HEIDELBERG, Germany – How do you build a galaxy? For centuries, astronomers have studied these glittering clouds of stars through telescopes. Today, a new generation of scientists is attempting something even more ambitious: creating them from scratch inside supercomputers. This September, hundreds of the world's leading theoretical astrophysicists will converge in Heidelberg for "Building Galaxies from Scratch II," a pivotal conference focused on the monumental challenge of simulating the birth and evolution of galaxies.
The goal is as grand as it is complex: to start with the recipe of the early universe—a hot, nearly smooth soup of particles after the Big Bang—and write computer code so powerful and detailed that it can accurately spin up the majestic spirals and glowing ellipticals we see today. Researchers describe this as one of the most significant challenges in all of theoretical astrophysics, requiring them to bridge gaps of billions of years in time and millions of light-years in space, all within a virtual box.
"These aren't just video game graphics," explains Dr. Andrew Pontzen, one of the conference's invited speakers. "We are trying to encode the laws of physics—gravity, the behavior of gases, the life cycles of stars—and see if they naturally give rise to a universe that looks like our own. It's the ultimate test of our understanding of cosmology."
The Foundation
All modern cosmological simulations begin with a common foundation known as the Lambda-CDM (ΛCDM) model, often called the standard model of big bang cosmology. This model posits that the cosmos is composed of three primary ingredients: ordinary matter (the stuff of stars, planets, and us), cold dark matter (an invisible substance that shapes the universe through gravity), and dark energy (a mysterious force driving the accelerated expansion of space).
Simulations start by seeding a virtual volume of space with tiny density fluctuations, consistent with the ΛCDM model and observations of the cosmic microwave background. Then, they press "play," letting supercomputers calculate how gravity amplifies those seeds over billions of simulated years. Dark matter clumps first into a cosmic web of filaments and halos. The ordinary gas, drawn into these halos, cools, condenses, and eventually forms stars.
The Devil in the Details: Why It's So Hard
The simple act of gravity pulling gas together is just the beginning. The real difficulty, and the core focus of the Heidelberg conference, lies in the intricate "baryonic physics", the complex behavior of the normal matter that lights up the sky.
"A galaxy. It's a dynamic ecosystem," says Dr. Maria Werhahn, another conference speaker. " Stars are born, they shine, they blow powerful winds, and they die in colossal supernova explosions. Supermassive black holes at the centers of galaxies swallow material and fire out unimaginable amounts of energy. All of this 'feedback' acts like a galactic regulator, heating gas, pushing it around, and shutting down star formation."
Capturing these interconnected processes, from the explosive death of a single star to the sweeping effect of a black hole's jets across an entire galaxy, is the frontier of the field. Researchers are now working to add even more layers of realism, including the roles of magnetic fields and cosmic rays.
Connecting the Virtual to the Real
The true test of a simulation is how well its virtual galaxies match real ones. The field has moved far beyond just comparing shapes. Today, scientists use a technique called "synthetic observation," in which they process the raw data from their simulations to create what a virtual galaxy would look like through a real telescope, such as the Hubble or the James Webb Space Telescope (JWST).
A cutting-edge study published in January 2026, for example, used a high-resolution simulation of a galaxy similar to the well-known "Phantom Galaxy," NGC 628, to understand how the tangled filaments of gas and dust seen by JWST directly lead to the formation of star clusters. This direct, one-to-one comparison with iconic JWST images is revolutionizing the field.
"Before, we might compare general statistics. Now, we can literally point to a filament in a JWST image and find its counterpart in our simulation," says Dr. Ralph Pudritz, a co-author of the NGC 628 study. "This allows us to reverse-engineer the physics that must be happening."
Looking to the Future
The "Building Galaxies from Scratch II" conference will also serve as a workshop for the next generation of cosmic simulation. With even more ambitious projects on the horizon, scientists are turning to innovative tools.
One promising area is machine learning. Teams are now training artificial intelligence on vast suites of simulations to help calibrate the complex feedback models or to analyze the overwhelming amount of data these virtual universes produce. Another innovative technique is "genetic modification," where scientists subtly tweak the initial conditions of a simulated galaxy to run controlled experiments, asking questions like, "What if this galaxy had experienced one less collision?"
The discussions in Heidelberg will shape the road for these future simulations. The goal is no longer just to create galaxies that look plausible, but to build predictive digital universes, tools so reliable they can guide astronomers to new discoveries about the real cosmos and help understand the remaining mysteries of dark matter and dark energy.
As the field stands on the brink of this new era, the conference embodies a profound shift from simply observing the universe to actively engineering it in silicon, in a relentless quest to understand our own origins.
Source:
https://www.sciencealert.com/ambitious-survey-hints-at-tantalizing-new-theories-on-dark-energy
https://iopscience.iop.org/article/10.3847/1538-4357/ae285a
https://wwwmpa.mpa-garching.mpg.de/conf/bugs2026/index.html