ABACUS: An electronic structure analysis package for the AI era.

Area of Science:

• Computational Physics
• Materials Science
• Quantum Chemistry

Background:

• First-principles calculations are crucial for understanding material properties.
• Existing software often lacks flexibility or integration with modern computational techniques.
• The demand for accurate and efficient simulations is growing.

Purpose of the Study:

• To introduce ABACUS, an open-source software for advanced electronic structure calculations.
• To provide a versatile platform for integrating diverse computational methods.
• To facilitate the generation of machine learning potentials using high-performance computing.

Main Methods:

• Density Functional Theory (DFT) and molecular dynamics simulations.
• Compatibility with plane wave and numerical atomic orbital basis sets.
• Integration of various DFT flavors (Kohn-Sham, stochastic, orbital-free, real-time TDDFT).

Main Results:

• ABACUS enables efficient first-principles calculations.
• It provides a platform for generating large datasets for machine learning potentials.
• The software interfaces with multiple AI-assisted packages (DeePKS-kit, DeePMD, DP-GEN, etc.).

Conclusions:

• ABACUS is a powerful, open-source tool for computational materials science.
• Its flexibility and AI integration accelerate materials discovery and simulation.
• It supports high-performance computing for large-scale applications.