The computational part of the thesis is the investigation of titanium chloride (II) as a potential catalyst for the bis-silylation reaction of ethylene with hexaclorodisilane at different levels of theory. Bis-silylation is an important reaction for producing bis(silyl) compounds and new C-Si bonds, which can serve as monomers for silicon containing polymers and silicon carbides. Ab initio calculations on the steps involved in a proposed mechanism are presented. This choice of reactants allows us to study this reaction at reliable levels of theory without compromising accuracy. Our calculations indicate that this is a highly exothermic barrierless reaction. The TiCl 2 catalyst removes a 50 kcal/mol activation energy barrier required for the reaction without the catalyst. The first step is interaction of TiCl 2 with ethylene to form an intermediate that is 60 kcal/mol below the energy of the reactants. This is the driving force for the entire reaction. Dynamic correlation plays a significant role because RHF calculations indicate that the net barrier for the catalyzed reaction is 50 kcal/mol. We conclude that divalent Ti has the potential to become an important industrial catalyst for silylation reactions.;In the programming part of the thesis, parallelization of different quantum chemistry methods is presented. The parallelization of code is becoming important aspect of quantum chemistry code development. Two trends contribute to it: the overall desire to study large chemical systems and the desire to employ highly correlated methods which are usually computationally and memory expensive. In the presented distributed data algorithms computation is parallelized and the largest arrays are evenly distributed among CPUs. First, the parallelization of the Hartree-Fock self-consistent field (SCF) method is considered. SCF method is the most common starting point for more accurate calculations. The Fock build (sub step of SCF) from AO integrals is also often used to avoid MO integral computation. The presented distributed data SCF increases the size of chemical systems that can be calculated by using RHF and DFT. The important ab initio method to study bond formation and breaking as well as excited molecules is CASSCF. The presented distributed data CASSCF algorithm can significantly decrease computational time and memory requirements per node. Therefore, large CASSCF computations can be performed. The most time consuming operation to study potential energy surfaces of reactions and chemical systems is Hessian calculations. The distributed data parallelization of CPHF will allow scientists carry out large analytic Hessian calculations.