A team including UC Berkeley researchers has found more than 1,200 potential gravitational lenses, celestial objects that have proven to be a powerful addition to astronomers’ toolkits and could help demystify the dark matter that most of the universe is composed of.

Gravitational lenses are astronomical phenomena where two galaxies or other large objects are aligned relative to a focal point, bending light through their gravitational fields in a way that creates multiple images of the galaxy, according to a press release by the Lawrence Berkeley National Laboratory, or Berkeley Lab. Since dark matter can only be observed indirectly through its gravitational effects, such lenses allow astrophysicists to track its quantity and distribution.

In May, the researchers found 355 lenses and expanded the search to more obscure parts of the data, such as lower-resolution imaging and galaxies of uncertain shape, according to Xiaosheng Huang, UCSF associate professor and lead author.

“(Gravitational lensing) was something that was hypothesized 60 years before it was ever seen,” said David Schlegel, a senior scientist in Berkeley Lab’s physics division who co-leads the related Dark Energy Spectroscopic Instrument, or DESI. “Einstein kind of hypothesized in the 1920s that we could see this effect … but now, we’re finding thousands of them.”

The images were discovered using machine learning techniques to scour an image survey of the universe taken for the DESI project, encompassing images that cover “about half the sky,” according to Schlegel. The team has already begun searching for more lenses in the survey’s newest release, according to Huang.

Huang said the team plans to chart the distribution and intensity of light being emitted by these lenses and higher-resolution images to determine whether these systems can truly be used as lenses. The team then plans to construct mathematical models for the systems in order to make “serious progress” toward determining the nature of dark matter, Huang added.

According to Schlegel, gravitational lenses have far broader applications than hunting for dark matter, and more lenses mean that researchers can be more selective in choosing which to use for modeling and better measure the scale and expansion of the universe.

UC Berkeley students played “significant roles” in these discoveries, according to Huang. Campus sophomore Saurav Banka provided human inspection of the lensing candidates suggested by the machine learning model to improve it. Banks said his participation in the research has inspired him to pursue a doctorate.

Campus junior Andi Gu has been with the project for a year and a half and works to develop techniques to improve the efficiency of the machine learning model.

“It’s very exciting — it’s definitely a privilege as an undergrad to get this unique opportunity,” Gu said. “My PI (Huang) has impressed upon me a sense of responsibility in that we need to make sure our results are thoroughly vetted before we put them out because they can have serious impacts on the direction of the field.”