Visualizations of Galactic Formations During the Epoch of Reionization
Jennifer Marie Deger ABSTRACT In this paper, the reionization and structure formation of the cosmos is examined using state-of-the-art computation and visualization technology. The starting point for gathering particle data was the Cosmic Microwave Background, relic radiation left over from the Big Bang. The supercomputer Titan created the data and put it in a format that could be downloaded onto weaker machines, for our purposes. This data was then consolidated and manipulated so that the evolution of features of the cosmos could be seen on a small, detailed scale, which allows for close observations and analysis. The results yielded depict many examples of the innate relationship between ionization and energy density, as well as how these features are related to temperature and photon density. Using a stereoscopic display, dark matter collapse was simulated in 3D so that the details of the connection between gravity and energy, mainly how the gravity of dense regions affects structure formation by causing dark matter to collapse and form halos, could be seen from an advantageous perspective. Overall, these simulations allow us to better understand how the cosmos has evolved over billions of years in order to give rise to stars and planets capable of supporting life.

1.

Introduction

All of the energy that forms the universe was at one time concentrated at a single point. This extremely concentrated point of energy expanded in what is known as the Big Bang. Blumenthal et al. (1986) Hyperinflation caused the tiny inhomogeneities in the original concentrated point to be amplified on a massive scale. Therefore, even though the young, expanding universe was mostly homogenous, it did have regions that were slightly denser than others. The Cosmic Microwave Background is evidence of such irregularities: although it is smooth for the most part, there are irregularities in radiation strength significant enough to be noticeable, even if it is just barely so. Seljak & Zaldarriaga (1996) Some regions of the universe were bound to be denser than others due to the tiny fluctuations in uniformity on a quantum scale. The over-dense areas of the universe eventually collapsed under their own gravity, creating the first stars and ending the cosmic dark ages. Miralda-Escude et al. (2000) These stars provided the radiation energy necessary to reionize neutral hydrogen atoms. Reionization finished sweeping the universe about 1 billion years after the Big Bang. Gnedin (1999) By learning about reionization, we can reveal much more about our universes early conditions and how structure formation occurred. Because all of the matter and energy was at one time concentrated at a single point that rapidly inflated, the quantum-sized inconsistencies present in that point became blown up to a scale so enormous that they were able

–2– to play a role in our universe as seeds of structure formation. Umemura et al. (1993) Patches that were slightly denser will over time became much denser than the space surrounding them. A region that has just slightly more density will pull in more and more matter, resulting in a snowball effect. The end result is that we have walls, or portions of the universe rich in energy and matter, and voids, places that are relatively devoid of either. There are many regular, detectable trends when it comes to the early formation of cosmological structures. One of the best ways to identify and study these trends is to visualize different structure formations and find features they share and patterns they follow. There are clear relationships between temperature, photon density, and ionization of hydrogen atoms that can be detected in many different regions of the universe. I will discuss these in more detail in the first and second examples of this paper. By creating simulations of how these features of the universe have evolved over time, we can find out more about how exactly the universe became what it is today. The patchiness of reionization, caused by the formations of the first galaxies, is best analyzed as it evolves over time. By creating movies of the reionization of the universe from a redshift of 150 to a redshift of 4.2, we can see approximately a billion years worth of evolution of a small piece of our universe. Even these small pieces require some of the most powerful computing machines in the world. In addition to visualizing reionization, we created simulations of dark matter structure formation. By rotating around a cube that shows how structures have developed over time, we are able to see the evolution in detail and from many different perspectives at once. This allows for a more accurate analysis of the given data. Dark matter collapse is examined in more detail in the last two examples in this paper.

2.

Methodology

The data required to simulate large-scale structure formations during the Epoch of Reionization was acquired using numerical, gravitational N-body, hydrodynamics, and radiative transfer methods. Battaglia et al. (2013) Large supercomputers were needed in order to carry out these functions, as the sheer amount of data involved would overwhelm lesser machines. The supercomputer Titan at the Oak Ridge National Laboratory was used to create our simulations. The box we visualized has a volume of (64 Mpc/h)3 and contains 40963 N-body particles, or dark matter mass elements. Rosdahl et al. (2013) This particle data provided us with the ability to track energy density and how dark matter halos have evolved over time. In addition to examining density evolution, we tracked hydrogen gas evolution and the formation of stars from said gas. Hydrodynamical equations were used to compute the ionization state of the gas. In addition, star deaths were recorded in the data. This data was consolidated and put into a format that could be downloaded onto less powerful computers. We downloaded the data onto our local machines via the Texas Advanced Computing Centers Stampede supercomputer. Two ionization movies were created using an IDL script, one of the entire square of given data and one on a smaller, closer-up scale. For the dark matter visualization videos, which rotate about cubes as opposed to simply watching a slice of the

–3– cube evolve, more complex methods were used. We ran Splotch with several different scene generator files in IDL in order to determine color scheme, camera angles, and camera path. Although each scene generator file was applied to the same cube of data, the movies created by each appear very different from one another due to the differing colors and perspectives. These dark matter movies were formatted so that they could be viewed in 3D on a stereoscopic display. By running Splotch with an eye separation of 1.8, we created a left and right channel for each movie.

