demo a 16 qubit on chip photonic quantum computer using measurement based architecture

[NoMercyElliot54]

Hefei Guizhen Chip Technology and USTC have demonstrated a 16 qubit on chip photonic measurement based quantum computing system that uses high dimensional path encoding rather than the more common approach of adding physical qubits one at a time

The system routes four photons through 16 waveguide paths and reportedly achieved a 98.7 percent success rate running Grover's search algorithm, which is a solid result for a photonic platform even accounting for the fact that Grover's algorithm on a small qubit count is a fairly standard benchmark rather than a demonstration of real world advantage

Measurement based quantum computing is architecturally distinct from the gate based approach most people are familiar with from IBM and Google's superconducting efforts, instead of applying sequential gates you prepare a large entangled resource state and then perform measurements that effectively steer the computation, which can be a more natural fit for photonic hardware

Using path encoding to get 16 effective qubits out of just four physical photons is the clever part here, it sidesteps some of the multi photon entanglement generation problems that have historically limited photonic quantum computing scale, though the tradeoff is that photon loss becomes an even bigger concern since losing a single photon can wipe out more encoded information than in a simpler encoding scheme

The result is still pending peer review, which matters a lot in a field where impressive sounding demonstrations sometimes do not fully hold up once independently verified, so I would treat the 98.7 percent figure as promising rather than confirmed until that process plays out

China's photonic quantum computing effort has been quietly building for years and this feels like a meaningful data point that it is staying competitive with the superconducting and trapped ion approaches getting most of the Western attention and funding

Newsquantumcomputingreport.com

[Daz]

Path encoding to multiply effective qubits out of fewer physical photons is a clever trick

[FinnHalliday]

Photon loss being the tradeoff makes sense, you are packing more information density per photon so losing one hurts more

[Gerrard98]

Photonic has real advantages for room temperature operation too which gate based superconducting just cannot match

[Sookie]

Fair, this is a proof of architecture demonstration more than a claim of practical advantage

[Indexer Tundra]

Agreed, Chinese quantum announcements especially tend to get hyped before the paper actually clears review

[Kayla82]

Probably does at some point, routing complexity is usually the limiting factor in these path encoding schemes

[PhotonBurst16]

Still no peer review on this which is worth flagging every time these announcements come out

[Joanne94]

98.7 percent success rate on Grover's is solid for this scale but worth remembering Grover's on 16 qubits is not exactly pushing the boundary of useful computation yet

[BinaryMonk95]

Good to see multiple national programs pushing different qubit modalities instead of everyone converging on one approach too early

[Jonathan_Repetto]

Curious how this scales past 16 qubits, does the waveguide routing complexity blow up nonlinearly

[Kev5]

Measurement based approaches are underrated compared to how much attention gate based superconducting gets in Western coverage