[G8b_run] Fwd: g8b TOF paddle issue
Eugene Pasyuk
pasyuk at jlab.org
Wed Mar 20 12:01:42 EDT 2013
So far 7 people responded to doodle poll. The earliest best time slot at the moment is Tuesday 12:30 JLab time.
-Eugene
----- Original Message -----
> From: "Eugene Pasyuk" <pasyuk at jlab.org>
> To: "g8b run" <g8b_run at jlab.org>
> Sent: Tuesday, March 19, 2013 4:36:38 PM
> Subject: [G8b_run] Fwd: g8b TOF paddle issue
>
>
>
>
> ----- Forwarded Message -----
> From: "Eugene Pasyuk" <pasyuk at jlab.org>
> To: "Michael Dugger" <dugger at jlab.org>
> Cc: "Ken Livingston" <Kenneth.Livingston at glasgow.ac.uk>, "Franz
> Klein" <fklein at jlab.org>, "Brian Vernarsky" <soulish at gmail.com>,
> "Volker Crede" <crede at fsu.edu>, "Patrick Collins"
> <pcollins at jlab.org>, "Charles Hanretty" <hanretty at jlab.org>
> Sent: Tuesday, March 19, 2013 4:21:10 PM
> Subject: Re: g8b TOF paddle issue
>
> Let's have a meeting. I set up doodle:
> http://doodle.com/ycqshd8fpnda66mx
>
> -Eugene
>
> ----- Original Message -----
> > From: "Michael Dugger" <dugger at jlab.org>
> > To: "Volker Crede" <crede at fsu.edu>
> > Cc: "Ken Livingston" <Kenneth.Livingston at glasgow.ac.uk>, "Franz
> > Klein" <fklein at jlab.org>, "Brian Vernarsky"
> > <soulish at gmail.com>, "Eugene Pasyuk" <pasyuk at jlab.org>
> > Sent: Tuesday, March 19, 2013 3:36:51 PM
> > Subject: Re: g8b TOF paddle issue
> >
> >
> > Hi,
> >
> > This could be scary.
> >
> > From what I understand, this problem will create phi dependent
> > inefficiencies. This will especially true for single track events.
> > However,
> > if an analysis uses the data to self analyze the phi dependent
> > efficiencies (e.g. forming quantities like (Y_PARA -
> > Y_PERP)/(Y_PARA
> > +
> > Y_PERP)), then the recooking of data will only help add statistics
> > for
> > those type of analyses.
> >
> > Since I have two channels of g8b that have gone through analysis
> > review,
> > and Volker is either finished with analysis review, or nearly so,
> > it
> > is
> > important that we determine if those analysis can remain as is.
> >
> > If it is determined that all of the analyses must use recooked
> > data,
> > we
> > need to do the recooking as soon as possible.
> >
> > Take care,
> > Michael
> >
> > On Tue, 19 Mar 2013, Volker Crede wrote:
> >
> > > Hi all,
> > >
> > > It sounds really scary. Yes, we should meet and discuss this.
> > >
> > > Volker
> > >
> > >
> > > On Mar 19, 2013, at 12:34 PM, Michael Dugger <dugger at jlab.org>
> > > wrote:
> > >
> > >> Hi,
> > >>
> > >> Having a meeting makes sense.
> > >>
> > >> Take care,
> > >> Michael
> > >>
> > >> On Tue, 19 Mar 2013, Ken Livingston wrote:
> > >>
> > >>> Hi All,
> > >>> It does indeed look like we cooked with the version that was in
> > >>> /home/clasg8/g8/packages/
> > >>> where, as Brian says, we had
> > >>> #define SC_TDC_MAX 4096
> > >>>
> > >>> if I look at set_env_g8b.sh see
> > >>> export CLAS_PACK=/u/home/${RUN_USER}/g8/packages
> > >>>
> > >>> I think we do need to consider recalibrating the TOF and
> > >>> recooking the data. At least starting with one coherent peak
> > >>> setting.
> > >>>
> > >>> What do you all think?
> > >>> We should at least have a meeting to discuss this.
> > >>>
> > >>> Cheers,
> > >>> Ken
> > >>>
> > >>>
> > >>>
> > >>>
> > >>>
> > >>> On 19/03/13 13:43, Franz Klein wrote:
> > >>>> sorry, replied only to Eugene ...
> > >>>> ----- Forwarded Message -----
> > >>>> From: "Franz Klein" <fklein at jlab.org>
> > >>>> To: "Eugene Pasyuk" <pasyuk at jlab.org>
> > >>>> Sent: Tuesday, March 19, 2013 9:38:42 AM
> > >>>> Subject: Re: g8b TOF paddle issue
> > >>>> Yes, Eugene, I worked on the copy_pipeline.cc routine in the
> > >>>> beginning of g9a ... mostly because the SC, EC banks took the
> > >>>> first TDC entry from SCT, ECT instead of the one which is in
> > >>>> the 'right' (expected) range.
> > >>>> So maybe I'll take a look first into the g8b version of this
> > >>>> routine as well as SCT, ECT, REF.
> > >>>> ----- Original Message -----
> > >>>> From: "Eugene Pasyuk" <pasyuk at jlab.org>
> > >>>> To: "Franz Klein" <fklein at jlab.org>
> > >>>> Cc: "Michael Dugger" <dugger at jlab.org>, crede at fsu.edu, "Brian
> > >>>> Vernarsky" <soulish at gmail.com>, "Kenneth Livingston"
> > >>>> <Kenneth.Livingston at glasgow.ac.uk>
> > >>>> Sent: Tuesday, March 19, 2013 9:29:45 AM
> > >>>> Subject: Re: g8b TOF paddle issue
> > >>>> One thing which comes to mind is that at some point after
> > >>>> pipeline TDC were installed we discovered that there is some
> > >>>> issue in on-line procedure of matching TDC and ADC from SC and
> > >>>> SCT banks and remaking of raw SC bank. The procedure was
> > >>>> revisited and as I recall we used new matching procedure in
> > >>>> user_ana for g9a data. May be this is the reason for g8b
> > >>>> problem?
> > >>>> -Eugene
> > >>>> ----- Original Message -----
> > >>>>> From: "Franz Klein" <fklein at jlab.org>
> > >>>>> To: "Kenneth Livingston" <Kenneth.Livingston at glasgow.ac.uk>
> > >>>>> Cc: "Michael Dugger" <dugger at jlab.org>, "Eugene Pasyuk"
> > >>>>> <pasyuk at jlab.org>, crede at fsu.edu, "Brian Vernarsky"
> > >>>>> <soulish at gmail.com>
> > >>>>> Sent: Tuesday, March 19, 2013 9:23:00 AM
> > >>>>> Subject: Re: g8b TOF paddle issue
> > >>>>> Ken,
> > >>>>> unfortunately I don't remember whether we checked or changed
> > >>>>> the cut
> > >>>>> on overflows. We took the code from eg3 (Joern) which assumes
> > >>>>> TDC
> > >>>>> entries in the range ~1000 to ~6000 (depending on particle
> > >>>>> and
> > >>>>> REF
> > >>>>> signal).
> > >>>>> However, Brian found out quite some time ago that often in
> > >>>>> multi-track events many track candidates weren't getting into
> > >>>>> time-based tracking due to missing timing information from ST
> > >>>>> and/or
> > >>>>> SC.
> > >>>>> For sure we made the mistake to cut out all SC paddles in the
> > >>>>> reconstruction for which the calibration didn't work out
> > >>>>> (instead of
> > >>>>> letting students worry about SC paddle cuts later in their
> > >>>>> analysis). In that sense it might be a good idea to recheck
> > >>>>> ST
> > >>>>> and
> > >>>>> SC calibrations and maybe rerun user_ana for some coh.edge
> > >>>>> setting
> > >>>>> on the farm.
> > >>>>> Maybe as a first step, Brian should tar his source code he
> > >>>>> copied
> > >>>>> from the clasg8 account and we verify whether it is the code
> > >>>>> which
> > >>>>> was used for g8b reconstruction.
