Gravitational potential in halos without the contribution from particles external to the halo
Justo Antonio Gonzalez Villalba
10 May '22
Dear Illustris-TNG team,
I would like to calculate the total potential energy of gas particles in a Illustris-TNG300-1 halo, but without the contribution from particles external to the halo.
Is there a way to correct the Potential value in the Halo cuts to remove the contribution of particles external to the halo?
Thank you!
Dylan Nelson
10 May '22
As you say, the Potential field in the snapshots is a global quantity, and there isn't any way to separate it into different contributions.
If you need a local measurement of the potential, it would make sense to simply compute this directly (e.g. using only halo particles, one halo at a time - this is not too expensive).
I see, then in order to calculate the potential I have a couple of questions regarding softening lengths:
In Nelson et al. 2018 the DM/Stars softening for TNG-300-1 at z=0 is 1.48kpc, but in the rest API (https://www.tng-project.org/api/TNG300-1/) we have softening_stars_comoving=2.0. If I multiply this by hubble 0.6774 then I get 1.3548kpc. Where is the difference coming from?
In Pillepich+ (2018) there is a comment that the gravitational forces become Newtonian at 2.8 x epsilon. So I assume for distances greater then 2.8 x epsilon I should calculate the potential as plain Newtonian and not Plummer-Equivalent?
Also, in the rest API we have softening_stars_max_phys=1.0 but both values mentioned above are greater than this threshold. Is the threshold really enforced?
I would probably suggest to ignore softenings for the potential calculation, as a first step.
For Illustris and TNG the softenings are comoving until z=1, then fixed. So this value 2.0 ckpc/h is the prior. Note that you divide by h, rather than multiply, to remove the hubble parameter from code lengths. So the box size of TNG100 is 75 Mpc/h = 110 Mpc. Then 1.0 / 0.6774 = 1.48.
So I have tried some quick tests with small halos, for example halo 1010 from TNG-300-1 snapshot 99 (z = 0 / a = 1)
I calculate the halo-only potential in the position of the less bounded dark matter particle (with id 7805765839). The halo-only potential is -159054.84 (km/s)^2 and the tabulated potential is -1944362.75 (km/s)^2. Therefore there is an offset of 1785307.88 (km/s)^2 which I assume to be due to the overall matter distribution in the box.
Then I calculate the sum of the potential energy off all DM particles in the halo, after removing the background offset, and I compare this total potential energy to the kinetic energy, which I calculate subtracting GroupVel from the tabulated particle velocities. This gives me a ratio of (Kinetic Energy)/(Potential Energy) = 0.07, that is quite far from what I would expect from a virialized halo (T= -V/2).
It seems all halos are in this state, where the kinetic energy is much lower than half of the potential energy from the hosting halo. What do you think about this?
Dear Illustris-TNG team,
I would like to calculate the total potential energy of gas particles in a Illustris-TNG300-1 halo, but without the contribution from particles external to the halo.
Is there a way to correct the Potential value in the Halo cuts to remove the contribution of particles external to the halo?
Thank you!
As you say, the
Potential
field in the snapshots is a global quantity, and there isn't any way to separate it into different contributions.If you need a local measurement of the potential, it would make sense to simply compute this directly (e.g. using only halo particles, one halo at a time - this is not too expensive).
I see, then in order to calculate the potential I have a couple of questions regarding softening lengths:
In Nelson et al. 2018 the DM/Stars softening for TNG-300-1 at z=0 is 1.48kpc, but in the rest API (https://www.tng-project.org/api/TNG300-1/) we have softening_stars_comoving=2.0. If I multiply this by hubble 0.6774 then I get 1.3548kpc. Where is the difference coming from?
In Pillepich+ (2018) there is a comment that the gravitational forces become Newtonian at 2.8 x epsilon. So I assume for distances greater then 2.8 x epsilon I should calculate the potential as plain Newtonian and not Plummer-Equivalent?
Also, in the rest API we have softening_stars_max_phys=1.0 but both values mentioned above are greater than this threshold. Is the threshold really enforced?
Thanks!
I would probably suggest to ignore softenings for the potential calculation, as a first step.
For Illustris and TNG the softenings are comoving until z=1, then fixed. So this value
2.0 ckpc/h
is the prior. Note that you divide by h, rather than multiply, to remove the hubble parameter from code lengths. So the box size of TNG100 is75 Mpc/h = 110 Mpc
. Then1.0 / 0.6774 = 1.48
.Hi Dylan,
So I have tried some quick tests with small halos, for example halo 1010 from TNG-300-1 snapshot 99 (z = 0 / a = 1)
I calculate the halo-only potential in the position of the less bounded dark matter particle (with id 7805765839). The halo-only potential is -159054.84 (km/s)^2 and the tabulated potential is -1944362.75 (km/s)^2. Therefore there is an offset of 1785307.88 (km/s)^2 which I assume to be due to the overall matter distribution in the box.
Then I calculate the sum of the potential energy off all DM particles in the halo, after removing the background offset, and I compare this total potential energy to the kinetic energy, which I calculate subtracting GroupVel from the tabulated particle velocities. This gives me a ratio of (Kinetic Energy)/(Potential Energy) = 0.07, that is quite far from what I would expect from a virialized halo (T= -V/2).
It seems all halos are in this state, where the kinetic energy is much lower than half of the potential energy from the hosting halo. What do you think about this?