Hot massive stars are the kitchens of the Universe, where heavy elements are cooked up. They end their lives in supernova explosions. To understand their recipe, astronomers use spectroscopy — the analysis of the rainbow band of light passed through a prism. Previously, deciphering such a 'menu' took hours.
Now the SATURN program does it instantly. The neural network is trained on millions of simulated spectra and, like an expert taster, determines the temperature, metal content, and rotation speed of a star from the 'flavor' of light. Each dark line in the rainbow is a 'note' of a specific element. Remarkably, the network has never seen real stars — it was trained on computer models, but now it handles real data just as confidently.
With new sky surveys collecting spectra of millions of stars, manual analysis is impossible. Automated analysis will reveal how elements, such as carbon, spread through the Galaxy, enriching future planets.
The star Zeta Ophiuchi rotates so rapidly that a full spin takes less time than an average workday — just a few hours (the Sun completes a rotation in a month).