arXiv:2610.05005
Arbitrary-phase CZ gate and remote readout across spatially separated superconducting nodes
Anatoly Kulikov, Josua D. Schär, Alonso Hernández-Antón, Aleksandr Grigorev
Abstract
Establishing a complete set of gates across spatially separated quantum nodes is fundamental for distributed quantum computation and communication networks. One of the approaches to do so consists of employing a superconducting device as a universal quantum node that is capable of storing and processing quantum information, and traveling microwave photons as means of communication. In this work, we realize an entangling gate across a primitive two-node superconducting circuit QED network consisting of two dilution cryostats, A and B, connected via a 30-m-long cryogenic quantum link. We perform a bidirectional photon transfer protocol, where a microwave photon emitted by the communication qubit at A acquires a conditional phase based on the state of qubit B. We vary the communication photon frequency, hence tuning the conditional phase acquired during the reflection off the remote node, effectively implementing an arbitrary-phase CZ gate. We use this entangling operation to perform single-shot readout of the qubit in node B from node A. An interesting feature of our scheme is that it does not require precise synchronization of the active node with the remote node; in particular, remote readout can be performed without any active participation by the remote node. These protocols add to the quantum communication toolbox with superconducting circuits and provide a way of realizing basic distributed quantum computing operations.
AI Summary
The paper investigates whether itinerant microwave photons can mediate a deterministic entangling operation and remote readout between superconducting nodes without synchronized control at both ends. The experiment connects transmons in separate dilution refrigerators through a 30 m superconducting waveguide. A shaped photon emitted at node A reflects from a dispersively coupled resonator at node B, acquiring a qubit-state-dependent phase, and returns for reabsorption at A. Node B requires no active participation during either operation. An amplitude- and frequency-calibrated Raman emission protocol compensates drive-induced AC Stark shifts and tunes the photon frequency across approximately 20 MHz. Ramsey measurements establish conditional-phase tunability over $[\pi/2,0.95\pi]$, rather than a full arbitrary-phase range. Two-qubit tomography after applying the operation to $|++\rangle$ yields fidelities of 67–71% relative to ideal states with the corresponding conditional phases; these are output-state fidelities, not process fidelities. Round-trip photon recovery reaches 67% in the initial characterization and remains above 57% across the frequency-tuning measurements. Using the conditional phase for remote readout gives 82.8% single-shot assignment fidelity. Repetition improves this to 88.2% with five-shot majority voting and 89.4% with seven-shot weighted decoding. Remote lifetime and Ramsey measurements demonstrate integration as a measurement primitive. The stated dominant limitation is photon loss in connections between chips and waveguide, rather than propagation through the waveguide. Additional errors include transmon relaxation, approximate photon-shaping calibration, scattering-induced wavepacket distortion, and qubit-state-dependent return delays. Remote-qubit relaxation limits gains from repeated readout; state-dependent delays intrinsically constrain absorption away from the equal-delay operating point.
Error found: The experimentally demonstrated conditional-phase range is approximately [π/2, 0.95π], rather than unrestricted arbitrary-phase coverage or an exact π CZ. The tomography caption assigns phases increasing with photon detuning, contradicting the decreasing trend in the Ramsey calibration with which those phases purportedly agree. The reported 67–71% fidelities concern output states for the single input |++⟩, not experimentally established gate-process fidelities. The stated node-B parameters predict a maximum phase contrast of approximately 0.775π in the supplied single-pole model, and the proposed explanation for the measured 0.95π is not quantitatively evaluated.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.