APA-Crossing Muons, Shower Recon Mike Wallbank University of She - - PowerPoint PPT Presentation

APA-Crossing Muons, Shower Recon

Mike Wallbank University of Sheffield 20/7/16

M Wallbank (Sheffield)

Update

• A few weeks ago, I started looking at through-going muons which pass through the APA frames

as a ‘way in’ to some 35ton data analysis — gave initial talk at this meeting on 8th June.

• Been a bit lot busy recently so little progress! I’m back working full-time now.
• Two points from last talk:
• The 35t was the first LArTPC with readout planes within an active volume so this is the first

time we can test these particular detector effects.

• Initially, I am looking at validating the T0 given by the counters using these through-going
• tracks. Will possibly lead to more studies…
• Not sure how well this method will work — but it’s giving some interesting results…
• Also have my wild dream of trying to reconstruct some showers…

2

M Wallbank (Sheffield)

What I’ve Done/Doing

• Sliced and filtered on events with a NS/EW trigger such that the track passes through the APAs.
• Runs 14427-14428, 14435-14470 (Alex’s good runs).
• Note last talk I was only using NS triggers. However, I had an issue with track length in the

SDV — if using EW counters close to the frames I’m guaranteed a good amount of hits.

• Initial look at the T0:
• Take tracks which pass through APAs and find the T0 as given by the counters;
• Try varying the T0 within a range around the counter T0 to see if the tracks line up any

better;

• Use this to determine if the counter T0 is systematically offset in some way.

3

M Wallbank (Sheffield)

Idea:

• The T0 has the effect of

making tracks from the different drift volumes misaligned.

• Correcting for T0 will join

the segments into a single straight line.

• In this first instance, this is

what I’m attempting to do!

4

T0

M Wallbank (Sheffield)

Hit Selection

• Using a similar method to Dom and probably many others!

5

T0 ‘correction’

M Wallbank (Sheffield)

‘Correct’ T0 Determination

• Fit a straight line through all the points;
• Determine the distance of each hit from this

central axis;

• chi-square:


• distance of hit from axis is taken as
• bserved — expected;
• RMS of these distances taken as error.
• Vary the T0 over a range and determine chi-

sq for each:

6

nhits

X

i

⇣oi − ei σi ⌘2

Split with 300 pre-trigger ticks

M Wallbank (Sheffield)

T0 Distribution

• I’ve attempted to automate this process so a distribution of T0 can be considered… My

jobs are still running though :(

• I’ll update next week (by which time they will have hopefully finished!).
• Gut feeling is this method appears reliable enough to make some conclusions… but the

initial results are quite a way off suspected.

7

M Wallbank (Sheffield)

Finding Systematics?

• I feel there’s something biasing the result more than the method… fair?
• (I know we need more events! I’m working on it…)
• Ghost triggers are discounted, varying the drift velocity (Tingjun suggested this could be

a large source of systematic uncertainty) doesn’t seem to make huge differences…

8

• Split with 300 pre-trigger ticks —> expect a T0 of

~300.

• From the chi squared, it appears to favour ~365, and

in the range 355 — 398 (+-1 chi-square).

M Wallbank (Sheffield)

Drift Velocity Uncertainty

• I’ve tried varying the drift velocity to see what effect this has on the uncertainty…
• Here’s +- 20%…

9

• 20%

389 375—422 +20% 358 348—363

• Even 2x nominal 250V/cm drift velocity converges around 350 ticks.

M Wallbank (Sheffield)

Where From Here?

• Obviously I’ll have to wait to run this over more events to see if this is the general trend.
• I can’t see any immediate issues with the method however… Happy to be disagreed with

though!

• Any other ideas of things that can be affecting these results?

10

M Wallbank (Sheffield)

Aside: Standard Reconstruction

• In my attempts to reconstruct some showers over the past few days I’ve run into a few issues with

the reconstruction.

• The main problem concerned the code unsticker and the waveform filter configurations:
• The module label used by the filterwf module is RawDigitModuleLabel (and also accepts the

instance as a separate parameter)

• Unless some configuration was happening that I’ve completely missed, this meant that the

standard reco file used the default RawDigitModuleLabel configured in dunetpc (SplitterInput:TPC) so ignored the unsticker completely.

• I changed the standard files in develop yesterday.

11 I’d really like to remove this parameter for consistency… thoughts?

M Wallbank (Sheffield)

Summary

• Was hoping to have a lot more to show today but jobs are running, the problems with the

shower reconstruction are taking a while to understand.

• Hopefully next week I’ll have all this! … :)

12