What I've tried to set this up is an initial model with these features:
1) Have the stars start out in the plane (zero height).
2) Give each star a random vertical velocity drawn from a gaussian with a
dispersion \sigma_z^2 = h_z^2 8 \pi G \rho
where \rho is the background mass density producing the background potential
(\rho = \rho_{disk} + \rho_{bulge} + \rho_{DM halo} ).
3) The ratio of vertical velocity dispersion to radial dispersion,
\sigma_z / \sigma_R = some value between 0.5 and 0.78. The ratio is independent
of galactic radius. The radial velocities are random ones drawn from a gaussian
of the given dispersion.
4) The velocity dispersion in the direction of galactic rotation, \sigma_{\phi},
is found from \sigma_{\phi} = \sigma_R \kappa / (2 \Omega), where
\kappa is the radial epicyclic frequency and \Omega is the angular rotation
rate of the disk (\kappa and \Omega come from the rotation curve of the
background potential). The phi velocities are random ones drawn from a gaussian
of the given dispersion about a mean = circular velocity so that the stars will
start out at their guiding centers.
Unfortunately, this setup produces a disk after integrating for 1-2 Gyr that is
thinner in the inner part than the outer part. I want the thickness to be
independent of radius.
The background potential includes a disk with \rho_{disk}(R,z) \propto
\exp{-R/h_r} sech^2 (z/z_0). This vertical profile is for an isothermal disk
and z_0 not necessarily equal to the scale height h_z of the stars.
Any ideas of what I'm doing wrong in this setup? Please email me any
suggestions. Thanks!
Nick
strobel@astro.washington.edu