It’s good to Talk! Why we need to hear from you to find Space Warps
Some of you may be wondering what happens to an image after you hit “Next” and why “Talk”ing about your lens candidates is important, so here’s a brief explanation!
WHAT HAPPENS TO THE IMAGES YOU DON’T MARK?
Each night, we retire images from the pool based on your collective classifications. If the community together says no (i.e. by enough people not placing a marker on the image), we throw out the image so that we can focus your classifications on fresh data and images that might contain gravitational lenses. After only five weeks, you guys have made an astonishing 5.2 million classifications. This means we’ve been able to already reject about 60% of the total CFHT Legacy Survey as not containing gravitational lenses!
WHAT HAPPENS TO THE IMAGES YOU DO MARK?
When you mark an image two things happen. First, we record your mark in our database so we can compare it with what other people thought. Second, that image is automatically saved into your Talk profile under a collection called “My Candidates”. Talk allows you to discuss your interesting candidates with the rest of the Space Warps community. It’s great to see so many discussions happening there already, so please keep talking! Talking in Space Warps is an essential part of refining the list of plausible candidates, which is explained next.
HERE’S HOW YOU CAN HELP
As we work our way through the images, it looks as though we are going to end up with a sample of a few thousand lens candidates from your markings. That’s great – it means Space Warps is a very effective filter! But a few thousand is still several times more than the number of actual lenses we expect – so we’ll need to investigate the images of the candidates further before presenting them to the rest of the astronomy community. This is where you, and Talk, can really help us out!
BECOME CURATORS OF YOUR LENS COLLECTIONS
If you see a lens candidate, either when browsing Talk, or while you are marking, that you would like to see investigated further, make a “collection” called ‘Probable Lens Candidates’ and add this object to it! Remember, you can also add images you think are the most likely lens candidates from your automatically filled ‘My Candidates’ collection. Then, later on, you might do some further investigation of the images in your collection – or someone else in the collaboration might do it, after browsing your collections. Either way, collecting the candidates is the first step. You can start a discussion about any candidate or collection any time, and ask the Space Warps community to share their thoughts.
WHAT HAVE YOU FOUND SO FAR?
We’ve just started looking at the most commonly marked images, and there are some promising candidates already being discussed in Talk. Some of these are previously known lens candidates: as you may know, the CFHT Legacy Survey has been searched using automated computer algorithms. We’ve started to label the candidates from those searches in Talk, you’ll see the label “Known Lens Candidate” at the bottom right of the image in the individual object page of Talk. As well as the labels, Budgieye has done a phenomenal job in collecting known CFHT-LS lens candidates from the research literature in a dedicated discussion board. Much like the tricky simulations, some of these known candidates may be difficult to spot.
Most excitingly, some of you have started discussing a few lens candidates that we think have been missed by the algorithms – watch this space for a special post about these potential new lens candidates next week!!!
HOW TO GET STARTED IN TALK
If you want some top tips on using Talk, please visit the discussion board (thanks Budgieye!)
Thanks again for your phenomenal work – and let’s get Talking!!!
Space Warps 
Hi Aprajita,
Thanks very much for all your and team work.
It’s good to follow your advice to clean up our candidates, even if simulations are put in another “harvest bin”.
As an example I classified 44000+ images having 3200+ potential lenses of which 224 candidates, as of today. I know I’ve missed may be few hundreds simulations!
I selected probable worth 32 and began looking at filters fits files!
Space Warps has the potential to end up with three or four hundred good lenses in addition to the Anupreeta’s list of 127 published. Sadly that total seems too much for a spectroscopic follow-up, isn’t it?
To go further let us know when you’ll give us the astrometry reduction for mining CFHTLS data.
Cheers
c_cld
Hi Claude
Thanks to you – Wow, 44000+ images, that’s unbelievable! Do you know that that’s about the size of one of our ten dataset releases that you’ve gone through as a single user. Others that may be reading this comment should not be discouraged from looking at a much smaller number. Even inspecting 20-40 images by any user is very valuable information for us.
Going back to your point Claude, if we do end up with three to four hundred highly probable candidates that would be great, but I expect the number to be closer to ~200. Having said that, yes spectroscopic follow-up of such a large sample would be difficult, but not impossible if spread over many facilities and several years. In addition to the sheer number, we also need to consider the brightness of the lenses and the images of the back ground galaxy to check feasibility of the spectroscopic follow-up. Some source will simply be too faint to confirm in a reasonable time. Imaging data is much more sensitive than spectroscopy.
Our goal, by all of us talking, is to discuss and then whittle down the list of plausible candidates to the most probable. Before we do any follow-up we also need to check the feasibility of the proposed lens system through modelling. This means constructing a model that recreates the lensed images for a given lens. While spectroscopically following-up ~few hundres sources will be difficult, modelling of order than number could be feasible. I think we will start a discussion on candidate modelling soon, so please watch out for that.
Many thanks,
Aprajita.
“… we throw out the image” – I sure hope not!
Please tell me that you are not destroying the hard-won imaging data; rather, you are removing the data which is the particular images in which SW zooites have found no lens candidates from the SW database (the image data is preserved, and presumably available for others to use for research of their own), right?
We simply remove these images from the classification stream for Space Warps. All of the CFHT-LS data will always exist for anyone to look at.