Constant offset compared to linear theory when computing initial conditions power spectrum
Rui Lan Zhang
18 Jun
Using the coordinates of the dark matter particles in the TNG300-2 initial conditions file obtained from this page (https://www.tng-project.org/data/docs/background/#sec1),
I tried computing the power spectrum using nbodykit. However, when I compare the result to the Eisenstein&Hu power spectrum (z=127), the one computed from the TNG initial conditions is approximately 5% lower. Do you know why this might be?
When I use the same procedure to compute the power spectrum of dark matter particles from the snapshot data then the results agree well with the Eisenstein&Hu power spectra at the corresponding redshifts.
Dylan Nelson
19 Jun
The particles in the initial conditions are "total matter particles", not DM particles (as in the snapshots). Does this matter?
Rui Lan Zhang
19 Jun
Actually, when I compute the power spectrum for the snapshots I use the dark versions. In this case the particles in the initial conditions and the snapshot should match?
Dylan Nelson
20 Jun
The initial conditions for the "Dark" and "baryonic" simulations, of the same box+resolution, are exactly the same.
Perhaps the answer is simply that the TNG transfer function is not equal to the Eisenstein & Hu. The TNG transfer function is based on a CAMB simulation (this is described in some detail in Pillepich et al. 2018).
Rui Lan Zhang
21 Jun
Would you be able to share the power spectrum that was used for the initial conditions, so that I can compare with the one I computed?
Thank you very much for sharing the power spectrum data.
Please may I ask what the normalisation is for this power spectrum?
The amplitude seems a lot higher than the power spectrum I computed from the dark matter particles.
Dylan Nelson
29 Jun
This is an output from CAMB, I expect it is in the standard normalization of CAMB?
Using the coordinates of the dark matter particles in the TNG300-2 initial conditions file obtained from this page (https://www.tng-project.org/data/docs/background/#sec1),
I tried computing the power spectrum using nbodykit. However, when I compare the result to the Eisenstein&Hu power spectrum (z=127), the one computed from the TNG initial conditions is approximately 5% lower. Do you know why this might be?
When I use the same procedure to compute the power spectrum of dark matter particles from the snapshot data then the results agree well with the Eisenstein&Hu power spectra at the corresponding redshifts.
The particles in the initial conditions are "total matter particles", not DM particles (as in the snapshots). Does this matter?
Actually, when I compute the power spectrum for the snapshots I use the dark versions. In this case the particles in the initial conditions and the snapshot should match?
The initial conditions for the "Dark" and "baryonic" simulations, of the same box+resolution, are exactly the same.
Perhaps the answer is simply that the TNG transfer function is not equal to the Eisenstein & Hu. The TNG transfer function is based on a CAMB simulation (this is described in some detail in Pillepich et al. 2018).
Would you be able to share the power spectrum that was used for the initial conditions, so that I can compare with the one I computed?
I've placed it (the input to N-GenIC) at https://www.tng-project.org/files/data/input_spectrum_PLANCK15.txt.
Thank you very much for sharing the power spectrum data.
Please may I ask what the normalisation is for this power spectrum?
The amplitude seems a lot higher than the power spectrum I computed from the dark matter particles.
This is an output from CAMB, I expect it is in the standard normalization of CAMB?