Quantum Gate Teleportation For Scalable Quantum Computing
Gate teleportation eliminates the 10-fold overhead of circuit breaking, enabling scalable quantum computing.
Despite high error rates in monolithic systems and limited qubit counts, distributed quantum computation can overcome the limitations of conventional quantum processors. DQC could lead to real quantum computation by integrating smaller quantum processing units (QPUs) into a larger, scalable system.
This goal requires effective interconnects and communication protocols, so researchers are comparing circuit cutting (classical links) and entanglement-based gate teleportation (quantum links) for non-local quantum operations. Hardware developments may make gate teleportation the best method for producing sophisticated multipartite entangled states, replacing circuit cutting.
Scalability through Modular Quantum Computing
A single, massive quantum computer that can solve complex problems is hindered by signal routing, fabrication yield, cryostat size for superconducting qubits, and trapped ion heating.
These issues are solved by modular quantum computing's networked QPUs for sophisticated calculations. Connectivity requires microwave-to-optical (M2O) transducers and photonic interconnects to convey quantum information between modules.
Two basic strategies have been studied to simplify module computation. First, circuit cutting breaks a large quantum circuit into smaller subcircuits that run on individual QPUs. Next, extensive standard post-processing merges the results.
Gate teleportation uses quantum links to remotely execute gates across qubits in different modules utilizing shared entangled pairings, or Bell pairs. Oxford physicists have recently teleported quantum gates like the controlled-Z gate between physically distant trapped-ion modules, allowing entire algorithms like Grover's search to be completed.
Remote Gates vs. Circuit Cuts
Researchers created Greenberger-Horne-Zeilinger (GHZ) state primitives of multipartite entangled states spread across processor modules to mimic remote gate teleportation and circuit cutting. For comparison, Hellinger fidelity, a performance parameter that measures ideal and simulated probability distribution similarity, was used.
The main overhead for circuit cutting is sampling fractured subcircuits. It's important to note that the number of CNOT gate-cuts needed to reproduce the original circuit increases exponentially, lowering circuit cutting quality. It causes exponential sampling and standard post-processing costs.
Remote gate teleportation, which uses TeleGate, does not always have exponential sampling overhead. The quantum interconnect's precision, specifically the noise from M2O transducers used to entangle superconducting qubits via optical lines, severely restricts its performance.
Finding the Break-Even Point
A comparative simulation identified regimes where each technique performed well. Remote gate performance degrades drastically when transducer noise contribution exceeds, deteriorating with larger GHZ circuit sizes. In the low domain, local two-qubit gate errors (set as 0.98 in the model) reduce remote gate fidelity more than transducer noise.
A ten-fold decrease in M2O transducer noise added figures is needed for gate teleportation to overcome circuit cutting. Lowering this noise threshold to (0.01, 0.1) would improve remote gates.
This increase is significant for generating sophisticated multipartite entangled situations. As GHZ circuit size increases, circuit cutting's exponential sampling overhead reduces fidelity for a fixed shot budget.
As the circuit size increases, remote gates' noise threshold relaxes, making them more advantageous than circuit cutting for generating these intricate entangled states.
Network-Aware Hybrid Strategy
The hardware metrics motivate a network-aware hybrid quantum-classical processing technique and near-term quantum interconnects. This method uses sparse quantum links and cutting techniques to reduce quantum runtime while maximizing quantum link and classical circuit cut benefits.
A greedy algorithm dynamically chooses distant gates or gate cuts based on Bell pairs and fidelity comparison. If a Bell pair is available and a quantum connection has greater fidelity, the distant gate is used; otherwise, a gate cut is made, which may increase the shot budget for that link.
This research will improve by comparing remote gates to circuit cutting for methods used in variational quantum circuits (VQC), quantum chemistry, and quantum machine learning. By co-optimizing transducer hardware metrics, such as conversion efficiency, noise, and cutting-edge protocols, researchers expect to develop scalable distributed quantum computation.
