[Solid_tracking] some initial Qsq / alignment studies

[Zhiwen Zhao]

I see
Maybe it's worthy to discuss it in a meeting later.
It would be more effective than email.

Zhiwen


On 6/18/2015 8:52 AM, Robert Michaels wrote:
> Thanks for the reply.  Paul Souder explained to me that in order
> to get an acceptable error in Qsq it is well-known (by some)
> that I need to use a reconstruction based on the GEM hits which
> have correction functions that Richard Holmes formulated.
>
> I suppose that using the result of Weizhi's track fitting
> would be equivalent and eventually necessary for solving
> the pattern-recognition problem in a multitrack situation.
>
> - Bob
>
> On Wed, 17 Jun 2015, Zhiwen Zhao wrote:
>
>> hi, Bob
>>
>> I am CCing this to solid_tracking at jlab.org because only a few of us
>> are on the list.
>> also because the emaillist will keep a copy of email in archive.
>>
>> The fake hits in space where there's no GEM plane is because current
>> GEMC use the hit processing "flux" to record the hits I gave you.
>> If one low energy tracks go forward, then bend back by field and thus
>> pass a GEMC plane twice or more, the hit position will be average out
>> between all hits.
>> "flux" is not the real GEM digitization, so just ignore those tracks
>> for now.
>>
>> I didn't find the comment about pt on 201 page of latest pCDR, but it
>> should be proportional for sure.
>>
>> So when you fit the helix from the hits on GEM1,2,3 to determine
>> radius of curvature and thus Pt, is field a fitting parameter or is it
>> given as a constant?
>> See can see the field variation from plots in pCDR
>>
>> The file I gave you have particles vertex distributed evenly along
>> 40cm target length.
>> When you determine theta by fit to hit on GEM 1,2,3 in r VS z, how did
>> you use the vertex information?
>>
>> Does thetadiff.gif have 0.36 degree shift you mentioned already? it
>> still doesn't sit at 0 now.
>>
>> How do you get "systematics are about 3% in Qsq"?
>>
>> I am not sure if qsq_compare.gif is best way to compare.
>> how about Qsq different VS p and theta in 2D plot?
>>
>> I guess is the final question is
>> if GEM alignnment tolerance gives Qsq error larger than tracking
>> error, we don't need very good tracking method for the study.
>> if GEM alignnment tolerance gives Qsq error smaller than tracking
>> error, we have push tracking error by using better method in order to
>> see the other effect.
>>
>> Thanks
>>
>> Zhiwen
>>
>> On 6/10/2015 2:57 PM, Robert Michaels wrote:
>>> I didn't use the solid_tracking list because I couldn't
>>> figure out who is on that list, and I don't think this
>>> email needs to go the entire world.  Too many questions.
>>>
>>> The goal of this study is to find how the GEM alignnment tolerance
>>> affects the systematic error in four-momentum squared Qsq.
>>> At this point, I have constructed Qsq, and I mainly need to
>>> tweak the GEM positions to measure how it affects the errors.
>>> Here is an update.
>>>
>>> I use the GEM hits to reconstruct Qsq using two things:
>>>
>>> 1. scattering angle theta:
>>>     tan(theta) = slope of R vs Z for GEM hits, where R = sqrt(X^2+Y^2).
>>>     I restrict the fit to the first 3 GEM chambers where B is uniform.
>>>
>>> 2. Transverse momentum Pt is proportional to the radius of curvature
>>>     of the helix.  Using the constraint that the beam goes through
>>>     the origin, I derived an expression for R based on the GEM hits.
>>>     This results in the following correlation, which is admittedly
>>>     imperfect:
>>>
>>> http://userweb.jlab.org/~rom/solid/ptvsrad.gif
>>>
>>> One really silly problem is that the Monte Carlo gives me hits
>>> in GEMs for radii like 20 cm and 15 m, but I think these are outside
>>> the physical size of the GEM, so there must be a misunderstanding.
>>>
>>> Nevertheless, based on 1 and 2, I tried to reconstruct Q^2 and
>>> compare to MC as shown below.
>>>
>>> First, the scattering angle compared to MC.
>>>
>>> http://userweb.jlab.org/~rom/solid/thetadiff.gif
>>>
>>> In order to align the peak at zero I had to adjust the reconstructed
>>> angle by 0.36 degrees; perhaps because something is unknown about
>>> the geometry.
>>>
>>> Reconstructed Qsq vs MC Qsq
>>> http://userweb.jlab.org/~rom/solid/qsq_compare.gif
>>>
>>> The systematics are about 3% in Qsq at the moment.
>>>
>>> For the purpose of the study (of GEM alignment) I don't need
>>> zero error, I mainly need to look for deviations that occur with
>>> misalignments.  Of course, it would be nice if the initial systematics
>>> (presently 3%) would be smaller, but I assume these will reduce
>>> when we use better track fitting.
>>>
>>> I've made an initial study of the sensitivity to the GEM
>>> chamber position, but I don't fully trust the results yet,
>>> so I'll inform you about that later.
>>>
>>> Also, I don't undestand the comment on page 201 of the pre-CDR that
>>> the radius is inversely proportional to Pt.  Isn't it proportional,
>>> not inversely proportional ?  Am I missing some elementary physics
>>> here ?
>>>
>>> For example, see this treatise about charge particles moving in a
>>> uniform B field:
>>> http://www.worldscientific.com/doi/pdf/10.1142/9789812798657_bmatter
>>>
>>> -------------------------------------------------------
>>> Robert W. Michaels, Staff Scientist
>>> http://userweb.jlab.org/~rom    (757) 269 7410
>>> Thomas Jefferson National Accelerator Facility
>>> 12000 Jefferson Ave, Newport News, VA 23606 USA
>>>
>>>
>>>
>>