An open-source framework for fast-yet-accurate calculation of quantum mechanical features

We present the open-source framework kallisto that enables the efficient and robust calculation of quantum mechanical features for atoms and molecules. For a benchmark set of 49 experimental molecular polarizabilities, the predictive power of the presented method competes against second-order perturbation theory in a converged atomic-orbital basis set at a fraction of its computational costs. Robustness tests within a diverse validation set of more than 80,000 molecules show that the calculation of isotropic molecular polarizabilities has a low failure-rate of only 0.3 %. We present furthermore a generally applicable van der Waals radius model that is rooted on atomic static polarizabilites. Efficiency tests show that such radii can even be calculated for small- to medium-size proteins where the largest system (SARS-CoV-2 spike protein) has 42,539 atoms. Following the work of Domingo-Alemenara et al. [Domingo-Alemenara et al., Nat. Comm., 2019, 10, 5811], we present computational predictions for retention times for different chromatographic methods and describe how physicochemical features improve the predictive power of machine-learning models that otherwise only rely on two-dimensional features like molecular fingerprints. Additionally, we developed an internal benchmark set of experimental super-critical fluid chromatography retention times. For those methods, improvements of up to 17 % are obtained when combining molecular fingerprints with physicochemical descriptors. Shapley additive explanation values show furthermore that the physical nature of the applied features can be retained within the final machine-learning models. We generally recommend the kallisto framework as a robust, low-cost, and physically motivated featurizer for upcoming state-of-the-art machine-learning studies.

Compact representation of generalized molecular polarizabilities and efficient calculation of polarization energy in an arbitrary electric field

Generalized polarizabilities and the molecular charge distribution can describe the response of a molecule in an arbitrary static electric field up to second order. Depending on the expansion functions used to describe the perturbing potential, the generalized polarizability matrix can have rather large dimension (~1000). This matrix is the discretized version of the density response function or electronic susceptibility. Diagonalizing and truncating it can lead to significant (over an order of magnitude) speed-up in simulations. We have analyzed the convergence behavior of the generalized polarizability using a plane wave basis for the potential. The eigenfunctions of the generalized polarizability matrix are the natural polarization potentials. They are potentially useful to construct efficient polarizability models for molecules.

How well do self-interaction corrections repair the over-estimation of molecular polarizabilities in density functional calculations?

We examine the effect of removing self-interaction error (SIE) on the calculation of molecular polarizabilities in the local spin density (LSDA) and generalized gradient approximations (GGA). To this end, we...

Simplified Ab Initio Molecular Dynamics-Based Raman Spectral Simulations

We describe a simplified approach to simulating Raman spectra from ab initio molecular dynamics (AIMD) calculations. The protocol relies on on-the-fly calculations of approximate molecular polarizabilities using the well-known sum over orbitals (as opposed to states) method. This approach bypasses the more accurate but computationally expensive approach to calculating molecular polarizabilities along AIMD trajectories, i.e., solving the coupled perturbed Hartree–Fock/Kohn–Sham equations. We demonstrate the advantages and limitations of our method through a few case studies targeting molecular systems of interest to surface- and/or tip-enhanced Raman spectroscopy practitioners.

Universal Scattering of Ultracold Atoms and Molecules in Optical Potentials

Universal collisions describe the reaction of molecules and atoms as dominated by long-range interparticle interactions. Here, we calculate the universal inelastic rate coefficients for a large group of ultracold polar molecules in their lower ro-vibrational states colliding with one of their constituent atoms. The rate coefficients are solely determined by values of the dispersion coefficient and reduced mass of the collisional system. We use the ab initio coupled-cluster linear response method to compute dynamic molecular polarizabilities and obtain the dispersion coefficients for some of the collisional partners and use values from the literature for others. Our polarizability calculations agree well with available experimental measurements. Comparison of our inelastic rate coefficients with results of numerically exact quantum-mechanical calculations leads us to conjecture that collisions with heavier atoms can be expected to be more universal.