arXiv:2610.01772
Efficient quantum phase estimation with adaptive entanglement-assisted Hadamard test
Hengzhun Chen, Benchi Zhao, Yingzhou Li
Abstract
The entanglement-assisted Hadamard test (EHT) is a practical method for estimating a quantum phase by amplifying the phase signal. However, the feasible amplification is fundamentally limited by the accuracy of the reference phase, such that the method is inefficient in the high-precision regime. In this work, we propose an algorithm, called adaptive entanglement-assisted Hadamard test (AEHT), that iteratively refines the reference phase, enabling progressively stronger amplification as the iteration goes by. We further consider the imperfect eigenstate preparation scenario, where a systematic bias is unavoidable when estimating the quantum phase with the conventional EHT. Such a bias can be suppressed by the proposed AEHT. Moreover, taking physical implementation into consideration, we adopt device-restart count to measure the cost of quantum phase estimation, rather than shot count. The numerical experiments confirm the effectiveness of the proposed AEHT compared with conventional methods under this measure. By unlocking the full amplification power of large entangled states, this work offers an efficient method to estimate high-precision quantum phase on near-term quantum processors.
AI Summary
The paper studies phase estimation using GHZ-assisted Hadamard tests when limited reference-phase accuracy restricts usable amplification. An $m$-qubit GHZ ancilla and $m$ parallel eigenstate copies produce a parity signal $\sin[m(\theta-\vartheta)]$, with branch ambiguity requiring $m|\theta-\vartheta|<\pi/2$. The proposed adaptive protocol updates the reference after each round, contracts its certified uncertainty as $\Delta_t=\gamma^t\Delta_0$, and increases amplification approximately as $m_t=\chi/\Delta_t$, where $\chi<\pi/2$. Finite-sample guarantees follow from Hoeffding bounds and controlled inverse-sine inversion. For exact eigenstates and uncapped amplification, reaching accuracy $\epsilon$ takes $O(\log(1/\epsilon))$ rounds and shots up to doubly logarithmic factors. A hardware cost model packs $\lfloor Q/[m(n_{\rm sys}+1)]\rfloor$ independent shots into each device restart. Under this model, adaptive restart cost scales as $\widetilde O(1/\epsilon)$, versus $O(1/\epsilon^2)$ for fixed amplification. For imperfect preparation, the signal becomes $\rho^m\sin[m(\theta_\psi-\vartheta)]$. Given $\rho\ge\rho_0$, the amplitude-unaware estimator has bias bounded by $(1-\rho_0^m)\tan(m\Delta)/m$. Adaptive reference updates suppress this algorithmic bias, but not the difference between the effective phase $\theta_\psi$ and the target eigenphase. Numerical comparisons use a one-qubit phase unitary and a 2,500-qubit device model; noisy circuit sampling caps amplification at eight. Finite hardware caps restore quadratic precision scaling, with a constant-factor advantage over fixed amplification. The analysis assumes an initial phase-error bound and, for imperfect preparation, an amplitude lower bound. No hardware demonstration is reported.
Error found: For fixed imperfect-state amplitude lower bound 0<ρ₀<1, the protocol’s bias certificate imposes the precision-independent amplification ceiling m ≤ ln(1−c_bias)/ln(ρ₀), potentially preventing amplification from reaching the hardware limit. After hardware saturation, the stated equal per-round failure allocation introduces a growing log T factor in adaptive restart cost, invalidating the claimed asymptotically persistent constant-factor advantage over fixed-amplification EHT. The imperfect-state numerical discussion mistakes certified upper bounds on bias for unattainable accuracy floors, although its own SHT example has actual infinite-shot bias approximately 0.00273 rather than the bound approximately 0.00405.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 18 pages, 7 figures