The first LensZoo project preview: beat the robots of the CFHT Legacy Survey!

Anupreeta More, Surhud More and Phil Marshall

Gravitational lensing is a spectacular phenomenon found in the Universe.  Predicted by Fritz Zwicky in the 1930’s, galaxies and clusters of galaxies acting as lenses are not just beautiful to look at but they also have plethora of applications, including revealing the whereabouts of the elusive Dark Matter. Gravitational lenses are rare objects since we require the foreground and background galaxies to be aligned on the sky to within a few thousandths of a degree.

Over the coming decade, larger and larger imaging surveys will map out ever wider and deeper regions of the Universe. This means we should be able to find more gravitational lenses, but it also means that we will have increasing amounts of data to inspect in order to find them. As a result, we would like to automate the process of finding gravitational lens systems from these vast treasure troves of data. However, as you know, discovering gravitational lens systems requires some skill, and the lens candidates need to satisfy a varied set of criteria before they can be tagged as promising lens systems. Our brains are more suited to carry out such tasks than are simple computer algorithms, so it makes sense for humans to look at the candidates that the robots flag as interesting. However, so far, astronomers have had difficulties in building robots that are capable of finding all the different kinds of lens systems that are potentially interesting. This is partly because we have not yet discovered very many lenses, nor exhaustively cataloged all the things that look like lenses but are not lenses in reality. To understand how to make the robots work better, we need to jump in to the data alongside them!

Our first project in the Lens Zoo is going to be a slightly unusual one, in that it’s focus will be on beating a lens finding robot, rather than checking through its outputs. We are going to use the optical and infrared data from the Canada-France-Hawaii Telescope Legacy Survey (CFHTLS) for this project. With the help of computer algorithms called “ArcFinder” and “RingFinder”, we have found a sample of lens candidates from the CFHTLS – but we know that these algorithms don’t do a very complete job. Opportunity knocks!  We would like the citizen scientists of the Lens Zoo to help us search the images of the CFHTLS to discover a variety of lenses which were missed by our robots.

The CFHTLS spans an area of about 170 square degrees of sky. It’s images are both higher resolution than those of the SDSS, with median seeing in the i’ band of around 0.7 arcsec, and deeper (i’ < 24.5 magnitudes) – which means that more gravitational lenses should be visible per square degree. The picture on the left shows a lens from CFHTLS that the ArcFinder did spot, a small galaxy group that is lensing a background star-forming blue galaxy. On the right is the SDSS image of this system, to show you the difference in image quality.








We have two goals for this project. First, we want to find all the gravitational lenses that the aforementioned algorithms missed – perhaps because the sources are quasars, or distant red galaxies, or because the lenses are complex, or confusing. Second, we want to catalog all the objects that look like lenses, but are not: these “false positives” will make an important training set for us to test our improved robots on.  This will be the first time that this survey’s images will have been exhaustively inspected – so there are bound to be some surprises!


6 responses to “The first LensZoo project preview: beat the robots of the CFHT Legacy Survey!”

  1. Jean Tate says :

    So, how do we do that (“search the images of the CFHTLS to discover a variety of lenses which were missed by our robots”)?

  2. drphilmarshall says :

    In short: by examining images! 🙂 We have some things to figure out – like how big a patch of sky to show at a time, how to stretch, scale and colour the image to make the lensed features stand out, how to teach new users what lenses look like as quickly as possible, and so on – but the basic zoo concept, of citizen scientists inspecting images, works well for us lens hunting. We’ll blog more about the above challenges in the coming weeks!

    • Jean Tate says :

      I actually meant, where can I – and other zooites – go to start examining images? From your reply, I gather the answer to that question is something like “nowhere … yet, because we haven’t got the system up and running yet, but stay tuned!”

      Jean (Staying Tuned) Tate

  3. drphilmarshall says :

    Yes, you’ve got it: the site is being built this autumn, so we’re thinking about how it should work now. Please do stay tuned – it’s likely you can help us get the design of the new zoo right!

  4. Thomas says :

    Great to see things moving along 🙂

  5. mocne strony www says :

    Itts like you rread my mind! You seewm to know
    a lot about this, like you wrote the book
    in it or something. I think that you could do with a few pics to drivce tthe message home a little
    bit, but instead of that, this iss great blog.
    An excellent read. I’ll certainly be back.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: