| Various computational methods have been used to generate potential energy surfaces, which can help us simulate and interpret how atoms or molecules behave during a chemical reaction. For accurate work, ab initio wavefunction methods have traditionally been used, which have some disadvantages. For example, highly accurate methods scale poorly with system size ( n7 or higher) and are mostly not well parallelized for calculations with multiple processors. One alternative method that has more favorable scaling with system size and is well parallelized is a computational technique called quantum Monte Carlo (QMC). QMC methods scale with the number of electrons as n3 and have been found to scale almost linearly with the number of processors, even beyond 500,000 cores. However, despite the favorable scaling towards large systems, the cost of QMC methods is relatively expensive for small systems. Small systems nevertheless make important benchmarks necessary for the new methods to gain acceptance. Thus, it was determined to study QMC methods in a few benchmark systems in order to assess its accuracy and routine applicability.;It was found that QMC methods can be very accurate comparing well with experimental measurements and other high-level ab initio methods. Benchmark calculations with QMC produced realistic spectroscopic parameters for CO and N2. However, for small system sizes, they are relatively very expensive to perform with the cost being orders of magnitude higher than traditional methods. Consequently, their use in small systems will likely most often be restricted to only a few geometrical points of interest, unlike traditional methods. Nevertheless, deep insight into the electronic structure of a system can be obtained. |
