Dear all,
here are some questions Flemming sent me regarding the report you have
all, no doubt, already read. I hope that my answers are clear
and to the point. Further questions are more than welcome.
In what follows, I enclose FV's questions follewed by my answers,
which I set off by >
Enjoy!
Fig 1. Does not show the tpm2 tracks. I know it is likely mostly (0) 1
and a few two's
>The mean number of tracks (after requiring at least one in the
>event!) is 1.19, while the mean number of hits (also requiring at
>least one track) is 1.6. I'll check the number of hits with no
>requirement on tracks, but it will of course be much larger.
Fig 4. says run 2256 though this is not given in the text of good runs.
>Yes, I looked at 2256, as well as the others. The only ps file I had
>showing adc's was from 2256, but this doesn't change in the (very
>few!) runs I have looked at.
This brings up the issues of agreeing to a set of comon runs. We
(FR,EJK and I) had a discussion today on )this item. I will propose to
choose
a) a set of zero field runs 90, 60 and 45 deg for MRS
4 and 5 deg for FS
>Sounds good.
b) A set of B<>0 A and B pol for 90, 45, 60 deg as well as 4,5 that we
choose as a standard so far. The runs picked so far is with no
special consideration, apart from the 2235, 2238 being early high
statistics runs.
>It would be particularly nice to use runs with downscaled ZDC
(i.e. not so many trigger 4 events).
Q: how much statistics do you need (More is better of course)
>This is a good question...my guess is that to do reasonable
>calibrations one needs ~200 counts in the TDC spectra. Because the
>target is not fixed, this means that we really need ~200 *tracks*
>reconstructed to the vertex (to obtain track lenght and p). I've not
>done any serious estimates yet, but it doesn't look like it will be a
>problem in the FFS. The MRS may be more problematic, though, since
>the number of TPM2 (and thus global) tracks is lower. Coupled with
>this is the fact that TOFW has twice as many slats as TOF1.
>If we blindly say that we lose a factor of two for #of tracks and two
>for #of slats, we'll need 4 times the number of MRS events than FFS.
>Of course, when the entire FS is operational, the numbers will be
>closer. So...what does this tell us? In run 2513, I sorted ~40K
>events to obtain the plot shown in figure 7. Note that this has
>no further vertex cut than the (minimal) requirement that trigger 6
>fired. Placing a +/-2ns cut on the (as yet uncalibrated!)
>BBL(fastest)-BBR(fastest) reduces the number of counts by a factor of
>5-10 (and cleans the spectrum a bit as well.). On this basis, I
>would estimate that a calibration of TOF1 requires on the order of
>40-50K events with the same trigger mix as run 2513, and on the order
>of 200K events with for TOFW. My estimate for TOF1 is the lower
>limit, and the for the TOFW is too uncertain to worry too much
>about.
>So the short answer is I need twice the statistics, whatever they are;-)
Comment: I have written a TrackSelect.C Program/Macro that realy helps
a lot and propose to put a set of files (in root format) that has
i) vertex -20<v<20 and ii) at least one track in tpm2
This reduces the size a lot, of course it is biased because tracking
is done , and may have non-optimal parameters; for studies it is
probably ok.
>For what I am looking at, bias is really OK, if not preffered. For
>TOF calibrations and PID (testing in particular!) we want tracks, and
>the more the merrier. Well, at least up to the point where the
>tracking can no longer handle them.
>Where does one find the secret treasure of flemming code?
Q: Do you understand the shape of the 'closest hit plot (Fig 3) that
seems already to deviate from
godd gaussian distruibution at teh 10% level of the peak. Does this
indicate missing tracks, or large background in TOF. Certainly for
TOFW the #hits is much greater than tracks!
>Note that this is for TOF1 not TOFW. I think that the misidentifying
>the "best" hit which is done very simply. I just loop over all hits,
>and for each hit I loop over tracks and plot the difference in x
>position. Since, for some events, both of these numbers are very
>large (>10!), I guess we get some junk. I'll do this again, but only
>look at events where there are fewer than X tracks in T2 and see how
>it looks. Also, I never remove hits from the loop, it could be that
>the same hit gets used more than once.
>Actually, I thought that this plot was very reassuring, since it
>shows that *most* of the local tracks in T2 actually point at hit in
>TOF1, and that the geometry is consistent, at least between T2 and
>TOF1.
> I am sending this to the list, because you have brought up
> some good questions, and it is easier to send this to everyone than
>add to my previous document.
>The other reason for sending this to the list is to announce that you
> have won the Weekly Carlsberg.
>What, you no doubt ask, is the Weekly Carlsberg?
>This is the prize to the person with the first pertintent and
> substantive comments to the HEHI weekly report.
> Sadly, this time you were the only contestant.
>Thanks for your input!
Flemming
------------------------------
Flemming Videbaek
Physics Department
Brookhaven National Laboratory
tlf: 631-344-4106
fax: 631-344-1334
videbaek@bnl.gov
Ian--------------------------------------------------------------------
| I.G. Bearden |
| Niels Bohr Institute Tb 3 email: bearden@nbi.dk |
| Blegdamsvej 17 phone: (+45) 35 32 53 23 |
| København Ø FAX: (+45) 31 42 10 16 |
| Danmark |
----------------------------------------------------------------------
This archive was generated by hypermail 2b29 : Wed Sep 13 2000 - 03:53:22 EDT