1. Search for A’ from p0  A’ g
T. Spadaro
M. Mirra
G. Rovelli

2. Goal of the present analysis
Detect dark photons, feebly coupled to SM g via: L = e Amn Fmn
Produce dark photons from the decay chain: K+  p+p0, p0  A’ g
Assume invisible decay of the A’ (or extremely long-lived A’)
Signature: one photon, missing mass peaking around the A’ mass
Data sample: 4 runs of period A (6610, 6614, 6625, 6670)
Reconstruction revision r1421
MC sample: private production, 11 mass bins from 30 to 130 MeV
Reconstruction revision r physics-list bug fix
06/04/17
Exotic physics meeting - CERN

3. Status of A’ invisible searches
Recent results: NA64, Babar; in future, expect results from Belle/Belle-II
06/04/17
Exotic physics meeting - CERN

4. Selection algorithm: “p0 tag”
pnn L0 trigger used for signal search
control trigger used to normalize signal counts to the p0 produced
One-track-selection
One good quality downstream track connected to CHOD candidate, 15 < p < 35 GeV
Association to 1 and only 1 upstream track from GTK: |TGTK – TEvent(RICH)| < 400 ps
Upstream and downstream tracks approach each other to better than 5 mm
Point of closest approach: 115 < Zvertex < 165 m, to reduce early-decay effect
Kaon momentum cut: 72 < pK < 78 GeV
Pion positive identification:
RICH: single ring association
!MUV3: no hit connected in time and space wrt expected impact
LKr, MUV1, MUV2 combined info: R. Aliberti’s tool, P(p) > 0.8
06/04/17
Exotic physics meeting - CERN

5. Selection algorithm: “p0 tag”
Cut on the squared missing mass < (PK - Pπ)2 < GeV2
(PK - Pπ)2 (GeV2)
Total number of events selected with control trigger:
> considering the downscale factor this leads to ~207 × 106 p0
06/04/17
Exotic physics meeting - CERN

6. Selection algorithm: “one g”
One-photon selection:
No signal from the LAV, SAC and IRC systems must be present with a time consistent with the TEvent(RICH) (standard tools)
Exactly one standard cluster with E > 1.5 GeV and -5 ns < TLKr – TEvent(RICH) < 10 ns
Missing momentum pmiss = pK – pp - pg must point to LKr
Cut on additional veto:
No activity in the LKr corresponding to auxiliary clusters of total energy above 1 GeV around the expected missing momentum extrapolated point;
No RICH hits in time in addition to pion track hits
Isolation: distance photon-expected vs photon found > 20 cm
Remain with events from PNN trigger corresponding to a single photon inefficiency ~ 8 × 10-4
06/04/17
Exotic physics meeting - CERN

7. Data control trigger: 2 γ on LKr, simulate 1g loss
Expected signal
Search for signal in missing mass windows, from pmiss = pK – pπ - pγ
Mmiss2 = Emiss2 – pmiss2
Data control trigger: 2 γ on LKr, simulate 1g loss
MC:
- mA’ = 30 MeV
- mA’ = 60 MeV
- mA’ = 90 MeV
Mmiss2 (GeV2)
06/04/17
Exotic physics meeting - CERN

8. Exotic physics meeting - CERN
Master formula
Determine number of signal events:
Correct for one-photon selection (esel) and trigger (etrg) efficiencies
Count number of signal events (nsig) in a mass-dependent Mmiss2 window
Account for mass-window acceptance (emass)
Normalize to number of p0 counts from control trigger, downscale corrected
Assumptions under above formula:
Acceptance for single-track selection exactly cancels (that must be true)
Assume control trigger efficiency for one-track events = 1
Assume nominal downscale when correcting (control trigger “re-start” to be checked)
To evaluate e, account for phase space factor ~(p*)3 = (1 – mA’2/mp02)3
06/04/17
Exotic physics meeting - CERN

9. Efficiency corrections: one-g selection
Efficiency to detect one g cluster with missing momentum towards LKr
Drop at 130 MeV, understood from kinematics consideration
General behavior reproduced using toy-MC simulation
Multiple energy scales because of:
g in LKr acceptance + isolation wrt p cluster + missing momentum pointing to LKr
Include 12% inefficiency from IRC, SAC, LKr random veto losses
Efficiency from MC
MA’ (MeV)
06/04/17
Exotic physics meeting - CERN

