Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 First look at new MC files First look at reconstruction output from the newly- generated “mock-data” MC files. –These contain the following improvements:

Similar presentations


Presentation on theme: "1 First look at new MC files First look at reconstruction output from the newly- generated “mock-data” MC files. –These contain the following improvements:"— Presentation transcript:

1 1 First look at new MC files First look at reconstruction output from the newly- generated “mock-data” MC files. –These contain the following improvements: Generator: NEUGEN v3 with INTRANUKE Detector simulation: DetSim + various geometry improvements Up-to-date NuMI fluxes I have run Standard Reconstruction on ~20000 events from the new far detector MC file and have compared the results with previous studies. Some differences in the processing include the use of different ROOT and minossoft versions – not yet clear whether this has a significant effect on the result. D. A. Petyt 4 th Feb ‘04

2 2 Truth quantities – left: old file, right: new file True neutrino energy y distribution Some enhancement evident in this region New file has a much flatter y distribution – is this an expected result of the recent changes to NEUGEN?

3 3 y versus E – top: old file, bottom: new file 0-2 GeV >4 GeV 2-4 GeV 0-2 GeV Flatter y distribution implies that NC/CC separation will be more difficult in new MC file

4 4 Truth quantities contd. True muon momentum True shower energy These differences can be explained by the flatter y distribution in thenew MC file No obvious effect of INTRANUKE in this plot – presumably washed out by the change in the y distribution?

5 5 Some numbers Track-finding efficiency is higher in the new MC file, at the expense of a much higher probability of finding a track in NC events –What is the cause of this change? –One possible contribution – apparent 8 plane tracking threshold in old file compared to 6 plane threshold in new (see PDF plots later in this talk). However, it seems that this can only partially explain the difference. QuantityOld fileNew file Number of CC events 1482015164 CC events in fid vol (true vtx) 1026810653 CC events with reco. track 9087 (88.5%)9580 (89.9%) Number of NC events 49683743 NC events in fid vol (true vtx) 34642641 NC events with reco. track 691 (19.9%)1777 (67.2%)

6 6 PDFs used for event selection – old file Number of planes in track Fraction of digits in track Track pulse height per plane CC NC

7 7 PDFs used for event selection – new file Number of planes in track Fraction of digits in track Track pulse height per plane What’s happening in this variable? Why does hit fraction cut off at 0.8?

8 8 A closer look at PDF variable #2 Hit fraction for new file seems stuck below 0.8. Why is this? –My guess is that this is due to many low pulse height hits in the event (a.k.a. cross-talk) that are not included in the track. These hits are not present (or are at a much lower level) in the old file Fraction of digits in track, plotted for y<0.2 OLDNEW

9 9 New variable Since cross-talk hits have low pulse height, a more reasonable track- like estimator in this case is the fraction of total pulse height contained in the track. The two plots below show the distribution of this variable for CC events in the old file (left) and the new file (right). The cut-off at 0.8 in the new file is eliminated. The low ‘shoulder’ in the right-hand distribution is more promiment, but this is likely to be a manifestation of the flatter y-distribution in the new file OLDNEW

10 10 PDFs – old file, with pulse height fraction variable

11 11 PDFs – new file

12 12 PDFs with 50 plane cut – old file

13 13 PDFs with 50 plane cut – new file

14 14 The initial conclusions of this study are: –The new MC file has a flatter y-distribution (is this expected from the recent NEUGEN updates?) –NC and CC distributions are less distinct as a result – NC/CC separation will be more difficult (this was also the conclusion of a similar NEUGEN2/NEUGEN3 comparison I did with off-axis MC) –The presence of what I claim are cross-talk hits makes pulse height-based variables preferable to hit-based variables. –No obvious signatures of INTRANUKE effects yet, although the ntuple format is not ideally suited to this (can’t study pion multiplicities for example). I haven’t looked at event separation/selection efficiencies yet, although this is the logical next step –Expect efficiencies/purities to be somewhat worse for the reasons stated above Summary and future work


Download ppt "1 First look at new MC files First look at reconstruction output from the newly- generated “mock-data” MC files. –These contain the following improvements:"

Similar presentations


Ads by Google