Hi,
below are the minutes of yesterday's analysis meeting.
Only in case of a particular need we will hold a meeting next Wednesday
2006/9/27. In any case there will be a rehearsel for Bill Franklin's
presentation for SPIN2006, either at 13:30 or directly after the
eventual meeting.
We still have not decided on moving the meeting schedule.
Latest suggestion is every other Thursday at 9am. Let me know if you can
accomodate this with your schedules.
Best regards,
Michael
Minutes
-Status v3_4_21 recrunch (MK)
+See ppt
+Recrunch of runs 7002-16506 complete using run-by-run time
calibration generated by Mark
+Presently evaluating quality, see ppt
+Looking at three time correlations, for (e,e'p), (e,e'gamma) and
(e,e'n) events: if calib is correct, all measured time diffs should
agree with the time diff calculated from tracking, for all detector
pairings (16x16 TOFsxTOFs and 16x72 TOFxNeutron bars)
+Results are complete for hydrogen runs 12144-13278 and look very good,
(mean,sigma) for deltaT(e'p) without any further cuts except vertex
and track charge gives (-2.4ns/1.8ns) in v20, becoming (0.2ns/0.9ns) in
v21!!!
+Will announce victory (and update th links in RECRUNCHDIR) after having
completed the quality checks on the entire dataset
-Comparison data/Montecarlo for d(e,e'p) (AD)
+See files in folder DEGRUSH
+Montecarlo vs data for several variables for 2005 deuterium
+Normalized MC to maximum bin in each panel
-> Better uses only one global normalization, this gives a better
feeling for local inefficiencies
+Data with Cerenkov cut, inefficient (nonexistent) for rear angles,
->relax this cut for this investigation
+Considered MC uses Geant only for acceptance test but tossed
variables otherwise. Therefore, resolution (convolution with
estimated resolution function) has to be adjusted manually
+vs electron momentum: quite some disagreement in higher Q2 bins,
could be due to cerenkov cut
+vs. phi: disagreement at forward angles already present in AM's
thesis
+This MC choice was made to facilitate the variation of input
parameters (GEn,GMn,GEp,GMp) that affect event weights. This way
only one common event list needs to be generated
->event-weighted versus reconstructed MC
+As the event-weighted MC has no energy loss, data yield should be
energy-loss corrected (in average) before comparing with MC, all the
multiple scattering and Landau broadening then needs to be modeled
manually to give a realistic response (Gaussian smearing).
+EG has used Geant-MC to determine a parameterization of the energy
loss effect as a function of particle's energy, will send the result
in a separate email
+At least one detailed MC including reconstruction of propagated
events for one parameter choice of G_E,M^p,n with all multiple
scattering and eloss effects turned on in propagation and average
Eloss corrections applied in reconstruction is necessary to evaluate
the quality of the event-weighted MC (->EG)
+need to look at variables such as theta_e, theta_p, TOFnum, CERnum,
missing momentum components longitudinal, sideways within the
scattering plane and out-of scattering plane (in the physics
coordinate frame) for data/MC comparison
+When Eloss is applied to data and resolution modeled in MC, the
remaining shifts between MC and and data can be used to parameterize
kinematic corrections for the d(e,e'p) channel
-(e,e'n) analysis (EG)
+Showing asymmetries and MC for v3_4_21 2005-deuterium (see ppt)
+v21 data doesn't look drastically different from v17 except at
certain locations. However, both do not look "satisfactory".
+There are regions of disagreement between MC and data regardless of
what is assumed for GEn, resulting in a chi^2/d.o.f. at the minimum
much larger than 1, also the chi2 distributions do not look
parabolic in those cases (1st and 2nd bin)
+There is a technical problem in regard of activation of the proton
veto from the wire chamber hit multiplicity. Since the
lrd-recrunches (v17-21) are done from the DST which only contains track
information, multiplicity variables are unavailable unless the
ntuples were generated with lrn from raw data. Therefore,
lrn-generated ntuples (v14,=ANALDIR) need to be accessed to process the
multiplicities in parallel with the events from the lrd-generated ntuples.
+Earlier on the phone we discussed a number of steps and detailed
checks that need to be done regarding:
+Data/MC yield comparisons in various variables, such as z,
theta(e,n), phi(e,n), mom(e,n), pmiss (also components);
+(e,e'n) rate monitor (after good cuts) vs run number and
vs. detector channels;
+Not mix things when doing comparisons between version
(e.g. different runlists)
+On time correction for neutron: Tnn is based on tdcdiff
measurement and pathlengths for electron and neutron. The
corrected electron track along with the neutron angles allow to
calculate Tnncalc. Comparison of Tnncalc with Tnn should be
used for event selection instead of missingmass (this way
negative Tnn's or shorter than photon-Tnn in case of a
miscalibration would not be rejected through the Mmiss evaluation)
+In the previous version of correcting Tnn, the difference
Tnn-Tnncalc was first parameterized as a function of Q2 and then
this function was used to correct Tnn event-by-event (i.e. the
correction itself was averaged)
+Plan: use Tnncalc for further processing of quantities that
depend on neutron momentum (this way the correction is
event-by-event and not averaged). Hope is that the uncertainty
of Tnncalc is less than for Tnn from the time measurement.
+A way to verify that Tnncalc has a better resolution than Tnn
would be to compare reconstructed vs. tossed MC values of these
two variables (if there wasn't the pathlength problem in MC).
+Also, the MC for the first and second bin look suspicious,
contracting to zero at pmiss ~=0.16-0.20, but not in the third and
fourth bin. VZ tried to convince me that this is real (or at least
in Arenhoevels calculations), but a zero-sensitivity to a variation
of GEn is hard to understand. Anyway it is important to also compare
the MC yields with data yields for the same bins/plots!
-aob
+status prl: confirmed receipt; 4 lines too long
+collaboration meeting on Friday 2006/11/17
+new meeting schedule: proposed every other Thursday 13:30
with occasional Bates Lunch seminars with 3pm limit to commute to
campus for departmental colloquium. HOWEVER: Adam may have classes
from noon on, and there may be senior campus lunch meetings.
Therefore we consider every other Thursday at 9am.
LET ME KNOW IF YOU HAVE OBJECTIONS!
Regards,
Michael
+-------------------------------------+--------------------------+
| Office: | Home: |
|-------------------------------------|--------------------------|
| Dr. Michael Kohl | Michael Kohl |
| Laboratory for Nuclear Science | 5 Ibbetson Street |
| MIT-Bates Linear Accelerator Center | Somerville, MA 02143 |
| Middleton, MA 01949 | U.S.A. |
| U.S.A. | |
| - - - - - - - - - - - - | - - - - - - - - -|
| Email: kohlm@mit.edu | K.Michael.Kohl@gmx.de |
| Work: +1-617-253-9207 | Home: +1-617-629-3147 |
| Fax: +1-617-253-9599 | Mobile: +1-978-580-4190 |
| http://blast.lns.mit.edu | |
+-------------------------------------+--------------------------+
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:33 EST