Event Topologies K + has two main decay modes + (BR = 64%) - - PowerPoint PPT Presentation

Selecting p→ ത νK Events Through K→μν and K→π+π0 Decay Chains

Dan Pershey Feb 6, 2019

Event Topologies

❑K+ has two main decay modes

• μ+ν (BR = 64%) – monochromatic muon
• π+π0 (BR = 21%) – totally reconstructable – can construct K+ invariant mass

❑Both have same topology – two Bragg peaks facing in the same direction

• Track is monochromatic, while a π0 comes out opposite the π+ in the π+π0 case

2

K+ e+ μ+ K+ γ γ π+ e+ μ+

Reconstructing GLOT’s

❑K+→μ+→e+ (or K+→π+→μ+→e+) events produce three contiguous tracks ❑Aim to “glue” together a set of ordered, reconstructed tracks (from pmtrack)

• Separated into longest track, and activity in front of or behind the longest track
• I’ve been referring to them as GLued Oriented Tracks (GLOT’s)

3

“Pre” segment “Post” segment

dEdx Traces from GLOT’s

❑Developed an analysis just focusing on the dEdx traces

• Currently only using hits reconstructed from the induction plane – room for improvement

4

Longest track Post segment Pre segment Track Length from Reco K+ Decay L = 0 L is negative, counting down from L = 0 dEdx (TPC charge)

Overall Event χ2

❑Since we only have three particles, and one is monochromatic, all events should

line up nicely

❑We can make signal templates from MC, with L=0 at kink between the longest

and pre portions, then make signal templates from MC

❑Calculate a χ2/dof for each event by dotting its dE trace into these templates

5

Track Length from Reco K+ Decay Track Length from Reco K+ Decay

K→μν sample K→π+π0 sample

Mass Fitting for GLOT Track Segments

❑Bethe-Bloch formula for ionization energy loss is parameterized by charged

particle mass and charge

❑Fitting to observed dE/dx gives you a best fit mass of the track

• Can run the fitter both “forward” and “backward” along the track
• Can get a sense of which direction the particle was tracked by the Δχ2 in each of these

two scenarios

• Important! For signal events, K and μ/π Bragg peaks are in the same direction

6

Fitting a hand- selected signal K+ track

7

Pre segment: true K

Bethe-Bloch Fit – Signal (K->μν)

mpre = 363 MeV mpre = 387 MeV Δχ2 = +87.5 mlong = 102 MeV mlong = 10 MeV Δχ2 = +21.6 Pre Segment Longest Track

mlong = 115 MeV mlong = 83 MeV Δχ2 = +92.3 mpre = 905 MeV mpre = 774 MeV Δχ2 = -55.7

Bethe-Bloch Fit – νμ CC Background

8

Pre segment: true proton Pre Segment Longest Track

Selecting Proton Decay Events – Fitted track masses

❑To start with a preselection, require that the longest track and pre portions

have a reconstructed mass between 10 and 2000 MeV

• Gives a 48% efficiency hit – mostly boils down to requiring 2+ tracks in the event

❑Require each of these sections reconstruct near physical values

• 50 < masslong < 300 MeV
• 100 < masspre < 800 MeV

9

Selecting Proton Decay Events – Fitted track direction

❑Again, put cuts on the longest and pre portions ❑The longest portion doesn’t give much discriminating power, but cut on it

anyway – it’s physical requiring the two to be similarly-directed

❑Require

• Δχ2

long > 0 and Δχ2 pre > 0

10

Selecting Proton Decay Events – μLL

❑Measure the “likeness” of the event to the dE trace to the MC prediction for

signal

❑Split the sample into two signal samples (for K→μν and K→π+π0):

• μLL < 1.3
• μLL > 1.3 && ππLL<2.3

11

μ sample Goes to ππLL cut ππ sample

Proton Decay Sample

❑We’re left with two samples proton decay events ❑Only atmospheric neutrino bkg included – very likely to dominate

• Only one bkg event selected (trkl = 21cm) –bkg prediction here is the track length after

masslong, masspre, and μLL/ππLL cuts, but normalized to give the total expected bkg

12

K→μν sample K→π+π0 sample

Tot signal: 2.46 Tot bkg: 1.5 Tot signal: 0.41 Tot bkg: 0

Calculating a Sensitivity

❑In the K→μν sample, optimize a signal range by optimizing S/sqrt(B)

• 48 < Track Length < 58 cm
• 1.939 signal events and 0.145 bkg events
• Efficiency: 17.4% for K→μν, or 11.1% of all p→ ത

νK events are selected in signal region

❑In the K→π+π0 sample, no bkg passed our cuts due to limited MC, so take

• 26 < Track Length < 36 cm
• 0.219 signal events
• Efficiency: 6.0% for K→π+π0, or 1.3% of all p→ ത

νK events are selected in signal region

❑Define sensitivity as the maximum half life that we would exclude to 90%,

assuming we see 0 events in 400 kton-years of data

• Pois(0 | N(τ)) = e-N(τ) = 0.1 → N(τ) = ln(10) = 2.303
• Notice above quoted numbers add up to 2.303, were calculated at our sensitivity

❑Sensitivity: τp/Br(p→ ത

νK) > 8.98e33 years (> 1e34 years at 446 kton-years)

• Or, τp/Br(p→ ത

νK) > 8.14e33 just using K→μν sample

13

Summary

❑We can select p→ ത

νK decays through K→μν and K→π+π0 topologies

• Selects 12.4% of p→ ത

νK decays

❑Most effective discriminator we have is the forward-backward Δχ2 fit to the

Bethe-Bloch prediction for dE/dx for the K candidate

❑Cutflow has some nice properties

• Cuts are all physically motivated – helpful for understanding systs. on our efficiency
• Track length not wrapped into the selection – largely responsible for why we can select

both μ and ππ topologies

❑Definitely room for improvement

• Incorporate induction plane hits into analysis
• Look for events where
• Look for ππ events where the K has a longer track than the π
• Tag K’s from ππ events by the invariant ππ mass in cases where the K isn’t tracked

14

Single Selected Background

❑Only one bkg event remains, has a track length of 21 cm

15

μLL

More Typical Bkg

16

μLL

More Typical Bkg

17