Pre meeting disussion of HST data. The April 8th SN9784 HST image we had analyzed was the one that somebody over at HST had crrej'ed and combined. As it turns out, they had already done a sky subtraction, whereas for the rest of the dates we did not. This changed the charge transfer efficiency calculation, so that the april 8 HST point on SN9784 is now 5% dimmer than it was before. This needs to go to Peter to put into the data file so that the lightcurve can be refit. The numbers are now more inline with what would be expected.
Greg notes that the whole thing is a little worrisome, since there is apparently some doubt whether or not the whole charge transfer efficiency thing is applicable. Saul says that we are going to say in the paper that this is preliminary data analysis anyway.
Gerson shows the hubble plot with all of the SNe for which we have light curves so far. Right now, there are 32 of them (including the first seven). He's only but in B&H extinction in cases where the extinction was of order 0.1. The trend seems to be towards 0 Omega for Lambda=0. Without the stretch correction, Omega comes out somewhat higher, and there is _less_ scatter than there is with the stretch correction. (Oh well.) There is an inconsistency in the stretch correction; for the old SNe, no stretch/magnitude correction was done where the stretch was outside the range seen in nearby SNe. For the more recent SNe, the stretch was applied each time.
There is a funny bump at cz~5.13, based on a few (5 or 6) SNe, which has been there before (it looks like they are 95 SNe, though at least one 97 is in there as well). This bump also seems to mostly go away when you don't correct for stretch.
There are two points for which MU is plotted, not MB. They are two where there is not currently an I-band calibration, so you can't get MB directly.
Oddballs; 971. Comes out with a very narrow stretch. There are a couple of dates of points which seem to confine it to low stretch (came out 0.54). The spectrum of this one should be inpsected to see if this is either a funny looking one, or if there is a reason to believe that this one might be a screwy SN.
Another oddball: 976. Stretch is large, comes out at 1.12 (not outlandish, and within errorbar of 1). Chi2 of the fit is bad (nearly 2). I-band is horrendous. 9733, comes out way to dim for its redshift. How good is the redshift on this one? The extinction could be nailing us again, ad the I-band is pretty ugly so we can't easily measure it that way. Chi2 is too low. (Gerson says that for a good SN, the Chi2 comes out close to 1.) Look at the spectrum of this; maybe it's not a good Ia.
Saul thinks we should go through every SN to decide which ones we think we believe, which ones we are worried about. Probably the best way to do this is not to look at our final result and then inspect the outliers. We should get a catalog of each SN with an evaulation of what we believe on each of these. Evaluate lightcurve, fit to same, spectrum. (Make a chart.) This should be put into the SN tracking system. We should try to cull out a group that doesn't need much more work, so we can plot the ones we believe in.
Sebastien: Working on paper with Bruce about gamma ray bursts. Looking at data from March run. A month before, somebody discovered a putative optical counterpart to to a GRB. Sebastien is looking at our data for similar sorts of events with the idea of setting limits on how often that sort of thing happens by chance. There is a 5 day gap between March 5 and March 10. Sebastien has scanned about a square degree, and has not found any event corresponding to this; he's only found two asteroids, and two SNe we already knew about. He comes out with a probability of 95% that there were less than 3 events in 1 square degree in that time. Right now he's putting up tables of fields and tuning the numbers. He's thinking about doing the same thing in the I-band, but isn't yet sure how well this will work. He's also wondering about doing this with several new fields instead of two new fields. This gets you on the order of 0.5 magnitude deeper, although for reasons that Sebastien doesn't understand, you can't go as bright (i.e. the saturation limit gets more stringent). Could it be a hardwired saturation number that's not dealing with the summed news correctly? There may be a set welldepth for just one image, or perhaps two images, which is not being increased as you sum in more images. Really, you want to compare individual images to their welldepth, rather than the sum.
Greg says that Tony Tyson (~3rd author) has published some limits on variability, perhaps a year ago. It would be worth looking at this to see how much he did, what parameter space, to find out how much more we have to do to have a meaningful limit here.
9571, we started talking about. R-band light got brighter by a factor of 3, but the lightcurve is too narrow. The spectrum does not look like a SN which is 3x brighter than a host galaxy. The galaxy has lines at z=0.86, but there is no supernova spectrum; it's just a garden variety galaxy spectrum. The theory most of us talk about is that it's a microlensing event. One question is whether or not it is achromatic. (Apparently Alex Filippenko has Keck time which he would take a spectrum of if he was convinced it was achromatic.)
