Particle-in-cell (PIC) is the most used algorithm to perform self-consistent
tracking of intense charged particle beams. It is based on depositing
macro-particles on a grid, and subsequently solving on it the Poisson equation.
It is well known that PIC algorithms occupy intrinsic limitations as they
introduce numerical noise. Although not significant for short-term tracking,
this becomes important in simulations for circular machines over millions of
turns as it may induce artificial diffusion of the beam.
In this work, we present a modeling of numerical noise induced by PIC
algorithms, and discuss its influence on particle dynamics. The combined effect
of particle tracking and noise created by PIC algorithms leads to correlated or
decorrelated numerical noise. For decorrelated numerical noise we derive a
scaling law for the simulation parameters, allowing an estimate of artificial
emittance growth. Lastly, the effect of correlated numerical noise is
discussed, and a mitigation strategy is proposed.Comment: 14 pages, 12 figure