PARSEC.py: A Python‐Based Real‐Space Kohn–Sham Density Functional Theory Code Accelerated by Machine Learned Charge Density

Abstract

ABSTRACT PARSEC.py is a Python‐based real‐space Kohn–Sham density functional theory (real‐space KS‐DFT) framework designed to provide a user‐ and developer‐friendly platform for first‐principles electronic‐structure simulations. Discretization on real‐space grids eliminates basis‐set approximations, while enabling systematic control of numerical accuracy and large‐scale parallelization. Building upon the theoretical foundations of the original PARSEC code, the framework leverages the Python scientific ecosystem to support modular development and integration with modern machine learning (ML) tools. PARSEC.py incorporates ML‐predicted density as an alternative initial guess for the superposition of atomic density (SAD) in self‐consistent field (SCF) iterations. This capability provides a framework for integrating ML models into electronic‐structure calculations and offers a pathway for accelerating SCF convergence. We present the theoretical framework, key algorithms, and performance characteristics of PARSEC.py using both SAD and ML initializations.