The Ultimate Duo: Astrophysics and…Slime Mold?

Title: Filaments of the Slime Mold Cosmic Web and How They Affect Galaxy Evolution

Authors: Farhanul Hasan, Joseph N. Burchett1, Douglas Hellinger, Oskar Elek, Daisuke Nagai, S. M. Faber, Joel R. Primack, David C. Koo, Nir Mandelker, and Joanna Woo

First Author’s Institution: New Mexico State University, Las Cruces, New Mexico, USA

Status: Submitted to arXiv (preprint) and The Astrophysical Journal [open access]

The Structure of the Universe

Just like us, galaxies exist in the context of all that came before them. While we often think of them as isolated structures, floating in the black void of space, galaxies are actually a part of a grander structure: the so-called Cosmic Web. This is a structure where galaxies clump together in clusters and groups in the “nodes” of the web, and are connected by longer “filaments”, like the threads of a cobweb. We see this structure very clearly in simulations, but observing it directly is more complicated. A lot of the foundation of the web takes the form of dark matter (which we can’t see) and very diffuse gas (which is extremely difficult to see). 

Understanding the structure of the Cosmic Web often requires a level of reconstruction, where we take objects that are easier to find, like galaxies, and extrapolate about their surrounding environment. One of the best tools we have to test these methods are cosmological simulations, as they provide information about both the hidden and visible structure. In today’s paper, researchers used the IllustrisTNG simulation to model the location of filaments using algorithms trained on the locations of galaxies. 

Cue the Mold

A key element of this work was the comparison of two different algorithms: DTFE and MCPM. DTFE is a widely used tool that breaks the volume of the simulation into tetrahedra with galaxies placed at the vertices. The gradients of the density field this creates are then calculated, which can be used to outline the structure of the web. MCPM stands for Monte Carlo Physarum Machine, named after Physarum polycephalum (aka slime mold). Slime mold has been used in a range of fields due to its ability to create complex networks as it searches for food. The MCPM algorithm mimics the behavior of slime mold, with galaxies being treated like ‘food sources’ and larger galaxies corresponding to more food. Then ‘agents’ move towards the ‘food’ in discrete time steps as guided by several probability distributions. This creates a density field that can again be used to find filaments in the cosmic web. Today’s paper took the galaxies in Illustris TNG and used both DTFE and MCPM to map out the filamentary structure of the simulation.

Mold Beats Humans

So how did these two algorithms compare? Well we can test them by comparing their results to the dark matter distribution in Illustris TNG. Figure One shows the density distributions at a redshift of zero. 

The DTFE panel shows a lot of the tesselation inherent in the algorithm’s structure, and while it does a decent job in high density regions, MCPM is much more aligned with the dark matter structure, even down to low density filaments.

In Figure Two, the dark matter distribution from the simulation is overlaid with the actual filamentary structures identified by the algorithms. The filaments found by MCPM have a much higher persistence (essentially meaning their detection is more significant and robust) and MCPM just found more filaments than DTFE did. As apparent in both of these figures and through the statistical analysis in the paper, the MCPM did a much better job of mapping out the cosmic web than its more widely used counterpart. 

But Why Do We Care?

Slime mold mapping out the universe is fun, but it can also tell us some interesting things about how galaxies evolve. Because they were able to identify where filaments were, the research team could look at how galaxy properties were related to the distance to those filaments and their general environment. DTFE suggested some correlations, but with the ability of MCPM to identify more filaments a lot of the statistical dependencies on filament distance disappeared. However, some relationships persisted with both algorithms, such as satellite galaxies having a larger dependence on the environment than central galaxies. They also found that galaxies are more likely to be quenched (no longer actively forming stars) when close to high density filaments as identified by MCPM, although this result only held at low redshift, with the correlation fading in earlier simulation outputs. Combined, these results suggest that filaments may play a role in star formation over cosmic time, and powerful, accurate algorithms like MCPM are an important tool for these studies.

Astrobite edited by Karthik Yadavalli

Featured image credit: Skylar Grayson