Using artificial intelligence to chart the universe

Scientists routinely use large telescopes to scan the sky, mapping the coordinates and estimating the distances of hundreds of thousands of galaxies and so enabling scientists to map the large-scale structure of the Universe. But the distribution they see is intriguing and hard to explain, with galaxies forming a complex ‘cosmic web’ showing clusters, filaments connecting them, and large empty regions in between.

The driving force for such a rich structure is gravitation. Around 5 percent of the cosmos appears to be made of ‘normal’ matter that makes up the stars, planets, dust and gas we can see and around 23 percent is made up of invisible ‘dark’ matter. The largest component, some 72 percent of the cosmos, is made up of a mysterious ‘dark energy’ thought to be responsible for accelerating the expansion of the Universe. This Lambda Cold Dark Matter (LCDM) model for the universe was the starting point for the work of the Potsdam team.

Measurements of the residual heat from the Big Bang – the so-called Cosmic Microwave Background Radiation or CMBR – allow astronomers to determine the motion of the Local Group, the cluster of galaxies that includes the Milky Way, the galaxy we live in. Astronomers try to reconcile this motion with that predicted by the distribution of matter around us, but this is compromised by the difficulty of mapping the dark matter in the same region.

“Finding the dark matter distribution corresponding to a galaxy catalogue is like trying to make a geographical map of Europe from a satellite image during the night which only shows the light coming from dense populated areas”, says Dr Kitaura.

His new algorithm is based on artificial intelligence (AI). It starts with the fluctuations in the density of the universe seen in the CMBR, then models the way that matter collapses into today’s galaxies over the subsequent 13700 million years. The results of the AI algorithm are a close fit to the observed distribution and motion of galaxies.

Dr Kitaura comments, “Our precise calculations show that the direction of motion and 80 percent of the speed of the galaxies that make up the Local Group can be explained by the gravitational forces that arise from matter up to 370 million light years away. In comparison the Andromeda Galaxy, the largest member of the Local Group, is a mere 2.5 million light years distant so we are seeing how the distribution of matter at great distances affects galaxies much closer to home.

Our results are also in close agreement with the predictions of the LCDM model. To explain the rest of the 20 percent of the speed, we need to consider the influence of matter up to about 460 million light years away, but at the moment the data are less reliable at such a large distance.

Despite this caveat, our model is a big step forward. With the help of AI, we can now model the universe around us with unprecedented accuracy and study how the largest structures in the cosmos came into being.”

Since 2011 Francisco Kitaura has been working at the AIP. His publication is available online on http://arxiv.org/abs/1205.5560and will soon be published in Monthly Notices of the Royal Astronomical Society (MNRAS).