Coadds vs templates

[rbiswas4]

One of the issues we discussed in the meeting was whether templates used in difference imaging and co-adds were identical products.

The context of this issue (during the meeting) was understanding the appropriate template for difference imaging in the N+1 th year. Ideally (not considering DCR) , one would like a low noise template implying a coadd of as many visits in the area as possible. For SN, this probably implies coadds for all but the current season. For different regions in the sky this corresponds to the N year coadd and in others it could be the N-1/2 (or some fraction) year coadd, because SN in the early part of the N+1th year could already be in the N th year coadd. This is only a problem for the early years.
There could be other arguments around DCR for arguing for different kinds of templates.

@SimonKrughoff confirms that the LSST definitions allow for a DI template and the coadds released to be different products.

This is supported in the DPDD :
Section 4.4.3:

Templates for difference image analysis will be created by coadding 6-months to a year long groups of visits. The coaddition process will take care to remove transient or fast moving objects (eg., asteroids) from the templates. Difference image analysis will use the appropriate template given the time of observation, airmass, and seeing.

I think this is pretty explicit about the fact that Templates are potentially different products from Coadds, and the project has a place for having the best template which may be different from the coadd.

Further, the coadds are described in Section 5.4.3 : This is also interesting, including the non-public (maybe virtual) different time-scale coadds.

[drphilmarshall]

Thanks for chasing this up, Rahul! So: what templates would you like to be made for Twinkles 1? a) Is there a particular year that you want to focus on? For example, if you want to emulate SN light curve analysis in DR3, we could make a year 1+2 template and then difference years 3-10 against it, to make 8 seasons worth of SN lightcurves. b) Do you want to emulate each data release? That would mean making the year 1 template and differencing year 2 against it, then the year1+2 template and differencing year 3 against it, ... then the year 1:K template and differencing year (K+1) against it, ... then the year 1:9 template and differencing year 10 against it, to make 10 seasons of realistic SN light curves each with its own realistic template image. My feeling is that option b) is most informative but is also the most fiddly for you and Tony to set up. Let's talk about this at 1pm?

…

On Wed, May 3, 2017 at 11:35 AM, rbiswas4 ***@***.***> wrote: One of the issues we discussed in the meeting was whether templates used in difference imaging and co-adds were identical products. The context of this issue (during the meeting) was understanding the appropriate template for difference imaging in the N+1 th year. Ideally (not considering DCR) , one would like a low noise template implying a coadd of as many visits in the area as possible. For SN, this probably implies coadds for all but the current season. For different regions in the sky this corresponds to the N year coadd and in others it could be the N-1/2 (or some fraction) year coadd, because SN in the early part of the N+1th year could already be in the N th year coadd. This is only a problem for the early years. There could be other arguments around DCR for arguing for different kinds of templates. @SimonKrughoff <https://github.com/SimonKrughoff> confirms that the LSST definitions allow for a DI template and the coadds released to be different products. This is supported in the DPDD <https://docushare.lsstcorp.org/docushare/dsweb/Get/LSE-163> : Section 4.4.3: Templates for difference image analysis will be created by coadding 6-months to a year long groups of visits. The coaddition process will take care to remove transient or fast moving objects (eg., asteroids) from the templates. Difference image analysis will use the appropriate template given the time of observation, airmass, and seeing. I think this is pretty explicit about the fact that Templates are potentially different products from Coadds, and the project has a place for having the best template which may be different from the coadd. Further, the coadds are described in Section 5.4.3 : This is also interesting, including the non-public (maybe virtual) different time-scale coadds. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#446>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AArY94-GL6QNLTFRjqteQY4L_EWbrt4sks5r2Mj7gaJpZM4NPyeB> .