Hello, everybody.
> Flemming Videbaek
> FV- This is a note I wrote the previous weekend doing some additional
> checking on the dN/dEta
> from Beam-Beam counters.
>
> The eta range is divided into very small bins (0.02). I would argue
> this is in fact much too small. The
> extension of a big tube is about .3 units. This might still work ok
> (see comment above). For the most inner
> tube a change in 5 cm changes eta by 0.023, and with a vertex sigma of
> 2-3 cm this is comparable. There
> may be good reasons to look at this for systematic effects, but final
> presentation should be no finer than .1
> in my opinion.
I absolutely agree with that. This small range was introduced in
order to better study this famous "jump" in dN/dEta spectrum in the
Right Array. If I remember it correct, number n (# of subintervals in
0.1 unit of Eta) should be easy to change
> The analysis code fills event-by-event dN/dEta distributions (each
> with a .02) bin, and evaluate in the very
> end the means. This is integrated over all tubes at this point.
>
> I had a concern about the binning (bin size 13 units) but made some
> checks with ROOT histograms and
> convinced my self that mean, and RMS is calculated properly as long
> there is no over or underflows.
>
> It would be most useful to have all tubes separated in the analysis at
> this level (eta binning) so each of these
> can be inspected and compared to the other tube(s) that have same
> geometry. This way we can also in the
> end look at the extracted dN/dEta for say the 8 left tubes- calculate
> a statistically proper Mean for each and
> Study the deviation between each measurement to get a better
> understanding of the statistical vs systematic
> error.
I've done it many times and will implement these minor changes in
my code in /brahms_app/yb_app to make it available for everybody, once
I'm back from my short vacation (Dec. 27). Statistical vs. Systematic
errors understanding is #1 on my to-do-ASAP list.
> From the RMS and # entries we would expect the stat. Error to be
> around +-2 out of 300. The syst. Error
> from above is also very small. I will tentatively as also said at many
> occasions that the errors shown in the
> plot so far do not reflect the facts. The statistical errors are much
> smaller maybe 1-2% and we cannot
> arbitrarily assign a large systematic error.
>
> A crucial missing part is to evaluate the BB dN/dEta using the Tiles
> as an event selection e.g. with JH's
> newest selection module.
Well, may by this is #1 on to-do-ASAP list as well ;-)))
> The other very open issue is to
> a) resolve the Geant puzzles that YB has presented at the
> collaboration meeting
> b) Understand quite a better the tube response within geant particular
> in regard to delta-rays that can
> be incorporated into geant- cuts and treatment in what materials are
> difficult to assess without
> quite a long study
> c) We have seen that these kind of inclusion can change the background
> correction by perhaps from
> .62 to .58 with some pieces included. What are the important one? And
> how do one check the
> calculations is reasonable converged?
> Some tentative conclusions:
> I think there too much to be done on a rapid time-scale to make this
> well understood.
> I do not think there is any trivial error in what is the main code for
> raw-data analysis, and the most
> important.
I am very happy to hear, that there's no major trivial mistakes. As for
the rest of the analysis, I still hope to finalize it, including
everybody's satisfaction with the results by mid Jan. (QM). Although, I
don't know what was the outcome of the QM talk selection process. Are we
going to present dN/dEta on the "united" plot?
Thanks.
Tired of non-stop snowfall for last 5 days in Russia and eager to work,
Yury Blyakhman.
This archive was generated by hypermail 2b29 : Mon Dec 18 2000 - 09:55:41 EST