Optimized readout strategies for neutral atom quantum processors

TL;DR

This work develops a quantitative framework to optimize readout in neutral-atom quantum processors by balancing readout fidelity and atomic retention, introducing the quantum circuit iteration rate and normalized quantum Fisher information as throughput metrics. It builds a physical model of heating and loss during photon scattering, derives readout fidelities for SPD and qCMOS detectors, and evaluates adaptive readout strategies that enable repeated task executions without frequent reloading. The findings indicate that information acquisition rates ranging from hundreds to thousands of hertz are achievable under realistic $^{87}$Rb parameters, depending on collection efficiency, trap depth, and cycle time. These insights offer practical guidance for scalable, high-throughput neutral-atom processors in sensing, simulation, and near-term quantum algorithms.

Abstract

Neutral atom quantum processors have emerged as a promising platform for scalable quantum information processing, offering high-fidelity operations and exceptional qubit scalability. A key challenge in realizing practical applications is efficiently extracting readout outcomes while maintaining high system throughput, i.e., the rate of quantum task executions. In this work, we develop a theoretical framework to quantify the trade-off between readout fidelity and atomic retention. Moreover, we introduce a metric of quantum circuit iteration rate (qCIR) and employ normalized quantum Fisher information to characterize system overall performance. Further, by carefully balancing fidelity and retention, we demonstrate a readout strategy for optimizing information acquisition efficiency. Considering the experimentally feasible parameters for 87Rb atoms, we demonstrate that qCIRs of 197.2Hz and 154.5Hz are achievable using single photon detectors and cameras, respectively. These results provide practical guidance for constructing scalable and high-throughput neutral atom quantum processors for applications in sensing, simulation, and near-term algorithm implementation.

Figures (8)

