Q-Pilot: Field Programmable Qubit Array Compilation with Flying Ancillas

Neutral atom arrays have become a promising platform for quantum computing, especially the field programmable qubit array (FPQA) endowed with the unique capability of atom movement. This feature allows dynamic alterations in qubit connectivity during runtime, which can reduce the cost of executing long-range gates and improve parallelism. However, this added flexibility introduces new challenges in circuit compilation. Inspired by the placement and routing strategies for FPGAs, we propose to map all data qubits to fixed atoms while utilizing movable atoms to route for 2-qubit gates between data qubits. Coined flying ancillas, these mobile atoms function as ancilla qubits, dynamically generated and recycled during execution. We present Q-Pilot, a scalable compiler for FPQA employing flying ancillas to maximize circuit parallelism. For two important quantum applications, quantum simulation and the Quantum Approximate Optimization Algorithm (QAOA), we devise domainspecific routing strategies. In comparison to alternative technologies such as superconducting devices or fixed atom arrays, Q-Pilot effectively harnesses the flexibility of FPQA, achieving reductions of 1.4×, 27.7×, and 6.3× in circuit depth for 100-qubit random, quantum simulation, and QAOA circuits, respectively.

Quantum computing (QC) hardware has seen rapid scaling, with superconducting systems offering up to 433 qubits [1-4], and neutral atom arrays reaching 1000+ qubits [5,50]. Utilizing these machines requires mapping qubits in a quantum program/circuit to physical qubits on the QPU, typically constrained by limited connectivity given by a coupling graph. For example, Fig. 1a illustrates a simple QPU with four physical qubits connected in a ring. 2-Q entangling gates, crucial for quantum programs, are restricted to adjacent physical qubits (e.g., (𝑝 0 , 𝑝 1 )). Consider a quantum program with gates CZ(𝑞 0 , 𝑞 1 ), CZ(𝑞 1 , 𝑞 2 ), and CZ(𝑞 2 , 𝑞 0 ). In Fig. 1b, the initial qubit mapping is 𝑞 𝑖 ↦ → 𝑝 𝑖 for 𝑖 = 0, 1, 2. While this mapping supports the first two gates, CZ(𝑞 2 , 𝑞 0 ) involves non-adjacent 𝑝 2 and 𝑝 0 . Here, a SWAP gate is inserted to route qubits, transforming the mapping. However, SWAP is costly: it can increase circuit depth, leading to more decoherence noise, and typically requires three 2-Q entangling gates, accumulating gate errors. Given the current QPUs' relatively high noise levels, as quantum circuits grow, it becomes crucial that compilers minimize the overheads incurred by mapping and routing to optimize performance [7