Alex (Conley) speaks. He's still working with the Keck images. He's noticed that the APM zeropoint is very different from one image to the next... he also has problems with standard star calibrations, as a lot of the standard stars are callibrated. He's currently looking to see how much lots of individual stars vary between images. Most of the stuff which wasn't working from last week is working now. However, all the magnitudes are screwed up, so the fake stuff still isn't quite producing consumable results.
Alex mentions that a sort of "year 2000-like" bug was found, when the date turned out to be the name of the image number.
Saul mentions that as of the last few days, the estimate of how many supernovae their out to be in the Keck fields has gone up to about 1/2 a supernova per field. (What does a half of a supernova look like?) This is bigger than the old numbers because previously we didn't use what we currently believe for our cosmology. The volume at high redshift is bigger than it "used to be". Well, Greg says we shouldn't get into this because it still needs to be done right. Me, I want to do the search and find out empirically what the rate is....
Kirsten speaks. Her code now runs in parallel. (With what? Eating dinner, Saul suggests.) If only the supercomputer wasn't so overloaded.... "From here, Peter says we're going to graph stuff." Peter: "Yeah." Kirsten: "Chi-square vs. Everything." Peter: "Yeah." The whole purpose of all of this is to try to match one spectrum with another, so that we can take an unknown spectrum and determine things like redshift, epoch, type, host galaxy contamination, etc. (Maybe even reddening....)
Peter and Greg got into some discussion about reddening (and Peter thinking that most of them are reddened), but I missed a lot of it because I had to go get my mouse. Peter seems to be saying that people are finding lots of Ia supernovae with reddening. Greg objects that then Ia's have to be predominantly a disk population, if most of them are reddenend. Another objection is that perhaps people are just starting to find supernovae on the far side of the disk.
Rob talked about the database. He's starting to redo some of our software to use an SQL database (using Postgres) instead of NetCDF files for the summary files. This rasied a lot of questions from a lot of quarters. Carl, in particular, was adamant that we should use Sybase, not something else. Rob's feeling is that the more proprietary systems we wed ourselves to, the harder and more expensive it gets to keep our software going and the less likely it gets that our software might ever run anywhere else. The other issue is that Saul is worried that the performance of Postgres will be terrible, given that it was way back when when he tried it on the Vax. Rob believes that today's computers are up to the task.
Peter is doing work for everybody else and nothing for himself. He discussions various issues of supercomputer time use and getting the NERSC computers to the point where they can actually run his supernova code. The core of the problem seems to be I/O... the nodes don't have enough memory, so he has to swap and read and write things.
Greg talks about the nearby search. We got a message from Anna from Portugal that we were turned down for the YALO time. We were hoping to get roughly 2 weeks of queue scheduled time, which was going to be one of the most efficient nearby search photometric followup scopes. It's a surprise that we didn't get this. Greg says now we are pursuing whether any of these queue scheduled people would be just as happy on another 1m telecope somewhere. We could then trade time, and get time on YALO. Our deal is that the queue scheduling on YALO is nice... we don't have to deal with the problem of scheduling scopes for periodic followup, and we don't have to have somebody sit in Chile to go to the telescope every three days. Saul says he got a message that apparently some people would be willing to change time... but this is preliminary. We shall see what happens with this, this is all a new development.
Greg says we also lost some HST time. We had some NICMOS orbits messed up earlier due to some sort of exception. Three orbits on our highest redshift supernovae had data lost altogether. Originally, we were given these orbits back, so that we could still use them. Then, there were one or two exposures (not orbits) on one of the other supernovae due to faults in the HST system somewhere. Greg was told to file an "HOPR" request to get the time back. Greg talked to Doug about this, who had already credited us for those three orbits. When he sent in the request about these other partial orbits, Doug and the people who told us we'd get the full orbits reinstated indicated that they didn't realize the policy had been changed. Now, if your program is 75 percent or better complete, you don't get your orbits back. So, in trying to get two partial orbits back, we lost three complete orbits. Greg speculates that if we hadn't tried to get the partial orbits back, they might never had noticed.... He is still going to try to appeal this.