3. 3.1.

Results

Up-Close Ionization

Fig. 1.— The image on the right depicts the beginnings of ionization (bright spots have been ionized, dark regions are still neutral at this point in time). The image on the left depicts the same cut-out region after all hydrogen atoms have been ionized. These figures depict the ionization of a small sub-section of the data box that we analyzed at two different time points: at the very beginning of ionization and after ionization of the region is completed. The red areas represent the densest regions, followed by yellow, then aqua, and finally blue, where density is lowest. The bright spots represent regions that have been ionized. The image on the left depicts ionization in its early stages, when there were only a few observable patches. Notice how there are a few spots from which ionization spread. These spots also have the highest densities, as can be seen from the clumps of red and yellow structures. The dark portions of this cutout are not nearly as rich in density as the light portions. This is to be expected, as the radiation energy necessary for ionization originates from stars, which form where density is highest. The second depicts the same field of view, but after all of the neutral hydrogen atoms have been ionized. Not only is the entire cutout now ionized, but the density structures have evolved so

–4– that there is less patchiness and larger, more pronounced patches throughout. The gravitational pull of dark matter has created super-dense regions, such as the vertical red streak that can be seen in the center of the second image. As energy and matter cannot be created or destroyed, these super-dense regions had to gain more mass from somewhere else. They did not just fabricate their own mass from nothing. Notice how in the first image, there are many spots of slightly dense regions throughout the cut-out. These spots are mostly yellow, with a few red centers here and there. Although they are numerous, because they are small they do not have any kind of significant gravitational pull. Instead of pulling on other matter, they are drawn in by neighboring larger clumps, such as the aforementioned vertically-running streak seen in this simulation. The second image has far fewer small clumps floating around than the first image because over time, the densest areas have sucked in any matter within the reach of their gravitational pulls. The empty areas are much more expansive in the second image because the small groups of matter have been pulled elsewhere. Over time, the density-poor areas get poorer and the density-rich areas get richer. In the second image, the ring of energy left over from a supernova is visible. This star death occurred because of the gravitational collapse of a massive star. The release of huge amount of energy left an imprint that is traceable in simulations such as this one. In order for a supernova to occur, there has to be a very extreme concentration of energy. The part of the cutout where this supernova took place also happens to be one of the parts from which ionization originated. This is no coincidence. It would be expected that the most energy-dense regions not only would catalyze ionization with their radiation effects but would also be the most likely regions in which supernova would occur. Energy is a requirement both for ionization and for the formation and death of massive stars. Therefore, it is unsurprising that the same portion of our cutout was responsible for both the origins of ionization and for a stars explosion.

3.2.

Wholebox Ionization

The following images depict the same cut-out region at a redshift of 4.2, or approximately 1.5 billion years after the Big Bang. There is an evident relationship between ionization, photon density, and temperature. Notice how there is an over-dense region, or bright red spot, in the bottom left-hand corner of the ionization box. In the same place is a region of low photon density, denoted by a light orange spot in the bottom left-hand corner of the photon density box. Although this is only one spot, it provides evidence for the rest of the cut-outs following the same pattern. The relationship is most easy to detect in this one highly-characterized spot. One can reasonably assume that the pattern is repeated throughout the box. It makes sense that a region with a high density of energy and matter would have a low density of photons because regions such as this would be ionized more quickly than less dense regions. Since the hydrogen atoms have been separated into protons and electrons in these regions, they have more free particles. More free particles means more scattering of light, which affects photon density.

–5–

Fig. 2.— This figure depicts a cut-out region that has fully undergone ionization. The red regions are the densest, while the darker blue regions are the least dense.

Fig. 3.— The image on the left depicts the photon density of the same cut-out region. The image on the right depicts temperature. There is also a clear connection between dark matter density and temperature than can be observed with these visualizations. Temperature is essentially a measure of kinetic energy, so it would be expected that regions with greater energy density would give off higher temperatures. This trend is apparent in the cut-outs above. For example, observe the aforementioned bright red spot in the bottom left-hand corner of the ionization box. There is a region of higher than average temperatures visible in the temperature box as a bright red spot as well. The laws of physics also support this data because dense regions would be expected to produce more radiation. This radiation is what caused ionization, so it is no surprise that the same areas were the first to be ionized.

–6– 3.3. Dark Matter Condensation

Fig. 4.— These images depict the same cube of particle data at the very beginning of structure collapse, contrasted with the same cube after structure collapse has taken place.

