[Olympus] Status of results and timeline

Mon Aug 8 16:49:24 EDT 2016

Dear Norik,

	I strongly disagree with your criticisms of the analyses by Brian and Rebecca and the procedure we have put forward for determining the OLYMPUS result for publication in PRL.

	For your first point.  If Brian, Rebecca, Axel, Jan, me, you, or anyone else use exactly the same analysis procedure as any of the given analyses; the result would be the same.  How would it be different except within some precision of the computer on which it was run?  Such an exercise would be meaningless.  On the monthly or weekly meetings the students routinely state the location of their analysis routines in the repository.  Anyone can look at the code and run it for themselves.  Please do so.  Let us know if you get a different result.

	That said you do not appreciate how complicated the tracking, event reconstruction, particle identification, detector acceptance and efficiency simulation, luminosity determination, etc. is.  It has taken the MIT group (with help from Lauren at ASU and Dmitry at Mainz) three years to get to this point with little help from anyone else in the collaboration.  That is not the way we wanted to do it; but it was the way we had to do it because there was little or no one else to do the work.

	Each student is doing their own analysis.  Please read Chapter 6 of Brian’s or Rebecca’s thesis.  Each approaches event selection, particle identification, etc. and then apply selection cuts and background subtraction in a different manner.  We do not expect exact agreement when they form the ratio.  However, that they agree so closely is encouraging and gives us confidence in the final result.  The difference is a systematic uncertainty related to the procedure that must be included in the accounting for the final uncertainty.  The differences arise not only from applying the procedure to the data but also from applying the same procedure to the MC data used to normalize the yields.

	Imagine two simple cuts: reconstructed beam energy and coplanarity.  Apply the same cuts but it different order.  The results will be different because of the tails of the distributions are different and contain different amounts of background and signal.  One tries to maximize signal and minimize background.  The students each approach it differently and get slightly different results.  It’s not surprising.  

	All students are using the same run list for their analysis and will bin the data in the same Q^2 and epsilon bins for comparison.

	For your point regarding Brian’s highest Q^2 point near 2 GeV^2 on page 45 of his defense.  You are right, the point is centered about 1.02.  BUT it is statistically consistent with 1.03.  If we look at slide 44.  We don’t believe the structure we see: drop at 1.45 GeV^2 followed by two higher points and then a drop again at 1.85 followed by another rise.  We don’t claim that we have measured resonances, structure, or anything like that in the ratio.  It’s not statistically significant.  What we do see is that both figures seem to agree with Bernauer’s curve (as much as it pains me to admit it) which rises to 1.03.  Or if you fit a straight line to the data Q^2 > 0.6 I expect it would be very close to 1.03 at 2 GeV.  There is nothing wrong with Brian’s statement.

	As for your Q^2 ratio plot no one understood it or understood what it meant to show or how this provided some criteria of the data quality.  No one at today’s meeting wished to discuss it.  Is this a good example or a bad example with a 10% variation between beam species. What tracked data sample was it run on?  Which runs?  What was the event selection?  How was background subtracted?  Did you normalize to the corresponding MC tracked simulation?  Where is your code?  What ratio did you get?

	In two weeks I hope we will show results from at least three different analyses.  We will compare results for leptons going into the left sector with the right sector.  And we will combine the results to show what we hope will be the ratio result from OLYMPUS.

	I apologize if this response seem harsh in any way.  That is not my intention.  But we have been working on the analysis for three years and want to finish and move on.

                                                     Cheers,
                                                             Douglas

26-415 M.I.T.                                  Tel: +1 (617) 258-7199
77 Massachusetts Avenue                        Fax: +1 (617) 258-5440
Cambridge, MA 02139, USA                       E-mail: hasell at mit.edu

