Search for $CP$ violation in $\Xi_c^+\rightarrow pK^-\pi^+$ decays using model-independent techniques

A first search for $CP$ violation in the Cabibbo-suppressed $\Xi_c^+\rightarrow pK^-\pi^+$ decay is performed using both a binned and an unbinned model-independent technique in the Dalitz plot. The studies are based on a sample of proton-proton collision data, corresponding to an integrated luminosity of $3.0~{\rm fb^{-1}}$, and collected by the LHCb experiment at centre-of-mass energies of $7$ and $8~\rm TeV$. The data are consistent with the hypothesis of no $CP$ violation.


Introduction
The non-invariance of fundamental interactions under the combination of charge conjugation and parity transformation, known as CP violation (CP V ), is a key requirement for the generation of the baryon-antibaryon asymmetry in the early Universe [1,2]. In the Standard Model (SM) of particle physics, CP V is included through the introduction of a single irreducible complex phase in the Cabibbo-Kobayashi-Maskawa (CKM) quark-mixing matrix [3,4]. The amount of CP V predicted by the CKM mechanism is not sufficient to explain a matter-dominated universe [5,6] and other sources of CP V are required. The realization of CP V in nature has been well established in the K-and B-meson systems by several experiments [7][8][9][10][11][12][13]. The LHCb experiment has observed for the first time CP V in the charm-meson sector as the difference of the CP asymmetries between the two-body decays D 0 → K − K + and D 0 → π − π + [14]. A similar study using Λ + c to pK − K + and pπ − π + has found no evidence for CP V [15]. Indeed, so far CP V has never been observed in any baryon system. Evidence for CP V in the b baryon sector reported by the LHCb collaboration in [16] has not been confirmed with more data [17]. Further measurements of processes involving the decay of charm hadrons can shed light on the origin and magnitude of CP V mechanisms within the SM and beyond.
In two-body decays of charm hadrons, CP V can manifest itself as an asymmetry between partial decay rates. Multi-body decays offer access to more observables which are sensitive to CP -violating effects. For a three-body baryon decay the kinematics can be characterised by three Euler angles and two squared invariant masses forming the Dalitz plot [18]. The Euler angles are redundant if all initial spin states are integrated over. Interference effects in the Dalitz plot probe CP asymmetries in both the magnitudes and phases of the amplitudes. In three-body decays there can be large local CP asymmetries in the Dalitz plot, even when no significant global CP V exists. A recent example has been measured in the decay B + → π + π − π + [19].
This article describes searches for direct CP V in the SCS decay Ξ + c → pK − π + produced promptly in pp collisions. The Λ + c → pK − π + decay is used as a control mode to study on data the level of experimental asymmetries that pollute the measurement. In this paper, the symbol H + c is used to refer to both Ξ + c and Λ + c . It is assumed that the polarisation of charm baryons produced in pp collisions is sufficiently small to justify the integration over the Euler angles. This measurement uses pp collision data, corresponding to an integrated luminosity of 3 fb −1 , recorded by the LHCb detector in Run 1. About 1 fb −1 is collected in 2011 at a centre-of-mass energy of 7 TeV and 2 fb −1 are collected in 2012 at a centre-of-mass energy of 8 TeV. The magnetic field polarity is reversed regularly during the data taking in order to minimise effects of charged particle and antiparticle detection asymmetries. Approximately half of the data are collected with each polarity.
There is presently no successful method for computing decay amplitudes in multi-body charm decays, which could provide reliable predictions on how the CP asymmetries vary over the phase space of the decay. This situation favours a model-independent approach, which looks for differences between multivariate density distributions for baryons and antibaryons. Therefore, in this article searches for CP V are performed through a direct comparison between the Dalitz plots of Ξ + c and Ξ − c decays using a binned significance (S CP ) method [35] and an unbinned k-nearest neighbour method (kNN) [36][37][38][39], both of which are model independent.

