Saul starts by showing off a figure for the paper; the HST Data on 9784 is really a lot better than the ground based data, in terms of the S/N. (Better than a factor of 10 higher S/N than ground based? Eyeball guess.)
Some discussion over procedure for producing the plots and fits for 9784... use just HST or HST+ground for finding the various parameters. See the paper for more information; the assertion is that any way of doing it amounts to a convoluted procedure, and the one chosen was in some sense the simplest one.
(Note: for joint fits, where the correlated errors put in properly to reflect the fact that the HST and ground based data were calibrated to each other through the two zeropoints?)
SN9784 paper: purpose is a demonstration paper, good quality data at large z, this is more important than the actual omega/lambda. Holding it up is getting the numbers signed off by Peter... perhaps the numbers will be finished today, Gerson says (he's talked to Peter). Note that because of the uncertainty on the color, there should actually be a high error on the extinction, which leads to a high error on the magnitude. There is an argument to be made that because the host is so piddily the extinction is probably small. (NOTE -- what about B&H extinction? Peter has that information, but it's not in the table that Don reads. The B&H extinction has not yet been included into the SN9784 numbers.) Can also point out that with NICMOS data we ought to be able to do a better job measuring extinction.
Group's original plan years ago when we started all this: we wanted to determine the _apparent_ magnitude of various SNe to 0.15mag.
Alex was working on Omega/Lambda. For a single SN, for a given OmegaMass there is a degeneracy in OmegaLambda -- i.e. two values of Lambda that give the same distance modulus for the same SN at a given Omega-Mass. Note that for Alex's confidence limit plots, they were calculated assuming it was OK to allow Omega to go negative. Issues about whether or not one should scale the confidence regions to reject the unphysical (Omega-Mass < 0) regions, or just allow an error bar that extends into the unphysical region. Don says there is no widely accepted procedure for doing this; in the literature, there is a Bayesian cutoff which invokes "two tooth-fairies". Effectively, you renormalize the distribution after rejecting the unphysical region. There are a number of assumptions which go into this... (see RPP).
The general conclusion seems to be not to cut out the unphysical regions, but to just say what our data tells us and quote an error bar that dips into negative Omega-Mass.
"The Bayesian stuff is incorrect. Well, it's not justifiable, anyway."
-D. Groom.
"Believe your experiment; don't get your mind screwed up with theory."
-D. Groom.
All of these results from Alex were lumped with the first five. Saul wonders, do we want to do this, or do we want to do it alone? Should we keep throwing the first five in, or should henceforth we cut out the first five? Perhaps this isn't such a different subsample from the first five that it's worth keeping the first five in. (Malmquist bias is one worry, as is the systematic that the lightcurve software has changed a little bit.) Additionally, people seem unhappy to give a very different value for this paper and for the second paper....
One point is that this is a redshift at which you can break the Omega/Lambda degeneracy. This is a followup on the original proposal that we ought to be able to do that once we started getting a range of redshifts. I.e., we can get good data (with HST and Keck) on SNe at high enough redshifts to close the confidence levels on the Omega/Lambda plane. So, as long as we keep cranking, we're gonna get us a good measurement of Omega-Mass and Lambda.
Perhaps we should (in a rapidly bloating letter) show two plots: one is the confidence regions with 1 SN, the other is the confidence region with the first five plus the new one. Lets us show that we close the contours with just one SN by itself, but also lets us pretend that our new results are not that inconsistent with our old results. (Or, two different intervals on one plot?) Go ahead and show plots extending into negative Omega regions... or just put in a caption that it extends to negative. The latter. Everybody seems to think the latter. For now.
Robert is working on fiducial objects with Rob. Should talk to Alex to find out what are the statistical tests he did before on the fiducials, to make sure we don't all of a sudden come to different conclusions from what we did before. (Or, if we do, so we can be aware of it and understand why.)
Don, CCDs. Surface leakage is getting below what conventional CCDs have, but as a result the CCDs now have grey pixels and afterimages. Don is also look at the "cosmic rays," most of which seem to be radioactivity rather than cosmic ray muons.
Carl says that runs are divided up so that you can easily find asteroids for HOU. Do we have anybody lined up to take our next SN search data and get it to the students quickly so that they can find SNe and asteroids? (The latter being easier, and things that we throw out.) Ideally we want to do this quickly.
Carl also mentions the thought of looking through longer baseline data to look for varying AGN. Greg pointed out to Carl earlier that Tyson has published a paper with a stringent limit on the variation of normal data. General long-term variability studies. Also, our oddball events (e.g. 9571).
Craig, still working on initial photometry vs. quality of spectra. Isobel and Craig agreed on a ranking scheme of the SNe. Preliminarily, most of the good spectra come from >100% Percent Increase (using the results from the lightcurves). Below 100%, you get a mix, and below 90%< you get very few excellent ones (but you still get some passable ones).
Gerson talks about colors. He has the distribution of R-I vs. stretch, and Bmax-Vmax (or B-V at Bmax? Which is the statistic that Don's program returns? We need to figure this out!) for the 95 and 96 SNe. These are the color of the SN. There is a correlation, low stretch has a redder color (same as nearby SNe, depending on what it is that we actually measured... if it's B-V at Bmax, then they should all be the same). There are a few which have considerable reddening, but it looks like most of them are not reddened. (I.e. probably most SNe are not too terribly extincted.) However, error bars are giant (of order +/- 0.2 for each SN). Correlation is weak, and the thumb test tends to make it go away.