> > >>>>> Cheers,
> > >>>>> Franz
> > >>>>> ----- Original Message -----
> > >>>>> From: "Ken Livingston" <Kenneth.Livingston at glasgow.ac.uk>
> > >>>>> To: "Brian Vernarsky" <soulish at gmail.com>
> > >>>>> Cc: "Michael Dugger" <dugger at jlab.org>, "Eugene Pasyuk"
> > >>>>> <pasyuk at jlab.org>, crede at fsu.edu, "Franz Klein"
> > >>>>> <fklein at jlab.org>
> > >>>>> Sent: Tuesday, March 19, 2013 8:18:45 AM
> > >>>>> Subject: Re: g8b TOF paddle issue
> > >>>>> Hi Brian,
> > >>>>> There's certainly something funny going on there which we
> > >>>>> need
> > >>>>> to get
> > >>>>> to
> > >>>>> the bottom of.
> > >>>>> However, I'm not sure your diagnosis is correct.
> > >>>>> As far as I can see, when I log on to the clasg8 account cd
> > >>>>> $CLAS_PACK
> > >>>>> directory is set to
> > >>>>> /home/clasg8/packages, not /home/clasg8/g8/packages/ where
> > >>>>> you
> > >>>>> mention
> > >>>>> finding the header file with the old version of sc.h.
> > >>>>> In the clasg8 account version of $CLAS_PACK/include/sc.h the
> > >>>>> relevant
> > >>>>> line is
> > >>>>> #define SC_TDC_MAX 8192 /* pipeline */
> > >>>>> Do any of the rest of you have any opinion or memory on that?
> > >>>>> Regards,
> > >>>>> Ken
> > >>>>> On 18/03/13 22:03, Brian Vernarsky wrote:
> > >>>>>> Hello all,
> > >>>>>> In trying to finalize my results for the spin density matrix
> > >>>>>> elements
> > >>>>>> for the omega I have run into a serious issue with the g8b
> > >>>>>> dataset.
> > >>>>>> When I project out my spin density matrix elements I found
> > >>>>>> that the
> > >>>>>> unpolarized elements, which can be compared to previous
> > >>>>>> experiments
> > >>>>>> (g11a most notably), were slightly offset with respect to
> > >>>>>> those
> > >>>>>> previous measurements. I have shown these elements at
> > >>>>>> previous
> > >>>>>> CLAS
> > >>>>>> Hadron meetings so some of you may be aware of this.
> > >>>>>> I traced the source of my problem back to the fact that the
> > >>>>>> phi
> > >>>>>> distributions of the particles were skewed and this was not
> > >>>>>> properly
> > >>>>>> reflected in the monte carlo. Looking further into the
> > >>>>>> issue
> > >>>>>> I
> > >>>>>> discovered that the start counters have a serious issue with
> > >>>>>> detecting
> > >>>>>> individual pions; the efficiency varies wildly from paddle
> > >>>>>> to
> > >>>>>> paddle.
> > >>>>>> However, since we had a one-track trigger and I am looking
> > >>>>>> for
> > >>>>>> three-track events with a proton, this was not the source of
> > >>>>>> my
> > >>>>>> issue,
> > >>>>>> however it could be a serious issue for anyone looking into
> > >>>>>> single
> > >>>>>> pion reactions. The source of my issue is particles that
> > >>>>>> have
> > >>>>>> a
> > >>>>>> TOF
> > >>>>>> paddle set to 0, indicating there is a problem with them.
> > >>>>>> As
> > >>>>>> you
> > >>>>>> can
> > >>>>>> see in the enclosed file phi_distributions_g8b.pdf the
> > >>>>>> actual
> > >>>>>> phi
> > >>>>>> distribution for each particle (in black) looks to be ok,
> > >>>>>> however
> > >>>>>> when
> > >>>>>> we remove the particles with TOF paddle = 0 (in red) and are
> > >>>>>> left
> > >>>>>> with
> > >>>>>> just the particles with a "good" TOF paddle (in blue) there
> > >>>>>> is
> > >>>>>> a
> > >>>>>> pretty severe problem, mostly for the pions.