The other group's z=0.97 spectrum. They saw an O2 line at this high redshift... then, later, in 1.5 hours on keck at decent seeing, they found nothing at all. The lightcurve looks allright for a SN Ia at this redshift. Why would the O2 line go away? Odd. Perhaps Brian just had better conditions the first time around. What could this object be? Of course, we probably arent' supposed to know all this, but Peter has spies everywhere.
Apparently also they are going to try a technique with taking some host galaxy spectrum by shifting the charge down the slit so that you can measure and get rid of the fringes. This is supposed to clean up the near-IR. We should ponder this and think about whether or not something like it would be worthwhile to do at our next run.
Robert Quimby, fiducials. Looking at new/ref fiducial ratio as a function of color. It looks like the ratio does not depend on fiducial color, which is very nice. This should be documented, and written up, so we know in the future that we've concluded this, and also how to check it for new telescopes. Rob, on the whole, is breathing a sigh of relief that he can continue to blithely use all of the fiducials to figure out how much ref to subtract from each new in the lightcurve software.
Also, Robert has written up documentation on cleaning images. When he finalizes this, it will be made available. On this topic, Greg looked back at the original big paper, where we say that after surfacing we get the paper down to 2%. He thinks we should do better. Is 2% a worst case? Matthew should be asked about this.
Alex, fits, templates. "Nothing very exciting." Some talk about error in data and error in model, combining to get a best fit. Story is you can't really use a chi2 fitting because there are no longer perfect Gaussian distributions, so you can't just "log out the e". ("I don't know if anybody follows what I'm saying." --A.kim.) Under assumption of normailty, he attempted an integral in 20 dimensions with matrices and other ugly stuff, and Mathematica barfed at it. Perhaps somebody with good mathematical ability and a strong stomach could do it by hand.
Other status, split Hamuy sample into two subgroups. Build the model on one half, test the model on the other half. Right now running to see what the answers are. He is still finding bugs in the template being made out of the full sample.
Mike says that things are going well. They are getting a lot of cataloges of a lot of different fields. In terms of dealing with overlap, Saul says he thinks the easiest thing to do is make a gigantic array of sky, gridding it into 1'x1' bins. Then, for each of these little squares on the sky, write into it the deepest image which covers that square (or, perhaps, total depth, or both). At the end, go back and look at this array to see what you've got. Note that we're talking about any array which is 600x360x60 =36000x360=10^7 grid points.... that's a LOT of computer memory.
Alex Filppenko is working on the spectrum data paper from our <=96 SNe. Apparently he should have something to show in time for Santa Barbara. Where we have good spectra well from max, the spectrum features are matching the aging well with consideration of time dilation. For host galaxy subtraction, A.F. believes that templates will work better than host spectra, for simple S/N reasons.
Santa Barbara: two days Alex and Saul want to go to, another day Peter will stay. August 6 and 7 have photometry stuff and talks on supernova lightcurves. Saul was originally thinking about trying to schedule another two-day meeting after the meeting, but it didn't work out for reasons of lack of hotel rooms. Saul orgainzed a two-hour session. Most of the rest of the meeting is explosion scenarios of various modellers. Peter says we may have something to say about this because we have some early photometry on some SNe. So, it sounds that only three from our group are going. There are thoughts about maybe we should have more there just to show the colors; we'd have to find our own hotel rooms somehow.
z=0.83 paper. Saul has written a version of the paper which has _only_ 9784, not 9621. We will try this draft out to see what everybody thinks. The idea is that it makes a cleaner, punchier, shorter paper. We refer to the fact that we've got other high-z SNe, but only really pay a lot of attention to ... the notes stopped in the middle of the sentence here
Time dilation. Gerson is stuck on writing the words; the data looks ready, he says.
Theoretically Alex will also work on getting SNMinui ready for his templates. Greg should get involved in this.
Color terms in calibratoin and instrumental corrections: Rob and Peter were doing something inconsistent, and are not any more.
Craig %Inc (returned by searchscan). vs. Isobel spectrum rank. No good way to set a cutoff. Using the measure of %Inc which is from where there is no SN light to the search, you can start to distinuish. There are no excellent ones below 100%.
Monday at 1:00 there will be a tiem for Peter and Alex to present to us what they will talk about at Santa Barbara.