Distribution of diaSources in Dp0.3 visits

Hi, I was taking a closer look at DP0.3 diaSources in visits with large numbers of SSO detections (i.e. near the ecliptic) and the spatial distrubution looks unusual to me. Is the star-like distribution in this figure to be expected? I would have expected the detections to fill the whole footrpint. Is this related to the known issue “The camera footprint was slightly inaccurate”?



Interestingly, in the above figure I have plotted the footprint outline and the detections are contained within the overlap of a footprint with rotSkyPos = theta and rotSkyPos = -theta.

Hi James,

Thanks for the plot. Yes, I would have said that this must be a manifestation of the known issue that you’ve linked to within the DP0.3 documentation. I believe that the relevant DP0.3 data set creators who might have more insight are at institutions currently on holiday until Monday 11/27, but I can ping them once they’re back for any further input they’d like to provide here.

Great thanks Aaron. I only came across this by accident and was mostly just wanting to check that this was a known effect. So not urgent at all while people are on holiday (I forgot it was Thanksgiving for USA folk)!

1 Like

@Gerenjie – might you have any further insight/commentary on this? For instance, is the “overlap of a footprint with rotSkyPos = theta and rotSkyPos = -theta” behavior mentioned by James universal across visits simulated for DP0.3? I was hoping to tag Pedro as well, but I can’t find a corresponding forum account. Thanks!

Woah, thanks @jrob93 !! The strange camera footprint was known and we have a new version of the simulator which generates a correct footprint, but we hadn’t managed understood what the strange shape was or pinned down the issue in our old software. This explaination makes a lot of sense to me, and that’s a very striking plot – I’m fully convinced that it’s a theta-and-negative-theta overlap bug.

1 Like

@jrob93 This is the most impressive piece of sleuthing I’ve seen in a while! We knew we had an incorrect footprint issue, but the old code was convoluted enough that we never figured out exactly how it was getting messed up.

Again, fabulous job!

1 Like

Neat, glad to be of service :slight_smile:

I only noticed this because I realised there was a theta/-theta issue when plotting the footprint in healpy gnomview (with RA increasing to the left) vs a regular cartesian plot (with RA increasing to the right). So maybe this arises for the old DP0.3 code as a similar clockwise/anti-clockwise rotation depending on the reference frame.

Thanks again for this post, @jrob93.

I’ve marked @Gerenjie 's response as the solution because it answered @jrob93’s questions and confirmed that the effect illustrated was the known issue, as suspected.