Reuse-Aware Compilation for Zoned Quantum Architectures Based on Neutral Atoms

TL;DR

ZAC minimizes data movement overhead between zones with qubit reuse, i.e., keeping them in the entanglement zone if an immediate entangling operation is pending, enabling advancements in quantum algorithms and applications.

Abstract

Quantum computing architectures based on neutral atoms offer large scales and high-fidelity operations. They can be heterogeneous, with different zones for storage, entangling operations, and readout. Zoned architectures improve computation fidelity by shielding idling qubits in storage from side-effect noise, unlike monolithic architectures where all operations occur in a single zone. However, supporting these flexible architectures with efficient compilation remains challenging. In this paper, we propose ZAC, a scalable compiler for zoned architectures. ZAC minimizes data movement overhead between zones with qubit reuse, i.e., keeping them in the entanglement zone if an immediate entangling operation is pending. Other innovations include novel data placement and instruction scheduling strategies in ZAC, a flexible specification of zoned architectures, and an intermediate representation for zoned architectures, ZAIR. Our evaluation shows that zoned architectures equipped with ZAC achieve a 22x improvement in fidelity compared to monolithic architectures. Moreover, ZAC is shown to have a 10% fidelity gap on average compared to the ideal solution. This significant performance enhancement enables more efficient and reliable quantum circuit execution, enabling advancements in quantum algorithms and applications. ZAC is open source at https://github.com/UCLA-VAST/ZAC

Figures (20)

• Figure 1: A comparison between a) monolithic and b) zoned architectures based on neutral atoms. Blue regions are illuminated by the Rydberg laser. The grey region in b) is not covered by the Rydberg laser. c) Fidelity breakdown for the monolithic architecture based on the results in Ref. tan2024enola. Side-effect noise (blue, Rydberg excitation of idling qubits) is significant.
• Figure 2: Our reference zoned architecture following Ref. bluvstein2024logical. a) Overview of the architecture consisting of a readout zone (orange) for qubit measurement, an entanglement zone (blue) for Rydberg entangling gates, and a storage zone (grey) for idle qubits. The entanglement zone consists of 7 rows of Rydberg sites, each contains 20 sites. The storage zone consists of 100$\times$100 storage traps. The zone separation is $d_\text{sep}$=10um. The red dashes denote a rearrangement job from the storage to the entanglement zone. b) Detailed layout in the entanglement zone. The dotted gray eclipses indicate Rydberg sites, each with two SLM traps separated by $d_\text{Ryd}$=2um. The blue discs indicate the Rydberg range for qubit interaction. The distance between two Rydberg sites is $d_\omega$=10um. This prevent unwanted interaction between qubits in different sites. The circles represent the empty traps, and the solid dots are the traps occupied by qubits. When the Rydberg laser is on, two qubits in the same Rydberg site perform a CZ operation, e.g., CZ($q_0$,$q_1$) and CZ($q_2$,$q_3$). c) Detailed layout in the storage zone, where the qubit separation is $d_s$=3um.
• Figure 3: Specification of AOD arrays, SLM arrays, zones, and architecture.
• Figure 4: Preprocessing: resynthesis, 1Q gate optimization, and as-soon-as-possible 2Q gate scheduling. The output is a list of gate stages.
• Figure 5: Initial placement. Qubits are represented by dots in different colors. Colored edges indicate the nearest Rydberg sites for each qubit, e.g., the nearest site for $q_0$ (red dot) is $\omega_{0,1}$. $\omega^\text{near}_{g_0}$ and $\omega^\text{near}_{g_1}$ are the nearest Rydberg site for $g_0$ and $g_1$, respectively. A possible movement in simulated annealing (dashed arrow) is to move $q_0$ to the site at $r_3$ and $c_9$ in the storage zone.