Several well-known programs generate pseudopotentials in a variety of formats, tailored to the needs of electronic-structure codes. While some generators are now able to output data in different bespoke formats, and some simulation codes are now able to read different pseudopotential formats, the common historical pattern in the design of those formats has been that a generator produced data for a single particular simulation code, most likely maintained by the same group. This implied that a number of implicit assumptions, shared by generator and user, have gone into the formats and fossilized there.
This leads to practical problems, not only of programming, but of interoperability and reproducibility, which depend on spelling out quite a number of details which are not well represented for all codes in existing formats.
PSML (for PSeudopotential Markup Language) is a file format for norm-conserving pseudopotential data which is designed to encapsulate as much as possible the abstract concepts in the domain's ontology, and to provide appropriate metadata and provenance information.
The software library libPSML can be used by electronic structure codes to transparently extract the information in a PSML file and adapt it to their own data structures, or to create converters for other formats.
A full description of PSML and its design principles has been published in:
A preprint is also available in arXiv.