Fig. 5.— These images depict a cube of particle data at the very beginning of structure collapse, contrasted with the cube after structure collapse has taken place. This is the same cube as was illustrated in the previous pair of figures, but with the brightness turned up and in a different color scheme as to gain a more well-rounded perspective of the structures. These images depict the dark matter condensation in a particular cube. The first images are

–7– of an earlier redshift than the second images. Structure formation is the primary subject of interest when studying visualizations such as these. The density distribution is far more homogenous at an earlier redshift. The web of dark matter distribution in the first image is more evenly spread out. If the density of any portion was to be measured, it would likely be close to the average density of the entire region. The second image, on the other hand, depicts a very inhomogeneous distribution. There are many dark matter halos, connected by dense filaments. These density-rich portions sharply contrast with the dark voids surrounding them. Taking the density at different cookie-cutter portions at this redshift would yield wildly different results depending on exactly where the density was measured. Although the image on the right is mostly uniform, especially in comparison with the image on the left, the careful observer will notice that there are in fact tiny inhomogeneities. These portions are just slightly denser than the space around them. Over time, they pull in more and more dark matter until they form the halos and filamentary structures visible in the second image. The pancaking effect of gravity is responsible for the filamentary nature of these structures.

3.4.

Dark Matter Condensation: A More Personal View

Fig. 6.— This image is taken from the perspective of inside the previously examined cube, as to give a closer look at how the filamentary structures and dark matter halos formed.

This image is a still-frame taken from a 3D movie where the camera flew through the box while the dark matter was evolving in time. Notice the bright spot in the upper half of the frame,

–8– and how it seems to be drawing other smaller bright spots toward it. This is a merging of dark matter halos. As described earlier, the densest regions are becoming increasingly denser by drawing in particles continuously. The space around these regions becomes empty, as can be seen by the pitch-black gapes around the edges of the frame. There are many other smaller, less pronounced mergers taking place throughout the cube. The filamentary structure throughout it evidence of this. The closer perspective pictured here provides an especially valuable observation of structure evolution

4.

Conclusion

The simulations created in this project accurately depict structure formation in the young universe according to what we know about physics and cosmology. The history of structure formation and the ionization of hydrogen atoms was visualized in such a way that we can further confirm that we are on the right track when it comes to understanding the origin of the cosmos. In the first example of ionization taking place in a smaller and more detailed cut-out, we saw how dark matter condensed and how this gravitational collapse is related to the ionization of hydrogen. The densest regions showed a tendency to ionize sooner than the emptier regions, and also to pull in matter from these emptier regions, thereby becoming greater and greater in mass. Sometimes they gained so much energy that structures collapsed in on themselves, as seen in the supernova in the first example. The second example depicted a relationship between temperature, photon density and ionization. The trends predicted by what we know about physical science hold true for the simulations in that regions of higher temperature and lower photon density tended to be the densest portions of the data box and also tended to ionize first. The third example provided a very beautiful visualization of how dark matter collapses in on itself to create dark matter halos and how dense regions become denser while empty regions become emptier. These visualizations also illustrated the filamentary structure of dark matter in the universe. Finally, the fourth example depicted a more close-up view of how the gravity from a dark matter halo pulls in particles from the space around it, creating walls and voids. All of these examples are related to one another in intrinsic ways, and their relationships tell us more about how the universe has evolved over billions of years in order to give rise to planets capable of supporting life forms such as human beings. One confounding factor that is hard to avoid when examining any aspect of the universe is that it is very difficult to take a large enough sample size to say for certain that the data examined is an accurate representation of what is going on in the rest of our expansive universe. The data sets taken in this project were small when compared to the entire cosmos, yet they required the most advanced supercomputers in the world to gather. The nature of mankind is to assume that we have the power to understand the universe around us, but in truth we can never begin to understand our origins or the origins of the universe. Any findings about reality are skewed by our narrow perspectives, our egos, and the knowledge that we are going to die. For thousands of years, all of humankind was convinced that the cosmos revolved around the earth. For all we know, we could

–9– be just as wrong about the universe now. The best we can do is use our primitive technology to try and understand how and why we exist. If nothing else, it is grounding to learn just how small our planet is. The priorities of our day-to-day lives are fleeting and insignificant beyond our mortal reckoning. However, this does not mean that we are unimportant or that our lives are meaningless. It is up to the individual to add as much personal meaning to his brief life as possible. Understanding the cosmos is certainly a noble pursuit, and one worthy of a lifetime of devotion. The intention to add to mankinds limited knowledge about the universe is exceedingly honorable. And if nothing else its cool as hell so we might as well try while we are alive.

REFERENCES Battaglia, N., Trac, H., Cen, R., & Loeb, A. 2013, ApJ, 776, 81 Blumenthal, G., Faber, S., Flores, R., & Primack, J. 1986, ApJ, 301, 27 Gnedin, N. 1999, Astronomical Society of the Pacific, 193, 598 Miralda-Escude, J., Haehnelt, M., & Rees, M. 2000, ApJ, 530 Rosdahl, J., Blaizot, J., Aubert, D., Stranex, T., & Teyssier, R. 2013, MNRAS, 436, 2188 Seljak, U., & Zaldarriaga, M. 1996, ApJ, 469, 437 Umemura, M., Loeb, A., & Turner, E. 1993, ApJ, 419

A This preprint was prepared with the AAS L TEX macros v5.2.