• Figure 1: Conceptual overview of optimized readout strategy. (a) A schematic diagram illustrating the photon scattering process during the readout stage. As the atom scatters more photons, the detector gathers more information, thereby improving readout fidelity. However, more scattered photons also result in greater atomic heating and increased atom loss probability. (b) Experimental sequences for two different readout approaches. The upper sequence employs a blow-away approach, in which atoms need to be re-preparation after readout. In contrast, the lower sequence utilizes a non-destructive readout technique, which preserves the atoms in the trap with high retention probability. This enables immediate execution of the next circuit iteration, effectively mitigating the detrimental impact of the lengthy loading time. (c) A logic diagram revealing the effect of various experimental parameters on readout performance. Varibles in the purple domain represent controllable experimental parameters: re-preparation duration ($t_\mathrm{dead}$), quantum task cycle time ($t_\mathrm{cycle}$), trap depth ($T_\mathrm{trap}$), initial atomic temperature ($T_\mathrm{i}$), atomic scattering rate ($R_\mathrm{sc}$), readout duration ($\tau$), and overall collection efficiency ($\eta$). The red domain illustrates the physical processes, where the total scattered photons ($N_\mathrm{sc}$) and the atomic retention probability ($P_\mathrm{ret}$) emerge as critical intermediate quantities. The yellow domain quantifies performance metrics: quantum circuit iteration rate (qCIR), readout fidelity ($\mathcal{F}$) and normalized Fisher information ($Q$). The logical relationships between variables, physical processes, and performance metrics are indicated by gray arrows.
• Figure 2: Trade-off between $N_\mathrm{sc}$ and $P_\mathrm{ret}$. (a) Maxwell-Boltzmann energy distribution for an atom at temperatures of $T_\mathrm{a}=100,\,150,\,200,\,250,\,300\,\mu\mathrm{K}$. The retention probability $P_\mathrm{ret}$ (loss probability $P_\mathrm{loss}$) is derived from the integral of the distribution below (above) the trap depth $T_\mathrm{trap}$, which is hinted by the blue (red) shaded area. Numerical values are calculated for an atom temperature of $300\,\mu\mathrm{K}$ and a trap depth of $2\,\mathrm{mK}$. (b) Dependence of retention probability $P_\mathrm{ret}=1-P_\mathrm{loss}$ on the temperature ratio $\xi=T_\mathrm{a}/T_\mathrm{trap}$. The inset shows several specific points for reference.
• Figure 3: Calculation results of $\mathcal{F}$ and $P_\mathrm{ret}$. (a) Photon-counting statistics for SPD readout with dark count $N_\mathrm{D,SPD}=0.05$ and atom fluorescence signal $\eta N_\mathrm{sc}=2.85$. (b) Photon-counting statistics for qCMOS camera readout with readout noise $N_\mathrm{QC}=360$, $\sigma_\mathrm{QC}=4$ and atom fluorescence signal $\eta N_\mathrm{sc}=28.5$. (c) Fidelity $\mathcal{F}$ versus readout stage duration $\tau$ for SPD and qCMOS. Curves are calculated with related parameters set to $(R_\mathrm{D,SPD},\,N_\mathrm{QC},\,\sigma_\mathrm{QC},\,\eta,\,R_\mathrm{sc})=(500\,\mathrm{Hz},\,360,\,4,\,0.3\,\%,9.5\,\mathrm{MHz})$. (d) Retention probability $P_\mathrm{ret}$ as a function of $\tau$ for shallow ($T_\mathrm{trap}=1\,\mathrm{mK}$) and deep ($T_\mathrm{trap}=5\,\mathrm{mK}$) optical traps with $T_\mathrm{i}=100\,\mu\mathrm{K}$ and $R_\mathrm{sc}=9.5\,\mathrm{MHz}$.
• Figure 4: Calculation results of the qCIR for a single atom by the adaptive strategy ($\mathcal{R}_1$). (a) $\mathcal{R}_1$ versus loss probability $P_\mathrm{loss}$. (b) $\mathcal{R}_1$ versus readout stage duration $\tau$ with two different trap depth of $T_\mathrm{trap}=1,\,5\,\mathrm{mK}$. Other related parameters are set to $(t_\mathrm{dead},\,t_\mathrm{cycle},\,T_\mathrm{i},\,R_\mathrm{sc})=(200\,\mathrm{ms},\,5\,\mathrm{ms},\,100\,\mu\mathrm{K},\,9.5\,\mathrm{MHz})$.
• Figure 5: The trade-off characteristics between $\mathcal{R}_1$ and $\mathcal{F}$ for SPD and qCMOS camera. The related readout device parameters are set to $(R_\mathrm{D,SPD},\,N_\mathrm{QC},\,\sigma_\mathrm{QC})=(500\,\mathrm{Hz},\,360,\,4)$, and related experimental parameters are set to $(t_\mathrm{dead},\,t_\mathrm{cycle},\,T_\mathrm{i},\,R_\mathrm{sc})=(200\,\mathrm{ms},\,5\,\mathrm{ms},\,100\,\mu\mathrm{K},\,9.5\,\mathrm{MHz})$. A colormap is employed to represent $Q$ under different $\mathcal{F}$ and $\mathcal{R}_1$, with contour lines plotted for reference. The optimal $Q_1$ are explicitly marked by hollow circles. (a) $(\eta,\,T_\mathrm{trap})=(0.3\,\%,\,1\,\mathrm{mK})$. For SPD, the optimal $Q_1=9.4\,\mathrm{Hz}$ with corresponding $\mathcal{F}=75.83\,\%$ and $\mathcal{R}_1=35.1\,\mathrm{Hz}$. While for qCMOS, the optimal $Q_1=4.8\,\mathrm{Hz}$ with corresponding $\mathcal{F}=99.63\,\%$ and $\mathcal{R}_1=4.9\,\mathrm{Hz}$. (b) $(\eta,\,T_\mathrm{trap})=(0.3\,\%,\,5\,\mathrm{mK})$. For SPD, the optimal $Q_1=146.4\,\mathrm{Hz}$ with corresponding $\mathcal{F}=94.04\,\%$ and $\mathcal{R}_1=188.8\,\mathrm{Hz}$. While for qCMOS, the optimal $Q_1=16.4\,\mathrm{Hz}$ with corresponding $\mathcal{F}=68.57\,\%$ and $\mathcal{R}_1=119.1\,\mathrm{Hz}$. (c) $(\eta,\,T_\mathrm{trap})=(1\,\%,\,1\,\mathrm{mK})$. For SPD, the optimal $Q_1=36.4\,\mathrm{Hz}$ with corresponding $\mathcal{F}=87.69\,\%$ and $\mathcal{R}_1=64.1\,\mathrm{Hz}$. While for qCMOS, the optimal $Q_1=5.10\,\mathrm{Hz}$ with corresponding $\mathcal{F}=99.42\,\%$ and $\mathcal{R}_1=5.2\,\mathrm{Hz}$. (d) $(\eta,\,T_\mathrm{trap})=(1\,\%,\,5\,\mathrm{mK})$. For SPD, the optimal $Q_1=193.8\,\mathrm{Hz}$ with corresponding $\mathcal{F}=99.56\,\%$ and $\mathcal{R}_1=197.2\,\mathrm{Hz}$. While for qCMOS, the optimal $Q_1=94.5\,\mathrm{Hz}$ with corresponding $\mathcal{F}=89.11\,\%$ and $\mathcal{R}_1=154.5\,\mathrm{Hz}$.