An international team led by researchers from the Institute of Cosmos Sciences of the University of Barcelona (ICCUB) has developed a new method that could significantly improve our understanding of the expansion of the Universe and the nature of dark energy. The work has been published today in Nature Astronomy.
The study presents a powerful framework called CIGaRS that allows scientists to extract much more information from exploding stars known as Type Ia supernovae, using mainly images rather than expensive spectroscopic observations. The results pave the way for making the most of the enormous amount of data expected from the next generation of astronomical surveys, especially the Vera C. Rubin Observatory.
Why supernovae matter for understanding the Universe
Type Ia supernovae are the explosive deaths of white dwarf stars. Because they tend to explode with almost the same intrinsic brightness, astronomers use them as “standard candles”: by comparing how bright they really are with how bright they look from Earth, scientists can measure cosmic distances.
This technique played a key role in the discovery that the expansion of the Universe is accelerating, a phenomenon attributed to dark energy, one of the biggest mysteries in modern physics.
However, there is a catch: not all Type Ia supernovae are exactly the same.
The problem: supernovae are affected by their environments
Over the last two decades, astronomers have found that the brightness of these supernovae depends slightly on the galaxies in which they explode. For example, supernovae in more massive or older galaxies tend to look a bit different from those in smaller or younger ones.
Until now, these effects have usually been corrected using simple, approximate adjustments. This can limit how precisely we can measure the distances to these supernovae.
A unified solution: modelling everything together
The new study tackles this problem head-on by modelling everything at once: the supernova explosions, the galaxies hosting them, dust that dims and reddens their light, how often supernovae occur over cosmic time and even the expansion of the Universe itself.
Instead of analysing each piece separately, the researchers built a single, self-consistent model that links all these elements physically and statistically.
“A powerful way of modeling the Universe is to simulate it ab initio in the computer using bayesian inference,” explains Raúl Jiménez (ICREA-ICCUB), co-author of the study. “This provides a way to vary all possible parameters at the same time to predict what Universe we live in. Furthermore, by having this capacity one can look into possible “unknown unknown” systematics to understand their effect. The impact of these systematics in our inference is arguably the most important missing ingredient in current approaches to model the Universe.”
Artificial intelligence meets cosmology
To make this ambitious approach computationally feasible, the team used a modern set of techniques known as simulation-based inference.
In simple terms, the method works like this:
- Scientists simulate many possible universes using physical models.
- A neural network (a type of artificial intelligence) learns how simulated data relate to the underlying physical parameters.
- The trained system can then infer those parameters directly from real observations.
This allows the analysis of tens of thousands of supernovae at once, something that would be impossible with traditional methods.
A key result: precise distances without spectroscopy
One of the most important outcomes is that the method can estimate galaxy distances (redshifts) very accurately using only images (Redshift is a measure of how much the light from a galaxy is stretched as the Universe expands. It tells us how far away, and how long ago, we are seeing it).
The new approach achieves a precision comparable to spectroscopic measurements, but without needing spectra. This is crucial because future sky surveys will discover millions of supernova candidates, while only a small fraction can realistically be studied with spectroscopy.
Preparing for the Rubin Observatory era
The Vera C. Rubin Observatory, currently under construction in Chile, will soon begin a ten-year survey of the sky, detecting an unprecedented number of supernovae. Around 99% of them will be observed only photometrically, meaning through images in different colours.
The CIGaRS framework is designed precisely for this scenario.
“Unlike other frameworks, which require analytic simplifications, our no-compromise end-to-end simulation-based inference approach is uniquely capable of extracting the full cosmological and astrophysical information from the Rubin Observatory's hard-earned data, while avoiding the pitfalls of selection and modelling biases.” says Konstantin Karchev (ICCUB-SISSA Trieste), lead author of the study.
Beyond cosmology: learning how stars explode
In addition to improving measurements of dark energy, the study also sheds light on how and when Type Ia supernovae form. By reconstructing how supernova rates depend on the ages of stars in galaxies, the model helps address long-standing questions about their progenitor systems.
The results show that combining physics-based modelling with artificial intelligence can overcome key limitations in current cosmological analyses. According to the authors, this approach could improve cosmological constraints by up to a factor of four compared to traditional methods that rely only on a small, spectroscopically observed subset of supernovae.
With the Rubin Observatory set to transform astronomy in the coming years, methods like CIGaRS ensure that we will be ready to fully understand the data and the Universe it reveals.
Reference:
Karchev, K., Trotta, R. & Jiménez, R. CIGaRS I: Combined simulation-based inference from Type Ia supernovae and host photometry. Nature Astronomy (2026). https://doi.org/10.1038/s41550-026-02842-5
