Argonne National Lab Simulation Tracks the Evolution of the Universe

Scientists at Argonne National Laboratory recently ran one of the most complex simulations of the evolution of the universe ever created. The purpose: To try and understand how the universe came to be and, in particular, to understand the mysterious influence of dark energy and dark matter—which makes up some 95 percent of everything—on its development.

The lead scientist on the project, Katrin Heitmann, joins us to talk about the big science being conducted just outside of Chicago.


"Chicago Tonight" spoke with Heitmann Monday afternoon by phone, prior to the on-air interview. Below, some highlights from the discussion.

What is the scientific value of doing the kinds of cosmological simulations that you do?

"First of all, what we want to understand in cosmology is the evolution of the universe—how it got to where it is today—as well as the make-up of the universe—what’s in it. In order to do that we are running very large surveys that basically map out the distribution of galaxies across the sky in a 3-D map. We have a certain understanding about the universe. We have understanding about the initial conditions and we have an understanding of how it evolved and an understanding of its make-up. What we want to do now with these simulations is exactly create this universe in our lab. So we build this model and we put it on a computer and evolve it forward, and now we have created a universe that we can look at and compare it to the real data."

Where are you drawing the raw data from to create the most realistic simulation?

"It’s a combination of theory and observations. Obviously, we have Einstein’s theory of relativity as a starting point, but we also have the observations so that we can make sure that what we get out of the computer simulations makes sense."

What period of time are you covering when you simulate the evolution of the universe?

"The universe’s age is roughly 13.7 billion years. We start at a couple of billion years after the Big Bang."

The simulation below shows a snapshot of the universe when it was 7.4 billion years old:


Understanding dark energy/dark matter is at the cutting edge of much scientific research. How much do we know about it and how do you factor it into your simulation?

"We don’t know what the origin of dark energy and dark matter actually is, but we have theories that point to something that we call the cosmological constant that was theorized by Einstein. Basically, what the dark energy does is change the expansion history of the universe and this expansion history is explicitly embedded in the code. So what I can do is actually impose a more complicated model than just the cosmological constant. What the upcoming simulations are trying to do is to see if there is a deviation from this very simple model that we have right now."

If we had been looking for dark matter/dark energy 10 billion years ago we wouldn’t have even known it was there. Does that mean there was some event that created it at that point or just that it wouldn’t have been detectable?

"That’s exactly the question that we want to understand. What we are mainly interested in is to follow gravity. On the scale that we are interested in, gravity is the major force that is important. So that’s a major part of our simulation, to simulate how matter evolves under the force of gravity."

When we see structures form in these simulations, if we were to zoom in to the highest level of detail, what would we see?

"In these simulations what we want to see is the distribution of galaxies and what you see in the simulations is actually the dark matter distribution. You see a lot of clumpiness in the simulations. The smallest clumps you would see hold a galaxy, but you do not see the galaxy itself in it. The problem is that if you wanted to do a simulation where you actually have galaxies form, then we can’t do it in as big a volume. So the simulations where you see galaxies and star formations are of a much smaller volume."

In the second simulation, a merger is depicted, triggering new star formation:


These simulations are run on supercomputers. In order of magnitude, compared to a PC, how much more powerful are the supercomputers that you are using?

"Basically, the Mira supercomputer at Argonne has 50,000 nodes and each of the nodes has roughly the computing power of a laptop. You could write a code for your laptop and actually run a simulation of a little universe on it, but the problem is that you want to analyze a much bigger volume of data than you could on your laptop. So you need to connect all these laptops and now you have to think about how one laptop talks to another—that’s the hard part about the supercomputing business."

Having done the simulation, how long will you be analyzing the data to see what you can learn from it?

"The analysis of these simulations normally lasts a few years, and whatever we learn from them, we use in the next simulation. Usually when you look at them you see things that you didn’t expect and you come up with new science programs based on that."


Related stories:

Physicist Lisa Randall Connects Dark Matter to Dinosaur Extinction

Particle physicist and New York Times bestselling author Lisa Randall joined us to discuss her new book "Dark Matter and the Dinosaurs: The Astounding Interconnectedness of the Universe."


New Brain Initiative Aims to Fully Map the Human Brain

An ambitious new government-led research initiative aims to fully map the human brain. Two leading neuroscientists join us to talk about this potentially groundbreaking project.