61b65061d775e708.tex
1: \begin{abstract}
2: 
3: 
4: Simulations of purely self-gravitating $N$-body systems are often 
5: used in astrophysics and cosmology to study the collisionless limit 
6: of such systems. Their results for macroscopic quantities should 
7: then converge well for sufficiently large $N$. Using a study of the evolution 
8: from a simple space of spherical initial conditions --- including a region
9: characterised by so-called ``radial orbit instability'' --- we
10: illustrate that the values of $N$ at which such convergence is
11: obtained can vary enormously. In the family of initial conditions we
12: study, good convergence can be obtained up to a few dynamical times
13: with $N \sim 10^3$ --- just large enough to suppress two body
14: relaxation --- for certain initial conditions, while in other cases
15: such convergence is not attained at this time even in our largest
16: simulations with $N \sim 10^5$. The qualitative difference is due to
17: the stability properties of fluctuations introduced by the $N$-body
18: discretisation, of which the initial amplitude depends on $N$. 
19: We discuss briefly why the crucial role which such fluctuations 
20: can potentially play in the evolution of the $N$ body system
21: could, in particular, constitute a serious problem in 
22: cosmological simulations of dark matter. 
23: \end{abstract}
24: