Darkish Vitality Survey finds lots of of latest gravitational lenses
It is relatively rare for a magical object from fantasy stories to have an analog in real life. A really working crystal ball (or palantir) would be useful for everything from military operations to checking grandma. While there is nothing to be able to observe the mundane daily routines, there is something equivalent for extraordinarily distant galaxies: gravitational lenses. Now, a team led by Xiaosheng Huang from Lawrence Berkeley National Laboratory (LBNL) and several universities around the world has published a list of more than 1200 new candidates for gravitational lenses.
Gravitational lenses occur when a massive object, such as a galaxy, is aligned directly between Earth and another massive object that is even further away. Although usually attributed to Einstein, they were first published by Orest Khvolson and Frantisek Link in 1924 and 1936, respectively, and arise from general relativity. This theory assumes that the light from the source (i.e. the object further away) bends around the gravitational weight of the object in the foreground.
The “molten ring” is one of the most complete Einstein rings (a kind of gravitational lens) ever discovered.
Photo credit: NASA
The effects of this bending can be fascinating, like the molten ring we reported on late last year. So far, however, there have only been a few hundred examples of such light distortions in the universe. The discovery of 1,200 new potential candidates is a huge boost to the overall catalog of these fascinating phenomena and potentially doubles the number of candidates ever found.
Part of the reason they are “candidates” rather than confirmed lenses is because the work of discovering them was performed by a computer-generated algorithm. The researchers developed a neural network-based algorithm that was originally developed for the Strong Gravitational Lens Finding Challenge. Data from the Dark Energy Camera Legacy Survey (DECaLS) was fed into the winning algorithm to get the new catalog.
UT video explaining the concept of the gravitational lens.
As any computer scientist will tell you, training sets are one of the most important inputs to a neural network program. The students have dealt with this important aspect. Students helped sort tens of thousands of images and eventually developed 632 previously discovered lenses and 21,000 non-lens images to ensure the algorithm can distinguish between real and imitation lenses.
This meticulous collaboration between the work and the algorithm included 60 “Grade A” lenses with the most pronounced lens-like features. Another 105 candidates in class B (slightly less pronounced) and 176 candidates in class C (weaker characteristics) round off the total. To prove that these candidates are indeed gravitational lenses, the team secured observation time using the Hubble Space Telescope, which began in 2019.
A gravitational lens found in the DESI data of four different background galaxies that appear as partial rings around the orange galaxy in the foreground.
Photo credit: DESI Legacy Imaging Surveys, Berkeley Lab, DOE, KPNO, CTIO, NOIRLab, NSF, AURA)
Your final results with the Hubble data have yet to be published. However, the LBNL researchers aren’t the only ones using this technique to look for gravitational lenses. A team from Australia published another list of potential lenses, and the LBNL team only found about 60% of what that team did. The rest was likely due to differences in training sets or algorithmic quirks. Obviously, there is much more to discover in the world of gravitational lenses.
If the researchers could use a crystal ball to find them all, it would make their job a lot easier. However, they are limited to terrestrial technologies such as the Vera C Rubin Observatory, which is scheduled to start operating in 2023. With the added power of observation from these new platforms, the team hopes to reach 1,000 new lens candidates, which will greatly increase the total number of these phenomena found so far. A real crystal ball could probably also tell you whether or not it is capable of this.
Learn more:
LBNL: AI finds more than 1,200 candidates for gravitational lenses
LBNL: Seeing the Universe through New Lenses
arXiv: Discovery of new powerful gravitational lenses in the DESI Legacy Imaging Surveys
NOIR Lab: doubling the number of known gravitational lenses
Mission statement:
Gravitational lenses compared side by side with the DECaLS survey and Hubble’s picture of them.
Image Credit: Dark Energy Camera Legacy Survey, Hubble Space Telescope
Like this:
Loading…
Comments are closed.