1. Forward-backward multiplicity correlations in PbPb, pPb and pp collisions from ATLAS Jiangyong Jia for the ATLAS collaboration 9/27-10/3, 2015 http://cds.cern.ch/record/2029370http://cds.cern.ch/record/2029370, ATLAS-CONF-2015-020 http://cds.cern.ch/record/2055672, ATLAS-CONF-2015-051http://cds.cern.ch/record/2055672

2. Origin of pseudorapidity correlations dN/dη shape reflects asymmetry in num. of forward/backward sources Sources can be wounded nucleons, or partons in nucleons (MPI) 2 n f ≠n b Goal: understand event-by-event fluctuation of dN/dη 1.Probes early time dynamics, longitudinal flow, final state effects. 2.Inputs for 3+1D hydrodynamic models W/O boost invariance matter p A Asymmetry seen directly in p+A collisions p+A A+A p+p Naturally exists in A+A collisions on EbyE bases!

3. Observables 3 Traditional observables: NbNb NfNf Alternative observables: 2D correlation function Correlation function disentangles dynamical fluctuation from statistical fluctuation Single particle distribution Lots of people worked on this E. L. Berger, NPB 85 (1975) 61

4. Observables 4 Traditional observables: NbNb NfNf Alternative observables: 2D correlation function Single particle distribution Mixed events may not match perfectly Remove residual centrality dependence See 1506.03496 and talk by Peng Huo Correlation function disentangles dynamical fluctuation from statistical fluctuation E. L. Berger, NPB 85 (1975) 61 Lots of people worked on this

5. Quantifying the shape fluctuations 5 a 1  fluctuation in the FB asymmetry of N(η), a 2  fluctuation in the width of N(η) EbyE fluct. expanded in Legendre polynomials Legendre expansion of two-particle correlations See Bzdak, Teaney 1210.1965 η/Y T1T1 T2T2

6. Pb+Pb Features of C N (η 1,η 2 ) Corr. Func.=short-range corr. (SRC) + long-range corr. (LRC) 6 CONF-2015-020 Correlation function Legendre spectra η - =η 1 -η 2 ~0 Short-range correlation Long range correlation η + =η 1 +η 2 ~0 & large |η - | Previously measured in Pb+Pb collisions Goal of this measurement: 1.Data-driven method to separate SRC from LRC 2.Compare SRC or LRC in three collision systems, pp,pPb,PbPb at similar N ch

7. Pb+Pb, p+Pb and pp datasets 7 How signals in three systems compare at same N ch ? Pb+Pb 2.76 TeV, 2010, MB p+Pb 5.02 TeV, 2013, MB+HMT p+p 13 TeV, 2015, MB+HMT Inner detector |η|<2.5 High-multiplicity track (HMT) trigger used to increase statistics Analysis carried out in many bins over Results presented as a function efficiency-corrected values b~1.19 for pp and ~1.29 for pPb and PbPb C N (η 1,η 2 ) calculated using charged particles p T >0.2 GeV.

8. Properties of short-range correlations CF for +- and ++,-- pairs separately. different SRC, similar LRC (based on ratio) 8 opp-charge same-charge Pb+Pb

9. Properties of short-range correlations CF for +- and ++,-- pairs separately. different SRC, similar LRC (based on ratio) 9 opp-charge same-charge LRC drops out of the ratio Pb+Pb

10. Properties of short-range correlations CF for +- and ++,-- pairs separately. different SRC, similar LRC (based on ratio) 10 opp-charge same-charge LRC drops out of the ratio Pb+Pb Amplitude quantified via η - peak width is constant in η + Shape of factorize in η + and η - Width Amplitude Shape along η +

11. Properties of short-range correlations CF for +- and ++,-- pairs separately. different SRC, similar LRC (based on ratio) 11 opp-charge same-charge LRC drops out of the ratio Pb+Pb Hence, δ SRC (η +,η - ) assumed to factorize in η +,η -. Key is to determine g(η - )! Width Amplitude

12. Estimate short-range correlations 12 Take C N (η - ) in |η + | η 0 =1.5, difference between data and fit in |η - |<2 is the g(η - ) for SRC Vary η 0 and the range of η + slices to check systematics. Obtain g(η - ) η + ≈0

13. Estimate short-range correlations Take C N (η - ) in |η + | η 0 =1.5, difference between data and fit in |η - |<2 is the g(η - ) for SRC Vary η 0 and the range of η + slices to check systematics. 13 The LRC, C N sub ( η +,η - ), has similar magnitudes between ++,-- and +- Obtain g(η - )

14. Results 14

15. Correlation functions in three systems 15 Short-range Long-range After SRC subtraction, similar LRC in all three systems Most FB asymmetry in pPb collisions is due to the SRC.

