arXiv:2611.04075
Runtime Compilation for Quantum Circuits with PQF-Enabled $R_z$ Synthesis
Siddharth Dangwal, Christopher Kang, Jad Habhab, Ritvik Sharma, Frederic T. Chong, Gokul Subramanian Ravi
Abstract
In Fault-Tolerant Quantum Computing (FTQC) using surface codes, reducing $T$-count is critical for efficient computation due to the high cost of $T$-state distillation. Probabilistic Quantum Circuit with Fallback (PQF) synthesis significantly lowers $T$-counts compared to standard protocols like Gridsynth, but introduces high circuit dynamism through extra gates inserted at runtime. This dynamism invalidates static compile-time routing, as these new $T$ gates require dynamic lattice surgery routing to establish non-intersecting ancilla paths to $T$-state factories while minimizing total execution time. To address this, we introduce $\text{R}^2$-PQF, a Runtime Routing protocol for PQF-synthesized quantum circuits. $\text{R}^2$-PQF dynamically routes gates layer-by-layer, computing paths during the execution of the previous logical timestep. To satisfy the strict computational window of $\sim t_{lattice\_surgery} + t_{reaction}$ and non-intersectionality constraints, we implement a Wavefront Breadth-First Search (W-BFS) and synthesise it on a 500MHz FPGA and 2 GHz ASIC. For a clock cycle of 2GHz, we achieve a median logical time-step of $1.51 μs$ for a 300 qubit Ising Hamiltonian circuit, demonstrating that $\text{R}^2$-PQF successfully resolves the vast majority of paths (with rare exceptions) within the execution window and enables the $T$-count benefits of PQF without latency penalties. Compared to a baseline stalling-based solution (the only no-runtime-recomputation solution), $\text{R}^2$-PQF speeds up circuit execution by $1.34\times$ on average and up to $1.76 \times$.
AI Summary
Probabilistic quantum circuits with fallback (PQF) reduce rotation-synthesis $T$ counts but require corrective gate sequences when measurements fail. These insertions can invalidate precomputed lattice-surgery routes. The paper asks whether routing can be recomputed during fault-tolerant execution without losing PQF’s resource benefits to scheduling delays. $R^2$-PQF incrementally reschedules dependency-ready gates, including inserted corrections, using bidirectional wavefront breadth-first search, heuristic route arbitration, and a retry stage. It computes the next layer’s routes while the current layer executes, with an initial stall after PQF failure. Evaluation combines circuit simulation for Hamiltonian simulation, VQE, QFT, and phase estimation on three layouts with FPGA and ASIC synthesis. Across benchmarks with 50–450 qubits at $p_{succ}=0.9$ and $\epsilon=10^{-13}$, reported schedule speedups average $1.34\times$ over serialized fallback execution and $1.25\times$ over a baseline that checks precomputed routes for conflicts; maxima are $1.76\times$ and $1.59\times$. Compared with statically compiled Gridsynth circuits, PQF circuits average $1.68\times$ faster. At the synthesized ASIC frequency of 2 GHz, median routing latency is $1.51\,\mu\mathrm{s}$ for a 300-qubit Ising circuit; at most $0.1\%$ of evaluated steps exceed the assumed $20\,\mu\mathrm{s}$ budget. Results depend on workload, layout, and timing assumptions. The 250 MHz FPGA exceeds that budget more often, requiring stalls. Repeated stochastic trials cover only circuits of at most 100 qubits, global-recompilation comparisons are limited by computational cost, and larger-scale timing estimates are extrapolations. Hardware results are synthesis-based rather than demonstrations integrated with a quantum processor.
Error found: The abstract's 1.51 μs 'median logical time-step' is actually the ASIC's classical routing latency, not the modeled quantum timestep of 20 μs at code distance 10. The abstract reports a 500 MHz FPGA implementation, but the methodology, hardware tables, and evaluated latencies use 250 MHz. The claim of operation 'without latency penalties' overlooks explicitly acknowledged post-failure stalls and occasional routing-deadline overruns, although the reported speedups include the post-failure stalls.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.