> > >>>>>> In looking at the TOF paddle = 0 events for the two pions
> > >>>>>> you
> > >>>>>> can
> > >>>>>> notice that they are almost the exact same, whereas the
> > >>>>>> proton
> > >>>>>> has
> > >>>>>> significantly less events with a "bad" TOF paddle. And
> > >>>>>> there
> > >>>>>> is a
> > >>>>>> definite phi dependence to the bad events. Certainly we
> > >>>>>> should
> > >>>>>> expect
> > >>>>>> some events with bad TOF paddles in every channel, however
> > >>>>>> this is
> > >>>>>> accounting for almost half of the total events in sector 5
> > >>>>>> for
> > >>>>>> the
> > >>>>>> pions. There is no physical reason for this to be so.
> > >>>>>> I have found the source of the problem to be in the TDCs of
> > >>>>>> the TOF
> > >>>>>> paddles. In the file tdcs_and_adcs_g8b.pdf I have shown the
> > >>>>>> values
> > >>>>>> from the TDCs and ADCs for the TOF paddles, in the order
> > >>>>>> TDCL,
> > >>>>>> ADCL,
> > >>>>>> TDCR, ADCR, for the proton (top), pi+ (middle) and pi-
> > >>>>>> (bottom),
> > >>>>>> integrated over all paddles and sectors for one amorphous
> > >>>>>> run
> > >>>>>> (48133).
> > >>>>>> The blue on each plot are "good" events and the red are
> > >>>>>> "bad", or
> > >>>>>> TOF
> > >>>>>> = 0, events. It is very evident from the TDC plots for all
> > >>>>>> particles
> > >>>>>> that something turns on at channel 4096. I have tried to
> > >>>>>> illustrate
> > >>>>>> that by drawing a vertical line at 4096 on the TDC plots.
> > >>>>>> The
> > >>>>>> ADC
> > >>>>>> plots show nothing out of the ordinary.
> > >>>>>> What happened is that before the g8b run the TDCs were
> > >>>>>> changed
> > >>>>>> from
> > >>>>>> a
> > >>>>>> 12-bit system to a 16-bit system, changing the maximum
> > >>>>>> channel
> > >>>>>> from
> > >>>>>> 4096 to 65536. However, this change was not made in the
> > >>>>>> code
> > >>>>>> as in
> > >>>>>> the g8b cooking code (found at /home/clasg8/g8/packages/)
> > >>>>>> the
> > >>>>>> file
> > >>>>>> include/sc.h we find
> > >>>>>> #define SC_TDC_MAX 4096
> > >>>>>> instead of 8192 like we find in other experiments after the
> > >>>>>> TDCs
> > >>>>>> were
> > >>>>>> changed. (It seems that while we effectively have 65,536
> > >>>>>> channels
> > >>>>>> in
> > >>>>>> practice we only use 8192.) Any particle that registered
> > >>>>>> above
> > >>>>>> 4096
> > >>>>>> on a TDC was deemed a "bad" event in that TDC. If the
> > >>>>>> particle had
> > >>>>>> a
> > >>>>>> "bad" TDCL but a "good" TDCR then it is still possible to
> > >>>>>> recover
> > >>>>>> the
> > >>>>>> events, and vice versa, though not all particles could be
> > >>>>>> recovered
> > >>>>>> in
> > >>>>>> this way. However, if both TDCs measured a value above 4096
> > >>>>>> then
> > >>>>>> the
> > >>>>>> particle was deemed "bad" and assigned TOF paddle = 0.
> > >>>>>> For a signifcant number of TOF paddles the range of TDC
> > >>>>>> channels
> > >>>>>> was
> > >>>>>> from 3800-4200, though some skewed higher or lower than
> > >>>>>> others.
> > >>>>>> For
> > >>>>>> paddles where the distribution was quite high as many as
> > >>>>>> half
> > >>>>>> of
> > >>>>>> the
> > >>>>>> pions were cut and a signficant number of protons were also
> > >>>>>> cut.
> > >>>>>> This
> > >>>>>> led to difficulty in calibrating the paddle and so the
> > >>>>>> entire
> > >>>>>> paddle
> > >>>>>> was cut. There were 31 paddles cut in g8b, more than 10% of
> > >>>>>> the
> > >>>>>> total
> > >>>>>> number of paddles.
