doi:10.1038/nindia.2016.151 Published online 17 November 2016
A technique to better understand how galaxies recede and influence the ongoing expansion of the universe has been developed by astrophysicists. Using an advanced neural network algorithm, they have generated accurate data about the luminosity of distant galaxies that are moving apart at an ever increasing speed1.
When a light-emitting object moves away from a viewer, its light waves shift towards the red end of spectrum, making it appears to be red. Distant galaxies that are moving away from us also appear red. This phenomenon of recession is known as red shift.
Applying an algorithm to an online database of celestial objects known as Sloan Digital Sky Survey, researchers from the Presidency University in Kolkata and the National Institute of Technology in Durgapur divided more than twenty thousand galaxies into clusters. Each cluster has between 50 and 100 galaxies.
Unlike existing algorithms that ignore uncertainties, the new algorithm takes into account the uncertainties in the measurements, making it an efficient tool. It also uses separate neural networks for the members of each galaxy cluster.
Compared to existing methods, the algorithm provides much better data about the distant galaxies. It is a simple code that can be easily run in any computer.
This technique will be very useful for astronomers given the large luminosity data, as well as forthcoming surveys that would scan space to detect and study distant galaxies, say the researchers.
1. Samui, S. et al.Photo-z with CuBANz: an improved photometric redshift estimator using clustering aided back propagation neural network. New. Astron. 51, 169-177 (2016)