Are you talking about the dark pixels? That may be a pixelization effect, though the others have something similar (if you look closely). Also, I'm not worried about dark pixels, since they don't contribute to the confusion limit. Remember that this is a log scale (spanning 4 orders of magnitude).
-Josh On Sun, Jan 17, 2016 at 2:46 PM, Peter Williams <[email protected]> wrote: > Hi Josh, > > The earth-rotation-synthesis version of the synthesized beam for > configuration (c) [Figure 5c, right panel] seems to have a grating pattern > in the sidelobes that's not present for configurations (a) and (b). I don't > feel like I have a great sense for what the practical effect of that might > be, but it seems potentially worrisome for imaging. Do you have a sense of > where that comes from? > > Peter > > On Sat, 2016-01-16 at 13:23 -0800, Josh Dillon wrote: > > Dear HERAtics, > > Aaron and I have been working on investigating how to arrange the > “inriggers" and outriggers for HERA. In particular, we’ve been interested > in two questions: 1) how do modifications to the configuration affect > imaging? and 2) how do they affect the ability to redundantly calibrate the > array (including outriggers) using Omnical? That work will become a paper > relatively soon (hopefully, before the proposal goes in). > > The key result is an idea for an array that includes a change to the array > core. Basically, I want to split it up into three segments, each relatively > displaced by less than the minimum baseline length. This is something we > should talk about as a collaboration, so I’d like to give everyone > something to think about before the next telecon. > > I’ve attached the current draft of that paper (which is just an abstract, > outline, figures, and captions right now) that helps lay out the argument. > Basically, the proposed configuration (which is configuration (c) in the > paper) is more robustly calibratable with Omnical, produces lower > calibration errors, and modestly suppresses sidelobes. > > The key figures to look at are 1, 2, 3, 4, 10, and 11. The first four show > the proposed configurations and the instantaneous baseline redundancy > patters they produce. Both configurations (b) and (c) are designed to tile > the uv plane nicely, putting new baselines where information is sparsest. > Figures 10 and 11 show the expected gain errors on the antennas after > logcal, both with and without the outriggers. The other figures have to do > with imaging, where the effect is more subtle, and are harder to understand > without the supporting text that I still need to write. > > Perhaps the most surprising result is that my original attempt to produce > perfectly tiled sub-grid baselines (see Figure 2b) actually leads to larger > calibration errors throughout the entire core by approximately 10%. That’s > because Omnical is trying to solve for a lot more visibilities with very > little information about them. That’s the main reason I think the > split-core is preferable to the kind of “inriggers" proposed in > configurations (a) and (b). The split core also allows for complete > simultaneous uv-coverage at triple density out to ~400 meters. > > In case anyone is curious or wants to try doing a calculation with them, > I’ve attached text files with positions of the split-core + outriggers > (HERA-361). I’ve also attached a slightly more symmetric version with 11 > core antennas removed (HERA-350), that currently Dave, Aaron, and I favor. > > Of course, I welcome any feedback. Enjoy your long weekend (of proposal > writing)! > > -Josh > >
