Predicting the Post-Hartree-Fock Electron Correlation Energy of Complex Systems with the Information-Theoretic Approach

Employing some simple physics-inspired density-based information-theoretic approach (ITA) quantities to predict the electron correlation energies remains an open challenge. In this work, we expand the scope of the LR(ITA) (LR means linear regression) protocol to more complex systems, including (i) 24 octane isomers; (ii) polymeric structures, polyyne, polyene, all-<i>trans</i>-polymethineimine, and acene; (iii) molecular clusters, such as metallic Be<i>_n</i> and Mg<i>_n</i>, covalent S<i>_n</i>, hydrogen-bonded protonated water clusters H⁺(H₂O)<i>_n</i>, and dispersion-bound carbon dioxide (CO₂)<i>_n</i>, and benzene (C₆H₆)<i>_n</i> clusters. With LR(ITA), one can simply predict the <i>post</i>-Hartree-Fock (such as MP2 and coupled cluster) electron correlation energies at the cost of Hartree-Fock calculations, even with chemical accuracy. For large molecular clusters, we employ the linear-scaling generalized energy-based fragmentation (GEBF) method to gauge the accuracy of LR(ITA). Employing benzene clusters as an illustration, the LR(ITA) method shows similar accuracy to that of GEBF. Overall, we have verified that ITA quantities can be used to predict the <i>post</i>-Hartree-Fock electron correlation energies of various complex systems.