Fast and accurate electronic structure methods for predicting molecular properties of large correlated systems

A good theoretical description of a large chemical system requires electronic structure methods that account for electron correlation with tractable computational effort. Part of this thesis deals with developing and implementing models that help to not only improve the second order Moller-Plesset (MP2) perturbative description of structure and energetics but also simultaneously reduce the computational effort required to evaluate the MP2 correlation energy. We present the single parameter 'opposite-spin' (OS) scaled techniques where only the alpha-beta component of the MP2 energy is scaled by an empirically determined factor (either constant or inter-electronic distance dependent) while the same-spin contribution is completely neglected. The resulting 'scaled opposite-spin' MP2 (SOS-MP2) and 'modified opposite-spin' MP2 (MOS-MP2) energies and analytical gradients are demonstrated to provide statistical improvement over MP2 in the description of a range of chemical and structural properties. We show that the OS energy and its first derivative can be computed with an effort that scales as only the fourth power with system size as opposed to the fifth order cost of traditional MP2 theory by employing a combination of auxiliary basis expansions and Laplace transformation techniques (without exploiting locality). We also propose another variation of the OS scheme, where optimal orbitals are determined self-consistently with the inclusion of OS correlation energy. We demonstrate that this orbital-optimized OS model (O2) virtually eliminates the spin-contamination and symmetry-breaking problems of the HF reference and provides orbital-optimized coupled-cluster doubles (OD) quality results for open-shell molecules with just fourth order computational effort while the latter scales with the sixth power of system size. In the second part of the thesis, we apply standard electronic structure methods like MP2 and density functional theory (DFT) to characterize the interaction of molecular hydrogen with models of potential hydrogen storage materials (HSM) such as metal-organic frameworks (MOFs). We show that the strength of the hydrogen adsorption plays a key role in determining the sorption/desorption characteristics of the HSM and is critically linked to the underlying factors that stabilize the interaction (such as electrostatics, dispersion and orbital interactions).