Simple Automatized Tool for Exchange-Correlation Functional Fitting

Density functional theory is, perhaps, the most popular and convenient tool in computational chemistry. DFT methods allow to solve different chemical tasks with a good balance of accuracy and computational time. Dozens of existing functionals cover majority of possible systems, and the development of new ones is still ongoing. However, despite the existence of different databases with accurate quantum-chemical data, the functional design remains a complicated and time-demanding task. Here we propose a novel approach simplifying and accelerating this process. The approach is based on Bayesian search with stochastic sub-sampling, that allows to consider 'history' of fitting steps, reduces computational time on each step and avoiding overfitting to training data. Besides general testing of the approach efficiency, we also showed an example of training specialized DFT functionals, outperforming popular ones. The approach is presented as a free code with built-in analysis tools. Using the code with appropriate reference database can help constructing DFT approximation for highly specialized task.