### what's the difference of the 'SubfindDMDensity' between PartType0,1,4

Di Yihuan
• 7 Jan '20

Hi TNG teams and other friends,
I wonder what's the difference of the 'SubfindDMDensity' between PartType0,1,4? if they are the same thing?

And does the Density and SubfindDensity in the PartType0 mean rho(x,y,z) (N,3)and rho(r)(N)?
if I try to picture a gas mean density profile with r, is it better to using SubfindDensity than Density?

Yihuan

Dylan Nelson
• 13 Jan '20

Hello,

Yes `SubfindDMDensity` is the same measurement for all particle types. Note that this is a local estimate, so it is valid at the position of that gas/DM/star particle.

All such values, including `Density` and `SubfindDensity`, are estimates at the precise `Coordinates` value (x,y,z) of that cell or particle.

There is no `rho(r)` data, you would need to compute this by first computing `r` of cells/particles (from the halo center), then binning.

For gas density, you have two options: use average/median values of `PartType0/Density` at the `r` you are interested in, or else sum all `PartType0/Masses` in a bin, i.e. spherical shell, around `r`, and then normalize by the volume of that bin.

Di Yihuan
• 13 Jan '20

I got it. Thanks again.

Yun Wang
• 18 Aug '20

Are 'SubfindDMDensity' and 'SubfindDensity' the same thing?

Dylan Nelson
• 18 Aug '20

Hi Yun,

No exactly, the first is "dark matter density" and the second is "total density" (i.e., also including baryons).

But since DM dominates, they are going to be generally quite similar.

Yun Wang
• 18 Aug '20

Do you mean that "SubfindDensity' contains all kinds of particle types?

Dylan Nelson
• 18 Aug '20

That's correct, all particle types = DM + baryons.

Yun Wang
• 18 Aug '20

So, does that mean I can use "SubfindDensity" and "coordinates" to compute the power spectrum of all matter (DM + baryons) directly?

Dylan Nelson
• 18 Aug '20

Hello,

It depends what your goal is, I would read again the description of `SubfindDensity` carefully. This is a local mass density estimate, at the location of each `Coordinates`, which is based on an adaptive-size smoothing kernel.

Personally I would not have thought to use it to compute a power spectrum, but instead the combination of `Masses` and `Coordinates` which gives the (un-smoothed) mass distribution within the simulation volume.

Yun Wang
• 18 Aug '20