Abstract
Simulations of stellar systems involve, first, creating a model and, subsequently, following its evolution through numerical integration of the equations of motion. The models can be generated from theoretical distributions obtained from the Newtonian potential, but the integration demands resorting to softened potentials in order to reduce relaxation effects. Usually, the difference between both potentials causes only negligible alterations in the model, but in some cases (e.g., cuspy distributions) it can substantially alter it. Here we present a model-independent simple method to correct this problem in simulations of spherical stellar systems.
