analiza aiqbench
This research introduces a cycle-accurate scheduler for Active Volume quantum architectures that explicitly models reaction latency and stale state lifetimes, offering significant runtime speedups over prior analytic estimation methods. By accurately capturing packing inefficiencies and memory requirements, the study provides a critical path toward optimizing compilation pipelines for large-scale logical qubit systems where workspace constraints become dominant.
A Scalable, Optimized Scheduler for the Active Volume Architecture
arXiv:2603.06376v2 Announce Type: replace Abstract: We improve the accuracy of Active Volume resource estimates by explicitly scheduling when Active Volume blocks execute and when their reactions complete. We present a cycle-accurate scheduler that assigns each logical qubit a role in every logical cycle, capturing execution effects such as packing inefficiencies, parallelization overhead, and memory requirements. In addition, the scheduler explicitly tracks reaction latency and stale state lifetimes, rather than assuming, as in prior analytic resource-estimation methods, that stale states can be absorbed into a fixed overhead. For a reduced phase precision $4\times4$ Fermi--Hubbard test circuit, we find a workspace-dependent transition to reaction-dominant execution at approximately $10^3$ logical qubits, beyond which stale state accumulation restricts the effective workspace and reduces the benefit of adding more qubits. Below this regime, for a full phase precision $4\times4$ instance, the scheduler predicts a $1.64\times$ runtime speedup relative to ARE, primarily due to a $1.80\times$ reduction in average memory occupancy. We also show that parallelization and reaction latency overhead grows with available workspace, inverting the trend of the \cite{Caesura25} ARE model. These results provide a path toward reaction-aware analytic resource estimates and a full compilation pipeline for Active Volume architectures.
Abstract:We improve the accuracy of Active Volume resource estimates by explicitly scheduling when Active Volume blocks execute and when their reactions complete. We present a cycle-accurate scheduler that assigns each logical qubit a role in every logical cycle, capturing execution effects such as packing inefficiencies, parallelization overhead, and memory requirements. In addition, the scheduler explicitly tracks reaction latency and stale state lifetimes, rather than assuming, as in prior analytic resource-estimation methods, that stale states can be absorbed into a fixed overhead. For a reduced phase precision $4\times4$ Fermi--Hubbard test circuit, we find a workspace-dependent transition to reaction-dominant execution at approximately $10^3$ logical qubits, beyond which stale state accumulation restricts the effective workspace and reduces the benefit of adding more qubits. Below this regime, for a full phase precision $4\times4$ instance, the scheduler predicts a $1.64\times$ runtime speedup relative to ARE, primarily due to a $1.80\times$ reduction in average memory occupancy. We also show that parallelization and reaction latency overhead grows with available workspace, inverting the trend of the \cite{Caesura25} ARE model. These results provide a path toward reaction-aware analytic resource estimates and a full compilation pipeline for Active Volume architectures.
arXiv:2603.06376 [quant-ph]
(or arXiv:2603.06376v2 [quant-ph] for this version)
https://doi.org/10.48550/arXiv.2603.06376 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 279 cuvinte