Shane Burns has now added the two most pressing high-z NICMOS observations (i.e. those we need to know soon whether or not we need final refs). Of the two, one we definitely need final refs, the other we may not need final refs. The second one is the only one Greg has looked at himself; Greg thought we'd probably want a final ref for. One question is if perhaps we have more of a wing on the psf than Tiny Tim suggests. However, Greg says that the extention is not symmetric, but rather looks like it might be a galaxy.
(Tiny Tim is the psf simulator for the HST.)
In short, we are probably going to have to use up a fair number of NICMOS orbits on final references.
Greg has no good news for us.
Greg talks about the fits again. Looking at some supernovae with large stretches and large uncertainties, Gerson and Greg realized that the way we were fitting the Hamuy supernovae (letting the offset float with a dataset truncated at 40 days) allowed too much freedom in the stretch. So, the fits have been done again with all of the data put back in, which tightened up the stretch. Finally, they decided to cut at 60 days, which gets past the kink in the lightcurve. Instead of the time of max being the constraining point for the stretch lever arm, it's the kink (for Hamuy supernovae which didn't go through max).
Don reminds us of the reason stuff was originally cut out. He says that the later points not only didn't fully agree with the template, but also varied supernova to supernova. Greg argues that this isn't as big an effect as what perhaps in the past had been believed. Greg says its more an issue of discrepant data points rather than a systematic supernova-wide discrepancy.
Peter is asleep. At least, his eyes are closed. Maybe he's just resting his eyes and is paying full attention.
Reynald (who is visiting this week) is worried about this change, and the resultant changes in the stretch. Greg says that the ones whose stretch changed a lot used to have big errorbars....
So, Greg does have good news after all. The stretch distribution of the Hamuy supernovae looks more like that of our SCP supernovae. Second, the fits that Peter has done have a chisquare that improves by 10 for 60 degrees of freedom (to 69 or some such). The reduced chisquare is something like 1.15 now (used to be 1.3).
"What has this done to the universe?"
Don Groom
Meanwhile, hearing the words pronounced and not seeing them written, Alex was confused about Hamuy vs. Minuit.
Gerson speaks. He's mostly spent his time redoing the fits. He is pleased that the distribution is now much more reasonable. (It will make his paper easier to write.) He's redone his table which has to go to Peter, and Peter is receiving his weekly (weakly?) request to redo the K-corrections, and Peter says he may do this next week.
Reynald is working on getting the final numbers for the rates. He's going to talk at a conference in Vancouver this Friday, to a bunch of Physicists, about our stuff.
Robert speaks. He's been busy programming. He rewrote distmod to have the equation of stat parameter w. He's writing another version that does looping between the variables in C instead of in IDL. He's now back working on a program that spits out zeropoints. It works its magic by looking at an image which is already calibrated by Rob and doing a ratio between images. Rob says that these zeropoints will be used in the next version of subtractimages and searchscan programs, which should be better than the apm zeropoints.
Greg says that Robert finished the Kron magnitudes but that he (Greg) hasn't parsed the information yet.
Patricia's program to fit Gaussians over... something? hosts?... is finished. Ah, yes, hosts, I assume, since Greg used the word "radial profile." Greg says that the idea is to take out the seeing... it's a way of doing a deconvolution that will give us a sort of a way to get a seeing corrected concentration index. She wants to give a talk to the group about this.
Carl is here today. He's going to tell us a couple of things. Regarding a telescope on the space station (a proposal that Carl put in), the plan is to put a 1500x1500 CCD on a 1m telescope on a Japanese somethingoranother. Carl says that this is starting to gather momentum. They are throwing pieces of Space Station hardware up at the end of this year, Carl says. He says that every image has to be shared with kids in a meaningful way.
Carl says that we finally got the engineering drawings for the Chew's Ridge site to the engineers. Apparently, we are still waiting for Forest Service approval. Things are going very slow with that, but Carl claims that they are semi-steady.
Don says that there's a little happening on the CCD front. We failed to get our MRI (whatever that is), and Don is very depressed about the possibility of things working. It means that the onus is on the current run to get something working. Right now, he is waiting for the etcher on campus, which has worked for 2 days out of the last (few?) months. He says that they are ready again with the mask, and waiting. Don says that Steve says it will take another month to get things working after that step. Don says that they are going ot meet with Stover to discuss the packaging issue (single CCD, not even mosaics). There are some real mechanical hassles associated with this.
Now we eat.