You are here

Massive galaxy cluster discovery sheds new light on how the largest structures in our universe are formed

Spotlight On:

Arif Babul
University of Victoria
Physics and Astronomy

Galaxy clusters are the largest objects in our universe. With masses comparable to a million-billion suns, they contain as many as 1,000 galaxies, vast amounts of dark matter, gargantuan black holes and X-ray-emitting gas that reaches over a million degrees Kelvin.

A dense concentration of galaxies is causing astrophysicists to question our understanding of how structures form in the universe. A study published in the journal Nature reports on a collection of galaxies observed when the universe was about one-tenth of its present age.

This particular galaxy cluster is so far away it takes 12.4 billion years for light from it to reach Earth. So while it was first observed last year, it was seen as it existed 1.4 billion years after the Big Bang—allowing astronomers a rare peek into galaxies forming in the early history of the universe.

A team of researchers from institutions including the University of Victoria (UVic), the National Research Council of Canada, Dalhousie University, the University of Illinois, and the Flatiron Institute, have discovered a cluster of galaxies assembling very differently and more quickly than current models would predict. The nascent cluster is made up of at least 14 galaxies packed into a region of space just three times the size of our Milky Way galaxy and each is forming stars up to 1,000 times faster than ours is forming them. Calculations indicate that the astronomers are witnessing the assembly of one of the most massive structures in the present-day universe.

“Having caught a massive galaxy cluster, and especially its gigantic central galaxy, in the throes of formation is spectacular in and of itself. But, the fact that this is happening so early in the history of the universe and in so dramatic a fashion poses a formidable challenge to our understanding of how and when structures in the universe form," said Arif Babul, an astrophysicist at the University of Victoria, and a co-author of the study. "Conventional wisdom suggests that the central cluster galaxy is assembled in dribs and drabs spread over 13 billion years, by cannibalizing smaller galaxies that venture too close. This discovery appears to upend this picture.”

The study was co-led by Scott Chapman, Research Officer at the National Research Council of Canada and Professor at Dalhousie University, and Tim Miller, as a Master’s student at Dalhousie University. It was co-authored by a team of scientists including Babul, who is a UVic Distinguished Professor; astrophysics doctoral student, Douglas Rennehan; UVic advanced research computing specialist, Belaid Moa; University of Illinois Professor, Joaquin Vieira; and Christopher Hayward, Associate Research Scientist at the Centre for Computation Astrophysics, Flatiron Institute.

To predict how the cluster will assemble, a team led by Babul used WestGrid and Compute Canada supercomputers to develop a numerical simulation. It showed that the core of the cluster will eventually combine into one giant galaxy, over the next billion years. Scientists anticipate that eventually this will prove to be one of the most massive clusters in the current-day universe.


This cluster system, originally discovered by the South Pole Telescope, was studied in detail using the Atacama Large Millimetre-submillimetre Array (ALMA). ALMA has the world’s most advanced radio astronomy detectors, developed by the National Research Council of Canada, which provided the detail necessary for the discovery of the system.

(Profile provided by the National Research Council of Canada. To read the full media release, click here.) 


Behind The Project: A Q&A with Belaid Moa

Belaid Moa, an Advanced Research Computing Specialist at UVic, is a member of the WestGrid / Compute Canada support team. He works with researchers both at UVic and across the country, connecting them with advanced research computing resources and services. Belaid constantly has projects on the go, whether it's meeting with users one-on-one, hosting training workshops to build researchers' computational skills, or helping users get the most from Compute Canada's national systems. In 2015, he received the Outstanding Achievement award through Compute Canada's Awards of Excellence prorgram, recognizing his passion and dedication for supporting advanced research computing.

Below is a short Q&A with Belaid about his work with Arif Babul and his team on their latest discovery. 

*    *    *   *    *    *   *    *    *   *    *    *   *    *    *   *    *    *   *    *    *     

Q: What was the main challenge facing Arif and his team? Why did they need to access WestGrid / Compute Canada systems?

Prof. Arif Babul and his team are in constant need of high performance resources to carry out their cosmological simulations focusing on the formation and evolution of structure — galaxies, galaxy groups and clusters of galaxies — in the Universe post-Big Bang. Because of the physical processes involved, these simulations span a huge dynamic range in space and time.

Moreover, they need to follow the evolution of both dark matter, the gravitationally dominant but otherwise invisible matter component, as well as baryons. Treating the baryons is particularly complicated. They are subject to numerous physical effects — radiative cooling, shock heating, turbulence, etc. — and can transform into stars and black holes, which themselves can impact their surroundings via radiation and powerful winds. Consequently, the simulations require intensive parallel computing.

A typical run can require more than 1,024 cores for two months or more to produce the required results, and detailed simulations can require several thousands of cores. These simulations generate large amounts of data and so the team also performs a lot of post-processing analytics that require intensive summarization and visualization. All of these require powerful clusters, which WestGrid/Compute Canada offer.

