Scalable codes for precision calculations of properties of complex atomic systems

TL;DR

This paper introduces scalable parallel atomic structure codes with improved load-balancing and memory management, enabling high-precision calculations of complex atomic systems previously limited by computational resources.

Contribution

Development of new parallel codes with efficient load-balancing and dynamic memory allocation for high-precision atomic calculations of complex systems.

Findings

Achieved near-perfect linear scalability with multiple cores.

Enabled calculations of all 60 electrons in Ir$^{17+}$ ion.

Predicted the $3C/3D$ line ratio in Fe$^{16+}$.

Abstract

High precision atomic data is indispensable for experiments involving studies of fundamental interactions, astrophysics, atomic clocks, plasma science, and others. We develop new parallel atomic structure codes and explore the difficulties of load-balancing in these codes. Efficient load-balancing of matrix elements for many-electron systems is very difficult due to the intrinsic nature of the computational methods used to compute them. By arithmetically selecting determinants for each core, we achieve very even workload distribution, and attain near-perfect linear scalability and efficiency with the number of cores. We also implement dynamic memory allocation to minimize memory usage and remove the need for users to set certain array parameters. Our newly developed codes enable computations that were not possible before due to lack of memory or prohibitive computation times, and allow a broader range of correlations to be investigated in a shorter period of time. This includes calculations correlating all 60 electrons in the highly charged Ir$^{17+}$ ion and calculations predicting the $3C/3D$ line intensity ratio in Fe$^{16+}$. Our new code package will also be used to produce large volumes of high precision atomic data for a new online portal being developed at the University of Delaware.