Notes:

• Operations have been ok past week. Cloudy at some point. SED machine working.
• Not sure how many SNeIa we added.
  • ZTF21aahdqrg: classified quickly as a SNIa, but is actually probably a SNIc.
  • ZTF21aagqdvr: could be a super-Chandra. These are overluminous, spectra show low velocities etc. Look funky, a bit like SNIaX (subluminous). This one has a PESSTO spectrum.
  • ZTF21aagnvvk: antoher potential super-Chandra (-20 around peak), gap due to bad weather
  • ZTF21aaglret: caught early
  • ZTF21aaglamm: has a ZTF sibling SN (Type II)
  • ZTF21aagkvqa: another one with ZTF sibling
• There is now a group on Fritz called "samehost"

• Paper on two SNe that are siblings (not twins)
• First one was dim (-> no spec), second one was bright (->spec)
• 0.6 arcsec separation corresponding to 0.7 kpc distance between the two SNe
• In the same host. If you think of local measurements, that should be almost 0 mass step. The distance should then be identical.
• We can work out salt2 parameters jointly. We assume prior on alpha from pantheon. Both x1 are 0, but c values are different: 0 for brightest vs 0.6 for faintest. The latter would not be part of the standard cosmology sample.
• We can constrain beta: 3.5 +- 0.5. Smaller than the value expected in MW (~4.1) but comprable to values in cosmological fits (~3.0)
• We assumed the size of the intrinsic disp as 0.1, but we can play around and it doesn’t change much. We can also remove prior on alpha from pantheon: no constraints on alpha, but beta still ok.
  • The reason is that if you think of the difference of mus, it doesn’t depend much on alpha, since x1 are similar, whereas c is diff.
• No spectrum of the dim SN, so we fit to different models and compute the BIC. Difference of 10 is a lot, here 100 for non Ia, so it seems safe.
• We have 10 siblings, but here only on 1. For now just a first result.
• Draft almost done

• Nothing overly exciting, kind of reinventing the wheel.
• We have the year 1 sample, great etc, but want a bigger one. Now taking everything in the data (816 with 331 spec) fit them with ipac FP, get 472 after cuts. on x1, c and x1_err.
• The goal is also to have better completeness. High x1 is a bit under-represented in the Y1 samples. We should try to avoid such biases.
• From the 470 objects, we go to PS1, go to images, do source extraction. Identify host galaxies and do cuts (DLR<5).
• At high z malmquist, so do z<0.08 by eye. Get 319 objects.
  • AG: you would want to do 0.04-0.05 if you want to be safe.
• Going from Gold sample to everything, nothing very exciting.
• If we then have GRIZY photo, use SED fitting techniques to get masses. We then see that there is correlation between mass and x1 (lower x1 for larger masses). Less correlation in color, likely scatter.
• Then if we don’t go to stellar mass, and go to color and position of the SN. From PS imaging, we can go to 2kpc AP, measure the color, k-correct them. Local 2 kpc U-R v Global U-R: well correlated, but there is scatter.
• If we do this, we have a tighter correlation than vs mass.
  • local U-R v x1, stronger correlation but not much tighter than the one with mass
  • local U-R v c: less scatter?
• Next thing is to figure out why we miss many objects vs suhail.
• AG: can we make some subset and say it is representative of higher z.
  • MS: running high z sample and we get a higher fraction of SF objects.
  • MS: What is the comparison of ZTF and existing low z samples? question: Is this representative low z vs high z and what is the difference.
• JJ: how dangerous is it to use (you are missing low mass host because of specZ cut). You do include the full sample, using SNID z.
  • MS: scatter on the mass is small. (…)
• JJ: We could narrow down the SNID z estimation, once we know the phases

• Local H0: the idea is to have good calibrator secondary. Right now below 40Mpc like Shoes and Carnegie-Chicago.
• We can look at entire 3 year, the nearby, if 30-35 Mpc range, try to get cepheid of TRGB distance or both. Then we could do comparison. One calibrator for the nice SN would be great.
• We have 13 in the distance range. Below 30 is 6, below 20 is 5. Examples
  • ZTF19aacgslb: 20 Mpc
  • ZTF19aatlmbo: 35 Mpc
  • ZTF20aavpnlv: 25 Mpc
• MR: maybe create a group on fritz. Do we want to collaborate with wendy and adam?
  • SD: Wendy interested in collaborating
• AG: do any have HST data?
  • SD: no cepheid, or TRGB, there is one with TF distance but huge error bar.

Mickael on Calibration:

• Get all the images from CCD7 for 3 month time. Use DASK to massively parallelize code.
• Query gaia and fit a psf, store the PSF, do statistical analysis with PSF and store it.
• 200 computers with 140Gb ram memory.
• Use ZIFF PSF fitter, with magnitude range and numpy_thread limited to 2. Then use dask to run on this list of « call » To Be Done (delayed).
• Then see a task graph, where you see getting data from IPAC. You can see which processes are used, the memory etc. Can then massively download and compute PSFs.
• Then downloaded the g band files for PSF shapes.
• Say 1257 files here and do binned stats. Get data, model and you see the rings. This is probably due to thickness, and we are thinking of incorporating in the PSF model. If single number, wrong at ±2%. In alert pipeline, looking at residual vs PS, see the same map. Not from PSF, but simply from photo.

• Jakob: upgrading UH88 (?) Might use some of that time if we have candidate 1a that are too faint.