A very large number of options is available for setting up a quantum chemical calculation, in particular when choosing the level of theory and the basis set. There is also a significant a priori uncertainty about the quality of the result obtained with each combination of the settings. Benchmarking on representative test sets has been a traditional approach to rank the quality of methods, basis sets, and other options. The number of various benchmarks, often involving similar or partly overlapping settings and test sets, has been growing steeply over the recent years, to the point when the sheer quantity and variety of benchmarks make it no easier to choose the best settings in practice. We have designed and implemented an expert system that analyses a database of benchmarks and makes a decision about which of them is more applicable to a given calculation type. The algorithm which is being developed within the quantum chemical program Jaguar will make it easier for non-experts to conduct quantum chemical calculations and will pave the way toward more automated quantum chemical solutions in applications.

Benchmarking the benchmarks: Navigating in the sea of quantum chemical information

248th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.