Hi all,
The *Big* wiggle problem in Eg1b (ie sector 5, 17.5<th<22.5 deg)
does not go away when I re-made the plots using events with
trigger bits 7 or 8 only (i.e. no Cherenkov in the trigger, only
E/P for PID cut). This seems to be something in the DC itself:
it shows up clearly as a several cm shift in reconstructed z,
and also in trl1_theta vrsus vertex theta.
For ruh 27323 (2 gev inbending), there also seems to be a
dead region in sector 3 for theta<20 deg. This dead region is
still there without the Cherenkov in the trigger or event selection.
This doesn't affect the asymmetry analysis, because these
events seem to just be gone, rather than shifted the wrong
kinematics.
For this run, I don't see any other *Big* DC problems:
certainly there can be some more suble effects that are hard
to see.
Yours, Peter
Prof. Peter Bosted
email: bosted@jlab.org
phone: (757) 269-5851
address: Jefferson Lab Suite 6, 12000 Jefferson Ave, Newport News VA 23606
On Fri, 29 Feb 2008, burkert wrote:
> Hi all, we should be careful when discussing the "wiggle problem". There
> are potentially a number of sources for "theta wiggles" and they depend
> on the specific configuration, e.g. inbending vs outbending electrons.
> (If you do see wiggles for non-electron tracks ignore the following, as
> such effects are more likely related to DC tracking aspects.)
>
> A 2 degree wiggle pattern seen for electrons can be the result of
> inefficiencies of the Cerenkov counter due to the mirror system. Since
> we have 18 mirrors covering about 9 - 45 degrees in geometrical angles,
> i.e. 36 degrees in Delta(theta) you do expect something like a 2deg.
> oscillation pattern just from the geometry of the mirror system if
> inefficient regions, such as fringes have not been eliminated carefully.
> This is particularly important for or out-bending electrons. I order to
> study the problem in for electrons in detail I suggest to make the
> following selections:
>
> 1) Look at in-bending electrons: Make narrow fiducial cuts that make
> absolutely sure that you avoid the fringe regions of the Cerenkov
> counter. Make a cut around the sector mid-plane to eliminate events near
> the mid-plane where the efficiency is known to be somewhat lower than in
> other regions. Then look for wiggles at fixed momentum.
>
> 2) Look at outbending electrons in eg1a,b: Here it is hard to avoid
> seeing wiggles from Cerenkov counter inefficiencies as the Cerenkov has
> been designed biased toward high efficiency for in-bending tracks and as
> a result has significant inefficiencies due to the mirror arrangement.
> This pattern has been studied by Vipuli in her PhD thesis and fiducial
> cuts have been developed to correct for it. This should, of course, not
> be the case for eg4 as this Cerenkov detector has specifically be
> designed for out-bending electrons, i.e. wiggles in eg4 must have a
> different origin.
>
>
> A wiggle pattern that is very different from 2 degree , or is seen for
> pions, is likely not related to the Cerenkov counter but more to effects
> that Sebastian mentioned earlier.
>
> Volker
>
>
>
>
> Sebastian Kuhn wrote:
> > Dear all,
> >
> > sorry for this "spam" mail, but several people have expressed interest
> > in our findings about "theta oscillations", which may be relevant also
> > for E4 and E1-DVCS.
> >
> > Shortly, the problem seems to be that with some CLAS experiments (with
> > targets upstream of the regular target position?), there is a clear
> > pattern in the angular distribution of scattered electrons (other
> > particles remain to be checked), in particular if one selects a tight
> > momentum region (e.g., 2.0 GeV/c - 2.2 GeV/c). This pattern can
> > (sometimes) be seen in theta-reconstructed as well as theta at the
> > first DC superlayer and sometimes even for various projections of y or
> > theta on the scintillator or EC plane.
> >
> > This problem was originally encountered during the analysis of EG1a
> > (see my message below for relevant links). Some additional plots can
> > be found at
> > http://www.jlab.org/Hall-B/secure/eg1/jyun/ecyi.gif
> > http://www.jlab.org/Hall-B/secure/eg1/alex/waves/oscillations.html
> > http://www.jlab.org/Hall-B/secure/eg1/tforest/Wiggle/Wiggle.html (for
> > explanation)
> >
> > The most recent analysis on EG1b data is from Nevzat Guler - see
> > http://www.jlab.org/Hall-B/secure/eg1/EG2000/nevzat/ThetaWigglePage_wfid/thetaWiggle.html
> >
> > Slava Tkachenko has looked at this for BoNuS - see
> > http://www.jlab.org/~slava/Check44/Check44_18/ThetaEL/0227/
> >
> > It is not clear what exactly is happening, but the most likely
> > explanation is that tracks are getting shifted out of the "valleys"
> > into the "peaks", leading to wrong theta assignment by about 1-2
> > degrees. (It is less likely that the valleys are just inefficiencies).
> > This may or may not affect inclusive data, but it might be useful to
> > ascertain this for each experiment.
> >
> > - Sebastian
> >
> >
> > Sebastian Kuhn wrote:
> >> Dear (former) EG1ers (and especially "oldtimers"),
> >>
> >> I am trying to remember what our final conclusion was regarding the
> >> infamous "wiggle/theta oscillation" problem with EG1a - see
> >> http://www.jlab.org/ccc/mail_archives/CLAS/eg1_run/msg00274.html to
> >> jog your memory. I went through all my accumulated notes but couldn't
> >> find any conclusive solution (and my memory is just not as good as it
> >> used to be... ;-)
> >> It appears possible that we have a similar problem in EG1b, see
> >> http://www.jlab.org/Hall-B/secure/eg1/EG2000/sharon/theta_compare/theta_compare_5_6.gif
> >> (and BoNuS, for that matter! ->
> >> http://www.jlab.org/~slava/Check44/Check44_18/ThetaEL/thetael_raw_0211_44_1_pre_el5gev.jpg
> >> ).
> >>
> >> Any recollections are welcome - Sebastian
>
>