Expanding the Neutral Atom Gate Set: Native iSWAP and Exchange Gates from Dipolar Rydberg Interactions

TL;DR

This work addresses expanding the neutral-atom gate set by exploiting dipole-dipole exchange between two Rydberg states to realize native iSWAP and parameterized exchange gates. The authors develop an optimal-control framework with global drives to map exchange dynamics into the qubit subspace, and perform extensive noise modeling (atomic motion, Rydberg decay, laser PSDs) plus a noise-aware pulse selection to identify robust protocols. They demonstrate high-fidelity iSWAP gates (fidelities exceeding 99.9% under realistic Sr-88 hardware) and analyze the contributions of different noise sources, revealing practical routes to robust two-qubit gates. The results suggest that dipolar exchange gates can significantly enrich the neutral-atom gate set, enabling faster operations and more efficient quantum circuits, with extensions to the entire U_XY(θ) family and integration of FRT into optimization.

Abstract

We present a native realization of iSWAP and parameterized exchange gates for neutral atom quantum processing units. Our approach leverages strong dipole-dipole interactions between two dipole-coupled Rydberg states, and employs optimal control techniques to design time-efficient, high-fidelity gate protocols. To minimize experimental complexity, we utilize global driving terms acting identically on all atoms. We implement a noise-aware pulse selection strategy to identify candidate protocols with reduced susceptibility to certain noise sources, then analyze their performance under realistic noise sources -- including atomic motion, Rydberg decay, and experimentally motivated laser phase and intensity noise. For a $^{88}$Sr-based architecture, we demonstrate fast iSWAP gate protocols which exceed fidelities of $99.9\%$ under realistic experimental conditions. These results pave the way for expanding the neutral atom gate set beyond typical Rydberg blockade-based entangling gates.

Figures (16)

• Figure 1: Optimal control framework for iSWAP gates with Rydberg Atoms. (a) We consider a pair of neutral atoms with two low-energy (meta-)stable qubit states $\ket{0}, \ket{1}$ and two high-energy Rydberg states $\ket{r}, \ket{r'}$, which are dipole coupled and undergo the exchange interaction, Eq. \ref{['eq:h_exchange']}. We consider global laser / microwave drives between qubit and Rydberg states using two different schemes A/B. (b) We use optimal control techniques to obtain high-fidelity pulse protocols for the iSWAP gate, starting from random ansatz pulse profiles. We use a regularization term in the cost function to obtain smooth pulses, which are easier to implement in an experiment. (c) A set of high-fidelity pulse candidates is then evaluated against a neutral-atom specific noise model to find the best performing pulse under experimentally realistic conditions. Our noise model contains noise from atomic motion in the traps, Rydberg decay, and laser phase and intensity noise modelled from PSDs.
• Figure 2: Optimal control results for iSWAP gate. (a-b) Gate infidelity of different optimal control runs versus unitless gate duration $\tau \Omega$ for Rabi modulated pulse for (a) driving scheme A, (b) driving scheme B [c.f. Fig. \ref{['fig:fig1_schemes']}]. (c-d) Same plots for phase modulated pulses for (c) driving scheme A, (d) driving scheme B. In all plots, each point corresponds to a single optimal control run with random initialization for a fixed gate duration $\tau$. Highlighted points and dashed lines are visual guides to the eye tracking the lowest infidelity pulses vs duration. Different colors indicate different interaction strengths $V_{\rm dipole}/\Omega$. For each scheme and modulation type we can clearly identify a strong drop of infidelity at a certain duration $\tau^*$, which indicates the corresponding iSWAP quantum speed limit for the given setting. For durations $\tau > \tau^*$ we obtain high-fidelity results with $1-\mathcal{F} < 10^{-10}$.
• Figure 3: Optimal control results for Rabi modulated drives for (a) Scheme A, (b) Scheme B with infidelity $1-\mathcal{F} < 10^{-5}$. We plot the unitless gate duration $\tau/\Omega$ as a function of the interaction strength $V_{\rm dipole}/\Omega$. High-fidelity, time optimal pulses are obtained for finite interaction strength $V_{\rm dipole} \sim \Omega$. For $V_{\rm dipole} \lesssim \Omega$ the gate duration is limited by finite interaction strength, for $V_{\rm dipole} \gtrsim \Omega$ it is limited by finite Rabi frequency.
• Figure 4: Possible atomic driving scheme proposals for typical $^{88}$Sr setups. (a) Driving scheme A, requiring a drive between $\ket{0}$ and $\ket{r'}$ can be implemented in the fine-structure qubit encoding with a two-photon drive between levels $^3P_0$ and $n^3P_0$. (b-c) Driving scheme B can be implemented with a MW drive between the Rydberg states $n^3S_1$ and $n^3P_0$, here shown for (b) the clock qubit encoding, and (c) the fine-structure qubit.
• Figure 5: Noise-aware pulse selection. (a) Pulse landscape for pulses with infidelities below $10^{-5}$, where the time spent in Rydberg manifold, $T_{\rm ryd}$ is plotted against the cumulative interaction $T_{\rm int} V_{\rm dipole}$, and the gate duration $\tau$ is color coded. We select three different pulse protocols with different properties. Their pulse profiles are shown as insets on top of the main figure. (b) Gate infidelity contributions of the atomic noise sources for the selected pulses. "Pulse 2" shows the overall lowest gate infidelity (black bars). (c-f) Dependency of the infidelity contributions as a function of the relevant duration parameter; (c) Interaction noise , (d) Doppler noise, (e) Decay noise.