[Olympus] momentum spectrum
Brian S. Henderson
bhender1 at MIT.EDU
Tue May 13 11:14:01 EDT 2014
Hello all,
To follow up, I was able to isolate the problem in the MWPC digitization =
which caused the strange MC momentum distributions in Kirill's =
presentation yesterday. The gist of the problem was that MC hits were =
smeared as single 2D hits, rather than as 1D hits in each plane (which =
is what happens in the real detector). This caused the hit error to be =
overestimated for certain combinations of all three wires, which led to =
the excessive broadening of the MC momentum distribution that we saw. =
This effect was hidden prior to the geometry change by the geometry =
errors to a certain extent, and does not show up when the GEMs are =
included since their resolution dominates and "over-smeared" MWPC hits =
have less effect on the tracking.
I have attached a plot showing the results for a single run (7909), with =
the quickest improvement to the MWPC digitization that I could think =
of. Rather than taking a single 2D hit in the MWPC middle plane as the =
old MWPC digitization did, I take the three individual plane hits, smear =
them individually, and reconstruct a 2D hit in the same fashion as is =
done in real data. As you can see, this improves the agreement between =
real data and MC significantly (although of course it can still be =
improved, especially since this was the quickest/easiest improvement I =
could think of).
This issue further highlights the need to move to the use of 1D MWPC =
hits only rather than 2D hits, as I have alluded to in previous =
meetings. I will make this my top priority so that analyses using the =
MWPCs can continue with minimal delay.
Brian
On 05/13/2014 09:38 AM, Brian S. Henderson wrote:
> Hello all,
>
> To clarify, I would expect the MC for MWPCs only to look more like what
> is shown on the second slide of the document Stan sent around (and the
> addition of external bremsstrahlung would help to fill in the low
> momentum tail, as has been seen in the past). On the first slide of
> what Stan sent around, you can see some results I ran yesterday for
> MWPC-only tracking. As you can see, the real data momentum distribution
> looks more like we would expect than MC. While the broadness is due to
> the low resolution of the MWPCs alone, there is a shift in the
> peak/shape/broadening that indicates a problem in the MC for the MWPCs,
> most likely in the digitization stage. While the MC I ran yesterday
> agrees with the distributions show by Kirril yesterday, I think the
> appropriate conclusion to draw is that the MWPCs are still not well
> modeled in the MC.
>
> Since the agreement between data and MC used to be better (as I alluded
> to yesterday and as shown by the plot included from Yuri on slide 2 of
> what Stan sent), my main guess is that I introduced some sort of bug
> into the MWPC digitization when I changed things to account for the
> recent geometry changes. I am currently re-examining the MWPC
> digitization and testing for possible problems.
>
> Brian
>
> On 05/13/2014 09:26 AM, Belostotski, Stanislav wrote:
>> Dear Brian and colleagues,
>> Thanks a lot for your plot. In view of our Monday discussion, the
>> result is as expected.
>> The broad Kirill's distribution is due to reconstruction done without
>> GEMs with MWPC only. The MWPC spatial resolution is about 0.5mm while
>> the total deflection of the lepton in the magnetic field is 2. to
>> 2.5mm, i.e. momentum resolution (without GEMs) must be very poor.
>> This has also been demonstrated by Yuri in his MC studies.
>> I included Brian's plot in a the attached file where various variants
>> (with or without GEMs reconstruction) are presented.
>> A good cross check is made.
>> With best regards, StanB
>>
>> On 13.05.2014 14:43, Brian S. Henderson wrote:
>>> Hello again,
>>>
>>> I have attached a set of plots that looks at the various reconstruction
>>> parameters for MWPC only tracking for two combined electron runs (7909
>>> and 7910) and two combined positron runs (7950 and 7951) with the
>>> corresponding MC data sets. The only thing like a cut applied was to
>>> accept only the single best fit track per event (with very loose cuts on
>>> the max chi^2 and number of iterations).
>>>
>>> As you can see, the momentum distributions from the MC look similar to
>>> what you showed yesterday. It looks like something has definitely
>>> changed in the MC since I last used MWPC only tracking, so I apologize
>>> for questioning your momentum distributions yesterday. Oddly, the data
>>> looks more like what I would expect than the MC (at least for the first
>>> 10 plots which are left arm), especially the missing energy distribution
>>> which peaks at around 250 MeV for the MC. This makes me think that
>>> there could be an issue in the way the MWPC digitization was changed to
>>> correspond with the GDML changes made to fix the coordinate system
>>> issues. It also could be that the estimate of the MWPC resolution
>>> currently used in the digitization is not a good model. I will take a
>>> look at this right away this morning to see if I can figure anything
>>> out.
>>>
>>> Brian
>>>
>>> On 05/12/2014 12:59 PM, Suvorov, Kirill wrote:
>>>> Hi Brian,
>>>>
>>>> Could you please send pictures of reconstructed momentum distribution
>>>> for radiative generator (without GEM digitization) and tell what cuts
>>>> did you use?
>>>>
>>>> best regards, Kirill
>>>>
>>>> p.s. We trying to understand why my pictures of momentum distribution
>>>> are worse.
> _______________________________________________
> Olympus mailing list
> Olympus at mit.edu
> http://mailman.mit.edu/mailman/listinfo/olympus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: digi_improvement.png
Type: image/png
Size: 23989 bytes
Desc: not available
Url : http://mailman.mit.edu/pipermail/olympus/attachments/20140513/496007f=
f/attachment.png
More information about the Olympus
mailing list