Overview

Pygadget is a python module for handling GADGET3 binary snapshots. Its goal is to provide a fast and intuitive way to access cosmological simulation data.

Requirements

• Numpy
• Pandas

Basic usage

Snapshots can be easily access using the Simulation class:

from pygadget import Simulation
snap = Simulation("filename")

Optionally one can signal the presence of blocks set in the makefile by passing one or more keyword arguments during object initialization. For example::

snap = Simulation("filename", pot=False, accel=False, endt=False, tstp=True)

After initialization the snapshot header information can be access as object attributes. For example:

snap.h

snap.omega_matter

snap.omega_lambda

snap.particle_numbers

For convenience a summary of the snapshot properties can be display by printing the Simulation class instance:

print(sim)

Reading blocks

For performance reasons blocks are only read on demand, for a specified particle type. The read_block() method is used for this task:

gas_pos = snap.read_block("pos", "gas")

Accepted keywords for block types are:

• "pos"
• "vel"
• "id"
• "mass"
• "u"
• "rho"
• "ne"
• "nh"
• "hsml"
• "sfr"
• "age"
• "metals"
• "pot"
• "accel"
• "endt"
• "tstp"
• "esn"
• "esncold"

And accepted keywords for particle types are:

• "gas"
• "halo"
• "disk"
• "buldge"
• "stars"
• "bndry"

Most block types return a single column Pandas Dataframe. The exceptions being "pos", "vel" and "accel" which return a dataframe with three columns ('x', 'y', 'z') representing Cartesian axis. And "metals" which return a DataFrame with 12 columns representing each of the chemical elements consider in the simulation.

Check the Pandas Intro to Data Structures for a primer on DataFrames manipulation and slicing.

For example

composition = snap.read_block("metals", "stars")
composition.He

Should return the helium mass content for every stellar particle in the snapshot. Similarly:

halo_pos = snap.read_block("pos", "halo")
pos.y

Should return the 'y' coordinate for every halo particle.

As blocks are index by particle ID, you should be careful on how you select rows. More information on the Idexing/Selection section of the Pandas documentation.

Subfind

The Subfind interface works very similarly to snapshots. The following code sets the subfind objects

sub = Subfind(basedir=dir, num=0, snap=snapshot)

where dir is the base directory for subfind folders, num the desired subfind folder and snapshot the associated Simulation object snapshot.

All subfind catalog properties can be access trough object attributes like

sub.nshubalos

sub.sublen

Subhalo particles can be read with

pos = sub.read_block_by_subhalo(block_type, particle_type, subhalo_number)

The output is the same as Simulation.read_block, but particles are filtered down to the ones in the required subhalo

A few extra methods are included

sub.optical_radius(subhalo, factor=0.83, rcut=30.0)

to calculate subhalos optical radius and

sub.mass_inside_radius(self, radius, subhalo, particle_keys=particle_keys)

to calculate the mass inside a give radius

Caveats

• No multi-file support yet
• Endianness was considered but is untested