The actual goal of this project, to find gravitational lensing evidence of a small dark matter halo from a database with tens of thousands of observations, is pretty ordinary as astrophysics goes, and it doesn't yet have any definitive results. Still, it is a worthwhile project that adds incrementally to what we know in a well focused way to expand the margins of our existing knowledge.
It illustrates the reality that the many modern telescopes in multiple frequencies that are now being used can collect a vast amounts of information. But this has made a lot of questions in astrophysics and astronomy "big data" problems.
With a smaller database, a single skilled research could personally review each one. This painstaking pouring over of data by a single highly trained scientist with the PhD in the relevant subfield of astronomy is how this kind of research got started. But it is impossible for a single astronomer to conduct the necessary fairly detailed analysis of each observation required for this kind of study, for such a large collection of data, in a reasonable amount of time. But timely analysis is necessary because the amount of data to review gets larger every month.
The firehose of incoming data is only getting stronger. For example, a new European Space Agency project targeted for the year 2045 will collect information on 10 to 12 billion new sources of light in the sky that are too faint to discern now.
The citizen science methodology used in this study is remarkable and exciting. It presents an alternative to statistical, machine learning, and supercomputing approaches to sorting through masses of data. Unlike these automated alternatives, this citizen science approach doesn't sacrifice the human judgment element of the process present when a single scientist analyzes a large, but tractable body of data.
In this case, twenty people, about a quarter of whom were scientists, about quarter of whom were graduate students, and about half of whom were undergraduates, mostly at the University of Crete, worked together to tackle the large dataset to identify 40 strong candidates out of 13,828 (many of which have multiple images at different wave lengths that had to be considered) including two particularly promising needles in the haystack.
It is a kind of project I am familiar with from my day job as an attorney, where, for example, I've had to mobilize similar numbers of people with similar skill levels, to review an entire room full of banker's boxes of not very well organized hard copy business records to locate a handful of key documents in complex securities fraud litigation.
The way this project managed to mobilize so many people to volunteer their time for this somewhat esoteric goal, hearteningly democratized this scientific endeavor and made this task possible to complete.
The paper and its abstract are as follows:
Dark Matter (DM) halos with masses below , which would help to discriminate between DM models, may be detected through their gravitational effect on distant sources. The same applies to primordial black holes, considered as an alternative scenario to DM particle models. However, there is still no evidence for the existence of such objects.
With the aim of finding compact objects in the mass range 10 -- 10, we search for strong gravitational lenses on milli (mas)-arcseconds scales (< 150 mas). For our search, we used the Astrogeo VLBI FITS image database -- the largest publicly available database, containing multi-frequency VLBI data of 13828 individual sources.
We used the citizen science approach to visually inspect all sources in all available frequencies in search for images with multiple compact components on mas-scales. At the final stage, sources were excluded based on the surface brightness preservation criterion. We obtained a sample of 40 sources that passed all steps and therefore are judged to be milli-arcsecond lens candidates.
These sources are currently followed-up with on-going European VLBI Network (EVN) observations at 5 and 22 GHz. Based on spectral index measurements, we suggest that two of our candidates have a higher probability to be associated with gravitational lenses.
C. Casadio, et al., "SMILE: Search for MIlli Lenses" arXiv: 2017.06896 (July 14, 2021) (accepted for publication).