Machine-readable tables of the source properties in Tables 3 and 4 can be found here. They will be kept up-to-date as we acquire more redshifts or sources, etc.
After a long time, I’m glad to finally get to share some results that have taken up a lot of my time for the last couple years. I’ve posted before about our group’s first efforts to model the effects of gravitational lensing for the sample of SPT objects using high-resolution ALMA imaging. The whole sample of 47 galaxies is now done!
What makes doing this modeling a bit difficult is that, unlike most optical telescopes, ALMA and other interferometers don’t actually observe an image of the sky. What you end up getting are many Fourier components of the sky, which you can invert to get an image. The problem with those images, though, is that every pixel is correlated with every other pixel (a single pair of antennas doesn’t just contribute at one location in the image, it contributes to all pixels in the image). This means that, unless you’re careful, the models you create and the uncertainties you derive can be wrong.
Modeling the visibilities directly solves this, and it also lets you incorporate a lot of the effects that can corrupt the data you get. For example, in the data we used for this paper, observations of each source were only a minute or two long. This means that there wasn’t enough time for a lot of atmospheric effects to average out (there wasn’t enough time for the wind to blow a patch of air across the entire array of dishes). Our models are able to account for this because we model at the visibilities directly.
Enough talk, time for pictures! Like I’ve said, the modeling is pretty sophisticated. Above this is one of my favorite galaxies, a source magnified by a group of several galaxies simultaneously. You didn’t hear it here, but I strongly advocate nicknaming this thing the Cosmic Clownface.
After doing all of these models, we end up with a lot of information about what our sources look like intrinsically – sizes, brightnesses, shapes, etc. Does the fact that we’re looking at lensed sources bias us in some way? For example, it’s well-known that sources which are intrinsically smaller can be magnified by a larger amount (this is because the region of high magnification is usually pretty small, near caustics). Most samples are still pretty small, but it looks like, as far as sizes are concerned, we don’t seem biased. At the right is a figure comparing the distribution of sizes of lensed objects from our sample and a comparable sample discovered by the Herschel satellite with some distributions of sizes of unlensed objects. As you can see, there’s not much difference. We speculate (who doesn’t love speculating?) that this might mean that the intrinsic size distribution of these galaxies is pretty narrow – narrow enough that the lensing size bias is hard to pick out. We’ll really only know this once more galaxies have been observed.
One cool aspect of having been part of a really extensive follow-up campaign over the last few years is that we have a lot of other information for these objects. In particular, we looked awhile ago at the [CII] fine structure line in many of our objects to try to understand why this line gets fainter relative to the dust emission when you look at very highly star forming objects. There’s been some hints in the literature that the number that actually matters isn’t the luminosity of the dust emission, but how compact that emission is. This was based mostly on a sample of very nearby star-forming galaxies observed with Herschel. Now that we know the intrinsic sizes of our objects, we can see if that relation still holds in the distant universe. Spoiler alert: It does! Not only does the [CII]/dust ratio still correlate with the compactness of the dust emitting region in our galaxies, the high-redshift objects fall on exactly the same relationship as the local ones! I was seriously impressed by how nicely this correlation lined up the first time I made this plot. It’s going to be really interesting to see if this relationship still holds within individual regions of galaxies, instead of averaging over the whole galaxy.