I am looking at the velocity dispersion for different types of particles and I notice that for different particle types (gas, wind/star, and dark matter), the data specifications always say, "The 3D velocity dispersion of all dark matter particles within a radius of SubfindHsml of this cell." for 'SubfindVelDisp'. So, it seems like for all these three different types, they all should return same set of velocity dispersion. However, I tried the velocity dipersion for these three different particles in one subhalo and they all return different set of velocity dispersion. So, I feel that the specifications don't all mean dark matter particles. Is there a possibility that this might be a typo?
I have changed the descriptions from "of this cell" to "of this particle" for DM, stars, and BHs, to avoid confusion.
But the description is correct as is - this value is a local measure of the DM velocity dispersion. It is not a subhalo integral quantity, as you are likely computing. (It would be rare that you would want to use SubfindVelDisp for anything. What you are calculating should be more similar to SubhaloVelDisp in the group catalogs, except split by particle type).
Thanks for the advice! I actually do use SubhaloVelDisp eventually. But I am still confused about the description. For instance, I think SubfindVelDisp under gas type category means to return the velocity dispersion of gas particles of that subhalo. Why the description mentioning dark matter particles?
You can certainly verify, by performing exactly the measurement as described. You should be able to recover the value (to within floating roundoff). This is simply a value calculated in the code that we have saved - the local DM velocity dispersion around the location of all particles/cells.
I probably see what my trouble is. Should we have a picture that a gas/star particle is consisted by hundreds of dark matter particles? So, the SubfindVelDisp calculates the velocity dispersion of all the dark matter particles inside the star/gas particles. Did I got it correctly? That's why it is rarely used!!
In addition, I have one more question. I am not familiar with the exact definition of 3D (peculiar) velocity dispersion. Is it Sigma_v3D = Sqrt[Sigma_vx^2 + Sigma_vy^2 + Sigma_vz^2]? and Sigma_v1D = Sigma_v3D/Sqrt? Thanks!
Dark matter particles are not inside other particles, but they are nearby (in space). And yes your conversion to 1D looks good.