Supporting Data for "Delta Learning and Its Applications in Computational Chemistry"

This is a Supporting File for the PhD thesis. The thesis is about a machine learning protocol called delta learning, which is used to calibrate computed low-fidelity results to their higher-fidelity counterparts.

The file contains a README file, listing all the data needed to plot the graphs consisted in the thesis. The file "dataset(main_folder)" contains all data and scripts. Running the *.py files will take in the *.txt files and plot out graphs. For running the scripts, Python with matplotlib is required, together with numpy and pandas.

Fig2.3abc are R^2 correlation graphs for SchNet-predicted, DFT-computed, and experimental heat of formation. Fig3.1 are graphs of R^2 correlation between MD computed, random-forest-predicted, and experimental open circuit voltages (OCVs). Fig3.2 are the OCV plots of three battery materials. Fig3.3 are the error analysis for the delta learning correction process. Fig3.4 are R^2 correlation between MD computed, auto-platform-predicted, and experimental OCVs.