Detector and simulation
The LHCb detector [40,41] is a single-arm forward spectrometer covering the pseudorapidity range 2 < η < 5. It is designed for the study of particles containing b and c quarks. The detector includes a high-precision tracking system consisting of a silicon-strip vertex detector surrounding the pp interaction region, a large-area silicon-strip detector located upstream of a dipole magnet with a bending power of about 4 Tm, and three stations of silicon-strip detectors and straw drift tubes placed downstream of the magnet. The tracking system provides a measurement of the momentum, p, of charged particles with a relative uncertainty that varies from 0.5% at low momentum to 1.0% at 200 GeV/c. The minimum distance of a track to a primary vertex (PV), the impact parameter (IP), is measured with a resolution of (15 + 29/p T ) µm, where p T is the component of the momentum transverse to the beam, in GeV/c. Different types of charged hadrons are distinguished using information from two ring-imaging Cherenkov detectors. Photons, electrons and hadrons are identified by a calorimeter system consisting of scintillatingpad and preshower detectors, an electromagnetic and a hadron calorimeter. Muons are identified by a system composed of alternating layers of iron and multiwire proportional chambers.
Samples of simulated events are used to optimise the signal selection, to derive the angular efficiency and to correct the decay-time efficiency. In the simulation, pp collisions are generated using PYTHIA [42] with a specific LHCb configuration [43]. Decays of hadronic particles are described by EVTGEN [44], in which final-state radiation is generated using PHOTOS [45]. The interaction of the generated particles with the detector, and its response, are implemented using the GEANT4 toolkit [46] as described in Ref. [47].

Selection of signal candidates
The online event selection is performed by a trigger consisting of a hardware stage, based on information from the calorimeter and muon systems, followed by two software stages. At the hardware trigger stage, events are required to have either muons with high p T or hadrons, photons or electrons with a high transverse-energy deposit in the calorimeters. In the first software trigger stage at least one good-quality track with a large p T is required. In the second software trigger stage, an H + c candidate is fully reconstructed by the association of three high-quality tracks forming a secondary vertex of the H + c candidate (SV) which must be well separated from any PV, and the tracks should not pointing to any PV. Requirements are also placed on p and p T of the H + c candidate; on the scalar sum of p T for the three tracks; on the particle identification criteria of the tracks; and on the direction vector from the associated PV to the H + c candidate decay vector and the SV, where the associated PV is that with the least IP χ 2 with respect to the H + c candidate. In the offline analysis, tighter selection requirements are placed on the trackreconstruction quality, the p T and p of the final-state particles. Additional requirements are also made on the SV fit quality, and the minimum significance of the displacement from the SV to any PV in the event. This reduces the contribution of charm baryons from b-hadron decays to less than 5% of the prompt signal. Fiducial requirements are imposed to exclude kinematic regions characterised by large detection asymmetries between particles and antiparticles. Reconstructed particles are accepted if there momenta are within a region defined by |p x | < 0.2p z and |p x | > 0.01p z , where p x and p z are the momentum components along the x and z axes 2 . Large detection asymmetries occur in certain kinematic regions because, for a given magnet polarity, particles of one charge with low p or flying with small polar angles may be deflected outside of the detector acceptance or into the LHC beam pipe, whereas particles of the opposite charge remain within the LHCb detector acceptance. About 25% of the selected charm-baryon candidates are rejected by these fiducial requirements. Differences in reconstruction efficiencies are also observed for candidates where p < 20 GeV/c for all charged tracks. These differences do not cancel by simply averaging the data acquired with opposite magnet polarities. To minimise the difference of the reconstruction efficiency for particles and anti-particles, the momentum of all tracks is required to be greater than 20 GeV/c. This requirement rejects about 20% of the selected charm-baryon candidates.
The distributions of the invariant-mass, M (pK − π + ), of selected Λ + c and Ξ + c candidates are presented in Figs

