UC San Diego

This is a study of an application of finite element methods designed for convection-dominated, time-dependent partial differential equations. Specifically, this work analyzes finite element discretizations that employ moving meshes in order to solve linear differential equations over space-time domains. These methods can lead to significant savings in computation costs for problems having solutions that develop steep moving fronts, as moving meshes have the ability to track these fronts continuously with a high concentration of nodes; this flexibility allows for much larger time steps than standard tensor product finite elements, while maintaining high resolution of fine structures that sweep through the spatial domain. The main results are a priori and a posteriori error bounds for some moving finite element methods of high order and general time-stepping schemes. These finite element methods follow a method of lines approach for propagating the solution in time, though the error analysis places a strong emphasis on the properties inherited by the finite element aspects of the discrete problem. Another focus of this work is to determine practical and efficient schemes for adaptive meshing and mesh motion. As a result of this research, a solver has been written in C++ that is applicable to time-dependent linear convection-diffusion-reaction equations with a single dimension for the spatial