> On Aug 8, 2016, at 1:51 PM, Norair Akopov <akopov at mail.desy.de> wrote:
> 
> 
> Dear Colleagues,
> 
> At first, my congratulations to Rebecca and Brian for successfully
> defended  theses :)
> 
> I had a look on their theses and again I recognized that we strongly need
> more strict and formal release procedure, which definitely supposes to
> have a comparison plot with the final ratio, BUT done with the same set od
> cuts, at least with the same binnng over Q2, \epsilon or lepton scatt.
> angle, then to see how the average values for Q2 and \epsilon differ for
> e-/e+ runs (do we need the center of bin  correction?). Supposing also Rebecca has plotted the final ratio with the stat. errors only (Brian has definitely noted that), I see
> that unfortunately, if we expect the systematics to be on
> the level of 1.5%, we can not claim about significant two photon effect at
> high Q2 (or low \epsilon) we have measured. BTW showing the R(Q2) on page
> 45 in his defense presentation , where the R(2.2) is 1.02, then Brian is
> writing in his preliminary conclusions that the ratio is rises up to 1.03
> at Q2=2.2? It's still Ok for the defense presentation , but not for
> a future journal paper :)
> 
> I was trying to introduce some quantitative criteria concerning the quality of analysis (Q2 plot), I'm not pretending this is the only possible criteria plot, but due to we are dealing with very strong elastic kinematics, we need to move from just visual level to more strict numerical criteria on how we are selcted the elastic events! It would be nice to see before the official release similar criteria plots from all independent analyzers.
> 
> 
> Best regard,
> Norik
> 
> On Mon, 8 Aug 2016, Douglas K Hasell wrote:
> 
>> Dear Colleagues,
>> 
>> 	Just to reiterate what was said during today’s collaboration meeting.
>> 
>> 	Rebecca Russell and Brian Henderson successfully presented and defended their theses last week.  Current versions of their theses are available on the OLYMPUS Wiki.  NOTE:  Please do not distribute these results publicly yet.  This is meant for internal consumption only.
>> 
>> 	Axel Schmidt will defend his thesis on Monday, 15 August, and this will also be available shortly after then.
>> 
>> 	Lauren Ice’s analysis is getting close to being final.
>> 
>> 	Colton O’Connor’s analysis will also be finished fairly soon.
>> 
>> 	Jan Bernauer will also have an independent analysis but won’t write a thesis for some strange reason.
>> 
>> 	The next meeting will be Monday, August 22 at 10:00 EST (16:00 CET).  It is hoped that we will have an initial good draft of the proposed PRL paper for release at that time.
>> 
>> 	The results shown in the PRL will be a combination of the completed analyses at that time.  Since the analyses are all using the same tracked data and MC; they are correlated and it is not possible to do a statistical average of the result.  However, we will combine them and assign the spread in the results as a systematic uncertainty from the analysis procedures.
>> 
>> 	There will not be a separate “write-up” of the analysis but there will be a presentation showing the analyses used in determining the results shown in the PRL and this will include a discussion of how the result was obtained including the uncertainties.
>> 
>> 	Details of the individual analyses are of course available in the theses.  Also there are internal reports on the Wiki with details of the luminosity determination used in the final analyses.  There are also various publications on the Wiki describing the detector and experiment, magnetic field mapping and parameterization, internal target, and SYMB.
>> 
>> 	The theses have details on the tracking, radiative corrections, target, magnetic field, luminosity determination, etc.
>> 
>> 	Following the presentation of the PRL paper on Monday August 22 we will ask for comments or questions until Tuesday, September 6 (Monday is Labour Day in North America) after which we hope to get approval from the group leaders to submit the PRL for publication and for presentation by Axel Schmidt at INPC 2016 in Adelaide 11-16 September.
>> 
>> 
>>                                                    Cheers,
>>                                                            Douglas
>> 
>> 26-415 M.I.T.                                  Tel: +1 (617) 258-7199
>> 77 Massachusetts Avenue                        Fax: +1 (617) 258-5440
>> Cambridge, MA 02139, USA                       E-mail: hasell at mit.edu
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> -------------- next part --------------
>> A non-text attachment was scrubbed...
>> Name: smime.p7s
>> Type: application/pkcs7-signature
>> Size: 1843 bytes
>> Desc: not available
>> Url : http://mailman.mit.edu/pipermail/olympus/attachments/20160808/b84084b7/attachment.bin
>> _______________________________________________
>> Olympus mailing list
>> Olympus at mit.edu
>> http://mailman.mit.edu/mailman/listinfo/olympus