COSMIC: Molecular Conformation Space Modeling in Internal Coordinates with an Adversarial Framework.

The fast and accurate conformation space modeling is an essential part of computational approaches for solving ligand and structure-based drug discovery problems. Recent state-of-the-art diffusion models for molecular conformation generation show promising distribution coverage and physical plausibility metrics but suffer from a slow sampling procedure. We propose a novel adversarial generative framework, COSMIC, that shows comparable generative performance but provides a time-efficient sampling and training procedure.

Comparing the accuracy of perturbative and variational calculations for predicting fundamental vibrational frequencies of dihalomethanes.

Three dihalogenated methane derivatives (CHF, CHFCl, and CHCl) were used as model systems to compare and assess the accuracy of two different approaches for predicting observed fundamental frequencies: canonical operator Van Vleck vibrational perturbation theory (CVPT) and vibrational configuration interaction (VCI). For convenience and consistency, both methods employ the Watson Hamiltonian in rectilinear normal coordinates, expanding the potential energy surface (PES) as a Taylor series about equilibrium and constructing the wavefunction from a harmonic oscillator product basis. At the highest levels of theory considered here, fourth-order CVPT and VCI in a harmonic oscillator basis with up to 10 quanta of vibrational excitation in conjunction with a 4-mode representation sextic force field (SFF-4MR) computed at MP2/cc-pVTZ with replacement CCSD(T)/aug-cc-pVQZ harmonic force constants, the agreement between computed fundamentals is closer to 0.

Tensor-structured coupled cluster theory.

We derive and implement a new way of solving coupled cluster equations with lower computational scaling. Our method is based on the decomposition of both amplitudes and two electron integrals, using a combination of tensor hypercontraction and canonical polyadic decomposition. While the original theory scales as O(N) with respect to the number of basis functions, we demonstrate numerically that we achieve sub-millihartree difference from the original theory with O(N) scaling.

Analytic energy gradient for the projected Hartree-Fock method.

We derive and implement the analytic energy gradient for the symmetry Projected Hartree-Fock (PHF) method avoiding the solution of coupled-perturbed HF-like equations, as in the regular unprojected method. Our formalism therefore has mean-field computational scaling and cost, despite the elaborate multi-reference character of the PHF wave function. As benchmark examples, we here apply our gradient implementation to the ortho-, meta-, and para-benzyne biradicals, and discuss their equilibrium geometries and vibrational frequencies.