Methods
The Dalitz plot for H + c → pK − π + is described by the squares of the invariant masses of two pairs of the decay products: M 2 (K − π + ) and M 2 (pK − ). Polarisation effects for the H + c baryons are neglected. Comparisons of the Dalitz plots of H + c and H − c candidates are performed using the binned S CP and the unbinned kNN methods, described in the following. For both the binned S CP and unbinned kNN methods, a signal of CP V is  established if a p-value lower then 3 × 10 −7 is found, corresponding to an exclusion of CP symmetry with a significance of five standard deviations. However, in case that no CPV is found, there is no model-independent mechanism for setting an upper limit on the amount of CP V in the Dalitz plot.

Binned S CP method
In the S CP method the Dalitz plots of particles and antiparticles are divided using an identical binning. The S CP method [35] has been used before for hypothesis testing in charm and beauty decays [39,[49][50][51][52]. This method is used to search for localised asymmetries in the phase space of the decay H + c → pK − π + and is based on a bin-by-bin comparison between the Dalitz plots of baryons, H + c , and antibaryons, H − c . For each bin i of the Dalitz plot, the significance of the difference between the number of H + c (n i + ) and where the factor α is defined as α = n + n − and n + , n − are the total number of candidates. This factor accounts for spurious asymmetries arising in the production of Λ + c or Ξ + c baryons, as well as in the detection of the final-state particles. The production and global detection asymmetries are assumed not to depend on the Dalitz plot position.
A numerical comparison between the Dalitz plots of the H + c and H − c candidates is made using a χ 2 test defined as This test is performed using a minimum of 10 H + c and 10 H − c candidates in each bin. A p-value for the hypothesis of no CPV is obtained considering that the number of degrees of freedom is equal to the total number of bins minus one, due to the constraint on the factor α of the overall H + c and H − c normalisation. In the hypothesis of no CPV, the S CP values are expected to be distributed according to the normal distribution with a mean of zero and a standard deviation of unity. In case of CPV, a deviation from the normal distribution is expected, generating a p-value close to zero.

Unbinned kNN method
The kNN method is based on the concept of a set of nearest neighbour candidates (n k ) in a combined sample of two data sets: baryons and antibaryons. As an unbinned method, the kNN approach is more sensitive to a CPV search in a sample with limited data, compared to that of the binned S CP method. The kNN method is used here to test whether baryons and antibaryons share the same parent distribution function [36][37][38]. To find the n k nearest neighbour events of each H + c and H − c candidates, an Euclidean distance between closest points in the Dalitz plot is used. A test statistic T for the null hypothesis is defined as where I(i, k) = 1 if the i th candidate and its k th nearest neighbour belong to the same sample of H + c or H − c candidates and I(i, k) = 0 otherwise. The test statistic T is the mean fraction of like-charged neighbour pairs in the sample of H + c and H − c decays. The advantage of the kNN method, in comparison with other proposed methods for unbinned analyses [36], is that the calculation of T is simple and fast and the expected distribution of T is well known. Under the hypothesis of no CPV, T follows a normal distribution with a mean, µ T , and a variance, σ T , where lim n,n k ,D→∞ with n = n + + n − and D = 2 is the dimensionality of the tested distribution. The convergence of the limit is so fast that it can be used to obtain a good approximation of σ T even for D = 2 for certain values of n + , n − and n k [36]. For n + = n − the mean µ T can be expressed as and is called the reference value, µ T R . For large n, µ T R asymptotically tends to 0.5.

