Gerson introduces Kyle Dawson who's a first year grad student interested in joining the SCP. He's especially interested in doing something with his hands.
Saul mentions that we have a review data coming up December 1 for SNAP/SAT. Among other people on the review committee are people like Persis Drell. Saul thinks that this is not a great thing because her paper was not very well considered, and if she feels like she has to stick to it, it could be messy.
Now we're getting into a discussion about what NGST is going to be. It has wandered off from the discussion of who might be on the SNAP review committee. The NGST people, it seems, are going to argue that they can do anything that SNAP would do with NGST.
First, he's going to say something about stretch. He wants to test how well we really know that Stretch applies to the earlier part of the lightcurve before maximum. The question he wants to ask is what happens if you say s=1 before maximum, and allow stretch to be fit after maximum. He compared a few different models: s=1 everywhere, s_e=1, s_l=fit (s_e=early stretch, s_l=late stretch), s_e=fit, s_l=fit, s=fit, and s_e and s_l fit independently. He lists the chisquare and DoF for the different ways of fitting it. se=1, sl=fit gives 801.9 and normal gives 800.3. For 796 degrees of freedom, he asserts that this is not that big a difference. (These are based on individual fits to each of our SNe.)
On to dust. He's been looking into placing constraing on far infrared background from FIRAS and source counts from SCUBA. He has fluxes (nu*I_nu) for the 850micron flux. There is an upper limit looking at the darkest pixels on the sky (7.8x10^-10 W/m^2/sr), not subtracting a galaxy. Subtract the galaxy, get 5.0+-1.2. Augirre has a model for how much FIR you would expect from dust which absorbs light, gets heated up, and radiates. Putting in enough uniformly spread dust to get the dimming of the supernovae with Omega_M=1, it takes 10*10^-10 in the same units. For Omega_M=0.2+dimming due to Dust, he only needs 5*10^-10 in the same units.
Next, he has point source counts from SCUBA, that amount to a few times 10^-10 W/m^2/sr. (1.2 for 2-10mJy, 2.6 for .5-10mJy, 3.6 for .25-20mJy; only the first number was available before Aguirre's paper. The last number is a bit of an extrapolation.) Error bars are to about 20%. These sources are localized sources; they can be galaxies and dust in galaxies. This is not the spread dust that could be faking a cosmological constant. Greg notes that the SCUBA resolution is low enough that Aguirre could point out that some of these may be 300kpc halos around galaxies. At that point, you'd have to worry about a filling factor; Greg guesses that they might still be low enough that there will be free lines of site, allowing us to use our dispersion argument.
FIRAS minus SCUBA suggests that the background from distributed dust must be no higher than 2.4x10^-10 or 1.4x10^-10 (+-1.3-1.4 in each case). It sounds like we're right on the edge of ruling out this model. Adding the SCUBA sources down to 0.5mJy just about does it enough to make us need some Lambda. These dimmer SCUBA sources are the data which is new since Aguirre's paper.
Greg wonders where the dust really peaks for Aguirre's model. He thinks that depending on where it peaks, there may already be limits which are better than FIRAS.
Next topic: reddening by grey dust. This is based on the Augirre suggestion that you take normal dust, and eliminate everything below some threshold, in this case, 100nm. (Things aren't sensitive to the upper cutoff, but are sensitive to the lower cutoff.) Tom shows a bunch of plots for E(x-lambda) vs. A_x, where x is B, V, R, I, J, and 1.5, and lambda=V, R, I, J, 1.5, 1.7, and K. Tom believes that 100nm is about as large as Aguirre was willing to say is plausible. The idea is that the dust is pushed out of galaxies by radiation pressure, and along the way they get sputtered. You can do a calculation for the lifetime, and how much you expect to get sputtered off. 50nm or 70nm might be more realistic. On the plots, he's plotted vertical lines of values of A_x that correspond to the amount of grey dust you need out to several given redshifts, for z=several values. (Once again, I probably ought to get the images to put into these notes, but I will probalby not do that.) The bottom line seems to be that E(x-lambda) stays less than 0.03 for even E(B-J) and E(V-J). (Tom says that all of this is in the observer frame; I'm not sure if/how redshifts were taken into account. These are all drawn for the case were dust completely elimiates lambda at Omega_M=0.2.)
(There is some discussion about the details of dust models, and whether this is based on the most up-to-date Aguirre models. I zoned a bit, I fear.)
Saul thinks it would be useful for the satellite proposal to do this the other way around. Suppose there is a little bit of grey dust out there, and we want to measure how much grey dust there is, and how much cosmological constant there is. (I.e., rather than trying to rule out grey dust as an alternative to Lambda.) How good a measurement of E(B-J) (say) do you need in order to measure this? For instance, a 15% measurement of the extinction at z=(0.4?), you need a 2% measurement of the magnitude difference.
Some discussion about star formation rates and how it goes up with redshift.
Tom says he'll have a chance to do only a little bit more on this before he goes; he may be able to do more of it after he goes. He'll be incommunicado from tomorrow until the 21st.