This AI adjusts the parameters of our Universe

Researchers are utilizing artificial intelligence to gain a better understanding of our Universe. This new approach promises significant breakthroughs.

Artificial intelligence unveils hidden secrets in the distribution of galaxies. A team at the Flatiron Institute has developed a method called SimBIG, capable of extracting crucial information from universal data.


SimBIG harnesses galaxy data to refine estimates of five of the six cosmological parameters. The results far exceed conventional methods, reducing uncertainty by half for some parameters. This increased precision is especially valuable as new cosmic surveys are poised to commence.

One of the main advances of this method lies in its ability to analyze small-scale galaxy distributions. Traditional methods were limited to larger scales, overlooking crucial information that SimBIG now captures. This innovation paves the way for a better understanding of dark matter and dark energy.

To train their model, researchers used a suite of simulations, creating 2,000 virtual universes with various configurations. This approach enabled the model to grasp how variations in galaxy distribution reflect cosmological parameters.

Once trained, the model was applied to a sample of over 100,000 real galaxies, achieving results so precise they equate to analyzing four times as many galaxies using traditional methods. This outstanding performance could play a key role in resolving the "Hubble tension," a major issue in cosmology.


The Hubble tension arises from divergent measurements of the Hubble constant, which describes the rate of the Universe's expansion. SimBIG could provide answers by offering more accurate estimates and potentially revealing new physics related to dark energy.

These efforts involve an international collaboration, with researchers from the Flatiron Institute, Princeton, the University of Waterloo, and other renowned institutions.

What is the "Hubble tension"?

The "Hubble tension" is a significant problem in cosmology. It refers to the divergence in estimates of the Hubble constant, a value that measures how fast the Universe is expanding.

Different measurement methods yield conflicting results. Observations of distant supernovae and fluctuations in the oldest light of the Universe produce estimates that differ by 10%. This divergence could signal a gap in our understanding of the Universe or even the existence of new physics.

New methods like SimBIG could help resolve this tension by providing more precise measurements and testing different hypotheses to explain this discrepancy.