Control mode, background and sensitivity studies
The S CP and kNN methods are tested using the Λ + c → pK − π + control mode where the CP asymmetry is expected to be null. The sidebands of Ξ + c → pK − π + candidates in the mass regions 2320 < M (pK − π + ) < 2445 MeV/c 2 and 2490 < M (pK − π + ) < 2650 MeV/c 2 are used to check that the background does not introduce spurious asymmetries. The sensitivity of the methods is estimated using pseudoexperiments. Both the S CP and kNN methods are checked to fulfill the following requirements: the method should not indicate the presence of a spurious asymmetry and confirm such a signal if present.
The measured total raw asymmetry is defined as  Figure 4: Measured values of A Raw in regions of Λ + c → pK − π + candidate decays for 2011 (stars) and 2012 (dots) data samples. R0 corresponds to full Dalitz plot and R2 is separated into R5 and R6, and these regions are correlated and separated by dashed lines. and it depends on the production asymmetry of H + c baryons and on the detection asymmetries that arise through charge-dependent selection efficiencies due to track reconstruction, trigger selection and particle identification. The measured value of A Raw in each region of the Dalitz plot of Λ + c → pK − π + decays is presented in Fig. 4. The measured A Raw value integrated over the Dalitz plot equals to −0.0230 ± 0.0016 and −0.0188 ± 0.0008 in the 2011 and 2012 data samples, where the uncertainties are statistical only. Within uncertainties, A Raw in all regions amounts to about −2%. There is no significant difference between the 2011 and 2012 data samples. Since the production and detection asymmetries of Λ + c baryons can depend on the baryon pseudorapidity, η, and p T , the dependence of A Raw in regions of the Dalitz plot is checked in bins of η and p T of the Λ + c baryon. It is observed that the value of A Raw globally changes from bin to bin of η and p T of the Λ c candidates, but for a given bin of η and p T a constant behaviour of A Raw in regions of the Dalitz plot is obtained.
In the S CP method the production asymmetry and all global effects are considered by introducing the α factor, following the strategy described in Sec. 4.1. The p-values obtained are larger than 58%, consistent with the absence of localised asymmetries. As an example, Fig. 5 shows the distribution of S i CP for Λ + c → pK − π + decays considering uniform binning, and for two granularities of the Daliz plot: 28 and 106 bins in the 2012 sample. Alternatively the Dalitz plot is divided into different size bins with the same population size in each bin. Typically, the p-values obtained are larger than 34%, consistent with the hypothesis of absence of localised asymmetries.
Following the strategy described in Sec. 4.2, the results of the kNN method in regions of the Dalitz plot for the Λ + c → pK − π + control mode are presented in Fig. 6, for n k = 50. The pulls, (µ T −µ T R )/∆(µ T −µ T R ), where ∆(µ T −µ T R ) is the uncertainty on the difference (µ T − µ T R ), are different from zero in all regions. The largest effect is observed when integrated over the full Dalitz plot. This asymmetry is an effect of a nonzero production asymmetry that is presented in Fig. 4 and discussed above. Pulls of the test statistic T , ((T − µ T )/σ T ), vary within −3 and +3, consistent with the hypothesis of absence of localised asymmetries in any region. The difference among data-taking years are consistent with statistical fluctuations. The signal yield in 2012 is twice than that in 2011.  illustrates how the larger 2012 data sample improves the power of the kNN method. In Run 2 (years of data taking 2016, 2017 and 2018) the yield is expected to be about three times larger than that from Run 1.
The interaction cross-section of charged hadrons with matter depends on the charged hadron momentum. As such, the detection asymmetries of the proton and kaon-pion systems are momentum dependent. Pseudoexperiments are performed to check whether the detection asymmetries related to particles reconstructed in the final state are or not generating a spurious CP asymmetry. The proton detection asymmetry varies from about 5% at low momentum to 1% at 100 GeV/c and is estimated using simulations. The kaon-pion detection asymmetry and its dependence on the kaon momentum is measured to vary from −1.4% at low momentum to −0.7% at 60 GeV/c [53]. The combined effect of the two asymmetries is found to cancel approximately and does not generate a spurious asymmetry.
These studies are repeated using the candidates in the sideband of the Ξ + c → pK − π + mass distribution. No spurious CP asymmetry is found for both methods. For further cross-checks, the control samples are divided according to the polarity of the magnetic field. The p-values are found to be distributed uniformly.
The expected statistical power of both methods is obtained by performing pseudoexperiments. A total one hundred samples of Ξ + c → pK − π + decays are generated each with a yield and purity equivalent to that observed in the combined 2011 and 2012 data samples, resulting in 200 000 Ξ + c decays generated in each pseudoexperiment. In this model, the two-dimensional Dalitz plots are generated assuming that the Ξ + c baryons are produced unpolarised. The model for Ξ + c → pK − π + decays is built by including the resonances observed in the data, using the same software as in Ref. [54]. The same resonances as described in Sec. 4.2 are included. The statistical powers of the two methods are found to be comparable. Both methods are sensitive to a 5% CP asymmetry in the K * (892) and ∆(1232) resonance regions, and signals with 3 and 5 sigma significances would be observed in 69% and 10% of the cases for the kNN method and 17% and 10% of the cases for the S CP method, respectively.

