[Hps-analysis] py/pz reconstruction

[Nelson, Timothy Knight]

Not entirely. However, there are a few possibilities.  First, low momentum tracks need better alignment than high momentum tracks to show this effect.  Keep in mind that alignment between the strips on the two layers that contain all the hits must be within the width of the spikes, which looks to be equivalent to better than 4 microns.  Second, multiple scattering assumed by GBL, larger for positrons (having lower momentum, on average), may smear this out. It’s also true that at low-momentum, the fact that tanLambda is dy/ds (and not dy/dz) begins to matter for seeing this effect. BTW, I’m not sure I understand why the positrons are peaked at lower abs(tanLambda) than the electrons.  What sample is this?  Also, the presence of many electrons that have lower abs(tanLambda) than our acceptance is puzzling to me.  Maybe we’d better see a comprehensive set of plots and an explanation of this sample before jumping to any more conclusions.

I should point out that it is also possible that the reason we see this on the top and not the bottom has to do with the ordering of the views (axial/stereo) which are opposite in the top vs. bottom.  This matters because of GBL, which places kinks at layers that tend to make tracks pass exactly through hits (since hit resolution is much better than multiple scattering-limited pointing resolution).

As I said, if you want to verify this hypothesis, you at least need to make some more plots.  So, in addition to my previous suggestion (excluding 1-hit clusters), another interesting one would be for low-momentum electrons. 

Tim

> On Apr 1, 2016, at 4:00 AM, Stepan Stepanyan <stepanya at jlab.org> wrote:
> 
> Is it clear then why positives do not see significant single hit clusters or "shockingly" good alignment?
> 
> Stepan
> 
> On 4/1/16 12:49 AM, Nelson, Timothy Knight wrote:
>> Hi Sebouh,
>> 
>> Pretty cool. I count 19 peaks in ~0.113 units of tanLambda, so each peak is ~0.113/19 = 0.000595 in tanLambda (which is dy/dz between the two hits in L1 and L2, assuming that this is the track state at the vertex).  For layers separated by dz=100 mm (e.g. layers 1 and 2), that means the peaks represent a difference in rise of 100mm*0.000595 = 60 microns.  That is the strip pitch.  For this pattern to appear, some things need to be true:
>> 
>> - a significant fraction of clusters have only one strip: true.
>> - angular alignment around the z-axis between L1 and L2 must be so good (shockingly good, actually) that the positions of strips are coherent independent of the position of the hits along the strips: apparently true in the top but not in the bottom.
>> 
>> So… here’s a test: make the same plots but require that tracks have more than one strip in the cluster (in either layer).  I believe that should kill this pattern.
>> 
>> Tim
>> 
>>> On Mar 31, 2016, at 8:04 PM, Sebouh Paul <sebouh.paul at gmail.com> wrote:
>>> 
>>> Here are the plots using tanLambda, for GBL tracks.
>>> 
>>> On Thu, Mar 31, 2016 at 7:27 PM, Nelson, Timothy Knight <tknelson at slac.stanford.edu> wrote:
>>> Hi Sebouh,
>>> 
>>> Can you please provide the tanLambda plot with appropriately fine binning?
>>> 
>>> Tim
>>> 
>>>> On Mar 31, 2016, at 3:17 PM, Sebouh Paul <sebouh.paul at gmail.com> wrote:
>>>> 
>>>> when using very fine binning, (less than 10 mrad) I see in the spectrum of py/pz (which is equivalent to tan_lambda/cos(phi0) )  a sort of regular ondulation, but only when looking at electrons in the top half of the detector from GBL tracks.
>>>> The attached screenshots are from v0 skim of run 5772 in pass6.
>>>> 
>>>> One of the attached screenshots is with the seed tracks (no ondulation), the second one is with gbl tracks (which has ondulations only in electrons on the top half of the detector), and the third one is a finer-binned zoom in of the second graph to highlight the ondulation.
>>>> 
>>>> Do any of you all have any idea what might be causing this?  This happens in both the DST and the LCIO, so it's probably not caused by the DST-maker
>>>> 
>>>> Note:  I also found that if I drop the cos(phi0) from the formula that I am plotting, I still see the ondulation, so this might just be something to do with tanLambda, and not necessarily the combined formula tanLambda/cos(theta)
>>>> 
>>>> <Screen Shot 2016-03-31 at 5.50.34 PM.png><Screen Shot 2016-03-31 at 5.56.34 PM.png><Screen Shot 2016-03-31 at 6.02.06 PM.png>_______________________________________________
>>>> Hps-analysis mailing list
>>>> Hps-analysis at jlab.org
>>>> https://mailman.jlab.org/mailman/listinfo/hps-analysis
>>> 
>>> <Screen Shot 2016-03-31 at 10.53.55 PM.png><Screen Shot 2016-03-31 at 10.57.40 PM.png><Screen Shot 2016-03-31 at 11.00.25 PM.png>
>> 
>> _______________________________________________
>> Hps-analysis mailing list
>> Hps-analysis at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/hps-analysis
> 
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis