SeQUeNCe.DQC adds full-stack simulation to distributed quantum computing research
A research paper posted to arXiv introduces SeQUeNCe.DQC, a new top-level module that simulates distributed quantum computing from circuit compilation through network execution. The module extends the SeQUeNCe package to compile monolithic circuits into topology-aware execution plans and coordinate their execution using existing quantum-network modules and protocols. It validates the approach by reproducing a two-node trapped-ion Grover experiment in simulation and comparing success probabilities with a reported 71% benchmark.
Distributed quantum computing connects multiple quantum processing units to overcome the scale limits of monolithic hardware. Building and testing physical testbeds is costly and difficult, making software simulation a vital approach for designing, evaluating, and optimizing architectures, algorithms, and protocols. The SeQUeNCe package is an open-source discrete-event quantum network simulator that already models photon transmission, heralding, entanglement swapping, and memory dynamics.
The paper’s contributions include the design and implementation of SeQUeNCe.DQC for full-stack distributed quantum computing simulation in SeQUeNCe. It validates the module by reproducing a two-node trapped-ion Grover experiment in simulation and comparing success probabilities with the reported 71% benchmark. It evaluates how compiler strategy, topology, and memory capacity jointly govern distributed circuit execution time and success rate. In these experiments, hybrid compilers successfully identified mixed TeleGate/TeleData execution plans that outperformed pure gate-teleportation strategies. By capturing the physical overhead of distributed quantum primitives, SeQUeNCe.DQC provides a rigorous platform for co-designing distributed quantum computing software and hardware architectures.
SeQUeNCe.DQC consists of a compiler, a controller, worker programs, and the existing SeQUeNCe quantum-network modules. The compiler takes a monolithic quantum circuit description and produces a topology-aware execution plan that assigns subcircuits to worker programs simulating quantum processing units based on the network topology of the simulated system. For each remote operation, the plan specifies whether to use TeleGate or TeleData primitives. The controller reads this plan and orchestrates the worker programs.
The design decision that enables the reported outcomes is the integration of network-aware compilation with protocol-level simulation within the SeQUeNCe framework. Earlier tools such as Interlin-q and SENDQ operated at the circuit level using simplified network abstractions that did not model noise or the probabilistic nature of entanglement generation. By contrast, SeQUeNCe.DQC models the repeated attempts required to establish remote entanglement, the classical signaling overhead, and the decoherence that occurs while qubits wait for entanglement. This fidelity allows the system to reveal trade-offs invisible in abstract models: a compilation that reduces the number of remote operations may still result in longer execution if it increases contention for communication memories, leading to more entanglement attempts and greater decoherence.
The hybrid compiler’s mixed TeleGate/TeleData plans succeed because they balance the cost of transmitting quantum gates against the cost of transmitting quantum data. Pure gate-teleportation strategies require a fresh high-fidelity entangled pair for each gate, incurring many entanglement attempts and classical messages. Mixed plans reuse established entangled pairs to teleport data blocks while teleporting only a subset of gates, reducing the number of entanglement cycles and the exposure of qubits to decoherence. The simulation quantifies these effects by measuring both execution time and success probability under varying topologies and memory capacities, showing that the mixed approach yields higher success rates in the evaluated scenarios.
The trade-off inherent in this design is computational cost. Discrete-event simulation of photon transmission, heralding, swapping, and memory decay for every remote operation generates a large number of events, especially as the number of quantum processing units and circuit depth increase. Consequently, scaling the simulator to larger networks or deeper circuits demands more processing power and memory, which may limit the speed of design iterations for very large distributed quantum computing systems.
The paper leaves open how the identified hybrid compilation strategies will perform when entanglement generation rates, memory lifetimes, and network contention differ from the simulated parameters. Determining whether the same advantage of mixed TeleGate/TeleData strategies persists under alternative physical constraints remains an unresolved question.