arXiv:2610.16401
Magic-Free Certification of Logical Error Rates in Universal Fault-Tolerant Circuits
Gebrat Porta-San Miguel, Shahram Panahiyan, Dieter Jaksch
Abstract
As quantum hardware enters the fault-tolerant regime, algorithms are beginning to be executed on encoded qubits, yet scalable certification of how well they run remains an open problem. Here we introduce magic-free certification, a certification protocol for universal fault-tolerant circuits that consumes no magic, the resource that generally places them beyond efficient classical simulation and is by far the most expensive element of a fault-tolerant computation. The protocol is code- and decoder-agnostic, runs interleaved with the target circuit on the same device, and has a sample complexity that grows linearly with the number of logical qubits and non-Clifford gates while remaining decoupled from the logical error rate itself. Given a predefined error tolerance $p_L^*$, the protocol determines at a chosen confidence level whether the device executes the target circuit at a logical error rate below $p_L^*$. The method relies on trap circuits, copies of the target with every magic state replaced by a stabilizer state, whose outcomes are deterministic and which, being indistinguishable from the target to stochastic Pauli noise, fail at the same rate. We illustrate the protocol in Steane-code simulations of a magic state injection gadget and a quantum Fourier transform.
AI Summary
The paper addresses certification of universal fault-tolerant circuits without classically computing their ideal outputs or consuming additional magic states. It constructs stabilizer-only trap circuits by replacing injected magic states with stabilizer states and sampling the injection corrections. Each trap prepares an input giving a deterministic logical Pauli measurement, so a flipped outcome directly detects an error. For $k_D$ data qubits and $k_T$ injections, the construction uses $3K$ measurement configurations, where $K=k_D+k_T$ counts injection ancillas even when hardware is reused. Pauli propagation, the fact that each nonidentity single-qubit Pauli anticommutes with two measurement axes, and a union bound give $p_L\leq\bar p_L=\frac12\sum_{i=1}^{3K}p(\mathcal T_i)$. A hypothesis test on pooled trap failures certifies a prescribed tolerance $p_L^*$, with shot scaling $O(K/p_L^*)$ for fixed confidence and acceptance margin. The budget is independent of the actual logical error rate, but diverges quadratically as the margin closes. Steane-code simulations illustrate a single injection gadget and a two-data-qubit QFT with three injections. For the latter, tolerance $10^{-5}$, 5% false-positive probability, and 90% acceptance probability at half the tolerance require approximately $1.9\times10^7$ shots. The guarantees assume Markovian stochastic Pauli noise, basis-independent preparation noise, a fixed decoder, and independently characterized magic states no noisier than their trap substitutes. Correlated executions can invalidate statistical confidence; Poisson calibration is approximate. The bound overcounts correlated and output-harmless errors. Residual coherent noise requires an independently characterized additive correction. No hardware demonstration is reported.
Central claim may not hold: The abstract claims equal target and trap failure rates, whereas the body establishes only an aggregate upper bound that can overcount errors and include errors harmless to the target. The advertised certificate requires assumptions beyond stochastic Pauli noise and excludes independent certification of the magic-state supply. The central theorem is false under A2 as written because a larger stabilizer-resource preparation error rate can cancel subsequent faults in the traps without canceling them in the target.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 21 pages, 3 figures