I am struggling to get an intuition for how the time non-dimensionalization parameter, “relaxation_time” (for NondimElasticQuasistatic) has an effect on the stability of my program, especially when we are already setting a specific time step. I am currently running a quasistatic, dynamic slip problem (governed by rate-and-state) with time dependent Dirichlet boundary conditions. It seems that the stability of the program depends on the ratio of the relaxation time and my time step size rather than the magnitude of my time step size itself. How is the relaxation time affecting the stability of my program?
The relaxation time in quasi-static problems corresponds to the time scale is used to nondimensionalize time. As a result, it also affects the scale used to nondimensionalize velocity or slip rate. Because we use an iterative solver, this also influences the solver parameters related to deciding whether a fault is locked or sliding. For more information see the tutorial on fault friction in our 2017 PyLith Tutorial.