> > >>>>>> For the pions the cut is pretty evenly distributed across
> > >>>>>> the
> > >>>>>> momentum
> > >>>>>> spectrum, but for protons it cuts only events above ~550 MeV
> > >>>>>> as can
> > >>>>>> be
> > >>>>>> seen in the file p_dist_bad_paddles.pdf, which shows the
> > >>>>>> momentum
> > >>>>>> distribution for protons that have been ascribed TOF paddle
> > >>>>>> =
> > >>>>>> 0.
> > >>>>>> Now,
> > >>>>>> this is a small fraction of the total number of proton
> > >>>>>> events,
> > >>>>>> but
> > >>>>>> it
> > >>>>>> is still significant.
> > >>>>>> Based on all of this I think that it is imperative that we
> > >>>>>> recook
> > >>>>>> the
> > >>>>>> g8b dataset. We will need to change the value of SC_TDC_MAX
> > >>>>>> and
> > >>>>>> then
> > >>>>>> recalibrate the TOF paddles; hopefully the ones that were
> > >>>>>> considered
> > >>>>>> "good" before will not need much work, but of the ones
> > >>>>>> considered
> > >>>>>> "bad" before most of them can almost certainly be salvaged.
> > >>>>>> This
> > >>>>>> recook is necessary because it is clear that this is
> > >>>>>> significantly
> > >>>>>> affecting the phi and momentum distributions of all
> > >>>>>> particles
> > >>>>>> and
> > >>>>>> thus
> > >>>>>> will affect all measurements that depend on phi, momentum,
> > >>>>>> TOF
> > >>>>>> or
> > >>>>>> particle ID. The plus side is that we can recover a large
> > >>>>>> number
> > >>>>>> of
> > >>>>>> paddles and thus increase our statistics significantly,
> > >>>>>> especially
> > >>>>>> with multi-particle events. Any other minor issues that
> > >>>>>> wouldn't
> > >>>>>> necessitate a recook on their own can also be addressed
> > >>>>>> before
> > >>>>>> we
> > >>>>>> start. And lastly I think that it would be a very good idea
> > >>>>>> to
> > >>>>>> include some more banks in the cooked files this time than
> > >>>>>> we
> > >>>>>> did
> > >>>>>> before; to find the TDC and ADC values I had to go to the
> > >>>>>> raw
> > >>>>>> files
> > >>>>>> without any PID info and then cross check between that and
> > >>>>>> the
> > >>>>>> cooked
> > >>>>>> file to make sure I had the right particle, certainly the SC
> > >>>>>> banks
> > >>>>>> should be included in the newly cooked files.
> > >>>>>> Let me know what you think, if you would like any additional
> > >>>>>> information or plots just ask.
> > >>>>>> Brian
> > >>>>> --
> > >>>>> =======================================================
> > >>>>> Ken Livingston
> > >>>>> Dept. of Physics & Astronomy, Tel: +44 141 330 6428
> > >>>>> University of Glasgow, Fax: +44 141 330 5889
> > >>>>> Glasgow G12 8QQ.
> > >>>>> Scotland. UK.
> > >>>>> =======================================================
> > >>>>> --
> > >>>>> ===============================================================
> > >>>>> Franz J. Klein, Associate Professor
> > >>>>> CUA, Department of Physics
> > >>>>> Washington, DC 20064
> > >>>>> office: Hannan Hall 206 phone: 202-319-6190
> > >>>>> or: Jefferson Lab,CC F-243 phone: 757-269-6672
> > >>>>> ---------------------------------------------------------------
> > >>>
> > >>>
> > >>> --
> > >>> =======================================================
> > >>> Ken Livingston
> > >>>
> > >>> Dept. of Physics & Astronomy, Tel: +44 141 330 6428
> > >>> University of Glasgow, Fax: +44 141 330 5889
> > >>> Glasgow G12 8QQ.
> > >>> Scotland. UK.
> > >>> =======================================================
> > >>
> > >
> > >
> >
> _______________________________________________
> G8b_run mailing list
> G8b_run at jlab.org
> https://mailman.jlab.org/mailman/listinfo/g8b_run
>
More information about the G8b_run
mailing list