10. Efficiency corrections: LKr veto
Require no additional LKr cluster from auxiliary reconstruction:
|dt| < 8 ns, distance to the expected g impact point < 40 cm, Ematch > 1 GeV
Reject 53% of remaining background (events with |M2miss| < GeV2), dominated by loss of low-energy clusters
2% random veto evaluated from control trigger + pp0 selection with p0  gg
Fraction of matched events
Fake veto probability
Standard reconstruction
Present default
Auxiliary reconstruction
Expected (GeV)
Eaux threshold (GeV)
06/04/17
Exotic physics meeting - CERN

11. Efficiency corrections: vetoes
Require no additional hit from RICH in time with the event
Effect: reject 5.5% of the residual background (events with |M2miss| < GeV2)
Random loss at 1% checked with data
From MC, inefficiency 0.5% 1.5% for MA’ = 30 MeV  130 MeV
In time RICH data hits to within 800 ps
distance to p ring center normalized to p ring radius
06/04/17
Exotic physics meeting - CERN

12. Efficiency corrections: Isolation cut
To avoid LKr-energy sharing issues, require > 20 cm distance between g and expected impact from pmiss to LKr:
Extremely effective for background in the negative tail (~ x10 wrt peak)
Pay a price in terms of efficiency, particularly at high A’ masses (as expected)
Fraction of rejected events
Efficiency from MC
Mmiss2 (GeV2)
MA’ (MeV)
06/04/17
Exotic physics meeting - CERN

13. Efficiency corrections: trigger
Trigger efficiency for signal obtained by reweighting MC kinematics:
1. Use data spectrum for pion energy release in LKr in momentum bins
2. Use probability from data for a photon to fire a newCHOD sector in energy bins
3. Combine g and p to simulate QX or !QX condition, apply L0 efficiency from data
Fraction of events with pnn trigger bit
Particles in !QX
0-g sample: two g’s on LAV
1-g sample: 1-g on LAV, !LAV12, 1g on LKr
2-g sample: !LAV, 2g on LKr, exclude condition with g’s in QX
For a given total energy, efficiency depends on number of particles in the LKr
Particles in QX
Total energy in the LKr (MeV)
06/04/17
Exotic physics meeting - CERN

14. Efficiency corrections: trigger
Linear dependence on the A’ mass:
Higher A’ masses, lower photon energy  higher efficiency for LKr L0
Positive bias from LKr-L0 malfunctioning important at low A’ masses
Efficiency from MC (reweighted w data input)
MA’ (MeV)
06/04/17
Exotic physics meeting - CERN

15. Exotic physics meeting - CERN
Expected background
Evaluate expected background from negative Mmiss2 side band
Assume purely symmetric resolution effects
Tested using data from control trigger with p0  gg: looks OK
 Use data for bkg estimate: same trigger, detector, and beam conditions
 Statistics of background subtraction same order of data
Data
Symmetrized data
Data – symmetrized data
Mmiss2 (GeV2)
Mmiss2 (GeV2)
06/04/17
Exotic physics meeting - CERN

16. Signal determination: the concept
To extract signal use “CLS method”:
Count data and background expected in 1-s windows around each A’ mass hypothesis
Statistics: Poisson(data | signal + background) / Poisson( data | background)
Vary the signal strength, evalute 90% CL coverage using toy-MC integration
To evaluate compatibility of limit wrt expected fluctuations:
“Expected” limit: assume no signal, vary data observed in toy-MC, evaluate limit spread
Example, MA’ = 90 MeV:
Count in < Mmiss2 < GeV2
12 observed in data
8 background expected
CLS = probability that signal is > S
Limit found
Median expected limit
1-s limit band
2-s limit band
90% CL coverage
Signal S
06/04/17
Exotic physics meeting - CERN

17. Signal determination: results
Limits observed are statistically compatible with fluctuations from the background-only hypothesis e
Limit if
Nobs = Nbkg
Limit found
1-s limit band
2-s limit band
MA’ (MeV)
06/04/17
Exotic physics meeting - CERN

18. Upper limit determination
NA62 limits in an interesting region (4 runs only used)
NA62
06/04/17
Exotic physics meeting - CERN

19. Exotic physics meeting - CERN
To-do list
Write extended documentation including all analysis details
Checking master formula assumptions can be synergic with pnn
using newly produced MC
from dedicated data-based studies
Check other background sources using newly produced MC
do not expect relevant contribution
Go for publication approval
06/04/17
Exotic physics meeting - CERN