Significance threshold for pseudopotential core overlap

Ascertain the threshold of pseudopotential core-radius overlap at which the pseudopotential approximation introduces significant errors in solid-state calculations, and quantify this threshold in terms of commonly used parameters (for example, the outermost core radius values provided in VASP POTCAR files).

Background

In validating hypothetical structures, the authors examine nearest-neighbor separations relative to pseudopotential core radii (e.g., VASP POTCAR RCORE) because overlapping pseudo-core regions can lead to errors, particularly with softer pseudopotentials. They report many structures in major databases exhibit such overlaps.

Despite this prevalence, the exact degree of overlap that results in materially significant errors remains unknown, complicating automated screening and dataset curation for training robust ML potentials.