I have a related question for clarification. Will the alert packets (sent to brokers within 60 seconds of image readout) include all three types of cutouts: visit/science, template/reference, and difference image (DI)? Or will they include only the template/reference and DI cutouts?
If the visit/science image cutout is not included, can we rely on the relation:
visit image cutout = reference image cutout + DI cutout, exactly?
Hi @deltasata, thanks for this question.
The plan is for alert packets to have only the difference and template cutouts, and yes the science cutout could be recreated from these two. For now the best reference for alert packet contents and the cutouts is Section 3.5 of the Data Products Definitions Document.
Iām going to mark this reply post as the solution but please donāt hesitate to create new topics with new questions at any time.
Hi @MelissaGraham, could you please confirm that the relation expressed by @deltasata will work for the alert packets? I mean, the difference image cutout will come with the same sign as the exposure?
I ask because e.g. in ZTF, difference image cutouts for negative differences come with all pixels multiplied by -1, so that there is always a bright source in the center (instead of a āholeā in the center when differences are negative).
Hello,
Referring to the nominal Alard & Lupton 98/2000 DIA algorithm, the relationship between science (N) and template (R) is:
D = N - k conv R,
where D is the science image, and k is a kernel. Will we receive the template (R), or the convolved template (k conv R) ?
Right, the difference is not a simple pixel-wise subtraction of the science and template. We are in fact going to send all three cutouts in the alert packetāI regret that this is not presently stated in the DPDD, but it is approved and reflected in our current alert schemas.
Thanks @ebellm. Will negative difference cutouts in alerts come with a ābrightā source in the center (as in ZTF) instead of the true negative flux?