Dark Matter Halo Tracking and Merger History Trees in Cosmology simulations

In cosmology, dark matter is a currently unknown type of matter theorized to account for a large part of the total mass in the universe. Estimated to constitute 83% of the matter in the universe, dark matter neither emits nor absorbs light or other electromagnetic radiation, and thus cannot be directly seen with telescopes. The only way of detecting dark matter is by observing the effect of its gravitational forces on stars and galaxies. Scattered in the form of clumps or in cosmological terms dark matter halos, the gravitational forces exerted by halos aid in the formation of clusters of galaxies. To understand the physics behind the formation of these structures and the galaxies within, cosmologists have to accurately study the evolution of dark matter halos over time and space. Understanding the behaviour of dark matter holds the key to unravelling numerous questions pertaining to galaxies and the whole universe. 

 

 

 

Cosmo

Figure 1: Hierarchy of the Universe (Please click image to enlarge)

The goal in computational cosmology is to simulate the universe down to each of the individual galaxies. Since the physics of galaxy formation is not known in detail, cosmologists perform a gravity- only simulation using super computers. The simulation formulated to compute the dark matter distribution covers a massive volume of (256 h-1Mpc)3 or (8.35x108 light-years)3 of the observable universe and evolves 256^3 dark matter tracer particles. These simulations determine the distribution of dark matter throughout the observable universe. The dark matter halos are then extracted from the simulations and galaxies are populated within these halos using various methods. Approximately 100 dark matter tracer particles can host a galaxy. Instead of the simpler Halo Occupation Distribution method, cosmologists can also use a more sophisticated method that takes into account the evolution history of these halos. The formation time of the halo (old or young) including events such as merging of halos, influences the galaxy population within. These models are checked by comparing their results with observations from survey telescopes which map the visible universe.

 

dark matter halo
Figure 2: A Dark Matter Halo with multiple subhalos (different colors)

Dark matter halos have a complicated substructure and they are dominated by primary halos known as satellite halos as shown in figure 2. Gravitational forces exerted by halos and satellite halos can give rise to various dynamic multi-level interactions. These satellite halos can merge with other satellite halos within the same host halo. They may also change their host halos by transferring from one host halo to a different host halo. In this work we apply a tracking algorithm to follow the evolution of each satellite halo and host halo from their birth until they either merge, split or die.

 

tracking halos
Figure 3: Tracking and visualizing the merger of two dark matter halos.

 

We track halos, satellite halos and individual dark matter tracer particles as the halos and satellite halos emerge, die, merge and continue to exist over time. This group tracking model is an extension to the feature tracking framework and also represents the evolution of halos and satellite halos in the form of a merger tree. We visualize the merger trees in 3D along with statistical halo information to provide a detailed overview of structural evolution.
 
To understand the physics behind galaxy evolution, cosmologists have to accurately know the evolution of the dark matter halos in terms of hierarchical merging at each time step. The hierarchy of merger of halos can be represented by merger trees. A Merger trees thus describes the sequence in which halos merge and grow. Every node represents a halo and an edge connected to the node tells about the halo’s descendents and ancestors. Traditional visualization of halo merger trees is achieved by representing halos in the form of circles. In this poster we extend the traditional representation by visualizing halos in 3D, thus providing useful particle distribution and velocity information. Using the output from the tracking modules, we classify whether a satellite halo or host halo is born, dies, splits, merges or continues. Depending on the classification, we update the merger tree for that specific satellite halo or host halo. The final result is thousand of merger trees for halos that exist at the last time step. Cosmologists can query for interesting statistical quantities and study the evolution of different halos by analyzing the merger tree. 
 
merger tree
Figure 4: Visualizing the merger tree of two dark matter halo merging.
 
Figure 4 shows a merger tree for the merging process of halo H69 and halo H134. The merger tree in figure 4(e) show the births of H69 and halo H134 along with continuation of halo H69. After a merger process between two structures is complete, the resultant merged structure retains the halo_Id or satellite_halo_Id of the structure with the smaller Id between the two participating structures. In figure 4(e), since halo H69 has the smaller Id, the merger leads to a resultant halo with halo_Id and color of H69. In figure 4(f), we also represent interactions of the inner satellite halos in the form of merger trees.
 
The merger trees will be useful for detailed analysis of halo evolution. Once the tracking is over we can visualize the merger tree of any halo. We studied different ways of visualizing the merger trees so as to convey maximum information to the cosmologists without overwhelming them. Click on the following link to know more about visualizing the merger trees:
Visualizing Halo Merger Trees
 
The merger tree data will also be used as an input to the software for galaxy distribution analysis, GALACTICUS. This implementation helps cosmologists understand the evolution and formation of galaxies with detailed information about their components and properties.

Some future directions for the merger trees: Merger Tree Visualization & Distributed Computing

 
We are grateful to the support of DOE #DE-FG02-09ER25977. This work is part of the SciDAC Institute for Ultra Scale Visualization.

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer