LETTER OPEN
doi:10.1038/nature14474

Observation of the rare B0
sRm1m2 decay from the

combined analysis of CMS and LHCb data
The CMS and LHCb collaborations*

The standard model of particle physics describes the fundamental
particles and their interactions via the strong, electromagnetic and
weak forces. It provides precise predictions for measurable quanti-
ties that can be tested experimentally. The probabilities, or branch-
ing fractions, of the strange B meson (B0

s ) and the B0 meson decaying
into two oppositely charged muons (m1 and m2) are especially inter-
esting because of their sensitivity to theories that extend the standard
model. The standard model predicts that the B0

s ?m1m2 and
B0?m1m2 decays are very rare, with about four of the former occur-
ring for every billion B0

s mesons produced, and one of the latter
occurring for every ten billion B0 mesons1. A difference in the
observed branching fractions with respect to the predictions of the
standard model would provide a direction in which the standard
model should be extended. Before the Large Hadron Collider (LHC)
at CERN2 started operating, no evidence for either decay mode had
been found. Upper limits on the branching fractions were an order
of magnitude above the standard model predictions. The CMS
(Compact Muon Solenoid) and LHCb (Large Hadron Collider beauty)
collaborations have performed a joint analysis of the data from
proton–proton collisions that they collected in 2011 at a centre-of-
mass energy of seven teraelectronvolts and in 2012 at eight teraelec-
tronvolts. Here we report the first observation of the B0

s ? m1m2

decay, with a statistical significance exceeding six standard deviations,
and the best measurement so far of its branching fraction.
Furthermore, we obtained evidence for the B0?m1m2 decay with
a statistical significance of three standard deviations. Both mea-
surements are statistically compatible with standard model predic-
tions and allow stringent constraints to be placed on theories beyond
the standard model. The LHC experiments will resume taking data in
2015, recording proton–proton collisions at a centre-of-mass energy
of 13 teraelectronvolts, which will approximately double the produc-
tion rates of B0

s and B0 mesons and lead to further improvements in
the precision of these crucial tests of the standard model.

Experimental particle physicists have been testing the predictions of
the standard model of particle physics (SM) with increasing precision
since the 1970s. Theoretical developments have kept pace by improving
the accuracy of the SM predictions as the experimental results gained in
precision. In the course of the past few decades, the SM has passed
critical tests derived from experiment, but it does not address some
profound questions about the nature of the Universe. For example, the
existence of dark matter, which has been confirmed by cosmological
data3, is not accommodated by the SM. It also fails to explain the origin
of the asymmetry between matter and antimatter, which after the Big
Bang led to the survival of the tiny amount of matter currently present
in the Universe3,4. Many theories have been proposed to modify the SM
to provide solutions to these open questions.

The B0
s and B0 mesons are unstable particles that decay via the weak

interaction. The measurement of the branching fractions of the very
rare decays of these mesons into a dimuon (m1m2) final state is espe-
cially interesting.

At the elementary level, the weak force is composed of a ‘charged
current’ and a ‘neutral current’ mediated by the W6 and Z0 bosons,

respectively. An example of the charged current is the decay of the p1

meson, which consists of an up (u) quark of electrical charge 12/3 of
the charge of the proton and a down (d) antiquark of charge 11/3. A
pictorial representation of this process, known as a Feynman diagram,
is shown in Fig. 1a. The u and d quarks are ‘first generation’ or lowest
mass quarks. Whenever a decay mode is specified in this Letter, the
charge conjugate mode is implied.

The B1 meson is similar to the p1, except that the light d antiquark
is replaced by the heavy ‘third generation’ (highest mass quarks)
beauty (b) antiquark, which has a charge of 11/3 and a mass of
,5 GeV/c2 (about five times the mass of a proton). The decay
B1R m1n, represented in Fig. 1b, is allowed but highly suppressed
because of angular momentum considerations (helicity suppression)
and because it involves transitions between quarks of different genera-
tions (CKM suppression), specifically the third and first generations of
quarks. All b hadrons, including the B1, B0

s and B0 mesons, decay
predominantly via the transition of the b antiquark to a ‘second gen-
eration’ (intermediate mass quarks) charm (c) antiquark, which is less
CKM suppressed, into final states with charmed hadrons. Many
allowed decay modes, which typically involve charmed hadrons and
other particles, have angular momentum configurations that are not
helicity suppressed.

The neutral B0
s meson is similar to the B1 except that the u quark is

replaced by a second generation strange (s) quark of charge 21/3. The
decay of the B0

s meson to two muons, shown in Fig. 1c, is forbidden at
the elementary level because the Z0 cannot couple directly to quarks of
different flavours, that is, there are no direct ‘flavour changing neutral
currents’. However, it is possible to respect this rule and still have this
decay occur through ‘higher order’ transitions such as those shown in
Fig. 1d and e. These are highly suppressed because each additional
interaction vertex reduces their probability of occurring significantly.
They are also helicity and CKM suppressed. Consequently, the
branching fraction for the B0

s?mzm{ decay is expected to be very
small compared to the dominant b antiquark to c antiquark transitions.
The corresponding decay of the B0 meson, where a d quark replaces the
s quark, is even more CKM suppressed because it requires a jump
across two quark generations rather than just one.

The branching fractions, B, of these two decays, accounting for
higher-order electromagnetic and strong interaction effects, and using
lattice quantum chromodynamics to compute the B0

s and B0 meson
decay constants5–7, are reliably calculated1 in the SM. Their values are
B(B0

s?mzm{)SM~(3:66+0:23)|10{9 and B(B0?mzm{)SM~
(1:06+0:09)|10{10.

Many theories that seek to go beyond the standard model (BSM)
include new phenomena and particles8,9, such as in the diagrams
shown in Fig. 1f and g, that can considerably modify the SM branching
fractions. In particular, theories with additional Higgs bosons10,11 pre-
dict possible enhancements to the branching fractions. A significant
deviation of either of the two branching fraction measurements from
the SM predictions would give insight on how the SM should be
extended. Alternatively, a measurement compatible with the SM could
provide strong constraints on BSM theories.

6 8 | N A T U R E | V O L 5 2 2 | 4 J U N E 2 0 1 5

*Lists of participants and their affiliations appear in the online version of the paper.

G2015 Macmillan Publishers Limited. All rights reserved

www.nature.com/doifinder/10.1038/nature14474


The ratio of the branching fractions of the two decay modes pro-
vides powerful discrimination among BSM theories12. It is predicted in
the SM (refs 1, 13 (updates available at http://itpwiki.unibe.ch/), 14,
15 (updated results and plots available at http://www.slac.stanford.
edu/xorg/hfag/)) to be R:B(B0?mzm{)SM=B(B0

s?mzm{)SM~
0:0295z0:0028

{0:0025. Notably, BSM theories with the property of minimal
flavour violation16 predict the same value as the SM for this ratio.

The first evidence for the decay B0
s?mzm{ was presented by the

LHCb collaboration in 201217. Both CMS and LHCb later published
results from all data collected in proton–proton collisions at centre-of-
mass energies of 7 TeV in 2011 and 8 TeV in 2012. The measurements
had comparable precision and were in good agreement18,19, although
neither of the individual results had sufficient precision to constitute
the first definitive observation of the B0

s decay to two muons.
In this Letter, the two sets of data are combined and analysed

simultaneously to exploit fully the statistical power of the data and
to account for the main correlations between them. The data corre-
spond to total integrated luminosities of 25 fb21 and 3 fb21 for the
CMS and LHCb experiments, respectively, equivalent to a total of
approximately 1012 B0

s and B0 mesons produced in the two experi-
ments together. Assuming the branching fractions given by the SM
and accounting for the detection efficiencies, the predicted numbers of
decays to be observed in the two experiments together are about 100
for B0

s?mzm{and 10 for B0 R m1m2.
The CMS20 and LHCb21 detectors are designed to measure SM phe-

nomena with high precision and search for possible deviations. The two
collaborations use different and complementary strategies. In addition to
performing a broad range of precision tests of the SM and studying the
newly-discovered Higgs boson22,23, CMS is designed to search for and
study new particles with masses from about 100 GeV/c2 to a few TeV/c2.
Since many of these new particles would be able to decay into b quarks
and many of the SM measurements also involve b quarks, the detection of
b-hadron decays was a key element in the design of CMS. The LHCb
collaboration has optimized its detector to study matter–antimatter
asymmetries and rare decays of particles containing b quarks, aiming
to detect deviations from precise SM predictions that would indicate
BSM effects. These different approaches, reflected in the design of the
detectors, lead to instrumentation of complementary angular regions
with respect to the LHC beams, to operation at different proton–proton
collision rates, and to selection of b quark events with different efficiency
(for experimental details, see Methods). In general, CMS operates at a
higher instantaneous luminosity than LHCb but has a lower efficiency
for reconstructing low-mass particles, resulting in a similar sensitivity to
LHCb for B0 or B0

s (denoted hereafter by B0
(s)) mesons decaying into two

muons.
Muons do not have strong nuclear interactions and are too mas-

sive to emit a substantial fraction of their energy by electromagnetic

radiation. This gives them the unique ability to penetrate dense mate-
rials, such as steel, and register signals in detectors embedded deep
within them. Both experiments use this characteristic to identify
muons.

The experiments follow similar data analysis strategies. Decays
compatible with B0

(s)?mzm{ (candidate decays) are found by com-
bining the reconstructed trajectories (tracks) of oppositely charged
particles identified as muons. The separation between genuine
B0

(s)?mzm{ decays and random combinations of two muons (com-
binatorial background), most often from semi-leptonic decays of two
different b hadrons, is achieved using the dimuon invariant mass,
mmzm{ , and the established characteristics of B0

(s)-meson decays. For
example, because of their lifetimes of about 1.5 ps and their production
at the LHC with momenta between a few GeV/c and ,100 GeV/c, B0

(s)
mesons travel up to a few centimetres before they decay. Therefore, the
B0

(s)?mzm{ ‘decay vertex’, from which the muons originate, is
required to be displaced with respect to the ‘production vertex’,
the point where the two protons collide. Furthermore, the negative
of the B0

(s) candidate’s momentum vector is required to point back to
the production vertex.

These criteria, amongst others that have some ability to distinguish
known signal events from background events, are combined into
boosted decision trees (BDTs)24–26. A BDT is an ensemble of decision
trees each placing different selection requirements on the individual
variables to achieve the best discrimination between ‘signal-like’ and
‘background-like’ events. Both experiments evaluated many variables
for their discriminating power and each chose the best set of about ten
to be used in its respective BDT. These include variables related to the
quality of the reconstructed tracks of the muons; kinematic variables
such as transverse momentum (with respect to the beam axis) of the
individual muons and of the B0

(s) candidate; variables related to the
decay vertex topology and fit quality, such as candidate decay length;
and isolation variables, which measure the activity in terms of other
particles in the vicinity of the two muons or their displaced vertex. A
BDT must be ‘trained’ on collections of known background and signal
events to generate the selection requirements on the variables and the
weights for each tree. In the case of CMS, the background events used
in the training are taken from intervals of dimuon mass above and
below the signal region in data, while simulated events are used for the
signal. The data are divided into disjoint sub-samples and the BDT
trained on one sub-sample is applied to a different sub-sample to avoid
any bias. LHCb uses simulated events for background and signal in the
training of its BDT. After training, the relevant BDT is applied to each
event in the data, returning a single value for the event, with high
values being more signal-like. To avoid possible biases, both experi-
ments kept the small mass interval that includes both the B0

s and B0

signals blind until all selection criteria were established.

a π+ → μ+ν

π+ W+

d

u

μ+

ν

b B+ → μ+ν

B+ W+

b

u

μ+

ν

c B0
s   → μ+μ–

B0
s

Z0

b

s

μ+

μ–

B0
s   → μ+μ– d

B0
s

 Z0

b

s

μ+

μ–

W–
t

B0
s   → μ+μ–

B0
s

W+b

s

μ+

μ–W–

t ν

e B0
s   → μ+μ–

B0
s

X+b

s

μ+

μ–W–

t ν

g

W+

f

t

s

B0
s   → μ+μ–

μ+

μ–

W–

X+
X0

b

B0
s

Figure 1 | Feynman diagrams related to the B0
s Rm1m2 decay. a, p1 meson

decay through the charged-current process; b, B1 meson decay through the
charged-current process; c, a B0

s decay through the direct flavour changing
neutral current process, which is forbidden in the SM, as indicated by a large red

‘X’; d, e, higher-order flavour changing neutral current processes for the
B0

s ?mzm{ decay allowed in the SM; and f and g, examples of processes for the
same decay in theories extending the SM, where new particles, denoted X0 and
X1, can alter the decay rate.

4 J U N E 2 0 1 5 | V O L 5 2 2 | N A T U R E | 6 9

LETTER RESEARCH

G2015 Macmillan Publishers Limited. All rights reserved

http://itpwiki.unibe.ch
http://www.slac.stanford.edu/xorg/hfag
http://www.slac.stanford.edu/xorg/hfag


In addition to the combinatorial background, specific b-hadron
decays, such as B0 R p2m1n where the neutrino cannot be detected
and the charged pion is misidentified as a muon, or B0 R p0m1m2,
where the neutral pion in the decay is not reconstructed, can mimic the
dimuon decay of the B0

(s) mesons. The invariant mass of the recon-
structed dimuon candidate for these processes (semi-leptonic back-
ground) is usually smaller than the mass of the B0

s or B0 meson because
the neutrino or another particle is not detected. There is also a back-
ground component from hadronic two-body B0

(s) decays (peaking
background) as B0 R K1 p2, when both hadrons from the decay are
misidentified as muons. These misidentified decays can produce peaks
in the dimuon invariant-mass spectrum near the expected signal,
especially for the B0 R m1m2 decay. Particle identification algorithms
are used to minimize the probability that pions and kaons are mis-
identified as muons, and thus suppress these background sources.
Excellent mass resolution is mandatory for distinguishing between
B0 and B0

s mesons with a mass difference of about 87 MeV/c2 and
for separating them from backgrounds. The mass resolution for
B0

s?mzm{ decays in CMS ranges from 32 to 75 MeV/c2, depending
on the direction of the muons relative to the beam axis, while LHCb
achieves a uniform mass resolution of about 25 MeV/c2.

The CMS and LHCb data are combined by fitting a common value for
each branching fraction to the data from both experiments. The branch-
ing fractions are determined from the observed numbers, efficiency-
corrected, of B0

(s) mesons that decay into two muons and the total
numbers of B0

(s) mesons produced. Both experiments derive the latter
from the number of observed B1 R J/y K1 decays, whose branching
fraction has been precisely measured elsewhere14. Assuming equal rates
for B1 and B0 production, this gives the normalization for B0 R m1m2.
To derive the number of B0

s mesons from this B1 decay mode, the ratio
of b quarks that form (hadronize into) B1 mesons to those that form B0

s
mesons is also needed. Measurements of this ratio27,28, for which there is
additional discussion in Methods, and of the branching fraction
B(B1 R J/y K1) are used to normalize both sets of data and are con-
strained within Gaussian uncertainties in the fit. The use of these two
results by both CMS and LHCb is the only significant source of correla-
tion between their individual branching fraction measurements. The
combined fit takes advantage of the larger data sample to increase the
precision while properly accounting for the correlation.

In the simultaneous fit to both the CMS and LHCb data, the branch-
ing fractions of the two signal channels are common parameters of
interest and are free to vary. Other parameters in the fit are considered
as nuisance parameters. Those for which additional knowledge is
available are constrained to be near their estimated values by using
Gaussian penalties with their estimated uncertainties while the others
are free to float in the fit. The ratio of the hadronization probability
into B1 and B0

s mesons and the branching fraction of the normaliza-
tion channel B1 R J/y K1 are common, constrained parameters.
Candidate decays are categorized according to whether they were
detected in CMS or LHCb and to the value of the relevant BDT dis-
criminant. In the case of CMS, they are further categorized according
to the data-taking period, and, because of the large variation in mass
resolution with angle, whether the muons are both produced at large
angles relative to the proton beams (central-region) or at least one
muon is emitted at small angle relative to the beams (forward-region).
An unbinned extended maximum likelihood fit to the dimuon invari-
ant-mass distribution, in a region of about 6500 MeV/c2 around the
B0

s mass, is performed simultaneously in all categories (12 categories
from CMS and eight from LHCb). Likelihood contours in the plane of
the parameters of interest, B(B0 R m1m2) versus B(B0

s?mzm{), are
obtained by constructing the test statistic 22DlnL from the difference
in log-likelihood (lnL) values between fits with fixed values for the
parameters of interest and the nominal fit. For each of the two branch-
ing fractions, a one-dimensional profile likelihood scan is likewise
obtained by fixing only the single parameter of interest and allowing
the other to vary during the fits. Additional fits are performed where
the parameters under consideration are the ratio of the branching

fractions relative to their SM predictions, SB0
(s)

SM:B(B0
(s)?mzm{)=

B(B0
(s)?mzm{)SM, or the ratioR of the two branching fractions.

The combined fit result is shown for all 20 categories in Extended
Data Fig. 1. To represent the result of the fit in a single dimuon
invariant-mass spectrum, the mass distributions of all categories,
weighted according to values of S/(S 1 B), where S is the expected
number of B0

s signals and B is the number of background events under
the B0

s peak in that category, are added together and shown in Fig. 2.
The result of the simultaneous fit is overlaid. An alternative repres-
entation of the fit to the dimuon invariant-mass distribution for the six

(MeV/c2)μ+μ–m
5,000 5,200 5,400 5,600 5,800

W
ei

gh
te

d
 c

an
d

id
at

es
 p

er
 4

0 
M

eV
/c

2

0

10

20

30

40

50

60 Data

Signal and background

μ+μ–→s
0

μ+μ–→
Combinatorial background

Semi-leptonic background

Peaking background

CMS and LHCb (LHC run I)

B0

B

Figure 2 | Weighted distribution of the dimuon invariant mass, mm1m2, for
all categories. Superimposed on the data points in black are the combined fit
(solid blue line) and its components: the B0

s (yellow shaded area) and B0 (light-
blue shaded area) signal components; the combinatorial background (dash-
dotted green line); the sum of the semi-leptonic backgrounds (dotted salmon

line); and the peaking backgrounds (dashed violet line). The horizontal bar on
each histogram point denotes the size of the binning, while the vertical bar
denotes the 68% confidence interval. See main text for details on the weighting
procedure.

7 0 | N A T U R E | V O L 5 2 2 | 4 J U N E 2 0 1 5

RESEARCH LETTER

G2015 Macmillan Publishers Limited. All rights reserved



categories with the highest S/(S 1 B) value for CMS and LHCb, as well
as displays of events with high probability to be genuine signal decays,
are shown in Extended Data Figs 2–4.

The combined fit leads to the measurements B(B0
s?mzm{)~

(2:8z0:7
{0:6) |10{9 and B(B0?mzm{)~(3:9z1:6

{1:4)|10{10, where the
uncertainties include both statistical and systematic sources, the latter
contributing 35% and 18% of the total uncertainty for the B0

s and B0

signals, respectively. Using Wilks’ theorem29, the statistical signifi-
cance in unit of standard deviations, s, is computed to be 6.2 for the
B0

s?mzm{ decay mode and 3.2 for the B0 R m1m2 mode. For each
signal the null hypothesis that is used to compute the significance
includes all background components predicted by the SM as well as
the other signal, whose branching fraction is allowed to vary freely. The
median expected significances assuming the SM branching fractions
are 7.4s and 0.8s for the B0

s and B0 modes, respectively. Likelihood
contours forB(B0 R m1m2) versusB(B0

s?mzm{) are shown in Fig. 3.
One-dimensional likelihood scans for both decay modes are displayed
in the same figure. In addition to the likelihood scan, the statistical
significance and confidence intervals for the B0 branching fraction are
determined using simulated experiments. This determination yields a
significance of 3.0s for a B0 signal with respect to the same null hypo-
thesis described above. Following the Feldman–Cousins30 procedure,

61s and 62s confidence intervals for B(B0 R m1m2) of [2.5, 5.6] 3

10210 and [1.4, 7.4] 3 10210 are obtained, respectively (see Extended
Data Fig. 5).

The fit for the ratios of the branching fractions relative to their SM
predictions yieldsSB0

s
SM~0:76z0:20

{0:18 andSB0

SM~3:7z1:6
{1:4. Associated like-

lihood contours and one-dimensional likelihood scans are shown in
Extended Data Fig. 6. The measurements are compatible with the SM
branching fractions of the B0

s?mzm{ and B0 R m1m2 decays at the
1.2s and 2.2s level, respectively, when computed from the one-
dimensional hypothesis tests. Finally, the fit for the ratio of branching
fractions yieldsR~0:14z0:08

{0:06, which is compatible with the SM at the
2.3s level. The one-dimensional likelihood scan for this parameter is
shown in Fig. 4.

The combined analysis of data from CMS and LHCb, taking advant-
age of their full statistical power, establishes conclusively the existence
of the B0

s?mzm{ decay and provides an improved measurement of its
branching fraction. This concludes a search that started more than
three decades ago (see Extended Data Fig. 7), and initiates a phase of
precision measurements of the properties of this decay. It also pro-
duces three standard deviation evidence for the B0 R m1m2 decay. The
measured branching fractions of both decays are compatible with SM
predictions. This is the first time that the CMS and LHCb collabora-
tions have performed a combined analysis of sets of their data in order
to obtain a statistically significant observation.

Online Content Methods, along with any additional Extended Data display items
andSourceData, are available in the online version of the paper; references unique
to these sections appear only in the online paper.

Received 12 November 2014; accepted 31 March 2015.

Published online 13 May 2015.

1. Bobeth, C. et al. Bs,d R l1l2 in the Standard Model with reduced theoretical
uncertainty. Phys. Rev. Lett. 112, 101801 (2014).

2. Evans, L. & Bryant, P. LHC machine. J. Instrum. 3, S08001 (2008).
3. Planck Collaboration, Ade P. A. R. et al. Planck 2013 results. XVI. Cosmological

parameters. Astron. Astrophys. 571, A16 (2014).
4. Gavela, M., Lozano, M., Orloff, J. & Pène, O. Standard model CP-violation and

baryon asymmetry (I). Zero temperature. Nucl. Phys. B 430, 345–381 (1994).
5. RBC–UKQCD Collaborations, Witzel, O. B-meson decay constants with domain-

wall light quarks and nonperturbatively tuned relativistic b-quarks. Preprint at
http://arXiv.org/abs/1311.0276 (2013).

6. HPQCD Collaboration, Na, H. et al. B and Bs meson decay constants from lattice
QCD. Phys. Rev. D 86, 034506 (2012).

7. Fermilab Lattice and MILC Collaborations, Bazavov A. et al. B- and D-meson decay
constants from three-flavor lattice QCD. Phys. Rev. D 85, 114506 (2012).

8. Huang, C.-S., Liao, W. & Yan, Q.-S. The promising process to distinguish super-
symmetric models with large tanb from the standard model: B R Xsm

1m2. Phys.
Rev. D 59, 011701 (1998).

0 0.2 0.4 0.6 0.8

–2
Δl

nL

0

2

4

6

8

10
SM

0 2 4 6 8

–2
Δl

nL

0

10

20

30

40
SM

→s(B0

0 1 2 3 4 5 6 7 8 9
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

68.27%

95.45%

99.73%

1 − 6.3×10 –5

1 − 5.7×10 –7
1 − 2×10 –9

SM

CMS and LHCb (LHC run I)
a b

c

μ+μ–) (10−9)

→s(B0 μ+μ–) (10−9)

→(B0 μ+μ–) (10−9)

→
(B

0
μ
+ μ

– )
 (1

0−
9 )

Figure 3 | Likelihood contours in the B(B0 R m1m2) versus
B(B0

s Rm1m2) plane. The (black) cross in a marks the best-fit central value.
The SM expectation and its uncertainty is shown as the (red) marker. Each
contour encloses a region approximately corresponding to the reported
confidence level. b, c, Variations of the test statistic 22DlnL forB(B0

s ?mzm{)

(b) andB(B0 R m1m2) (c). The dark and light (cyan) areas define the 61s and
62s confidence intervals for the branching fraction, respectively. The SM
prediction and its uncertainty for each branching fraction is denoted with the
vertical (red) band.

0 0.1 0.2 0.3 0.4 0.5

–2
Δl

nL

0

2

4

6

8

10

SM and MFV

CMS and LHCb (LHC run I)

Figure 4 | Variation of the test statistic 22DlnL as a function of the ratio of
branching fractionsR:B(B0 Rm1m2)/B(B0

s Rm1m2). The dark and light
(cyan) areas define the 61s and 62s confidence intervals forR, respectively.
The value and uncertainty forR predicted in the SM, which is the same in BSM
theories with the minimal flavour violation (MFV) property, is denoted with
the vertical (red) band.

4 J U N E 2 0 1 5 | V O L 5 2 2 | N A T U R E | 7 1

LETTER RESEARCH

G2015 Macmillan Publishers Limited. All rights reserved

www.nature.com/doifinder/10.1038/nature14474
http://arXiv.org/abs/1311.0276


9. Rai Choudhury, S. & Gaur, N. Dileptonic decay of Bs meson in SUSY models with
large tan b. Phys. Lett. B 451, 86–92 (1999).

10. Babu, K. & Kolda, C. F. Higgs-mediated B0 R m1m2 in minimal supersymmetry.
Phys. Rev. Lett. 84, 228–231 (2000).

11. Bobeth, C., Ewerth, T., Kruger, F. & Urban, J. Analysis of neutral Higgs-boson
contributions to the decays �Bs?‘z‘{ and �B?K‘z‘{ . Phys. Rev. D 64, 074014
(2001).

12. Buras, A. J. Relations between DMs,d and Bs,d R m�m in models with minimal flavor
violation. Phys. Lett. B 566, 115–119 (2003).

13. Aoki, S. et al. Review of lattice results concerning low-energy particle physics. Eur.
Phys. J. C 74, 2890 (2014).

14. Particle Data Group, Beringer, J. et al. Review of particle physics. Phys. Rev. D 86,
010001 (2012); 2013 partial update for the 2014 edition at http://pdg.lbl.gov.

15. Heavy Flavor Averaging Group, Amhis, Y. et al. Averages of b-hadron, c-hadron, and
t-lepton properties as of early 2012. Preprint at http://arXiv.org/abs/1207.1158
(2012).

16. D’Ambrosio, G., Giudice, G. F., Isidori, G. & Strumia, A. Minimal flavour violation: an
effective field theory approach. Nucl. Phys. B 645, 155–187 (2002).

17. LHCb Collaboration, Aaij, R. et al. First evidence for the decay B0
s ?mzm{ . Phys. Rev.

Lett. 110, 021801 (2013).
18. CMSCollaboration, Chatrchyan, S.et al. Measurement of the B0

s ?mzm{ branching
fraction and search for B0?mzm{with the CMS experiment. Phys. Rev. Lett. 111,
101804 (2013).

19. LHCb Collaboration, Aaij, R. et al. Measurement of the B0
s ?mzm{ branching

fraction and search for B0?mzm{decays at the LHCb experiment. Phys. Rev. Lett.
111, 101805 (2013).

20. CMS Collaboration, Chatrchyan, S. et al. The CMS experiment at the CERN LHC. J.
Instrum. 3, S08004 (2008).

21. LHCb Collaboration, Alves, A. A. Jr et al. The LHCb detector at the LHC. J. Instrum. 3,
S08005 (2008).

22. ATLAS Collaboration, Aad, G. et al. Observation of a new particle in the search for
the Standard Model Higgs boson with the ATLAS detector at the LHC. Phys. Lett. B
716, 1–29 (2012).

23. CMS Collaboration, Chatrchyan, S. et al. Observation of a new boson at a mass of
125 GeV with the CMS experiment at the LHC. Phys. Lett. B 716, 30–61 (2012).

24. Breiman, L., Friedman, J. H., Olshen, R. A. & Stone, C. J. Classification and Regression
Trees (Wadsworth International Group, 1984).

25. Freund, Y. & Schapire, R. E. A decision-theoretic generalization of on-line learning
and an application to boosting. J. Comput. Syst. Sci. 55, 119–139 (1997).

26. Hoecker, A. et al. TMVA: Toolkit for Multivariate Data Analysis. Proc. Sci. Adv.
Comput. Anal. Techn. Phys. Res. 040 http://pos.sissa.it/archive/conferences/050/
040/ACAT_040.pdf (2007).

27. LHCb Collaboration, Aaij, R. et al. Measurement ofb hadron production fractions in
7 TeV pp collisions. Phys. Rev. D 85, 032008 (2012).

28. LHCb Collaboration, Aaij, R. et al. Measurement of the fragmentation fraction ratio
fs/fd and its dependence on B meson kinematics. J. High Energy Phys. 4, 1 (2013);
updated in https://cds.cern.ch/record/1559262/files/LHCb-CONF-2013-
011.pdf.

29. Wilks, S. S. The large-sample distribution of the likelihood ratio for testing
composite hypotheses. Ann. Math. Stat. 9, 60–62 (1938).

30. Feldman, G. J. & Cousins, R. D. Unified approach to the classical statistical analysis
of small signals. Phys. Rev. D 57, 3873–3889 (1998).

Acknowledgements We express our gratitude to colleagues in the CERN accelerator
departments for the excellent performance of the LHC. We thank the technical and
administrative staff at CERN, at the CMS institutes and at the LHCb institutes. In
addition, we gratefully acknowledge the computing centres and personnel of the
Worldwide LHC Computing Grid for delivering so effectively the computing
infrastructure essential to our analyses. Finally, we acknowledge the enduring support

for the construction and operation of the LHC, the CMS and the LHCb detectors
provided by CERN and by many funding agencies. The following agencies provide
support for both CMS and LHCb: CAPES, CNPq, FAPERJ and FINEP (Brazil); NSFC
(China); CNRS/IN2P3 (France); BMBF, DFG and HGF (Germany); SFI (Ireland); INFN
(Italy); NASU (Ukraine); STFC (UK); and NSF (USA). Agencies that provide support for
CMS only are BMWFW and FWF (Austria); FNRS and FWO (Belgium); FAPESP (Brazil);
MES (Bulgaria); CAS and MoST (China); COLCIENCIAS (Colombia); MSES and CSF
(Croatia); RPF (Cyprus); MoER, ERC IUT and ERDF (Estonia); Academy of Finland, MEC,
and HIP (Finland); CEA (France); GSRT (Greece); OTKA and NIH (Hungary); DAE and
DST (India); IPM (Iran); NRF and WCU (Republic of Korea); LAS (Lithuania); MOE and
UM (Malaysia); CINVESTAV, CONACYT, SEP, and UASLP-FAI (Mexico); MBIE (New
Zealand); PAEC (Pakistan); MSHE and NSC (Poland); FCT (Portugal); JINR (Dubna);
MON, RosAtom, RAS and RFBR (Russia); MESTD (Serbia); SEIDI and CPAN (Spain);
Swiss Funding Agencies (Switzerland); MST (Taipei); ThEPCenter, IPST, STAR and
NSTDA (Thailand); TUBITAK and TAEK (Turkey); SFFR (Ukraine); and DOE (USA).
Agencies that provide support for only LHCb are: FINEP (Brazil); MPG (Germany); FOM
and NWO (The Netherlands); MNiSW and NCN (Poland); MEN/IFA (Romania); MinES
and FANO (Russia); MinECo (Spain); SNSF and SER (Switzerland). Individuals from the
CMS collaboration have received support from the Marie-Curie programme and the
European Research Council and EPLANET (European Union); the Leventis Foundation;
the A. P. Sloan Foundation; the Alexander von Humboldt Foundation; the Belgian
Federal Science Policy Office; the Fonds pour la Formation à la Recherche dans
l’Industrie et dans l’Agriculture (FRIABelgium); the Agentschap voor Innovatie door
Wetenschap en Technologie (IWT-Belgium); the Ministry of Education, Youth and
Sports (MEYS) of the Czech Republic; the Council of Science and Industrial Research,
India; the HOMING PLUS programme of Foundation for Polish Science, cofinanced
from European Union, Regional Development Fund; the Compagnia di San Paolo
(Torino); the Consorzio per la Fisica (Trieste); MIUR project 20108T4XTM (Italy); the
Thalis and Aristeia programmes cofinanced by EU-ESF and the Greek NSRF; and the
National Priorities Research Program by Qatar National Research Fund. Individual
groups or members of the LHCb collaboration have received support from EPLANET,
Marie Skłodowska-Curie Actions and ERC (European Union), Conseil général de
Haute-Savoie, Labex ENIGMASS and OCEVU, Région Auvergne (France), RFBR
(Russia), XuntaGal and GENCAT (Spain), Royal Society and Royal Commission for the
Exhibition of 1851 (UK). LHCb is also thankful for the computing resources and the
access to software R&D tools provided by Yandex LLC (Russia). The CMS and LHCb
collaborations are indebted to the communities behind the multiple open source
software packages on which they depend.

Author Contributions All authors have contributed to the publication, being variously
involved in the design and the construction of the detectors, in writing software,
calibrating sub-systems, operating the detectors and acquiring data and finally
analysing the processed data.

Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Readers are welcome to comment on the online version of the paper.
Correspondence and requests for materials should be addressed to
cms-publication-committee-chair@cern.ch and to
lhcb-editorial-board-chair@cern.ch.

This work is licensed under a Creative Commons Attribution-
NonCommercial-ShareAlike 3.0 Unported licence. The images or other

third party material in this article are included in the article’s Creative Commons licence,
unless indicated otherwise in the credit line; if the material is not included under the
Creative Commons licence, users will need to obtain permission from the licence holder
to reproduce the material. To view a copy of this licence, visit http://creativecommons.
org/licenses/by-nc-sa/3.0

7 2 | N A T U R E | V O L 5 2 2 | 4 J U N E 2 0 1 5

RESEARCH LETTER

G2015 Macmillan Publishers Limited. All rights reserved

http://pdg.lbl.gov
http://arXiv.org/abs/1207.1158
http://pos.sissa.it/archive/conferences/050/040/ACAT_040.pdf
http://pos.sissa.it/archive/conferences/050/040/ACAT_040.pdf
https://cds.cern.ch/record/1559262/files/LHCb-CONF-2013-011.pdf
https://cds.cern.ch/record/1559262/files/LHCb-CONF-2013-011.pdf
www.nature.com/reprints
www.nature.com/doifinder/10.1038/nature14474
mailto:cms-publication-committee-chair@cern.ch
mailto:lhcb-editorial-board-chair@cern.ch
http://creativecommons.org/licenses/by-nc-sa/3.0
http://creativecommons.org/licenses/by-nc-sa/3.0


METHODS
Experimental setup. At the Large Hadron Collider (LHC), two counter-rotating
beams of protons, contained and guided by superconducting magnets spaced
around a 27 km circular tunnel, located approximately 100 m underground near
Geneva, Switzerland, are brought into collision at four interaction points (IPs).
The study presented in this Letter uses data collected at energies of 3.5 TeV per
beam in 2011 and 4 TeV per beam in 2012 by the CMS and LHCb experiments
located at two of these IPs.

The CMS and LHCb detectors are both designed to look for phenomena beyond
the SM (BSM), but using complementary strategies. The CMS detector20, shown in
Extended Data Fig. 3, is optimized to search for yet unknown heavy particles, with
masses ranging from 100 GeV/c2 to a few TeV/c2, which, if observed, would be a
direct manifestation of BSM phenomena. Since many of the hypothesized new
particles can decay into particles containing b quarks or into muons, CMS is able to
detect efficiently and study B0 (5,280 MeV/c2) and B0

s (5,367 MeV/c2) mesons
decaying to two muons even though it is designed to search for particles with much
larger masses. The CMS detector covers a very large range of angles and momenta
to reconstruct high-mass states efficiently. To that extent, it employs a 13 m long, 6
m diameter superconducting solenoid magnet, operated at a field of 3.8 T, centred
on the IP with its axis along the beam direction and covering both hemispheres. A
series of silicon tracking layers, consisting of silicon pixel detectors near the beam
and silicon strips farther out, organized in concentric cylinders around the beam,
extending to a radius of 1.1 m and terminated on each end by planar detectors
(disks) perpendicular to the beam, measures the momentum, angles, and position
of charged particles emerging from the collisions. Tracking coverage starts from
the direction perpendicular to the beam and extends to within 220 mrad from it on
both sides of the IP. The inner three cylinders and disks extending from 4.3 to 10.7
cm in radius transverse to the beam are arrays of 100 3 150 mm2 silicon pixels,
which can distinguish the displacement of the b-hadron decays from the primary
vertex of the collision. The silicon strips, covering radii from 25 cm to approxi-
mately 110 cm, have pitches ranging from 80 to 183 mm. The impact parameter is
measured with a precision of 10 mm for transverse momenta of 100 GeV/c and 20
mm for 10 GeV/c. The momentum resolution, provided mainly by the silicon
strips, changes with the angle relative to the beam direction, resulting in a mass
resolution for B0

(s)?mzm{ decays that varies from 32 MeV/c2 for B0
(s) mesons

produced perpendicularly to the proton beams to 75 MeV/c2 for those produced at
small angles relative to the beam direction. After the tracking system, at a greater
distance from the IP, there is a calorimeter that stops (absorbs) all particles except
muons and measures their energies. The calorimeter consists of an electromag-
netic section followed by a hadronic section. Muons are identified by their ability
to penetrate the calorimeter and the steel return yoke of the solenoid magnet and
to produce signals in gas-ionization particle detectors located in compartments
within the steel yoke. The CMS detector has no capability to discriminate between
charged hadron species, pions, kaons, or protons, that is effective at the typical
particle momenta in this analysis.

The primary commitment of the LHCb collaboration is the study of particle–
antiparticle asymmetries and of rare decays of particles containing b and c quarks.
LHCb aims at detecting BSM particles indirectly by measuring their effect on
b-hadron properties for which precise SM predictions exist. The production cross
section of b hadrons at the LHC is particularly large at small angles relative to the
colliding beams. The small-angle region also provides advantages for the detection
and reconstruction of a wide range of their decays. The LHCb experiment21, shown
in Extended Data Fig. 4, instruments the angular interval from 10 to 300 mrad with
respect to the beam direction on one side of the interaction region. Its detectors are
designed to reconstruct efficiently a wide range of b-hadron decays, resulting in
charged pions and kaons, protons, muons, electrons, and photons in the final state.
The detector includes a high-precision tracking system consisting of a silicon strip
vertex detector, a large-area silicon strip detector located upstream of a dipole
magnet characterized by a field integral of 4 T m, and three stations of silicon strip
detectors and straw drift tubes downstream of the magnet. The vertex detector has
sufficient spatial resolution to distinguish the slight displacement of the weakly
decaying b hadron from the primary production vertex where the two protons
collided and produced it. The tracking detectors upstream and downstream of the
dipole magnet measure the momenta of charged particles. The combined tracking
system provides a momentum measurement with an uncertainty that varies from
0.4% at 5 GeV/c to 0.6% at 100 GeV/c. This results in an invariant-mass resolution
of 25 MeV/c2 for B0

(s) mesons decaying to two muons that is nearly independent of
the angle with respect to the beam. The impact parameter resolution is smaller
than 20 mm for particle tracks with large transverse momentum. Different types of
charged hadrons are distinguished by information from two ring-imaging
Cherenkov detectors. Photon, electron, and hadron candidates are identified by
calorimeters. Muons are identified by a system composed of alternating layers of
iron and multiwire proportional chambers.

Neither CMS nor LHCb records all the interactions occurring at its IP
because the data storage and analysis costs would be prohibitive. Since most
of the interactions are reasonably well characterized (and can be further
studied by recording only a small sample of them) specific event filters (known
as triggers) select the rare processes that are of interest to the experiments.
Both CMS and LHCb implement triggers that specifically select events con-
taining two muons. The triggers of both experiments have a hardware stage,
based on information from the calorimeter and muon systems, followed by a
software stage, consisting of a large computing cluster that uses all the
information from the detector, including the tracking, to make the final selec-
tion of events to be recorded for subsequent analysis. Since CMS is designed to
look for much heavier objects than B0

(s) mesons, it selects events that contain
muons with higher transverse momenta than those selected by LHCb. This
eliminates many of the B0

(s) decays while permitting CMS to run at a higher
proton–proton collision rate to look for the more rare massive particles. Thus
CMS runs at higher collision rates but with lower efficiency than LHCb for B0

ðsÞ
mesons decaying to two muons. The overall sensitivity to these decays turns
out to be similar in the two experiments.

CMS and LHCb are not the only collaborations to have searched for B0
s ?mzm{

and B0R m1 m2 decays. Over three decades, a total of eleven collaborations have
taken part in this search14, as illustrated by Extended Data Fig. 7. This plot gathers
the results from CLEO31–35, ARGUS36, UA137,38, CDF39–44, L345, DØ46–50, Belle51,
Babar52,53, LHCb17,54–57 CMS18,58,59 and ATLAS60.
Analysis description. The analysis techniques used to obtain the results presented
in this Letter are very similar to those used to obtain the individual result in each
collaboration, described in more detail in refs 18, 19. Here only the main analysis
steps are reviewed and the changes used in the combined analysis are highlighted.
Data samples for this analysis were collected by the two experiments in proton–
proton collisions at a centre-of-mass energy of 7 and 8 TeV during 2011 and 2012,
respectively. These samples correspond to a total integrated luminosity of 25 and 3
fb21 for the CMS and LHCb experiments, respectively, and represent their com-
plete data sets from the first running period of the LHC.

The trigger criteria were slightly different between the two experiments. The
large majority of events were triggered by requirements on one or both muons of
the signal decay: the LHCb detector triggered on muons with transverse
momentum pT . 1.5 GeV/c while the CMS detector, because of its geometry
and higher instantaneous luminosity, triggered on two muons with pT . 4 (3)
GeV/c, for the leading (sub-leading) muon.

The data analysis procedures in the two experiments follow similar strategies.
Pairs of high-quality oppositely charged particle tracks that have one of the
expected patterns of hits in the muon detectors are fitted to form a common vertex
in three dimensions, which is required to be displaced from the primary inter-
action vertex (PV) and to have a small x2 in the fit. The resulting B0

(s) candidate is
further required to point back to the PV, for example, to have a small impact
parameter, consistent with zero, with respect to it. The final classification of data
events is done in categories of the response of a multivariate discriminant (MVA)
combining information from the kinematics and vertex topology of the events.
The type of MVA used is a boosted decision tree (BDT)24–26. The branching
fractions are then obtained by a fit to the dimuon invariant mass, mmzm{ , of all
categories simultaneously.

The signals appear as peaks at the B0
s and B0 masses in the invariant-mass

distributions, observed over background events. One of the components of the
background is combinatorial in nature, as it is due to the random combinations of
genuine muons. These produce a smooth dimuon mass distribution in the vicinity
of the B0

s and B0 masses, estimated in the fit to the data by extrapolation from the
sidebands of the invariant-mass distribution. In addition to the combinatorial
background, certain specific b-hadron decays can mimic the signal or contribute
to the background in its vicinity. In particular, the semi-leptonic decays B0 R
p2m1n, B0

s R K2m1n, and L0
b?pm{�n can have reconstructed masses that are

near the signal if one of the hadrons is misidentified as a muon and is combined
with a genuine muon. Similarly the dimuon coming from the rare B0 R p0m1m2

and B1 R p1m1m2 decays can also fake the signal. All these background decays,
when reconstructed as a dimuon final state, have invariant masses that are lower
than the masses of the B0 and B0

s mesons, because they are missing one of the
original decay particles. An exception is the decay L0

b?pm{�n, which can also
populate, with a smooth mass distribution, higher-mass regions. Furthermore,
background due to misidentified hadronic two-body decays B0

(s)?hzh’{, where

h( 0)~p or K, is present when both hadrons are misidentified as muons. These
misidentified decays produce an apparent dimuon invariant-mass peak close to
the B0 mass value. Such a peak can mimic a B0 R m1m2 signal and is estimated
from control channels and added to the fit.

The distributions of signal in the invariant mass and in the MVA discriminant
are derived from simulations with a detailed description of the detector response

LETTER RESEARCH

G2015 Macmillan Publishers Limited. All rights reserved



for CMS and are calibrated using exclusive two-body hadronic decays in data for
LHCb. The distributions for the backgrounds are obtained from simulation with
the exception of the combinatorial background. The latter is obtained by inter-
polating from the data invariant-mass sidebands separately for each category, after
the subtraction of the other background components.

To compute the signal branching fractions, the numbers of B0
s and B0 mesons

that are produced, as well as the numbers of those that have decayed into a dimuon
pair, are needed. The latter numbers are the raw results of this analysis, whereas the
former need to be determined from measurements of one or more ‘normalization’
decay channels, which are abundantly produced, have an absolute branching
fraction that is already known with good precision, and that share characteristics
with the signals, so that their trigger and selection efficiencies do not differ sig-
nificantly. Both experiments use the B1 R J/y K1 decay as a normalization
channel with B(B1 R J/y (m1m2) K1) 5 (6.10 6 0.19) 3 1025, and LHCb also
uses the B0 R K1p2 channel with B(B0 R K1p2) 5 (1.96 6 0.05) 3 1025. Both
branching fraction values are taken from ref. 14. Hence, the B0

s R m1m2 branching
fraction is expressed as a function of the number of signal events (NB0

s ?mzm{ ) in the
data normalized to the numbers of B1 R J/y K1 and B0 R K1p2 events:

B(B0
s ?mzm{)~

NB0
s ?mzm{

Nnorm:
|

fd

fs
|

enorm:

eB0
s ?mzm{

|Bnorm:~anorm:|NB0
s ?mzm{ ð1Þ

where the ‘norm.’ subscript refers to either of the normalization channels. The
values of the normalization parameter anorm. obtained by LHCb from the two
normalization channels are found in good agreement and their weighted average is
used. In this formula e indicates the total event detection efficiency including
geometrical acceptance, trigger selection, reconstruction, and analysis selection
for the corresponding decay. The fd/fs factor is the ratio of the probabilities for a b
quark to hadronize into a B0 as compared to a B0

s meson; the probability to
hadronize into a B1 (fu) is assumed to be equal to that into B0 (fd) on the basis
of theoretical grounds, and this assumption is checked on data. The value of fd/fs 5

3.86 6 0.22 measured by LHCb27,28,61 is used in this analysis. As the value of fd/fs
depends on the kinematic range of the considered particles, which differs between
LHCb and CMS, CMS checked this observable with the decays B0

s ?J=yw and
B1 R J/yK1 within its acceptance, finding a consistent value. An additional
systematic uncertainty of 5% was assigned to fd/fs to account for the extrapolation
of the LHCb result to the CMS acceptance. An analogous formula to that in
equation (1) holds for the normalization of the B0 R m1m2 decay, with the notable
difference that the fd/fs factor is replaced by fd/fu 5 1.

The antiparticle �B0 (�B0
s ) and the particle B0 (B0

s ) can both decay into two muons
and no attempt is made in this analysis to determine whether the antiparticle or
particle was produced (untagged method). However, the B0 and B0

s particles are
known to oscillate, that is to transform continuously into their antiparticles and
vice versa. Therefore, a quantum superposition of particle and antiparticle states
propagates in the laboratory before decaying. This superposition can be described
by two ‘mass eigenstates’, which are symmetric and antisymmetric in the charge-
parity (CP) quantum number, and have slightly different masses. In the SM, the
heavy eigenstate can decay into two muons, whereas the light eigenstate cannot
without violating the CP quantum number conservation. In BSM models, this is
not necessarily the case. In addition to their masses, the two eigenstates of the B0

s
system also differ in their lifetime values14. The lifetimes of the light and heavy
eigenstates are also different from the average B0

s lifetime, which is used by CMS
and LHCb in the simulations of signal decays. Since the information on the
displacement of the secondary decay with respect to the PV is used as a discrim-
inant against combinatorial background in the analysis, the efficiency versus life-
time has a model-dependent bias62 that must be removed. This bias is estimated
assuming SM dynamics. Owing to the smaller difference between the lifetime of its
heavy and light mass eigenstates, no correction is required for the B0 decay mode.

Detector simulations are needed by both CMS and LHCb. CMS relies on
simulated events to determine resolutions and trigger and reconstruction effi-
ciencies, and to provide the signal sample for training the BDT. The dimuon
mass resolution given by the simulation is validated using data on J/y, U, and
Z-boson decays to two muons. The tracking and trigger efficiencies obtained
from the simulation are checked using special control samples from data. The
LHCb analysis is designed to minimize the impact of discrepancies between
simulations and data. The mass resolution is measured with data. The distri-
bution of the BDT for the signal and for the background is also calibrated with
data using control channels and mass sidebands. The efficiency ratio for the
trigger is also largely determined from data. The simulations are used to deter-
mine the efficiency ratios of selection and reconstruction processes between
signal and normalization channels. As for the overall detector simulation, each
experiment has a team dedicated to making the simulations as complete and
realistic as possible. The simulated data are constantly being compared to the

actual data. Agreement between simulation and data in both experiments is
quite good, often extending well beyond the cores of distributions. Differences
occur because, for example, of incomplete description of the material of the
detectors, approximations made to keep the computer time manageable, resi-
dual uncertainties in calibration and alignment, and discrepancies or limita-
tions in the underlying theory and experimental data used to model the
relevant collisions and decays. Small differences between simulation and data
that are known to have an impact on the result are treated either by reweighting
the simulations to match the data or by assigning appropriate systematic
uncertainties.

Small changes are made to the analysis procedure with respect to refs 18, 19 in
order to achieve a consistent combination between the two experiments. In the
LHCb analysis, the L0

b?pm{�n background component, which was not included in
the fit for the previous result but whose effect was accounted for as an additional
systematic uncertainty, is now included in the standard fit. The following modifica-
tions are made to the CMS analysis: the L0

b?pm{�n branching fraction is updated to
a more recent prediction63,64 of B(L0

b?pm{�n)~(4:94+2:19) |10{4; the phase
space model of the decay L0

b?pm{�n is changed to a more appropriate semi-
leptonic decay model63; and the decay time bias correction for the B0

s, previously
absent from the analysis, is now calculated and applied with a different correction
for each category of the multivariate discriminant.

These modifications result in changes in the individual results of each experi-
ment. The modified CMS analysis, applied on the CMS data, yields

B(B0
s ?mzm{)~(2:8z1:1

{0:9)|10{9 and B(B0?mzm{)~(4:4z2:2
{1:9)|10{10 ð2Þ

while the LHCb results change to

B(B0
s ?mzm{)~(2:7z1:1

{0:9)|10{9 and B(B0?mzm{)~(3:3z2:4
{2:1)|10{10 ð3Þ

These results are only slightly different from the published ones and are in agree-
ment with each other.
Simultaneous fit. The goal of the analysis presented in this Letter is to combine the
full data sets of the two experiments to reduce the uncertainties on the branching
fractions of the signal decays obtained from the individual determinations. A sim-
ultaneous unbinned extended maximum likelihood fit is performed to the data of
the two experiments, using the invariant-mass distributions of all 20 MVA discrim-
inant categories of both experiments. The invariant-mass distributions are defined
in the dimuon mass ranges mmzm{ g[4.9, 5.9] GeV/c2 and [4.9, 6.0] GeV/c2 for the
CMS and LHCb experiments, respectively. The branching fractions of the signal
decays, the hadronization fraction ratio fd/fs, and the branching fraction of the
normalization channel B1 R J/y K1 are treated as common parameters. The value
of the B1 R J/y K1 branching fraction is the combination of results from five
different experiments14, taking advantage of all their data to achieve the most precise
input parameters for this analysis. The combined fit takes advantage of the larger
data sample and proper treatment of the correlations between the individual mea-
surements to increase the precision and reliability of the result, respectively.

Fit parameters, other than those of primary physics interest, whose limited
knowledge affects the results, are called ‘nuisance parameters’. In particular, sys-
tematic uncertainties are modelled by introducing nuisance parameters into the
statistical model and allowing them to vary in the fit; those for which additional
knowledge is present are constrained using Gaussian distributions. The mean and
standard deviation of these distributions are set to the central value and uncer-
tainty obtained either from other measurements or from control channels. The
statistical component of the final uncertainty on the branching fractions is
obtained by repeating the fit after fixing all of the constrained nuisance parameters
to their best fitted values. The systematic component is then calculated by sub-
tracting in quadrature the statistical component from the total uncertainty. In
addition to the free fit, a two-dimensional likelihood ratio scan in the plane
B(B0 R m1m2) versus B(B0

s ?mzm{) is performed.
Feldman–Cousins confidence interval. The Feldman–Cousins likelihood ratio
ordering procedure30 is a unified frequentist method to construct single- and
double-sided confidence intervals for parameters of a given model adapted to
the data. It provides a natural transition between single-sided confidence intervals,
used to define upper or lower limits, and double-sided ones. Since the single-
experiment results18,19 showed that the B0 R m1m2 signal is at the edge of the
probability region customarily used to assert statistically significant evidence for a
result, a Feldman–Cousins procedure is performed. This allows a more reliable
determination of the confidence interval and significance of this signal without the
assumptions required for the use of Wilks’ theorem. In addition, a prescription for
the treatment of nuisance parameters has to be chosen because scanning the whole
parameter space in the presence of more than a few parameters is computationally
too intensive. In this case the procedure described by the ATLAS and CMS Higgs
combination group65 is adopted. For each point of the space of the relevant para-
meters, the nuisance parameters are fixed to their best value estimated by the mean

RESEARCH LETTER

G2015 Macmillan Publishers Limited. All rights reserved



of a maximum likelihood fit to the data with the value ofB(B0 R m1m2) fixed and
all nuisance parameters profiled with Gaussian penalties. Sampling distributions
are constructed for each tested point of the parameter of interest by generating
simulated experiments and performing maximum likelihood fits in which the
Gaussian mean values of the external constraints on the nuisance parameters
are randomized around the best-fit values for the nuisance parameters used to
generate the simulated experiments. The sampling distribution is constructed
from the distribution of the negative log-likelihood ratio evaluated on the simu-
lated experiments by performing one likelihood fit in which the value of B(B0 R
m1m2) is free to float and another with theB(B0 R m1m2) fixed to the tested point
value. This sampling distribution is then converted to a confidence level by evalu-
ating the fraction of simulated experiments entries with a value for the negative
log-likelihood ratio greater than or equal to the value observed in the data for each
tested point. The results of this procedure are shown in Extended Data Fig. 5.

31. CLEO Collaboration, Giles R. et al. Two-body decays of B mesons. Phys. Rev. D 30,
2279–2294 (1984).

32. CLEO Collaboration, Avery P. et al. Limits on rare exclusive decays of B mesons.
Phys. Lett. B 183, 429–433 (1987).

33. CLEO Collaboration, Avery P. et al. A search for exclusive penguin decays of B
mesons. Phys. Lett. B 223, 470–475 (1989).

34. CLEO Collaboration, Ammar R. et al. Search for B0 decays to two charged leptons.
Phys. Rev. D 49, 5701–5704 (1994).

35. CLEO Collaboration, Bergfeld T. et al. Search for decays of B0 mesons into pairs of
leptons: B0 R e1e2, B0 R m1m2 and B0 R e6m7. Phys. Rev. D 62, 091102 (2000).

36. ARGUS Collaboration, Albrecht H. et al. B meson decays into charmonium states.
Phys. Lett. B 199, 451–456 (1987).

37. UA1 Collaboration, Albajar C. et al. Low mass dimuon production at the CERN
proton-antiproton collider. Phys. Lett. B 209, 397–406 (1988).

38. UA1 Collaboration, Albajar C. et al. A search for rare B meson decays at the CERN
Sp�pS collider. Phys. Lett. B 262, 163–170 (1991).

39. CDFCollaboration, Abe F.et al. Search for flavor-changingneutral currentB meson
decays in p�p collisions at

ffiffiffi

s
p

5 1.8 TeV. Phys. Rev. Lett. 76, 4675–4680 (1996).
40. CDF Collaboration, Abe F. et al. Search for the decays B0

d?mzm{ and B0
s ?mzm{ in

p�p collisions at
ffiffiffi

s
p

5 1.8 TeV. Phys. Rev. D 57, 3811–3816 (1998).
41. CDF Collaboration, Acousta D. et al. Search for B0

s ?mzm{ and B0
d?mzm{ decays

in p�p collisions at
ffiffiffi

s
p

5 1.96 TeV. Phys. Rev. Lett. 93, 032001 (2004).
42. CDF Collaboration, Abulencia A. et al. Search for Bs R m1m2 and Bd R m1m2 decays

in p�p collisions with CDF II. Phys. Rev. Lett. 95, 221805 (2005).
43. CDF Collaboration, Aaltonen T. et al. Search for Bs R m1m2 and Bd R m1m– decays

with CDF II. Phys. Rev. Lett. 107, 191801 (2011).
44. CDF Collaboration, Aaltonen T. et al. Search for Bs R m1m2 and Bd R m1m2 decays

with the full CDF Run II data set. Phys. Rev. D 87, 072003 (2013).
45. L3 Collaboration, Acciarri M. et al. Search for neutral B meson decays to two

charged leptons. Phys. Lett. B 391, 474–480 (1997).

46. DØ Collaboration, Abbott B. et al. Search for the decay b R Xsm
1m2. Phys. Lett. B

423, 419–426 (1998).
47. DØ Collaboration, Abazov V. et al. A search for the flavor-changing neutral current

decay B0
s ?mzm{ in p�p collisions at

ffiffiffi

s
p

5 1.96 TeV with the DØ detector. Phys. Rev.
Lett. 94, 071802 (2005).

48. DØ Collaboration, Abazov V. M. et al. Search for B0
s ?mzm{ at DØ. Phys. Rev. D 76,

092001 (2007).
49. DØ Collaboration, Abazov V. M. et al. Search for the rare decay B0

s ?mzm{ . Phys.
Lett. B 693, 539–544 (2010).

50. DØ Collaboration, Abazov V. M. et al. Search for the rare decay B0
s ?mzm{ . Phys.

Rev. D 87, 072006 (2013).
51. BELLE Collaboration, Chang M. et al. Search for B0 R ,1,2 at BELLE. Phys. Rev. D

68, 111101 (2003).
52. BaBar Collaboration, Aubert B. et al. Search for decays of B0 mesons into pairs of

charged leptons: B0 R e1e2, B0 R m1m2, B0 R e6m7. Phys. Rev. Lett. 94, 221803
(2005).

53. BaBar Collaboration, Aubert B. et al. Search for decays of B0 mesons into e1e2,
m1m2, and e6m7 final states. Phys. Rev. D 77, 032007 (2008).

54. LHCb Collaboration, Aaij R. et al. Search for the rare decays B0
s ?mzm{ and B0 R

m1m2. Phys. Lett. B 699, 330–340 (2011).
55. LHCb Collaboration, Aaij R. et al. Strong constraints on the rare decays B0

s Rm1m2

and B0 R m1m2. Phys. Rev. Lett. 108, 231801 (2012).
56. LHCb Collaboration, Aaij R. et al. Search for the rare decays B0

s ?mzm{ and B0 R
m1m2. Phys. Lett. B 708, 55–67 (2012).

57. LHCb Collaboration, Aaij R. et al. Measurement of the B0
s ?mzm{ branching

fraction and search for B0 R m1m2 decays at the LHCb experiment. Phys. Rev. Lett.
111, 101805 (2013).

58. CMS Collaboration, Chatrchyan S. et al. Search for B0
s ?mzm{ and B0 R m1m2

decays in pp collisions at 7 TeV. Phys. Rev. Lett. 107, 191802 (2011).
59. CMS Collaboration, Chatrchyan S. et al. Search for B0

s ?mzm{ and B0 R m1m2

decays. J. High Energy Phys. 04, 033 (2012).
60. ATLAS Collaboration, Aad G. et al. Search for the decay B0

s ?mzm{ with the ATLAS
detector. Phys. Lett. B 713, 387–407 (2012).

61. LHCb Collaboration, Aaij R. et al. Updated average fs/fd b-hadron production
fraction ratio for 7 TeV pp collisions. http://cds.cern.ch/record/1559262 (LHCb-
CONF-2013-011, 2013).

62. De Bruyn, K. et al. Probing new physics via the B0
s ?mzm{ effective lifetime. Phys.

Rev. Lett. 109, 041801 (2012).
63. Khodjamirian, A., Klein, C., Mannel, T. & Wang, Y.-M. Form factors and strong

couplings of heavy baryons from QCD light-cone sum rules. J. High Energy Phys.
09, 106 (2011).

64. LHCb Collaboration, Aaij R. et al. Precision measurementof the ratio of the L0
b to �B0

lifetimes. Phys. Lett. B 734, 122–130 (2014).
65. ATLAS and CMS Collaborations. Procedure for the LHC Higgs boson search

combination in summer 2011. http://cds.cern.ch/record/1379837 (ATL-PHYS-
PUB-2011-011, CMS NOTE 2011/005, 2011).

LETTER RESEARCH

G2015 Macmillan Publishers Limited. All rights reserved

http://cds.cern.ch/record/1559262
http://cds.cern.ch/record/1379837


CMS Collaboration
V.Khachatryan1, A.M. Sirunyan1, A. Tumasyan1, W. Adam2, T. Bergauer2, M. Dragicevic2,
J. Erö2, M. Friedl2, R. Frühwirth2,204, V.M. Ghete2, C. Hartl2, N. Hörmann2, J. Hrubec2,
M. Jeitler2,204, W. Kiesenhofer2, V. Knünz2, M. Krammer2,204, I. Krätschmer2, D. Liko2,
I. Mikulec2, D. Rabady2,205, B. Rahbaran2, H. Rohringer2, R. Schöfbeck2, J. Strauss2,
W. Treberer-Treberspurg2, W. Waltenberger2, C.-E. Wulz2,204, V. Mossolov3,
N. Shumeiko3, J. Suarez Gonzalez3, S. Alderweireldt4, S. Bansal4, T. Cornelis4, E.A. De
Wolf4, X. Janssen4, A. Knutsson4, J. Lauwers4, S. Luyckx4, S. Ochesanu4, R. Rougny4,
M. Van De Klundert4, H. Van Haevermaet4, P. Van Mechelen4, N. Van Remortel4,
A. Van Spilbeeck4, F. Blekman5, S. Blyweert5, J. D’Hondt5, N. Daci5, N. Heracleous5,
J. Keaveney5, S. Lowette5, M. Maes5, A. Olbrechts5, Q. Python5, D. Strom5, S. Tavernier5,
W. Van Doninck5, P. Van Mulders5, G.P. Van Onsem5, I. Villella5, C. Caillol6, B. Clerbaux6,
G. De Lentdecker6, D. Dobur6, L. Favart6, A.P.R. Gay6, A. Grebenyuk6, A. Léonard6,
A. Mohammadi6, L. Perniè6,205, A. Randle-conde6, T. Reis6, T. Seva6, L. Thomas6,
C. Vander Velde6, P. Vanlaer6, J. Wang6, F. Zenoni6, V. Adler7, K. Beernaert7, L. Benucci7,
A. Cimmino7, S. Costantini7, S. Crucy7, S. Dildick7, A. Fagot7, G. Garcia7, J. Mccartin7,
A.A. Ocampo Rios7, D. Ryckbosch7, S. Salva Diblen7, M. Sigamani7, N. Strobbe7,
F. Thyssen7, M. Tytgat7, E. Yazgan7, N. Zaganidis7, S. Basegmez8, C. Beluffi8,206,
G. Bruno8, R. Castello8, A. Caudron8, L. Ceard8, G.G. Da Silveira8, C. Delaere8, T. du Pree8,
D. Favart8, L. Forthomme8, A. Giammanco8,207, J. Hollar8, A. Jafari8, P. Jez8, M. Komm8,
V. Lemaitre8, C. Nuttens8, D. Pagano8, L. Perrini8, A. Pin8, K. Piotrzkowski8, A. Popov8,208,
L. Quertenmont8, M. Selvaggi8, M. Vidal Marono8, J.M. Vizan Garcia8, N. Beliy9,
T. Caebergs9, E. Daubie9, G.H. Hammad9, W.L. Aldá Júnior10, G.A. Alves10, L. Brito10,
M. Correa Martins Junior10, T. Dos Reis Martins10, C. Mora Herrera10, M.E. Pol10,
P. Rebello Teles10, W. Carvalho11, J. Chinellato11,209, A. Custódio11, E.M. Da Costa11,
D. De Jesus Damiao11, C. De Oliveira Martins11, S. Fonseca De Souza11,
H. Malbouisson11, D. Matos Figueiredo11, L. Mundim11, H. Nogima11, W.L. Prado Da
Silva11, J. Santaolalla11, A. Santoro11, A. Sznajder11, E.J. Tonelli Manganote11,209,
A. Vilela Pereira11, C.A. Bernardes14, S. Dogra13, T.R. Fernandez Perez Tomei13,
E.M. Gregores14, P.G. Mercadante14, S.F. Novaes13, Sandra S. Padula13,
A. Aleksandrov15, V. Genchev15,205, R. Hadjiiska15, P. Iaydjiev15, A. Marinov15, S.
Piperov15, M. Rodozov15, G. Sultanov15, M. Vutova15, A. Dimitrov16, I. Glushkov16,
L. Litov16, B. Pavlov16, P. Petkov16, J.G. Bian17, G.M. Chen17, H.S. Chen17, M. Chen17,
T. Cheng17, R. Du17, C.H. Jiang17, R. Plestina17,210, F. Romeo17, J. Tao17, Z. Wang17,
C. Asawatangtrakuldee18, Y. Ban18, Q. Li18, S. Liu18, Y. Mao18, S.J. Qian18, D. Wang18,
Z. Xu18, W. Zou18, C. Avila19, A. Cabrera19, L.F. Chaparro Sierra19, C. Florez19,
J.P. Gomez19, B. Gomez Moreno19, J.C. Sanabria19, N.Godinovic20, D. Lelas20, D.Polic20,
I. Puljak20, Z. Antunovic21, M. Kovac21, V. Brigljevic22, K. Kadija22, J. Luetic22, D.
Mekterovic22, L. Sudic22, A. Attikis23, G. Mavromanolakis23, J. Mousa23, C. Nicolaou23,
F. Ptochos23, P.A. Razis23, M. Bodlak24, M. Finger24, M. Finger Jr.24,211, Y. Assran25,212,
A. Ellithi Kamel25,213, M.A. Mahmoud25,214, A. Radi25,215,216, M. Kadastik26,
M. Murumaa26, M. Raidal26, A. Tiko26, P. Eerola27, G. Fedi27, M. Voutilainen27,
J. Härkönen28, V. Karimäki28, R. Kinnunen28, M.J. Kortelainen28, T. Lampén28, K. Lassila-
Perini28, S. Lehti28, T. Lindén28, P. Luukka28, T. Mäenpää28, T. Peltola28, E. Tuominen28,
J. Tuominiemi28, E. Tuovinen28, L. Wendland28, J. Talvitie29, T. Tuuva29, M. Besancon30,
F. Couderc30, M. Dejardin30, D. Denegri30, B. Fabbro30, J.L. Faure30, C. Favaro30,
F. Ferri30, S.Ganjour30, A.Givernaud30, P.Gras30,G.HameldeMonchenault30, P. Jarry30,
E. Locci30, J. Malcles30, J. Rander30, A. Rosowsky30, M. Titov30, S. Baffioni31,
F. Beaudette31, P. Busson31, C. Charlot31, T. Dahms31, M. Dalchenko31, L. Dobrzynski31,
N. Filipovic31, A. Florent31, R. Granier de Cassagnac31, L. Mastrolorenzo31, P. Miné31,
C. Mironov31, I.N. Naranjo31, M. Nguyen31, C. Ochando31, G. Ortona31, P. Paganini31,
S. Regnard31, R. Salerno31, J.B. Sauvan31, Y. Sirois31, C. Veelken31, Y. Yilmaz31, A. Zabi31,
J.-L. Agram32,217, J. Andrea32, A. Aubin32, D. Bloch32, J.-M. Brom32, E.C. Chabert32,
C.Collard32, E.Conte32,217, J.-C.Fontaine32,217,D.Gelé32,U.Goerlach32,C.Goetzmann32,
A.-C. Le Bihan32, K. Skovpen32, P. Van Hove32, S. Gadrat33, S. Beauceron34,
N. Beaupere34, G. Boudoul34,205, E. Bouvier34, S. Brochet34, C.A. Carrillo Montoya34,
J. Chasserat34, R. Chierici34, D. Contardo34,205, P. Depasse34, H. El Mamouni34, J. Fan34,
J. Fay34, S. Gascon34, M. Gouzevitch34, B. Ille34, T. Kurca34, M. Lethuillier34, L. Mirabito34,
S. Perries34, J.D. Ruiz Alvarez34, D. Sabes34, L. Sgandurra34, V. Sordini34, M. Vander
Donckt34, P. Verdier34, S. Viret34, H. Xiao34, Z. Tsamalaidze35,211, C. Autermann36,
S. Beranek36, M. Bontenackels36, M. Edelhoff36, L. Feld36, A. Heister36, O. Hindrichs36,
K. Klein36, A. Ostapchuk36, F. Raupach36, J. Sammet36, S. Schael36, J.F. Schulte36,
H. Weber36, B. Wittmer36, V. Zhukov36,208, M. Ata37, M. Brodski37, E. Dietz-Laursonn37,
D. Duchardt37, M. Erdmann37, R. Fischer37, A. Güth37, T. Hebbeker37, C. Heidemann37,
K.Hoepfner37,D.Klingebiel37, S.Knutzen37, P.Kreuzer37,M.Merschmeyer37, A.Meyer37,
P. Millet37, M. Olschewski37, K. Padeken37, P. Papacz37, H. Reithler37, S.A. Schmitz37,
L. Sonnenschein37, D. Teyssier37, S. Thüer37, M. Weber37, V. Cherepanov38, Y. Erdogan38,
G. Flügge38, H. Geenen38, M. Geisler38, W. Haj Ahmad38, F. Hoehle38, B. Kargoll38,
T. Kress38, Y. Kuessel38, A. Künsken38, J. Lingemann38,205, A. Nowack38, I.M. Nugent38,
O.Pooth38, A. Stahl38, M. Aldaya Martin39, I. Asin39, N. Bartosik39, J. Behr39, U.Behrens39,
A.J. Bell39, A. Bethani39, K. Borras39, A. Burgmeier39, A. Cakir39, L. Calligaris39,
A. Campbell39, S. Choudhury39, F. Costanza39, C. Diez Pardos39, G. Dolinska39,
S. Dooling39, T. Dorland39, G. Eckerlin39, D.Eckstein39, T. Eichhorn39,G. Flucke39, J.Garay
Garcia39, A. Geiser39, P. Gunnellini39, J. Hauk39, M. Hempel39,218, H. Jung39,
A. Kalogeropoulos39, M. Kasemann39, P. Katsas39, J. Kieseler39, C. Kleinwort39, I. Korol39,
D. Krücker39, W. Lange39, J. Leonard39, K. Lipka39, A. Lobanov39, W. Lohmann39,218,
B. Lutz39, R. Mankel39, I. Marfin39,218, I.-A. Melzer-Pellmann39, A.B. Meyer39, G. Mittag39,
J. Mnich39, A. Mussgiller39, S. Naumann-Emme39, A. Nayak39, E. Ntomari39, H. Perrey39,
D. Pitzl39, R. Placakyte39, A. Raspereza39, P.M. Ribeiro Cipriano39, B. Roland39, E. Ron39,
M.Ö. Sahin39, J. Salfeld-Nebgen39, P. Saxena39, T. Schoerner-Sadenius39, M. Schröder39,
C. Seitz39, S. Spannagel39, A.D.R. Vargas Trevino39, R. Walsh39, C. Wissing39, V. Blobel40,
M. Centis Vignali40, A.R. Draeger40, J. Erfle40, E. Garutti40, K. Goebel40, M. Görner40,
J. Haller40, M. Hoffmann40, R.S. Höing40, A. Junkes40, H. Kirschenmann40, R. Klanner40,
R. Kogler40, J. Lange40, T. Lapsien40, T. Lenz40, I. Marchesini40, J. Ott40, T. Peiffer40,
A. Perieanu40, N. Pietsch40, J. Poehlsen40, T. Poehlsen40, D. Rathjens40, C. Sander40,

H. Schettler40, P. Schleper40, E. Schlieckau40, A. Schmidt40, M. Seidel40, V. Sola40,
H. Stadie40, G. Steinbrück40, D. Troendle40, E. Usai40, L. Vanelderen40, A. Vanhoefer40,
C. Barth41, C. Baus41, J. Berger41, C. Böser41, E. Butz41, T. Chwalek41, W. De Boer41,
A. Descroix41, A. Dierlamm41, M. Feindt41, F. Frensch41, M. Giffels41, A. Gilbert41,
F. Hartmann41,205, T. Hauth41, U. Husemann41, I. Katkov41,208, A. Kornmayer41,205,
E. Kuznetsova41, P. Lobelle Pardo41, M.U. Mozer41, T. Müller41, Th. Müller41,
A. Nürnberg41, G. Quast41, K. Rabbertz41, S. Röcker41, H.J. Simonis41, F.M. Stober41,
R. Ulrich41, J. Wagner-Kuhr41, S. Wayand41, T. Weiler41, R. Wolf41, G. Anagnostou42,
G. Daskalakis42, T. Geralis42, V.A. Giakoumopoulou42, A. Kyriakis42, D. Loukas42,
A. Markou42, C. Markou42, A. Psallidas42, I. Topsis-Giotis42, A. Agapitos43,
S. Kesisoglou43, A. Panagiotou43, N. Saoulidou43, E. Stiliaris43, X. Aslanoglou44,
I. Evangelou44, G. Flouris44, C. Foudas44, P. Kokkas44, N. Manthos44, I. Papadopoulos44,
E. Paradas44, J. Strologas44, G. Bencze45, C. Hajdu45, P. Hidas45, D. Horvath45,219,
F. Sikler45, V. Veszpremi45, G. Vesztergombi45,220, A.J. Zsigmond45, N. Beni46,
S. Czellar46, J. Karancsi46,221, J. Molnar46, J. Palinkas46, Z. Szillasi46, A. Makovec47,
P. Raics47, Z.L. Trocsanyi47, B. Ujvari47, N. Sahoo48, S.K. Swain48, S.B. Beri49,
V. Bhatnagar49, R. Gupta49, U.Bhawandeep49, A.K. Kalsi49, M. Kaur49, R. Kumar49,
M. Mittal49, N. Nishu49, J.B. Singh49, Ashok Kumar50, Arun Kumar50, S. Ahuja50,
A. Bhardwaj50, B.C. Choudhary50, A. Kumar50, S. Malhotra50, M. Naimuddin50,
K. Ranjan50, V. Sharma50, S. Banerjee51, S. Bhattacharya51, K. Chatterjee51, S. Dutta51,
B. Gomber51, Sa. Jain51, Sh. Jain51, R. Khurana51, A. Modak51, S. Mukherjee51, D. Roy51,
S. Sarkar51, M. Sharan51, A. Abdulsalam52, D. Dutta52, S. Kailas52, V. Kumar52,
A.K. Mohanty52,205, L.M. Pant52, P. Shukla52, A. Topkar52, T. Aziz53, S. Banerjee53,
S. Bhowmik53,222, R.M. Chatterjee53, R.K. Dewanjee53, S. Dugad53, S. Ganguly53,
S. Ghosh53, M. Guchait53, A. Gurtu53,223, G. Kole53, S. Kumar53, M. Maity53,222,
G. Majumder53, K. Mazumdar53, G.B. Mohanty53, B. Parida53, K. Sudhakar53,
N. Wickramage53,224, H. Bakhshiansohi54, H. Behnamian54, S.M. Etesami54,225,
A. Fahim54,226, R. Goldouzian54, M. Khakzad54, M. Mohammadi Najafabadi54,
M. Naseri54, S. Paktinat Mehdiabadi54, F. Rezaei Hosseinabadi54, B. Safarzadeh54,227,
M. Zeinali54, M. Felcini55, M. Grunewald55, M. Abbrescia57,58, C. Calabria57,58,
S.S. Chhibra57,58, A. Colaleo57, D. Creanza57,59, N. De Filippis57,59, M. De Palma57,58,
L. Fiore57, G. Iaselli57,59, G. Maggi57,59, M. Maggi57, S. My57,59, S. Nuzzo57,58,
A. Pompili57,58, G. Pugliese57,59, R. Radogna57,58,205, G. Selvaggi57,58, A. Sharma57,
L. Silvestris57,205, R. Venditti57,58, P. Verwilligen57, G. Abbiendi61, A.C. Benvenuti61,
D. Bonacorsi61,62, S. Braibant-Giacomelli61,62, L. Brigliadori61,62, R. Campanini61,62,
P. Capiluppi61,62, A. Castro61,62, F.R. Cavallo61, G. Codispoti61,62, M. Cuffiani61,62,
G.M. Dallavalle61, F. Fabbri61, A. Fanfani61,62, D. Fasanella61,62, P. Giacomelli61,
C. Grandi61, L. Guiducci61,62, S. Marcellini61, G. Masetti61, A. Montanari61,
F.L. Navarria61,62, A. Perrotta61, F. Primavera61,62, A.M. Rossi61,62, T. Rovelli61,62,
G.P. Siroli61,62, N. Tosi61,62, R. Travaglini61,62, S. Albergo64,65, G. Cappello64,
M. Chiorboli64,65, S. Costa64,65, F. Giordano64,205, R. Potenza64,65, A. Tricomi64,65,
C. Tuve64,65, G. Barbagli68, V. Ciulli68,69, C. Civinini68, R. D’Alessandro68,69,
E. Focardi68,69, E. Gallo68, S. Gonzi68,69, V. Gori68,69, P. Lenzi68,69, M. Meschini68,
S. Paoletti68, G. Sguazzoni68, A. Tropiano68,69, L. Benussi70, S. Bianco70, F. Fabbri70,
D. Piccolo70, R. Ferretti72,73, F. Ferro72, M. Lo Vetere72,73, E. Robutti72, S. Tosi72,73,
M.E. Dinardo75,76, S. Fiorendi75,76, S. Gennai75,205, R. Gerosa75,76,205, A. Ghezzi75,76,
P.Govoni75,76, M.T. Lucchini75,76,205, S.Malvezzi75, R.A. Manzoni75,76, A.Martelli75,76, B.
Marzocchi75,76,205, D. Menasce75, L. Moroni75, M. Paganoni75,76, D. Pedrini75,
S. Ragazzi75,76, N. Redaelli75, T. Tabarelli de Fatis75,76, S. Buontempo78, N. Cavallo78,80,
S. Di Guida78,81,205, F. Fabozzi78,80, A.O.M. Iorio78,79, L. Lista78, S. Meola78,81,205,
M. Merola78, P. Paolucci78,205, P. Azzi83, N. Bacchetta83, D. Bisello83,84, A. Branca83,84,
R. Carlin83,84, P. Checchia83, M. Dall’Osso83,84, T. Dorigo83, U. Dosselli83, M. Galanti83,84,
F. Gasparini83,84, U. Gasparini83,84, P. Giubilato83,84, A. Gozzelino83, K. Kanishchev83,85,
S. Lacaprara83, M. Margoni83,84, A.T. Meneguzzo83,84, J. Pazzini83,84, N. Pozzobon83,84,
P. Ronchese83,84, F. Simonetto83,84, E. Torassa83, M. Tosi83,84, P. Zotto83,84,
A. Zucchetta83,84, G. Zumerle83,84, M. Gabusi87,88, S.P. Ratti87,88, V. Re87, C.Riccardi87,88,
P. Salvini87, P. Vitulo87,88, M. Biasini90,91, G.M. Bilei90, D.Ciangottini90,91,205, L. Fanò90,91,
P. Lariccia90,91, G. Mantovani90,91, M. Menichelli90, A. Saha90, A. Santocchia90,91,
A. Spiezia90,91,205, K. Androsov93,228, P. Azzurri93, G. Bagliesi93, J. Bernardini93,
T. Boccali93, G. Broccolo93,95, R. Castaldi93, M.A. Ciocci93,228, R. Dell’Orso93,
S. Donato93,95,205, F. Fiori93,95, L. Foà93,95, A. Giassi93, M.T. Grippo93,228, F. Ligabue93,95,
T. Lomtadze93, L. Martini93,94, A. Messineo93,94, C.S. Moon93,229, F. Palla93,205,
A. Rizzi93,94, A. Savoy-Navarro93,230, A.T. Serban93, P. Spagnolo93, P. Squillacioti93,228,
R. Tenchini93, G. Tonelli93,94, A. Venturi93, P.G. Verdini93, C. Vernieri93,95, L. Barone97,98,
F. Cavallari97, G. D’imperio97,98, D. Del Re97,98, M. Diemoz97, C. Jorda97, E. Longo97,98,
F.Margaroli97,98, P.Meridiani97, F. Micheli97,98,205, S.Nourbakhsh97,98, G.Organtini97,98,
R. Paramatti97, S. Rahatlou97,98, C. Rovelli97, F. Santanastasio97,98, L. Soffi97,98,
P. Traczyk97,98,205, N. Amapane100,101, R. Arcidiacono100,102, S. Argiro100,101,
M. Arneodo100,102, R. Bellan100,101, C. Biino100, N. Cartiglia100, S. Casasso100,101,205,
M. Costa100,101, A. Degano100,101, N. Demaria100, L. Finco100,101,205, C. Mariotti100,
S. Maselli100, E. Migliore100,101, V. Monaco100,101, M. Musich100, M.M. Obertino100,102,
L. Pacher100,101, N. Pastrone100, M. Pelliccioni100, G.L. Pinna Angioni100,101,
A. Potenza100,101, A. Romero100,101, M. Ruspa100,102, R. Sacchi100,101, A. Solano100,101,
A. Staiano100, U. Tamponi100, S. Belforte104, V. Candelise104,105,205, M. Casarsa104,
F. Cossutti104, G. Della Ricca104,105, B. Gobbo104, C. La Licata104,105, M. Marone104,105,
A. Schizzi104,105, T. Umer104,105, A. Zanetti104, S. Chang106, A. Kropivnitskaya106,
S.K. Nam106, D.H. Kim107, G.N. Kim107, M.S. Kim107, D.J. Kong107, S. Lee107, Y.D. Oh107,
H. Park107, A. Sakharov107, D.C. Son107, T.J. Kim108, J.Y. Kim109, S. Song109, S. Choi110,
D. Gyun110, B. Hong110, M. Jo110, H. Kim110, Y. Kim110, B. Lee110, K.S. Lee110,
S.K. Park110, Y. Roh110, H.D. Yoo111, M. Choi112, J.H. Kim112, I.C. Park112, G. Ryu112,
M.S. Ryu112, Y. Choi113, Y.K. Choi113, J. Goh113, D. Kim113, E. Kwon113, J. Lee113, I. Yu113,
A. Juodagalvis114, J.R. Komaragiri115, M.A.B. Md Ali115, E. Casimiro Linares116,
H. Castilla-Valdez116, E. De La Cruz-Burelo116, I. Heredia-de La Cruz116,231,
A. Hernandez-Almada116, R. Lopez-Fernandez116, A. Sanchez-Hernandez116, S. Carrillo
Moreno117, F. Vazquez Valencia117, I. Pedraza118, H.A. Salazar Ibarguen118, A. Morelos
Pineda119, D. Krofcheck120, P.H. Butler121, S. Reucroft121, A. Ahmad122, M. Ahmad122,
Q. Hassan122, H.R. Hoorani122, W.A. Khan122, T. Khurshid122, M. Shoaib122,

RESEARCH LETTER

G2015 Macmillan Publishers Limited. All rights reserved



H. Bialkowska123, M. Bluj123, B. Boimska123, T. Frueboes123, M. Górski123, M. Kazana123,
K.Nawrocki123, K.Romanowska-Rybinska123, M.Szleper123, P. Zalewski123, G.Brona124,
K. Bunkowski124, M. Cwiok124, W. Dominik124, K. Doroba124, A. Kalinowski124,
M. Konecki124, J. Krolikowski124, M. Misiura124, M. Olszewski124, W. Wolszczak124,
P. Bargassa125, C. Beirão Da Cruz E Silva125, P. Faccioli125, P.G. Ferreira Parracho125,
M. Gallinaro125, L. Lloret Iglesias125, F. Nguyen125, J. Rodrigues Antunes125, J. Seixas125,
J. Varela125, P. Vischia125, S. Afanasiev126, P. Bunin126, M. Gavrilenko126, I. Golutvin126,
I. Gorbunov126, A. Kamenev126, V. Karjavin126, V. Konoplyanikov126, A. Lanev126,
A. Malakhov126, V. Matveev126,232, P. Moisenz126, V. Palichik126, V. Perelygin126,
S. Shmatov126, N. Skatchkov126, V. Smirnov126, A. Zarubin126, V. Golovtsov127,
Y. Ivanov127, V. Kim127,233, P. Levchenko127, V. Murzin127, V. Oreshkin127, I. Smirnov127,
V.Sulimov127, L.Uvarov127,S.Vavilov127,A.Vorobyev127,An.Vorobyev127, Yu.Andreev128,
A. Dermenev128, S. Gninenko128, N. Golubev128, M. Kirsanov128, N. Krasnikov128,
A. Pashenkov128, D. Tlisov128, A. Toropin128, V. Epshteyn129, V. Gavrilov129,
N. Lychkovskaya129, V. Popov129, I. Pozdnyakov129, G. Safronov129, S. Semenov129,
A. Spiridonov129, V. Stolin129, E. Vlasov129, A. Zhokin129, V. Andreev130,
M. Azarkin130, I. Dremin130, M. Kirakosyan130, A. Leonidov130, G. Mesyats130,
S.V. Rusakov130, A. Vinogradov130, A. Belyaev131, E. Boos131, M. Dubinin131,234,
L. Dudko131, A. Ershov131, A. Gribushin131, V. Klyukhin131, O. Kodolova131, I. Lokhtin131,
S. Obraztsov131, S. Petrushanko131, V. Savrin131, A. Snigirev131, I. Azhgirey132,
I. Bayshev132, S. Bitioukov132, V. Kachanov132, A. Kalinin132, D. Konstantinov132,
V. Krychkine132, V. Petrov132, R. Ryutin132, A. Sobol132, L. Tourtchanovitch132,
S. Troshin132, N. Tyurin132, A.Uzunian132, A. Volkov132, P. Adzic133,235, M. Ekmedzic133,
J. Milosevic133, V. Rekovic133, J. Alcaraz Maestre134, C. Battilana134, E. Calvo134,
M. Cerrada134, M. Chamizo Llatas134, N. Colino134, B. De La Cruz134, A. Delgado
Peris134, D. Domı́nguez Vázquez134, A. Escalante Del Valle134, C. Fernandez Bedoya134,
J.P. Fernández Ramos134, J. Flix134, M.C. Fouz134, P. Garcia-Abia134, O. Gonzalez
Lopez134, S. Goy Lopez134, J.M. Hernandez134, M.I. Josa134, E. Navarro De Martino134,
A.Pérez-CaleroYzquierdo134, J.PuertaPelayo134,A.QuintarioOlmeda134, I. Redondo134,
L. Romero134, M.S. Soares134, C. Albajar135, J.F. de Trocóniz135, M. Missiroli135,
D. Moran135, H. Brun136, J. Cuevas136, J. Fernandez Menendez136, S. Folgueras136,
I. Gonzalez Caballero136, J.A. Brochero Cifuentes137, I.J. Cabrillo137, A. Calderon137,
J. Duarte Campderros137, M. Fernandez137, G. Gomez137, A. Graziano137, A. Lopez
Virto137, J. Marco137, R. Marco137, C. Martinez Rivero137, F. Matorras137, F.J. Munoz
Sanchez137, J. Piedra Gomez137, T. Rodrigo137, A.Y. Rodrı́guez-Marrero137, A. Ruiz-
Jimeno137, L.Scodellaro137, I. Vila137,R.VilarCortabitarte137,D.Abbaneo138, E.Auffray138,
G. Auzinger138, M. Bachtis138, P. Baillon138, A.H. Ball138, D. Barney138, A. Benaglia138,
J. Bendavid138, L. Benhabib138, J.F. Benitez138, C. Bernet138,210, P. Bloch138, A. Bocci138,
A. Bonato138, O. Bondu138, C. Botta138, H. Breuker138, T. Camporesi138, G. Cerminara138,
S. Colafranceschi138,236, M. D’Alfonso138, D. d’Enterria138, A. Dabrowski138, A. David138,
F. De Guio138, A. De Roeck138, S. De Visscher138, E. Di Marco138, M. Dobson138,
M. Dordevic138, N. Dupont-Sagorin138, A. Elliott-Peisert138, G. Franzoni138, W. Funk138,
D. Gigi138, K. Gill138, D. Giordano138, M. Girone138, F. Glege138, R. Guida138,
S. Gundacker138, M. Guthoff138, J. Hammer138, M. Hansen138, P. Harris138,
J. Hegeman138, V. Innocente138, P. Janot138, K. Kousouris138, K. Krajczar138, P. Lecoq138,
C. Lourenço138, N. Magini138, L. Malgeri138, M. Mannelli138, J. Marrouche138, L. Masetti138,
F. Meijers138, S. Mersi138, E. Meschi138, F. Moortgat138, S. Morovic138, M. Mulders138,
L. Orsini138, L. Pape138, E. Perez138, L. Perrozzi138, A. Petrilli138, G. Petrucciani138,
A. Pfeiffer138, M. Pimiä138, D. Piparo138, M. Plagge138, A. Racz138, G. Rolandi138,237,
M. Rovere138, H. Sakulin138, C. Schäfer138, C. Schwick138, A. Sharma138, P. Siegrist138,
P. Silva138, M. Simon138, P. Sphicas138,238, D. Spiga138, J. Steggemann138, B. Stieger138,
M. Stoye138, Y. Takahashi138, D. Treille138, A. Tsirou138, G.I. Veres138,220, N. Wardle138,
H.K. Wöhri138, H. Wollny138, W.D. Zeuner138, W. Bertl139, K. Deiters139, W. Erdmann139,
R. Horisberger139, Q. Ingram139, H.C. Kaestli139, D. Kotlinski139, D. Renker139, T. Rohe139,
F. Bachmair140, L. Bäni140, L. Bianchini140, M.A. Buchmann140, B. Casal140, N. Chanon140,
G. Dissertori140, M. Dittmar140, M. Donegà140, M. Dünser140, P. Eller140, C. Grab140,
D. Hits140, J. Hoss140, W. Lustermann140, B. Mangano140, A.C. Marini140, M.
Marionneau140, P. Martinez Ruiz del Arbol140, M. Masciovecchio140, D. Meister140,
N. Mohr140, P. Musella140, C. Nägeli140,239, F. Nessi-Tedaldi140, F. Pandolfi140,
F. Pauss140, M. Peruzzi140, M. Quittnat140, L. Rebane140, M. Rossini140,
A. Starodumov140,240, M. Takahashi140, K. Theofilatos140, R. Wallny140,
H.A. Weber140, C. Amsler141,241, M.F. Canelli141, V. Chiochia141, A. De Cosa141,
A. Hinzmann141, T. Hreus141, B. Kilminster141, C. Lange141, B. Millan Mejias141,
J. Ngadiuba141, D. Pinna141, P. Robmann141, F.J. Ronga141, S. Taroni141, M. Verzetti141,
Y. Yang141, M. Cardaci142, K.H. Chen142, C. Ferro142, C.M. Kuo142, W. Lin142, Y.J. Lu142,
R. Volpe142, S.S. Yu142, P. Chang143, Y.H. Chang143, Y.W. Chang143, Y. Chao143, K.F.
Chen143, P.H.Chen143, C.Dietz143, U.Grundler143, W.-S. Hou143, K.Y.Kao143, Y.F. Liu143,
R.-S. Lu143, D. Majumder143, E. Petrakou143, Y.M. Tzeng143, R. Wilken143,
B. Asavapibhop144, G. Singh144, N. Srimanobhas144, N. Suwonjandee144,
A. Adiguzel145, M.N. Bakirci145,242, S. Cerci145,243, C. Dozen145, I. Dumanoglu145,
E. Eskut145, S. Girgis145, G. Gokbulut145, E. Gurpinar145, I. Hos145, E.E. Kangal145,
A. Kayis Topaksu145, G. Onengut145,244, K. Ozdemir145, S. Ozturk145,242, A. Polatoz145,
D. Sunar Cerci145,243, B. Tali145,243, H. Topakli145,242, M. Vergili145, I.V. Akin146,
B. Bilin146, S. Bilmis146, H. Gamsizkan146,245, B. Isildak146,246, G. Karapinar146,247,
K. Ocalan146,248, S. Sekmen146, U.E. Surat146, M. Yalvac146, M. Zeyrek146,
E.A. Albayrak147,249, E. Gülmez147, M. Kaya147,250, O. Kaya147,251, T. Yetkin147,252,
K. Cankocak148, F.I. Vardarlı148, L. Levchuk149, P. Sorokin149, J.J. Brooke150,
E. Clement150, D. Cussans150, H. Flacher150, J. Goldstein150, M. Grimes150, G.P. Heath150,
H.F. Heath150, J. Jacob150, L. Kreczko150, C. Lucas150, Z. Meng150, D.M. Newbold150,253,
S. Paramesvaran150, A. Poll150, T. Sakuma150, S. Senkin150, V.J. Smith150, K.W. Bell151,
A. Belyaev151,254, C. Brew151, R.M. Brown151, D.J.A. Cockerill151, J.A. Coughlan151,
K. Harder151, S. Harper151, E. Olaiya151, D. Petyt151, C.H. Shepherd-Themistocleous151,
A. Thea151, I.R. Tomalin151, T. Williams151, W.J. Womersley151, S.D. Worm151, M. Baber152,
R. Bainbridge152, O. Buchmuller152, D. Burton152, D. Colling152, N. Cripps152,
P. Dauncey152, G. Davies152, M. Della Negra152, P. Dunne152, W. Ferguson152, J.
Fulcher152,D. Futyan152,G.Hall152,G. Iles152,M. Jarvis152,G.Karapostoli152,M.Kenzie152,
R. Lane152, R. Lucas152,253, L. Lyons152, A.-M. Magnan152, S. Malik152, B. Mathias152,

J. Nash152, A. Nikitenko152,240, J. Pela152, M. Pesaresi152, K. Petridis152, D.M. Raymond152,
S. Rogerson152, A. Rose152, C. Seez152, P. Sharp152{, A. Tapper152, M. Vazquez Acosta152,
T. Virdee152, S.C. Zenz152, J.E. Cole153, P.R. Hobson153, A. Khan153, P. Kyberd153,
D. Leggat153, D. Leslie153, I.D. Reid153, P. Symonds153, L. Teodorescu153, M. Turner153,
J.Dittmann154,K.Hatakeyama154,A.Kasmi154,H. Liu154, T. Scarborough154,O.Charaf155,
S.I. Cooper155, C.Henderson155, P. Rumerio155, A. Avetisyan156, T. Bose156, C. Fantasia156,
P. Lawson156, C. Richardson156, J. Rohlf156, J. St. John156, L. Sulak156, J. Alimena157,
E. Berry157, S. Bhattacharya157, G. Christopher157, D. Cutts157, Z. Demiragli157,
N. Dhingra157, A. Ferapontov157, A. Garabedian157, U. Heintz157, G. Kukartsev157,
E. Laird157, G. Landsberg157, M. Luk157, M. Narain157, M. Segala157, T. Sinthuprasith157,
T. Speer157, J. Swanson157, R. Breedon158, G. Breto158, M. Calderon De La Barca
Sanchez158, S. Chauhan158, M. Chertok158, J. Conway158, R. Conway158, P.T. Cox158,
R. Erbacher158, M. Gardner158, W. Ko158, R. Lander158, M. Mulhearn158, D. Pellett158,
J. Pilot158, F. Ricci-Tam158, S. Shalhout158, J. Smith158, M. Squires158, D. Stolp158,
M. Tripathi158, S. Wilbur158, R. Yohay158, R. Cousins159, P. Everaerts159, C. Farrell159,
J. Hauser159, M. Ignatenko159, G. Rakness159, E. Takasugi159, V. Valuev159, M. Weber159,
K. Burt160, R. Clare160, J. Ellison160, J.W. Gary160, G. Hanson160, J. Heilman160, M. Ivova
Rikova160, P. Jandir160, E. Kennedy160, F. Lacroix160, O.R. Long160, A. Luthra160,
M. Malberti160, M. Olmedo Negrete160, A. Shrinivas160, S. Sumowidagdo160,
S. Wimpenny160, J.G. Branson161, G.B. Cerati161, S. Cittolin161, R.T. D’Agnolo161,
A. Holzner161, R. Kelley161, D. Klein161, D. Kovalskyi161, J. Letts161, I. Macneill161,
D. Olivito161, S. Padhi161, C. Palmer161, M. Pieri161, M. Sani161, V. Sharma161, S. Simon161,
Y. Tu161, A. Vartak161, C. Welke161, F. Würthwein161, A. Yagil161, D. Barge162, J. Bradmiller-
Feld162, C. Campagnari162, T. Danielson162, A. Dishaw162, V. Dutta162, K. Flowers162,
M. Franco Sevilla162, P. Geffert162, C. George162, F. Golf162, L. Gouskos162, J. Incandela162,
C. Justus162, N. Mccoll162, J. Richman162, D. Stuart162, W. To162, C. West162, J. Yoo162,
A. Apresyan163, A. Bornheim163, J. Bunn163, Y. Chen163, J. Duarte163, A. Mott163,
H.B. Newman163, C. Pena163, M. Pierini163, M. Spiropulu163, J.R. Vlimant163,
R. Wilkinson163, S. Xie163, R.Y. Zhu163, V. Azzolini164, A. Calamba164, B. Carlson164,
T. Ferguson164, Y. Iiyama164, M. Paulini164, J. Russ164, H. Vogel164, I. Vorobiev164,
J.P. Cumalat165, W.T. Ford165, A. Gaz165, M. Krohn165, E. Luiggi Lopez165,
U. Nauenberg165, J.G. Smith165, K. Stenson165, S.R. Wagner165, J. Alexander166,
A. Chatterjee166, J. Chaves166, J. Chu166, S. Dittmer166, N. Eggert166, N. Mirman166,
G. Nicolas Kaufman166, J.R. Patterson166, A. Ryd166, E. Salvati166, L. Skinnari166,
W. Sun166, W.D. Teo166, J. Thom166, J. Thompson166, J. Tucker166, Y. Weng166,
L. Winstrom166, P. Wittich166, D. Winn167, S. Abdullin168, M. Albrow168,
J. Anderson168, G. Apollinari168, L.A.T. Bauerdick168, A. Beretvas168, J. Berryhill168,
P.C. Bhat168, G. Bolla168, K. Burkett168, J.N. Butler168, H.W.K. Cheung168,
F. Chlebana168, S. Cihangir168, V.D. Elvira168, I. Fisk168, J. Freeman168, Y. Gao168,
E. Gottschalk168, L. Gray168, D. Green168, S. Grünendahl168, O. Gutsche168,
J. Hanlon168, D. Hare168, R.M. Harris168, J. Hirschauer168, B. Hooberman168,
S. Jindariani168, M. Johnson168, U. Joshi168, K. Kaadze168, B. Klima168, B. Kreis168,
S. Kwan168{, J. Linacre168, D. Lincoln168, R. Lipton168, T. Liu168, J. Lykken168,
K. Maeshima168, J.M. Marraffino168, V.I. Martinez Outschoorn168, S. Maruyama168,
D. Mason168, P. McBride168, P. Merkel168, K. Mishra168, S. Mrenna168, S. Nahn168,
C. Newman-Holmes168, V. O’Dell168, O. Prokofyev168, E. Sexton-Kennedy168,
S. Sharma168, A. Soha168, W.J. Spalding168, L. Spiegel168, L. Taylor168, S. Tkaczyk168,
N.V. Tran168, L. Uplegger168, E.W. Vaandering168, R. Vidal168, A. Whitbeck168,
J. Whitmore168, F. Yang168, D. Acosta169, P. Avery169, P. Bortignon169, D. Bourilkov169,
M.Carver169,D.Curry169, S.Das169,M.DeGruttola169,G.P.DiGiovanni169, R.D. Field169,
M. Fisher169, I.K. Furic169, J. Hugon169, J. Konigsberg169, A. Korytov169, T. Kypreos169,
J.F. Low169, K. Matchev169, H. Mei169, P. Milenovic169,255, G. Mitselmakher169,
L. Muniz169, A. Rinkevicius169, L. Shchutska169, M. Snowball169, D. Sperka169,
J. Yelton169, M. Zakaria169, S. Hewamanage170, S. Linn170, P. Markowitz170,
G. Martinez170, J.L. Rodriguez170, T. Adams171, A. Askew171, J. Bochenek171,
B. Diamond171, J. Haas171, S. Hagopian171, V. Hagopian171, K.F. Johnson171,
H. Prosper171, V. Veeraraghavan171, M. Weinberg171, M.M. Baarmand172,
M. Hohlmann172, H. Kalakhety172, F. Yumiceva172, M.R. Adams173, L. Apanasevich173,
D. Berry173, R.R. Betts173, I. Bucinskaite173, R. Cavanaugh173, O. Evdokimov173,
L. Gauthier173, C.E. Gerber173, D.J. Hofman173, P. Kurt173, D.H. Moon173, C. O’Brien173,
I.D. Sandoval Gonzalez173, C. Silkworth173, P. Turner173, N. Varelas173, B. Bilki174,256,
W. Clarida174, K. Dilsiz174, M. Haytmyradov174, J.-P. Merlo174, H. Mermerkaya174,257,
A. Mestvirishvili174, A. Moeller174, J. Nachtman174, H. Ogul174, Y. Onel174, F. Ozok174,249,
A. Penzo174, R. Rahmat174, S. Sen174, P. Tan174, E. Tiras174, J. Wetzel174, K. Yi174,
B.A. Barnett175, B. Blumenfeld175, S. Bolognesi175, D. Fehling175, A.V. Gritsan175,
P. Maksimovic175, C. Martin175, M. Swartz175, P. Baringer176, A. Bean176, G. Benelli176,
C. Bruner176, R.P. Kenny III176, M. Malek176, M. Murray176, D. Noonan176, S. Sanders176,
J. Sekaric176, R. Stringer176, Q. Wang176, J.S. Wood176, I. Chakaberia177, A. Ivanov177,
S. Khalil177, M. Makouski177, Y. Maravin177, L.K. Saini177, N. Skhirtladze177,
I. Svintradze177, J. Gronberg178, D. Lange178, F. Rebassoo178, D. Wright178,
A. Baden179, A. Belloni179, B. Calvert179, S.C. Eno179, J.A. Gomez179, N.J. Hadley179,
R.G. Kellogg179, T. Kolberg179, Y. Lu179, A.C. Mignerey179, K. Pedro179, A. Skuja179,
M.B. Tonjes179, S.C. Tonwar179, A. Apyan180, R. Barbieri180, G. Bauer180, W. Busza180,
I.A. Cali180, M. Chan180, L. Di Matteo180, G. Gomez Ceballos180, M. Goncharov180,
D. Gulhan180, M. Klute180, Y.S. Lai180, Y.-J. Lee180, A. Levin180, P.D. Luckey180, T. Ma180,
C. Paus180, D. Ralph180, C. Roland180, G. Roland180, G.S.F. Stephans180, K. Sumorok180,
D. Velicanu180, J. Veverka180, B. Wyslouch180, M. Yang180, M. Zanetti180, V. Zhukova180,
B. Dahmes181, A. Gude181, S.C. Kao181, K. Klapoetke181, Y. Kubota181, J. Mans181,
N. Pastika181, R. Rusack181, A. Singovsky181, N. Tambe181, J. Turkewitz181,
J.G. Acosta182, S. Oliveros182, E. Avdeeva183, K. Bloom183, S. Bose183, D.R. Claes183,
A. Dominguez183, R. Gonzalez Suarez183, J. Keller183, D. Knowlton183, I. Kravchenko183,
J. Lazo-Flores183, F. Meier183, F. Ratnikov183, G.R. Snow183, M. Zvada183, J. Dolen184,
A. Godshalk184, I. Iashvili184, A. Kharchilava184, A. Kumar184, S. Rappoccio184,
G. Alverson185, E. Barberis185, D. Baumgartel185, M. Chasco185, A. Massironi185,
D.M. Morse185, D. Nash185, T. Orimoto185, D. Trocino185, R.-J. Wang185, D. Wood185,
J. Zhang185, K.A. Hahn186, A. Kubik186, N. Mucia186, N. Odell186, B. Pollack186,
A. Pozdnyakov186, M. Schmitt186, S. Stoynev186, K. Sung186, M. Velasco186, S. Won186,

LETTER RESEARCH

G2015 Macmillan Publishers Limited. All rights reserved



A. Brinkerhoff187, K.M. Chan187, A. Drozdetskiy187, M. Hildreth187, C. Jessop187,
D.J. Karmgard187, N. Kellams187, K. Lannon187, S. Lynch187, N. Marinelli187,
Y. Musienko187,232, T. Pearson187, M. Planer187, R. Ruchti187, G. Smith187, N. Valls187,
M. Wayne187, M. Wolf187, A. Woodard187, L. Antonelli188, J. Brinson188, B. Bylsma188,
L.S. Durkin188, S. Flowers188, A. Hart188, C. Hill188, R. Hughes188, K. Kotov188,
T.Y. Ling188, W. Luo188, D. Puigh188, M. Rodenburg188, B.L. Winer188, H. Wolfe188,
H.W. Wulsin188, O. Driga189, P. Elmer189, J. Hardenbrook189, P. Hebda189, A. Hunt189,
S.A. Koay189, P. Lujan189, D. Marlow189, T. Medvedeva189, M. Mooney189, J. Olsen189,
P. Piroué189, X. Quan189, H. Saka189, D. Stickland189,205, C. Tully189, J.S. Werner189,
A. Zuranski189, E. Brownson190, S. Malik190, H. Mendez190, J.E. Ramirez Vargas190,
V.E. Barnes191, D. Benedetti191, D. Bortoletto191, M. De Mattia191, L. Gutay191, Z. Hu191,
M.K. Jha191, M. Jones191, K. Jung191, M. Kress191, N. Leonardo191, D.H. Miller191,
N. Neumeister191, B.C. Radburn-Smith191, X. Shi191, I. Shipsey191, D. Silvers191,
A. Svyatkovskiy191, F. Wang191, W. Xie191, L. Xu191, J. Zablocki191, N. Parashar192,
J. Stupak192, A. Adair193, B. Akgun193, K.M. Ecklund193, F.J.M. Geurts193, W. Li193,
B. Michlin193, B.P. Padley193, R. Redjimi193, J. Roberts193, J. Zabel193, B. Betchart194,
A. Bodek194, R. Covarelli194, P. de Barbaro194, R. Demina194, Y. Eshaq194, T. Ferbel194,
A. Garcia-Bellido194, P. Goldenzweig194, J. Han194, A. Harel194, A. Khukhunaishvili194,
S. Korjenevski194, G. Petrillo194, D. Vishnevskiy194, R. Ciesielski195, L. Demortier195,
K. Goulianos195, C. Mesropian195, S. Arora196, A. Barker196, J.P. Chou196, C. Contreras-
Campana196, E.Contreras-Campana196, D.Duggan196, D. Ferencek196, Y.Gershtein196,
R. Gray196, E. Halkiadakis196, D. Hidas196, S. Kaplan196, A. Lath196, S. Panwalkar196,
M. Park196, R. Patel196, S. Salur196, S. Schnetzer196, S. Somalwar196, R. Stone196,
S. Thomas196, P. Thomassen196, M. Walker196, K. Rose197, S. Spanier197, A. York197,
O. Bouhali198,258, A. Castaneda Hernandez198, R. Eusebi198, W. Flanagan198,
J. Gilmore198, T. Kamon198,259, V. Khotilovich198, V. Krutelyov198, R. Montalvo198,
I. Osipenkov198, Y. Pakhotin198, A. Perloff198, J. Roe198, A. Rose198, A. Safonov198,
I. Suarez198, A. Tatarinov198, K.A. Ulmer198, N. Akchurin199, C.Cowden199, J. Damgov199,
C. Dragoiu199, P.R. Dudero199, J. Faulkner199, K. Kovitanggoon199, S. Kunori199,
S.W. Lee199, T. Libeiro199, I. Volobouev199, E. Appelt200, A.G. Delannoy200, S. Greene200,
A. Gurrola200, W. Johns200, C. Maguire200, Y. Mao200, A. Melo200, M. Sharma200,
P. Sheldon200, B. Snook200, S. Tuo200, J. Velkovska200, M.W. Arenton201, S. Boutle201,
B. Cox201, B. Francis201, J. Goodell201, R. Hirosky201, A. Ledovskoy201, H. Li201, C. Lin201,
C. Neu201, J. Wood201, C. Clarke202, R. Harr202, P.E. Karchin202, C. Kottachchi
Kankanamge Don202, P. Lamichhane202, J. Sturdy202, D.A. Belknap203, D. Carlsmith203,
M. Cepeda203, S. Dasu203, L. Dodd203, S. Duric203, E. Friis203, R. Hall-Wilton203,
M. Herndon203, A. Hervé203, P. Klabbers203, A. Lanaro203, C. Lazaridis203, A. Levine203,
R. Loveless203, A. Mohapatra203, I. Ojalvo203, T. Perry203, G.A. Pierro203, G. Polese203,
I. Ross203, T. Sarangi203, A. Savin203, W.H. Smith203, D. Taylor203, C. Vuosalo203 &
N. Woods203

Primary affiliations
1Yerevan Physics Institute, Yerevan, Armenia. 2Institut für Hochenergiephysik der
OeAW, Wien, Austria. 3National Centre for Particle and High Energy Physics, Minsk,
Belarus. 4Universiteit Antwerpen, Antwerpen, Belgium. 5Vrije Universiteit Brussel,
Brussel, Belgium. 6Université Libre de Bruxelles, Bruxelles, Belgium. 7Ghent University,
Ghent, Belgium. 8Université Catholique de Louvain, Louvain-la-Neuve, Belgium.
9Université de Mons, Mons, Belgium. 10Centro Brasileiro de Pesquisas Fisicas, Rio de
Janeiro, Brazil. 11Universidade do Estado do Rio de Janeiro, Rio de Janeiro, Brazil.
12Universidade Estadual Paulista, Universidade Federal do ABC, São Paulo, Brazil.
13Universidade Estadual Paulista. 14Universidade Federal do ABC. 15Institute for
Nuclear Research and Nuclear Energy, Sofia, Bulgaria. 16University of Sofia, Sofia,
Bulgaria. 17Institute of High Energy Physics, Beijing, China. 18State Key Laboratory of
Nuclear Physics and Technology, Peking University, Beijing, China. 19Universidad de
Los Andes, Bogota, Colombia. 20University of Split, Faculty of Electrical Engineering,
Mechanical Engineering and Naval Architecture, Split, Croatia. 21University of Split,
Faculty of Science, Split, Croatia. 22Institute Rudjer Boskovic, Zagreb, Croatia.
23University of Cyprus, Nicosia, Cyprus. 24Charles University, Prague, Czech Republic.
25Academy of Scientific Research and Technology of the Arab Republic of Egypt,
Egyptian Network of High Energy Physics, Cairo, Egypt. 26National Institute of Chemical
Physics and Biophysics, Tallinn, Estonia. 27Department of Physics, University of
Helsinki, Helsinki, Finland. 28Helsinki Institute of Physics, Helsinki, Finland.
29Lappeenranta University of Technology, Lappeenranta, Finland. 30DSM/IRFU, CEA/
Saclay, Gif-sur-Yvette, France. 31Laboratoire Leprince-Ringuet, Ecole Polytechnique,
IN2P3-CNRS, Palaiseau, France. 32Institut Pluridisciplinaire Hubert Curien, Université
deStrasbourg,Université deHauteAlsace Mulhouse, CNRS/IN2P3,Strasbourg, France.
33Centre de Calcul de l’Institut National de Physique Nucleaire et de Physique des
Particules, CNRS/IN2P3, Villeurbanne, France. 34Université de Lyon, Université Claude
Bernard Lyon 1, CNRS-IN2P3, Institut de Physique Nucléaire de Lyon, Villeurbanne,
France. 35Institute of High Energy Physics and Informatization, Tbilisi State University,
Tbilisi, Georgia. 36RWTH Aachen University, I. Physikalisches Institut, Aachen, Germany.
37RWTH Aachen Un