Despite the major success modern Cosmology has made when it comes to predicting and explaining observed cosmic phenomena, there are still many questions left unanswered in the field. Understanding the growth and evolution of large-scale structure (LSS) formation, which we see as cosmic webs and voids, is one such big question. Due to the influence of dark matter (DM) and dark energy (DE), explaining LSS is not an easy task, especially due to the fact that there is no current consensus as to what exactly DM and DE are and what their properties may be. To study the evolution of LSS quantitatively, cosmologists have invoked the help of mock Universe simulation to help understand the evolution of matter distribution within the Universe. A limitation that comes with these simulations is the vast amount of resources needed to run them efficiently. When running these simulations, there are two main factors cosmologists try to optimize: simulation size and resolution, both of which come at the cost of the other. Higher-resolution simulations typically require smaller simulation sizes, while larger simulation sizes have to be run at lower resolution. If one would want a large simulation with high resolution, the amount of resources (run time, computer memory, money, etc.) needed increases exponentially. As a way to work around these limitations, ongoing research is being conducted to determine if Dark Matter Only Simulations (DMOS) are accurate enough for studying LSS. The appeal of this being DMOS allows for larger simulation sizes with higher resolution while requiring fewer resources to run. The main idea of this research is to see if DMOS data can be corrected to make them as accurate as Full Physics Simulations (FPS).

There are many properties of dark matter halos we can get from simulations, some of which include halo masses, position, and velocity along with things such as halo luminosity. We can also compute things like the power spectrum of dark matter halo distribution in space. For this project, the variable of interest is the halo mass. From analyzing FPS and DMOS simulation data sets, provided by the IllustrisTNG team (https://www.tng-project.org/), I was able to determine that halo masses produced by DMOS were on average around 15% higher than those produced by FPS. Along with this (and assuming the mass discrepancy is constant across other simulations at higher redshifts) I was able to find a correction formula (see Figure 1; the angle theta is the angle between the line of one-to-one correspondence and the line of actual correspondence)

Fig 1: Correction Formula: D is a DMOS data point and theta is the offset variable Made by Weston Schwartz, July 13th 2023

to apply to DMOS that would correct for the higher than expected masses (see figures 2 and 3; the magenta point is to track the overall transformation to the data set).

Following this, we later began analyzing data from the AbacusSummit simulation suite (https://abacussummit.readthedocs.io/en/latest/) to determine the impact of using this correction formula; the analyzed simulation set used the same cosmology (Plank 2018 ΛCDM) and same redshift (z=0.1)) as was used for IllustrisTNG). The appeal of using the data produced by the AbacusSummit team is that all of the simulations ran were DMOS only. Because this was a DMOS set, in order to test whether or not our correction formula is successful we compare the two-point correlation functions computed from data before and after correction (see Figure 4),

Fig. 4: Correlation Function of pre and post mass modification: Correlation function determines the probaility of finding a galaxy some distance away from another. The correlation function using modified masses is in strong comparison to observational results

as opposed to directly comparing the DMOS halo masses to the FPS halo masses as in the IllustrisTNG analysis. The results of applying this correction formula indicates that when applied to simulations that assume Plank 2018 ΛCDM cosmology at the redshift z=0.1 we can successfully reproduce halo masses generated by DMOS that are comparable to those produced by FPS.

Future work that is needed in order to conclude these methods of mass correction are viable for future simulations is to apply them to other simulations at different redshift values and for different cosmological models. Currently, this correction formula assumes a constant scaling factor, regardless of simulation type and/or redshift. More work is needed to verify whether or not this assumption is true. Furthermore, if shown to be true, more work is needed to determine the cause of consistently higher masses; If shown to not be true, then more work is needed to determine how this scaling factor changes between simulations and/or redshift. Along with determining how consistent this mass correction is, future work will also involve finding methods of correcting other data types associated with halos (position, velocity, luminosity, etc.).

This project was conducted by Weston Schwartz under the supervision and direction of Kansas State Physics professor, Dr. Lado Samushia.

 

Acknolwedgments

National Science Foundation-- K-State Physics Department and Faculty-- Kim Coy, Dr. Loren Greenman, Dr. J.T. Laverty, Dr. Lado Samushia

Final Presentation