Reading four qubits in two photons

Two photons, four qubits

The architectural choice in this experiment is to put four qubits into the time bins of two photons rather than distribute them across four separate photons. Each photon carries two qubits. A phase applied to the final pump pulse establishes the required cluster state during generation. This avoids adding a lossy controlled-phase gate after the quantum photons have been produced. The computational resource comes from denser use of the same particle’s temporal structure. The hardware economy starts here: the number of physical carriers need not equal the number of logical units they carry.^[1]

Information density does not itself protect against loss. Losing one photon removes both qubits carried by it. The same arrangement reduces carrier count while increasing the logical cost of a lost carrier. My reading of the gain is therefore the operation of a small cluster state together with fiber transmission and configurable measurement hardware. It does not establish a fault-tolerant network. The two-node experiment leaves multi-node loss management unresolved. Preserving a logical state and keeping a network useful after losses are separate engineering requirements.^[1]

Phase control as a measurement component

At the centre of the measurement apparatus is an electro-optic phase modulator between oppositely chirped fiber Bragg gratings. The first grating spreads the photon’s spectral components in time; an electrical phase pattern changes that expanded structure; the second grating recompresses it. The resulting interference supplies a beam-splitting operation between time bins. This creates a configurable optical component for different qubit measurements. Electrical control determines which components the measurement compares without first converting all the photon’s information into a classical number.^[1]

The component depends on maintained timing. Temperature changes delay in deployed fiber, and active optical-delay feedback compensated that drift in the experiment. Producing an entangled state was insufficient: the arriving photon’s temporal structure had to match the structure expected by the measurement component. Delay stabilisation illustrates why treating source and detector as independent boxes misses a system requirement. The link carries a timing relationship that must survive between state preparation and measurement. Component losses and unwanted mixing between time bins remain constraints on that relationship.^[1]

Separating distance from topology

Of the approximately 29.5 kilometres of optical path, 4.5 kilometres was deployed fiber and 25 kilometres was a spool. Both processing nodes remained in the same laboratory. These details prevent a long optical path from being equated with processors operating in distant cities. Variability in the deployed fiber nevertheless imposed a real experimental burden. The entanglement witness crossed the separability boundary, and entanglement in the remaining pair was tested after selected qubits were measured. The result supports basic measurement-based processing operations across the link; it does not demonstrate universal computation.^[1]

A useful criterion for assessing this apparatus is the integration of state preparation, fiber passage and selectable measurement within one system. The same evidence defines its boundary: loss of a two-qubit photon, optical insertion losses and the two-node topology remain unresolved constraints. The apparatus supplies a concrete answer to how densely encoded information is read. It does not independently answer how losses are managed in a larger network. Component reconfigurability and system resilience therefore cannot be treated as the same performance measure.^[1]