Binned S CP method
The binned S CP method is applied to look for local CP asymmetries in Ξ + c → pK − π + decays following the strategy described in Sec. 4.1. The measured p-values as well as the S i CP distributions are shown in Fig. 7 for the combined 2011 and 2012 data samples. Two binning schemes are tested: 29 and 111 uniform bins. The normalization factor α, defined in Eq. 1, is determined to be 1.029 ± 0.004. The measured p-values using a χ 2 test are larger than 32%, consistent with no evidence for CP V . The obtained S CP distributions agree with a normal distribution. It is also checked that the results in the 2011 and 2012 data samples are consistent with each other.

Unbinned kNN method
The unbinned kNN method is applied to look for CP asymmetry in Ξ + c → pK − π + decays, following the strategy described in Sec. 4.2. The results are presented in Fig. 8 for n k = 50 for the merged 2011 and 2012 data samples. The measured pull values, ((µ T −µ T R )/∆(µ T − µ T R )), are different from zero. The largest effect is observed integrated over the full Dalitz plot. This is due to the expected nonzero production and detector asymmetries, that is presented in Fig. 9. The measured A Raw is constant within uncertainties in all regions.
The pulls of the test statistic T , ((T − µ T )/σ T ), shown in Fig. 8 vary within −3 and +3, consistent with the hypothesis of absence of localised asymmetries. To check for any systematic effects the kNN test is repeated for the individual 2011 and 2012 data samples as well as for samples separated according to the polarity of the magnetic field. All obtained results are compatible within uncertainties and no systematic effects are observed.
Since the sensitivity of the method can depend on the n k parameter, the analysis is repeated with different values of n k from 10 up to 3000. Only T and σ T depend on n k . Pulls of statistic T are shown in Fig. 10. All results show no significant deviation from the hypothesis of CP symmetry.

Conclusions
Model-independent searches for CP violation in Ξ + c → pK − π + decays are presented using the binned S CP and the unbinned kNN methods. The Λ + c → pK − π + candidates and the sideband regions of Ξ + c → pK − π + candidates are used to ensure that no spurious charge asymmetries affect the methods. Both methods are sensitive to CP asymmetry larger than a 5% in the regions around the K * (892) and the ∆(1232). The obtained results are consistent with the absence of CP violation in Ξ + c → pK − π + decays.

Acknowledgements
We express our gratitude to our colleagues in the CERN accelerator departments for the excellent performance of the LHC. We thank the technical and administrative staff at the LHCb institutes. We acknowledge support from CERN and from the national agencies:  Figure 9: The measured A Raw in regions in signal Ξ + c → pK − π + candidate decays for the combined data collected in 2011 and 2012. R0 corresponds to full Dalitz plot and R2 is separated into R8 and R9, R10 is separated into R4 and R5, R11 is separated into R4, R5, R6 and R7, and these regions are correlated and separated by dashed lines.