Fundamental error bound for entanglement generation between interacting Rydberg atoms

Technical Summary: Fundamental Error Bound for Entanglement Generation between Interacting Rydberg Atoms

Problem Statement
Neutral atom arrays utilizing Rydberg excitations are a leading platform for quantum computing. The primary mechanism for generating entanglement and executing quantum gates involves laser-driven transitions to Rydberg states, which interact strongly via van der Waals or dipole-dipole forces. However, these Rydberg states possess finite lifetimes, leading to spontaneous decay at a rate $\Gamma$. While technical errors (laser noise, atomic motion, etc.) can theoretically be eliminated, the decay error remains a fundamental limitation. This error is proportional to the average time ($T_r$) atoms spend in the decaying Rydberg states during the operation. Previous work has identified time-optimal pulses, but a rigorous lower bound on the achievable error, and the specific laser pulses that approach this bound, had not been satisfactorily addressed.

Methodology
The authors employ a two-pronged approach combining analytical derivation and numerical optimization:

1. Analytical Derivation of the Lower Bound:

   • The system is modeled as two atoms ($A$ and $B$) with internal states including a ground state $|g\rangle$ and a Rydberg state $|r\rangle$, interacting via a dispersive Hamiltonian $H_{int} = B |rr\rangle\langle rr|$.
   • The decay error is defined as $E \simeq \Gamma T_r = \Gamma \int_0^T P_r(t) dt$, where $P_r(t)$ is the instantaneous Rydberg population.
   • To find the minimum $T_r$, the authors utilize the min-entropy $S = -\log_2 c_1^2$ (where $c_1$ is the largest Schmidt coefficient) as a measure of distance from a maximally entangled state.
   • By analyzing the time derivative of the min-entropy $\dot{S}$ under unitary evolution driven by the interaction, they derive a relationship between the population $P_r$ and the rate of entanglement generation $\dot{S}$.
   • They formulate an optimization problem to minimize the integral of the ratio $P_r / |\dot{S}|$ over the evolution of the min-entropy from $S=0$ (product state) to $S=1$ (maximally entangled state).

2. Numerical Optimization (GRAPE):

   • To identify physical laser pulses that approach the theoretical bound, the authors employ Gradient Ascent Pulse Engineering (GRAPE).
   • The system is reduced to a three-level ladder system within the symmetric subspace ($|gg\rangle, |W\rangle, |rr\rangle$), where $|W\rangle$ is the symmetric Bell state.
   • The optimization minimizes a cost functional $J(\gamma) = F - \gamma T_r$, balancing high fidelity ($F$) against low Rydberg population ($T_r$). The penalty parameter $\gamma$ is gradually reduced to prioritize fidelity once a low-population region is found.
   • The optimization varies the Rabi frequency $\Omega(t)$ and detuning $\Delta(t)$ over a finite duration $T$.

Key Contributions and Results

• Fundamental Error Bound: The authors analytically derive a rigorous lower bound for the decay error in preparing any maximally entangled state of two atoms:
  $$E \geq \eta_{min} \frac{\Gamma}{B}, \quad \text{where} \quad \eta_{min} = 1 + \frac{\pi}{2} \approx 2.57$$
  This bound assumes all technical errors are negligible and depends solely on the ratio of the decay rate $\Gamma$ to the interaction strength $B$. This refines previous estimates which suggested a looser bound of $E \gtrsim 1.05 \Gamma/B$.

• Tightness of the Bound: Using GRAPE optimization, the authors identify specific laser pulses that generate a maximally entangled (Bell) state with an error $E \approx 2.575 \Gamma/B$. This result is only $\sim 1\%$ above the theoretical limit, demonstrating that the bound is tight and realistically achievable.

• Optimal Pulse Characteristics:

  • The optimal pulses require a duration $T \gtrsim 5/B$ to approach the bound.
  • The pulses are non-trivial, involving time-dependent Rabi frequencies and detunings that minimize the time spent in the Rydberg state while maximizing the rate of entanglement generation.
  • Theoretical analysis reveals that the absolute limit ($\eta_{min}$) is only reached in the limit of infinite duration ($BT \to \infty$), but finite-duration pulses can closely approximate it.

• Comparison with Standard Protocols:

  • Standard protocols (e.g., a $\pi/2$ pulse followed by a free evolution of time $\pi/B$) yield an error $E \approx \pi \Gamma/B \approx 3.14 \Gamma/B$, which is $\sim 22\%$ higher than the fundamental bound.
  • The authors note that while their derived bound applies to the preparation of a specific Bell state, it does not automatically apply to "special perfect entanglers" (gates like CZ that maximally entangle all four orthogonal input product states simultaneously). For such gates, the average error remains higher (e.g., $\bar{E} \approx \pi \Gamma/B$ for two-level atoms), and the authors were unable to find pulses that reduce the average error of a CZ gate to the fundamental bound $\eta_{min} \Gamma/B$.

Significance and Claims
The paper claims to establish the fundamental physical limit for the fidelity of entanglement generation in Rydberg atom systems, constrained only by spontaneous decay and interaction strength. By proving that this bound is tight (achievable within 1%), the work provides a definitive benchmark for experimental performance.

The authors emphasize that their result:

1. Refines theoretical estimates: Corrects previous looser bounds and provides a mathematically rigorous derivation.
2. Guides experimental design: Identifies that smooth, optimized laser pulses can significantly outperform standard protocols, reducing the decay error contribution to gate infidelity.
3. Applies broadly: The derivation is general and applicable to other quantum information processing systems involving decay and relaxation.
4. Practical Implications: For realistic parameters (e.g., $^{87}\text{Rb}$ atoms at $n \approx 40-60$), the fundamental error bound suggests that gate errors below $10^{-4}$ are achievable even at room temperature, provided the interaction strength $B$ is sufficiently large and optimized pulses are used.

The paper concludes that while the fundamental bound for single-state preparation is reachable, achieving similar low errors for universal two-qubit gates (like CZ) remains a challenge, particularly in the strong blockade regime.