Detuning-robust Rydberg entangling gates from echoed pulses

Technical Summary: Detuning-Robust Rydberg Entangling Gates from Echoed Pulses

Problem Statement
Rydberg-based two-qubit gates in neutral atom arrays are primarily limited by technical imperfections, specifically laser intensity inhomogeneity and detuning errors. These detuning errors arise from frequency miscalibration, background electric field drift, intermediate state light shifts, and Doppler shifts due to finite atomic temperature. While quantum optimal control can suppress intensity inhomogeneity, a fundamental "no-go" theorem (Ref. [29]) establishes that no pulse can render a standard Controlled-Z (CZ) gate first-order insensitive to atomic detunings. This is because the geometric phase accrued in each computational sector is strictly positive and shifts proportionally to the time spent in the Rydberg state, making it impossible to cancel detuning-induced phase errors within a single gate sequence.

Methodology
The authors propose a gate design that circumvents the no-go theorem by shifting the burden of detuning sensitivity from the entangling phase to single-qubit operations, which are then cancelled via an echoed sequence.

1. Target Operation: Instead of targeting a standard CZ gate, the design targets a $\sqrt{\text{CZ}}$ gate (a two-qubit rotation angle of $\pi/2$). The goal is to ensure that the entangling phase $\theta_{11} - \theta_{10} - \theta_{01}$ remains stationary (first-order insensitive) to detuning, while allowing the individual geometric phases ($\theta_{01}, \theta_{10}, \theta_{11}$) to remain first-order sensitive.
2. Pulse Design Constraints: Using quantum optimal control (GRAPE algorithm), the authors optimize pulse shapes to satisfy three specific conditions derived from perturbation theory:
   • Rydberg Leakage Suppression: Ensure the amplitude of population remaining in the Rydberg state $|r\rangle$ is zero for all computational sectors ($|01\rangle, |10\rangle, |11\rangle$).
   • Symmetric Detuning Robustness: For symmetric detuning ($\Delta_1 = \Delta_2$), constrain the pulse such that the derivative of the entangling phase with respect to detuning vanishes.
   • Anti-symmetric Detuning Robustness: For anti-symmetric detuning ($\Delta_1 = -\Delta_2$), suppress the leakage into the dark state $|W^-\rangle$ to ensure the $|11\rangle$ sector evolution is insensitive to detuning.
3. Echoed Sequence: The optimized $\sqrt{\text{CZ}}$ pulse is embedded in an echoed sequence: $\sqrt{\text{CZ}} - X^{\otimes} - \sqrt{\text{CZ}} - X^{\otimes}$. The interleaved global single-qubit $X$ rotations exchange the qubit states $|0\rangle$ and $|1\rangle$. This structure ensures that all four computational sectors undergo the entangling dynamics. Crucially, the single-qubit $Z$-rotations induced by detuning errors in the first half of the sequence are cancelled by the second half, resulting in a maximally-entangling $ZZ$ gate that is locally equivalent to a CZ gate but robust against arbitrary detunings.
4. Erasure Conversion: The authors note that residual errors are dominated by population outside the computational subspace (Rydberg population). They propose converting this population into heralded erasure errors using fluorescence or autoionization detection, a technique compatible with metastable $^{171}\text{Yb}$ qubits.

Key Results

• Anti-symmetric Detuning Robust (ADR) Gate: Optimized for $\Delta_1 = -\Delta_2$. This gate maintains an infidelity below $10^{-3}$ over a detuning range of $|\Delta|/\Omega \leq 0.1$, significantly outperforming time-optimal (TO) pulses which require $|\Delta|/\Omega \leq 0.01$ for the same fidelity.
• Detuning-Robust (DR) Gate: Optimized for arbitrary two-atom detunings ($\Delta_1, \Delta_2$). While the error-free infidelity is slightly higher ($\sim 10^{-4}$) due to the stricter constraints, the DR gate eliminates the leading-order (quadratic) dependence of infidelity on detuning. It maintains infidelity below $10^{-3}$ for single-atom detunings up to $|\Delta_i|/\Omega \approx 0.06$, compared to $\lesssim 0.01$ for TO pulses.
• Erasure Conversion: Simulations show that converting residual Rydberg population to erasures significantly reduces the conditional infidelity. A single erasure-conversion step at the end of the echoed sequence is sufficient, as the echo structure dynamically decouples leaked amplitude from the second half of the sequence.
• Robustness to Noise: The gate design reduces requirements on laser frequency stability, intensity homogeneity, field calibration, and atomic temperature.

Significance and Claims
The paper claims to significantly reduce the experimental requirements for practical fault-tolerant quantum computing with neutral atoms. By demonstrating that the "no-go" theorem for detuning-robust CZ gates can be circumvented by relegating errors to single-qubit rotations and removing them via echoing, the authors provide a pathway to high-fidelity gates in the presence of realistic noise.

The authors emphasize that their results:

• Allow for a much wider range of single-atom detunings while maintaining infidelity below $10^{-3}$.
• Convert a large fraction of residual errors into detectable erasures, which are easier to correct in quantum error correction codes than Pauli errors.
• Are compatible with existing experimental architectures, such as the "zone" architecture using stimulated Raman transitions for the fast $X$ rotations.

The work does not claim to solve intensity inhomogeneity (which requires separate optimal control strategies) but focuses specifically on the detuning problem, noting that amplitude robustness could be combined with this approach in future work at the cost of longer gate durations. The authors also acknowledge that while the pulses are optimized for detuning, they respect finite modulator bandwidth constraints by construction.