What is dark matter?

pie_blueOver the last two decades, astronomy has experienced an exponential growth of data that has revolutionized our understanding of the universe. Namely, we accurately know its composition now: about 70% dark energy, 25% dark matter (one or many hitherto unknown particles), and only 5% of “normal” matter (or atoms, such as ourselves). We infer the existence of dark matter from its gravitational pull on the normal matter, but besides gravity, dark matter interacts with atoms and light very weakly, or not at all.

Simulating the invisible

Given that we cannot see dark matter, what do we know about its distribution in the Universe today? First, relic light from the big bang (called the cosmic microwave background) tells us that all matter was initially distributed evenly throughout the universe, but with tiny ripples: patches of slight over- and under-density. Second, we know that dark matter more or less only feels the force of gravity.

Combining our knowledge of the initial conditions in the Universe and the presumed physics of dark matter, we can simulate its evolution throughout the age of the Universe. We simplify the picture further by pretending that the atoms behave like dark matter. The simulations follow a virtual grid of a large number of virtual dark matter particles (a billion in this case) whose initial positions on the grid are perturbed to mimic the density ripples. We now run the simulation forward in time as shown in this video:


As the simulation speeds through the roughly 14 billion years since the Big Bang, we see how gravity causes the dark matter to collapse into a “cosmic web” of walls, filaments, and dense balls called “halos”. These halos are particularly important, as they are the sites where galaxies such as our Milky Way form. The video below zooms in on the formation of a single, very massive halo (using a different color scheme):


More images and videos of dark matter simulations can be found on Benedikt’s website. The simulations used here are described in much greater detail in this science paper. The simulations were run using the Gadget2 code by Volker Springel. The simulation images and videos shown on this website were produced using Phil Mansfield’s GoTetra code.