In quantum chemistry, scientists aim to solve the time-independent Schrödinger equation by employing a variety of approximation techniques whose accuracy are typically inversely proportional to their computational cost. This problem is amplified when it comes to chiral molecules, whose stereochemical assignments and associated chiroptical properties can be incredibly sensitive to small changes in their three-dimensional structure, requiring highly accurate theoretical methods. On the other hand, due to the polynomial scaling with system size, it is sometimes impractical to apply such methods to chemical compounds of broad scientific interest, especially when a multitude of low-energy conformations have to be accounted for as well. As a result, the assignment of absolute configurations to chiral compounds remains a tedious task. However, the characterization of these compounds is something that many different scientists are significantly invested in. The ultimate goal, then, is twofold: to gain useful insight by utilizing the electronic structure methods at your disposal while simultaneously developing new approximation techniques that can be used to push the boundaries on what is currently capable in computational chemistry. Therefore, we start by applying widely accepted density functional theory methods to predict optical rotations and electronic circular dichroism for naturally occurring antiplasmodial and anticancer drug candidates. We find that by comparing the computational results directly with those obtained through experimental measurement, we can provide reliable absolute config- uraitonal assignments to a variety of chiral compounds with numerous stereogenic centers. We also present the first ever prediction of vibrational circular dichroism with second-order Møller-Plesset perturbation theory. This extension opens the door to systematically improvable correlated wave function methods that can be employed when density functional theory fails or when higher accuracy results are required. / Doctor of Philosophy / Theoretical chemistry aims to draw a line from a molecule's three-dimensional structure to a set of physical observables, allowing for the efficient prediction of such properties. One family of chemical compounds for which this task becomes increasingly difficult is known as chiral molecules. A chiral compound is defined as one that has a non-superimposable mirror image. The concept of chirality is most tangibly seen with a pair of human hands, which demonstrate this same mirror-like behavior. In the same way that a person has left and right hands, a three-dimensonal handedness can be used to characterize many compounds that are essential to life including enzymes, sugars, and proteins. Although procedures have been developed to consistently isolate pure samples of such compounds, the correct identification of each hand poses a much larger task and costs the global pharmaceutical industry tens to hundreds of millions of dollars every year. As such, gaining insight about these incredibly valuable compounds and their associated properties is a never ending goal for many scientists. One such way to gain insight is through the direct comparison of experimental and calculated properties, namely chiroptical properties. These unique properties define how chiral compounds interact with light. While experimental scientists are limited in the degree to which they can probe a molecule's structure, theoretical chemists have the advantage of knowing the exact three-dimensional structure for the compound they are studying. On the other hand, theoretical chemists rely on comparison to experimental results to develop new methods or apply the available techniques to predict molecular properties. This work begins by attempting to match calculated properties to experimentally measured ones in order to confirm the detailed molecular structure of natural product drug candidates. Through multiple such computational studies, it is shown that the current methods are sometimes limited in the knowledge that they can provide. As a result, it is absolutely necessary to continue to improve on the existing methods. We go on to provide a first-of-its-kind implementation that allows for theoretical chemists to compare their results to experiment in a way that was not previously possible.

Les fonctionnelles de la densité couramment utilisées ont rencontrées un succès spectaculaire dans la modélisation des systèmes physiques, chimiques, et biologiques. Toutefois, elles se sont avérées inadaptées pour décrire certaines situations, comme par exemple les forces de dispersion de London ou les phénomènes de corrélation forte. Dans le cadre de cette thèse, nous nous sommes intéressés à des développements récents de la formulation de l’énergie de corrélation exprimée à partir du théorème de fluctuation-dissipation et connexion adiabatique, visant à pallier ces problèmes. En particulier, différentes implémentations des méthodes au-delà de l’approximation de la phase aléatoire, qui permettent la prise en compte de la contribution d’échange dans le calcul de l’énergie de corrélation, ont été comparées. Ensuite, afin de réduire drastiquement la complexité numérique, une procédure d’orthogonalisation des vecteurs utilisées pour représenter la matrice diélectrique a été développée. Ces méthodes ont ensuite été appliquées au calcul de l’énergie de liaison de petits complexes moléculaires. La formulation de l’énergie de corrélation de la théorie de perturbation de Møller-Plesset dans le contexte matrice diélectrique est aussi présentée et testée. En parallèle, des calculs utilisant les méthodes semi-empiriques numériquement efficaces ont été conduits sur trois ensembles de molécules afin d’en tester les performances concernant les énergies de liaisons en les comparant aux valeurs de références disponibles dans la littérature / Commonly used density functionals have encountered a spectacular success in the modelling of physical, chemical or biological systems. However, they have proven to be unsuitable to describe some situations, such as London’s dispersion forces or strong correlation behaviour. In this thesis, we have been interested in recent developments in the formulation of the correlation energy from the adiabatic connection fluctuation dissipation theorem, to overcome these problems. In particular, different implementations of methods beyond the random phase approximation, which allow to take into account the exchange contribution in the computation of the correlation energy, have been compared. Then, in order to drastically decrease the numerical complexity, an orthogonalization procedure of the vectors used to represent the dielectric matrix has been developed. Then these approaches were applied to the calculation of the binding energy of small molecular complexes. The formulation of the correlation energy of the Møller-Plesset perturbation theory within the dielectric matrix context is also presented and tested. In parallel, calculations using numerically efficient semi-empirical methods were conducted over three molecular sets in order to test their performances regarding the binding energies by comparing them to reference values available in the literature