arXiv:2610.09749
Automated reduction of fault-tolerant circuits
Hyeongjun Jeon, Jeonghoon Lee, Taehyun Kim
Abstract
We present an automated method for reducing fault-tolerant circuits through fault-equivalent rewrites. Starting from a known fault-tolerant circuit, the search applies non-reducing enabling rules to expose Bell-pair reductions, each of which removes an ancilla preparation and a CNOT gate. Because every search transition preserves fault equivalence, the resulting circuits inherit the fault-tolerance properties of the input circuit. Candidate circuits are evaluated using circuit-level Monte Carlo simulation. For Shor-style syndrome extraction with the $[[7,1,3]]$ code, our method reduces one syndrome-measurement round from 30 to 18 ancilla preparations and from 54 to 42 CNOT gates. At a physical two-qubit error rate of $p = 10^{-3}$, the optimized circuit lowers the logical error rate by approximately 21% for both logical basis states. The reduction ranges from 13% to 23% over a range of $p$ spanning two orders of magnitude. We also apply the method to Steane-based dynamic syndrome extraction constructed from Goto's verified logical-$\lvert 0 \rangle$ preparation, in which the verification qubit serves as a flag. The search selects a circuit using four ancillas and 14 CNOT gates, matching the resource counts of the previous design but with an improved CNOT depth. Under depolarizing idle noise at $3p/10$ the selected circuit lowers the logical error rate by approximately 15% for both logical basis states, and by approximately 10% at $p/10$. These results demonstrate that automated fault-equivalent rewriting can identify circuits with lower resource costs or improved logical performance without separately verifying the fault tolerance of every candidate.
AI Summary
Can fault-tolerant circuits be optimized automatically without rechecking fault tolerance for every candidate? This paper searches over circuit rewrites derived from an existing fault-equivalent ZX-calculus framework. Restricted commutation, basis swaps and target swaps expose Bell-pair reductions, each removing one ancilla preparation and one CNOT. Composing these transformations preserves fault equivalence and hence the input circuit’s fault-tolerance properties under the specified noise model. Search transitions combine enabling rewrites with a reduction, so resource counts decrease along every path. Two case studies use syndrome extraction for the $[[7,1,3]]$ Steane code. For Shor-style extraction, a full measurement round decreases from 30 to 18 ancilla preparations and from 54 to 42 CNOTs, while per-gadget CNOT depth increases from five to six. Circuit-level Monte Carlo simulations with flag-conditioned lookup-table decoding report approximately $21\%$ lower logical error at $p=10^{-3}$ and $13$–$23\%$ reductions over the scanned error rates; near-quadratic scaling supports suppression of single-fault logical failures. For Steane-based dynamic extraction, 391 candidates are generated. The selected circuit matches the previous design’s four ancillas and 14 CNOTs but reduces CNOT depth from eight to seven. Simulated logical error decreases by approximately $10\%$ or $15\%$, depending on depolarizing idle-noise strength. The search is restricted to selected rewrites and short enabling sequences, does not deduplicate circuits, and establishes no global optimality. Monte Carlo evaluation is expensive and statistically noisy. Higher-distance codes, other gadgets, hardware connectivity constraints and improved search objectives remain future directions.
Central claim may not hold: The Bell-reduction constraints permit measurement branches with identical retained parity but different surviving quantum states, so replacing the eliminated measurement by its outcome-zero effect does not establish the unconditional equivalence needed for the claimed fault-tolerance guarantee.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.