Hi Zaki, This is a very well written paper! I'm impressed how clearly you presented all these complex techniques. I have been thinking about learning how to create power spectra for a while, and reading your paper really helped me to understand the topic better. Here are my comments for the part of the paper that I find the most interesting:
I always put the text in your manuscript inside square brackets so you can search for them easily. Pg. 5: Missing "Fig." [The per integration solutions are then applied to visibilities and the results are shown in 4] Pg. 5: noise and variance means the same thing, so it should be either noise or the variance. [is the noise in the variance of the visibility] Pg. 5: Here the role of \chi^2 can use a bit more clarification. There are two types of bad things, both of which can be detected by \chi^2. A large \chi^2 can mean 1) bad fit, meaning the redundant baselines are not redundant, which can be caused by crosstalk. 2) Strong signal such as RFI, which can still be a very good fit, but due to strong signal, the noise would be underestimated, causing a high \chi^2. [corresponding to a bad fit] Pg. 5: What exactly is the quantity plotted for Fig. 5? Since they are always less than 1 I assume they are neither \chi^2 nor \chi^2/DoF. Is it \chi^2 before dividing noise? [for all frequencies for and entire day][Also caption of Fig. 5] Pg. 6: From what I can see you are saying that OMNICAL is the first calibration you performed on the raw data, as suggested by Fig. 2, which means OMNICAL is applied on raw data. However this makes Fig. 4 confusing because the raw data should look much messier than the left panel. So you should probably change the wording for Fig. 4 caption a little bit, to something like "previously used method (left) and OMNICAL (right)" [before (left) and after (right) the application of redundancy-based calibration] Pg. 6: Floating text [last one]] Pg. 6: Equation 6 looks almost mathematically incorrect, since B and I disappeared after the second equal sign, and B is not defined after the equation. [is a Blackman-Harris taper function to minimize band edge effects] Pg. 7: typo [that accounts for for shape of the source spectrum] Pg. 7: I'm curious: which days did you exclude and why? You don't have to explain in the paper but I'd like to know. [The full season was 135 days long; of these 124 days were included in the average.] Pg. 9: Fig. 9 looks a bit strange: there's no spacing between the bottom of its caption and the main texts. [Fringe-rate transform of visibilities measured] Pg. 9: Eq. 8 looks incorrect to me: are you sure Tsys and Trms are not swapped here? [Using these variances, we calculate the system temperature as a function of LST] Pg. 10: Unclear and a typo in "vecotrs": does two different redundant baselines mean "two baselines in the same redundant group" or "two baselines in different redundant groups"? [are visibility vecotrs that correspond to two different redundant baselines.] Pg. 11: Multiple typo/grammar: "what" should be "that" [In the description what follows we use the the] Pg. 11: Multiple typos in 'analysis' and 'spacing'. Also, this whole sentence is confusing, partly due to the use of 'but'. I'd say "In the following descriptions we use the 30\,m baseline as an example, and identical techniques are applied to the other two redundant baseline types..." [but we apply the same anaysis to the 30 m baselines with one row spaceings] Pg. 11: typo/grammar [which is looking ahead to the application of the method our data] Pg. 11: typo/grammar [The application of the covariance matrix to data has the same general suppression, however, different data sets have different covariances.] Pg. 12: typo [never comput the auto power] Pg. 12: typo [Taking the above two caveats in to consideration] Pg. 12: incomplete sentence [only the frequency-frequency covariances between channels] Again, very impressive work! Jeff On Sat, Feb 14, 2015 at 7:00 PM, Zaki Ali <[email protected]> wrote: > Hi All, > > Here is an updated draft with some of the comments from Gianni and Jonnie > included, amongst other things. > There is still a bit to get through, especially the power spectrum > analysis and calibration section. > I am still working through some of the sections and will keep updating > this draft. If you would like to stay up-to-date with the latest and > greatest, clone the repo. > I would like to get comments back by the end of the day Monday, and > comments on this draft are fine. > > I will be calling in to the datacon on Monday. Hopefully we will be able > to discuss some of the good comments that were made. > We can move it up if it would fit peoples schedule better and/or we think > we need more time. Otherwise 12:30 -2pm PST. > Some of the things I would like to discuss is the additive offset removal > in Omnical and quadratic estimator power spectrum analysis, to clarify a > few things. Hopefully, experts on these topics will be there. > > We want to have it submitted by Wednesday. This is a very fast turn around > and having everyones comments in a timely manner is crucial. > Attached is the pdf and the tex for inline comments. Clone the repo to > stay up to date. > > > I appreciate all the help and comments! > > Cheers, > Zaki > > > > On Feb 13, 2015, at 11:42 PM, Gianni Bernardi <[email protected]> > wrote: > > > Happy long weekend -- hope all of it's not spent writing. >> >> Jonnie >> >> > indeed forgot to ask... is anybody calling in on Monday? If we know in > advance that nobody won't, I'll save myself going to the office :-) > Thanks and happy long weekend! > > Gianni > > > > > >
