### Trying to draw some physically meaningful known graphs Hannah Ji
• 4
• 24 Aug '18

Hi, thank you for providing wonderful sets of data.

I'm currently trying to understand and learn how to extract information I want, and things don't go well as I expected. I tried two things:

One is the Hubble's law. I tried to get the famous v=H0 D graph, so initially I came up with an idea that maybe I should extract from 'Coordinates' and 'Velocities' from PartType1(or whatever), snap135 and then calculate the length of each vectors. But it didn't work, and I suddenly realized that I'm not sure where the origin of the 'Coordinates' values is, and that 'Velocities' values do include both recessional and peculiar velocities. Am I right about why I failed to draw Hubble graph? If not, what's wrong?(and where is the origin of 'Coordinates'?)

Another is Rotation Curve. To do this, I read data specifications and concluded that I need 'SubfindHsml' and 'SubfindVeldisp' values from 'PartType1', but failed once again. I think it's because 'SubfindVeldisp' is not rotational velocity...(but not sure) Which data should I look at to draw such graphs?

Guess I should mention that I'm using data from Illusris-3. Dylan Nelson
• 1
• 28 Aug '18

Hi Hannah,

Some points:

(1) There is effectively no origin for `Coordinates`, this is because the simulation volume is periodic. You should think of them only as relative to somewhere else in the box.

(2) The `Velocities` are comoving (as are the `Coordinates`), so they do not include a Hubble expansion term. If you want to calculate the velocity between two galaxies, for example, including the Hubble flow, then you have to add this in (e.g. roughly speaking as in `v_rel = v_1 - v_0 + H(z)*d` where d is the distance between them).

(3) For rotation or circular velocity curves, these don't have anything to do with the `Subfind*` fields (in general you can ignore these). Instead, e.g. for a circular velocity curve, you need to sum up the total enclosed mass as a function of distance away from a galaxy or halo center. So you will want to load the `Masses` and `Coordinates` of particles (of all types, but most importantly of the dark matter), transforming the latter into distances from the center (e.g. relative to `SubhaloPos`).

Hope that helps a little! Hannah Ji
• 6 Sep '18

Thank you Dylan! I tried what you said about Hubble law, and I have an additional question.

1. I did Hubble by taking a reference point, and calculating the distances and relative velocities(vector subtraction) from this point. I've got somewhat linear data distribution when I took zeroth component of 'PartType1' from snap_135.0.hdf5 for the reference (though the value of Hubble parameter was very different from what is known. Maybe there was a mistake!) But when I did this with zeroth of others( like snap_135.20.hdf5) as a reference, the result was too different...(negative linear relation). I want to know why.

2. And on the rotation curve, why am I supposed to load 'Masses'? Isn't rotation curve a relation between orbital velocity and distance from a halo center? Dylan Nelson
• 6 Sep '18

Hi Hannah,

1. Are you writing to plot Hubble's Law itself? There isn't much to be gained here, because the simulation itself assumes the expansion with a particular value for Hubble's constant - for Illustris, H0 = 70.4 km/s/Mpc.

2. This is the difference between a rotation curve, and a circular velocity curve. The latter is perhaps easier to start with, and also commonly extracted from simulations, but you could measure either. Hannah Ji
• 14 Sep '18

You mean plotting Hubble's law is meaningless here because the whole data was simulated according to hubble constant, right? Thank you, I didn't think of it... And as you said, I'll try circular velocity curve.