16. Legendre spectra in Pb+Pb The strong charge dependence  SRC contributes to all coefficients 16 Before subtraction Pb+Pb +- ++,-- Before SRC subtraction

17. Legendre spectra in Pb+Pb The strong charge dependence  SRC contributes to all coefficients 17 After SRC removal, results are independent of charge combinations! Pb+Pb +- ++,-- Before subtraction After subtraction Dominated by a 1 After SRC subtraction

18. Legendre spectra in three systems The strong charge dependence  SRC contributes to all coefficients 18 Before subtraction After subtraction Dominated by a 1 After SRC removal, results are independent of charge combinations! Pb+Pbp+Pb p+p

19. Projections of correlation function 19 Fit in limited η 1 and η 2 space, the systematics largely independent! Compare to a 1 obtained from Legendre expansion method η+η+ η-η- η -η η ref CF dominated by a 1, thus: similar to the CMS correlator for flow decorrelation 1503.01692

20. Projections of CF in p+Pb 20 Fit η - dependence Fit η + dependenceFit η dependence

21. Projections of CF in p+Pb 21 Fit η - dependence Fit η + dependenceFit η dependence Global a 1 LRC is ~symmetric between proton- and lead-going side!

22. N ch dependence of LRC (in terms of a 1 ) Calculated from Legendre expansion method for all-charge, opposite-charge and all-charge pairs 22 No dependence on charge combinations Pb+Pbp+Pb p+p

23. N ch dependence of SRC Quantify SRC using its average amplitude 23 Y=2.4

24. N ch dependence of SRC Quantify SRC using its average amplitude 24 Strong dependence on charge and system size Y=2.4 Pb+Pbp+Pb p+p

25. N ch dep. of SRC and a 1 in three systems They follow power-law function 25 Fit with SRC a 1 for LRC

26. Expected scaling behavior? SRC can be related to the number of sources n contributing to N ch LRC expected to be related to the asymmetry between n f and n b Assume independent cluster picture NPB85 (1975)61 : each source emits the same number of pairs and the number of sources follows Poisson fluctuations, then 26 Sources could be: wounded nucleons, or partons (via frag.), flux tubes, or final state resonance decays

27. N ch dep. of SRC and a 1 in three systems They follow power-law function 27 Fit with α smaller in pp  num. of cluster is smaller or cluster size is larger at same N ch ? SRC a 1 for LRC

28. Summary Two-particle correlation C N (η 1,η 2 ) measured in PbPb, pPb and pp collisions for p T >0.2 GeV and |η|<2.4 at similar event multiplicity N ch. Sum of a short-range component (SRC) + a long-range component (LRC). Data-driven method to separate SRC and LRC based on the fact that SRC differs between +- and ±± pairs, but not for LRC LRC is symmetric in all systems, but in pPb SRC is asymmetric for η and -η. LRC consistent with 1+  a 2 1  η 1 η 2,  a 2 1  is extracted via Legendre expansion as well as from the projections of the LRC in limited η 1,η 2 phase space N ch -scaling of LRC (via a 1 ) and SRC are studied LRC (via a 1 ) controlled by N ch, not by collision systems or charge combination SRC depends strongly on collision system and charge combination N ch dependence of LRC and SRC follows power-law with an index close to 0.5  relation to the number of sources for particle production? 28 Important inputs for 3+1D hydrodynamic models W/O boost invariance Insights on particle production, longitudinal transport, final-state effects See works by Bzdak, Bozek, Broniowski1509.02967,1509.04124;, Akihiko, Schenke 1509.04103

29. END 29

30. ALICE dN/dη distribution 30 Pb+Pb 2.76 TeV

31. Projections of CF in Pb+Pb Consistent a 1 at different slices  LRC dominated by a 1. 31 Consistent between four methods Fit η - dependence Fit η + dependenceFit η dependence

32. Projections of CF in p+p 32 Consistent a 1 at different slices  LRC dominated by a 1. Consistent between four methods Fit η - dependence Fit η + dependenceFit η dependence

33. a 1 from four methods (i) Quadratic fit to η- direction, (ii) Quadratic fit to η+ direction, (iii) Linear fit to r correlator, (iv) Legendre expansion 33 4 methods consistent with each other in all three collision systems Pb+Pbp+Pb p+p

34. Properties of SRC in p+Pb 34 Most of the FB asymmetry is in the magnitude of the SRC. η - -width of SRC is constant as a function of η +.

35. Estimation of the SRC in pPb Asymmetry between proton-going and lead-going direc is due to SRC LRC is very symmetric 35 The LRC, C N sub ( η +,η - ), has similar magnitude between charge combinations.

36. Legendre spectra before and after subtraction a n values significantly reduced after SRC subtraction Dominated by a 1. 36 Pb+Pbp+Pb p+p

37. Constructing correlation function Signal and Background distributions Tracks from events with similar N ch and z vtx Due to symmetry in PbPb or pp, only one quadrant is independent. η - =η 1 -η 2 >0, η + =η 1 +η 2 >0 In pPb collision, only one half is independent η - =η 1 -η 2 >0 But distribution can be symmetrized in any case to compare PbPb and pp 37

38. Summary of systematics 38 Evaluate based on ratios of correlation function Separately for & Uncertainty propagate to the a 1 calculated from four methods Correlation function a 1 from the four methods