Comparision Project of Cosmological Numerical Codes:

Hydro and shocks properties. 

Vazza,F., Dolag K., Ryu D., Brunetti G., Gheller C., Kang H., C.Pfrommer

Read the full paper here (Vazza et al.2011 MNRAS).


 More data, figures and updated resources can be found at

The goal of this project is to investigate how much the prediction of the shock front distributions (in strength and volume filling) by cosmological, hydrodynamic Al simulations depend on the underlying numerical methods. Therefore we will investigate 3 different cosmological hydro codes ( TVD-code, Gadget-2 and ENZO). We want to understand how much they differ already in their predictions of the low density, low temperature gas, as well as how the different ability of resolving low/high density regions and the structure within them influences the shock distribution. We also want to test internal versus different external shock detection methods (e.g. based on Density/Temperature versus velocity jumps).  

The Comparison Project focouses onto 2 different volumes:

Initial conditions

To follow a resonably resolved cosmological box with a massive cluster we setup a box of size 100 Mpc/h with a intentionally large value of sigma8 : Theri is a set of initial conditions produced by Dongsu Ryu for 64³, 128³, 256³ and 512³ dark matter particles, inclusive some description on the data format available at

Here you can find some theoretical predictions of (cumulative) massfunctions for the standard LCDM model (produced by K.Dolag) and the model addapte here.
First column contains log10(M) in units of Msol/h, second column has the prediction of cumulative mass function log10(N) using Press & Schechter in units of 1/(Mpc/h)³ and third column the same using Sheth & Tormen.

Here is an idl program to create Gadget initial conditions from the above dark matter particle positions, (K.Dolag)

The ICs for the gadget runs and the simulation data can be found on clx at CINECA under /clx/userhpe/hpedeue7/Data/ShockComp/



Dark matter particles distribution


Shown are maps of projections of dm particles form different resolutions and volumes.
Blue points are for the  64³ run, red for the 128³ one.

DM particle distribution
central slice of 100Mpc/h
30Mpc/h 'carrot'
64³128³  64³128³ 

Shown are maps of projections of dm particles form the different code: red=Gadget2,
blue=Gadget2 (grid-like IC), green=ENZO.

Mass functions:
good convergence just for M>5 x 10^13 M_sol/h clusters.

Gas distribution

Color Maps:

Shown are maps of projections/slices  for different resolutions. From left to right 64³,128³,256³ and 512³
Enzo is on top, TVD is in the center and SPH is at the bottom.

Gas density and gas temperature distributions

At the best resolution, most of the volume is converged in term of density, but
it is not converging in temperature.

Individual Clusters: entropy priofiles

While gas density and gas temperature are similar for all codes, gas entropy is not.
TVD and ENZO show a peak in entropy at 2R_vir, while GADGET shows a smooth transition.
This is true for all clusters and all resolutions.

Shock Waves

Color Maps:

Shocks were detected using the most suitable algorithm for every code:
Shown are maps of projections/slices  for different resolutions. From left to right 64³,128³,256³ and 512³,
at the Top is TVD, GADGET is in between and ENZO is at the bottom
The color scale for Mach number is shown in the figure (in log10(Mach)).

Energy Flux distribution for shocks:

The bulk of energy release is similar.
However the different code/methods different place.
TVD/ENZO show different trend compared to GADGET in the low density
regions of external shocks (M>50).

go home