Arif's team relies heavily of these systems and, in fact, they have held large allocations on WestGrid and Compute Canada systems for the last six years. In the case of proto-cluster, we had a hard deadline challenge as we needed to generate the simulation and visualization in a very short time. We utilized 512 cores on the new national system Cedar to produce the simulation and visualization results on time.

Q: How did you help Arif and his team? 

I have been assisting Arif's team through one-to-one sessions with the team members since 2012. I helped his postdocs and PhD students optimise their massively parallel cosmological simulation codes as well as implement an increasing array of physical effects so that they can run state-of-the-art simulations efficiently across multiple nodes on WestGrid / Compute Canada resources.

Sometimes their work must account for things like the characteristics of the computing platform, the available compilers, and other system features. For instance, porting from UVic's Nestor cluster to the Cedar system at Simon Fraser University posed a fair bit of challenge because scripts, etc. needed to be revised to match to the Cedar environment. Other times, we need to restructure parts of the code to make it more efficient and stable. The codes truly push the envelope. Rapid deposition of energy following the death of a star can easily trigger large-scale instabilities if not managed properly. Identifying the range of physical processes that ought to be part of any realistic simulation is definitely a challenge but implementing these in the code without compromising the high degree of parallelization is itself important.

Most recently, I have been working with Douglas Rennehan (Ph.D. Candidate) to treat the effects of turbulence in the simulations via large eddy approximation. I have also helped to develop the post-processing pipelines, which also must run in parallel mode, as well as adapt pre-existing codes for generating initial conditions to the available computing environment. The latter often involves understanding the structure of the code. The proto-cluster project dominated my one-to-one sessions with Douglas for nearly two months. During that time, I assisted with setting up the Gizmo software stack, helping to install and optimise the different post-processing and visualization tools (SUNRISE, yt-project, GALSTEP, etc.) and tutored Douglas in the use of Paraview and Visit for the visualization. We even modified the SPH filter in Paraview to work with our data. After setting up the simulations, we started with low resolution runs for quick testing and then moved to the large simulations and ran them on 512 cores.

We faced a number of challenges at the beginning. For instance, at one point, the simulations would abruptly die without rhyme or reason. It took some creative sleuthing to figure out why this was happening but I managed to fix the issue. My colleagues at Compute Canada and WestGrid, especially the Cedar admins, were invaluable in helping to tweak Cedar and its environment into a stable state so that we could run without major interruptions.

Q: Would this discovery have been possible without access to WestGrid / Compute Canada? 

The simulations that Professor Babul and his team carry out would not be possible without WestGrid / Compute Canada. In all likelihood, they would have had to join a foreign team as junior partners to be able to do what they are doing. With WestGrid / Compute Canada resources, they have their own leading-edge program and when they do collaborate, they do so as equal partners.

Access to WestGrid / Compute Canada resources was especially critical for the proto-cluster project. It allowed Douglas Rennehan to take the lead in carrying out the first simulation of the system. Given the hard deadlines and the amount of processing and post-processing involved, it would have been difficult to have the simulation and visualization done on time. Within the two months of running the simulations on Cedar, we produced all the necessary results.

Q: What did you enjoy most about working on this project?

The best part of working with Arif and his team is that they treat me not as a Tier 3 support person but as a valuable member of the team. And of course, their work is fascinating. The very idea of being able to replicate how the Universe has unfolded over the past 13 billion years on a supercomputer, and have it be right in a broad brush sense, is mind blowing. In the case of the proto-cluster project, the close relationship I have with the Arif's team was critical to do the necessary work in a very short period of time on a relatively brand new, powerful cluster. And the end result — compelling simulations and visualizations — have truly helped enhance the novelty of the observations.

The discovery itself is fascinating and I enjoyed being part of it. And the fact that the discovery, as accentuated by the movie made from the simulation outputs illustrates, challenges the conventional understanding makes it all the more intriguing — a riddle that I played a part, a small part but nonetheless a part, in bringing to light.


*    *    *   *    *    *   *    *    *   *    *    *   *    *    *   *    *    *   *    *    *   *   *    *    *   *    *    *   *   *    *    *   *    *    *   *   *    *    *   

Arif Babul NRC UVic galaxy discovery   

Image Caption

Computer-simulated image of the 14 galaxies packed in a region only three times the size of our Milky Way galaxy, as seen in the Atacama Large Millimeter Array (ALMA) observations. This unique cosmic conjunction is on the verge of coalescing into a massive galaxy cluster core only 1.4 billion years after the Big Bang (see the simulation movie for details). Credits: D. Rennehan, B. Moa, C Hayward / UVic, WestGrid, Compute Canada, Flatiron Institute.

Other background