[BLAST_ANAWARE] Minutes of the 2005/05/11 analysis meeting

From: Michael Kohl (kohlm@mit.edu)
Date: Wed May 11 2005 - 14:41:55 EDT


Attending AS,BT,EG,BF,TA,DH,MK

-AS raised the question how well the electron spin angle is known and
 if there is any helicity dependence on this.
 The Siberian snake provides a precise conservation of the spin angle.
 A polarized electron bunch is injected in average at some angle at
 the target (up to 10 degrees off the longitudinal direction). The
 Siberian snake rotates the azimuthal angles by 180 degrees every
 turn, so after two turns the orientation is identical to the
 original one. While the single-turn angular distribution of electron
 spins of a single bunch would in general show a finite angle,
 stacking of close to 2000 bunches in the ring in a single turn gives
 an envelope that peaks very close to zero. Averaging over many turns
 makes the envelope symmetric around zero. What matters is the
 time-averaged polarization, the angle of which averages to zero. The
 effective longitudinal polarization may be somewhat smaller than the
 absolute polarization of a single bunch by averaging over the width
 of the angular distribution, but for a width of up to 10 degrees the
 loss of effective polarization is very small. It is a feature of the
 snake: Either the polarization remains finite and at the same time the
 average angle of the envelope is zero, OR if the average angle is
 non-zero the polarization can't be kept up.
 Any helicity dependence can only introduced by the laser
 properties. While the average spin direction at the target cannot be
 helicity-dependent (because of the snakes) the degree of polarization
 for helicity +1 and -1 potentially could.

-Discussion about randoms:
 Generally, if one measures two time intervalls of two events with two
 clocks t1 and t2, both started with the same common start signal and
 stopped individually, then one may either have a correlation between
 the two (because the "event" is a true coincidence event with common
 origin), or one has no correlation between the two (because both
 individual stops arose from two different processes). While the
 former leads to a delta-function like behavior of t1-t2, the latter
 leads to a triangular distribution of t1-t2, if both t1 and t2 are
 uniformly distributed.
 The above consideration works if the "common start time" is really
 equal for the two events, and if the two stop times are independent
 from each other, and also independent of the start signal. Here comes
 the catch: In BLAST, "common start" is generated by the event itself,
 i.e. it does not come from a third source, like a start counter would
 be. Therefore, one of the two times t1,t2 is always "self-timed",
 i.e. one of the two times is not a uniform distribution. Hence, the
 method of randomizing the true sample by event-mixing would not work
 because t1,t2 are not independently measured (common start mode, no
 start counter).

 The conjecture of all the above considerations were the concerns on
 how to estimate the random contribution to (e,e'n) and how it would
 affect the asymmetries and the extraction of GEn. Evaluating the Tnn
 spectrum after all cuts except for neutron information, an average 6%
 of the events is considered random, as estimated from the number of
 events in the unphysical region of times < t_gamma.
 The source for most of the random counts are the forward neutron
 counters, i.e. the angle of the neutral track for e,e'n random
 coincidences is mostly constrained to forward angles. Vitaliy's
 argumentation is like the following: At low Q2 where most of the
 e,e'n yield comes from, most of the randoms are suppressed once the
 angle of the neutral track is taken into account (from pure geometry,
 without evaluating absolute time-of-flight of the neutral
 track). This is equivalent to saying "randoms happen to be at high
 missing momentum, and only low pmiss events enter the final sample".
 I.e. most of the randoms (>90%?) of the 6% are cut away by requiring
 pmiss<200 MeV/c. The fraction of randoms that survive the pmiss cut
 becomes gradually larger if one gets to higher Q2, equivalent to
 getting to more forward angles. Now, the argument is that randoms
 that survive the trigger requirement (charged veto) are neutral,
 i.e. protons from possible e,e'p or from pion channels can't
 contribute to randoms. The selection of a proper quasielastic
 electron in the first place makes it very likely that the random
 events under consideration are in fact quasifree scattering events
 from a (polarized) neutron in deuterium, where the original neutron
 has been missed by the neutron detectors because of inefficiency (the
 probability to *not* detect an impacting neutron is twice as high as
 to detect it) and where these candidate events still made it into the
 trigger because a random hit at some other neutron counter provided
 the trigger. This type of random events, however, would carry the
 dynamical information of a quasifree e-n scattering and would
 therefore exhibit the same asymmetry like a true coincidence. The
 fact that a different neutron counter from the one that was supposed
 to fire was hit, translates into a misreconstruction of missing
 momentum. Now, at pmiss <200MeV/c and at higher Q2, the asymmetry
 does not depend strongly on pmiss. Therefore, the "misreconstructed
 missing momentum" would impose only very little influence on the
 asymmetry and thus on GEn (at higher Q2). At lower Q2, the pmiss
 dependence is stronger, however the abundance of randoms is
 suppressed, as argued above.

 We discussed what else could be done to prove/disprove VZ's
 expectation that randoms either don't contribute significantly at low
 Q2 or they do contribute at high Q2 but there the assigned pmiss is
 not very different from the expected one, i.e. the asymmetries should
 be about the same for randoms and trues.
 One should evauluate the beam-target vector asymmetry as a function
 of the neutron time-of-flight Tnn. First of all one would see wether
 the unphysical events at Tnn<t_gamma exhibit an asymmetry at
 all. Secondly, one could check if the asymmetry undergoes any
 discrete change when entering the physical region for Tnn. Thirdly,
 one should do this once without constraining electron kinematics to
 be quasielastic and once with (Q2/(2mw)-1<delta). Finally, Looking
 at the asymmetry as a function of Tnn could be done for various Q2
 bins and/or various detector groups (L15,L20,NC).

 Finally, one more thought (not discussed in the meeting):
 =========================================================
 One could think of a high-Q2 experiment where the neutron detector is
 completely replaced with a veto counter only. Every
 quasielastic-electron event that has no hit in the veto counter
 (ToF+Wch) but with the momentum transfer vector pointing well into
 the veto acceptance would be considered as (e,e'n), and if the
 asymmetry dependence on the missing momentum is negligible, the
 missing-momentum yield-weighted asymmetry average would be close to
 the free-neutron case (or small missing momentum).
 This way one would have a "neutron efficiency" of 1 ... In fact, we
 have all of those events where the neutron detector did not fire
 due to inefficiency but where the quasielastic electron was present
 in the singles trigger types. While trigger 3 and 7 (with prescale
 factors 10 and 3) require the Cerenkov, triggers 4 (prescale 100,
 multi-TOF in one sector, none in the other and no Cerenkov) and 6
 (prescale 1000, recently changed to 100, ToF12-15 single without
 Cerenkov but with neutron hit in the same sector; in the left sector
 we removed the neutron counter requirement recently). This is also a
 possible application for the lead glass to help identifying
 quasielastic electrons at large angles (mostly trigger 6). In
 principle, this could boost our "(e,e'n)" yields at high Q2 by a
 factor 3 ....
 Let me know what you think!

Please also let me know if you prefer shorter minutes;-)

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 - 14:07:32 EST