Wailord: Parsers and Reproducibility for Quantum Chemistry
Data driven advances dominate the applied sciences landscape, with quantum chemistry being no exception to the rule. Dataset biases and human error are key bottlenecks in the development of reproducible and generalized insights. At a computational level, we demonstrate how changing the granularity of the abstractions employed in data generation from simulations can aid in reproducible work. In particular, we introduce wailord (https://wailord.xyz), a free-and-open-source python library to shorten the gap between data-analysis and computational chemistry, with a focus on the ORCA suite binaries. A two level hierarchy and exhaustive unit-testing ensure the ability to reproducibly describe and analyze \textquotedbl{}computational experiments\textquotedbl{}. wailord offers both input generation, with enhanced analysis, and raw output analysis, for traditionally executed ORCA runs. The design focuses on treating output and input generation in terms of a mini domain specific language instead of more imperative approaches, and we demonstrate how this abstraction facilitates chemical insights.
quantum chemistry, parsers, reproducible reports, computational inference
DOI10.25080/majora-212e5952-021