Global analysis of parton densities and fragmentation functions

Transcription

1 Global analysis of parton densities and fragmentation functions Carlota Andrés Jefferson Lab 18 JLab Users Group Meeting Newport News, Virginia, June 18-, 18

2 Why JAM? JAM: Jefferson Lab Angular Momentum Collaboration To study the quark and gluon structure of the nucleon by performing global fits of both spin-dependent ( distribution functions ( ) How? PDFs PDFs) and unpolarized parton Analyzing the impact of JLab in a rigorous way x b JLab DIS data: large, low-intermediate Q and W Framework: (NLO) collinear factorization Higher twist (HT) and Target Mass Corrections (TMC) needed at large! x b

3 Evolution of JAM Iterative MC fitting technique JAM15 IMC analysis + all available JLab data JAM15 Uses CJ1 NLO unpolarized PDFs u + and d + consistent with previous analysis JAM15 s + slightly harder Sato, Melnitchouk, Kuhn, Ethier,Accardi Phys. Rev. D 93, 745 (16)!3 Impact of JLab data

4 JAM16 First IMC analysis of FFs Only SIA included Q =1GeV Q =1GeV Q =1GeV JAM17 First (simultaneous) MC analysis of polarized PDFs and FFs Polarized SIDIS, polarized DIS and SIA included Q = m c Q = m b Q =1GeV Sign change JAM16 Q =1GeV JAM17 HKNS DSS Sato,Ethier,Melnitchouk, Hirai, Kumano and Accardi Phys. Rev. D 94, 1144 (16) JAM17 JAM17 FFs better agreement with other analysis!4 Ethier, Sato, Melnitchouk: Phys. Rev. Lett. 119, 131 (17)

5 JAM18!5

6 Motivation Knowing the limits in x and Q of collinear factorization Testing the universality of PDFs,FFs All the data must be studied using the same theoretical framework First step: (first) combined analysis of unpolarized PDFs and FFs!6

7 Setup: JAM18 data DIS : l +(p, d)! l + X Q 1 1 SLAC BCDMS NMC HERA Q = m c SIDIS : l +(p, d)! l + h + X Q x bj W > 1 GeV Q > 5 GeV COMPASS x bj Q E866 (pp) 5 E866 (pd) 15 W > 1 GeV x bj DY : p +(p, d)! l l + X SIA : e + + e Q z! h + X x F!7 1 1 Argus Belle BaBar TPC TASSO ALEPH DELPHI SLD OPAL z

8 Setup: theory All observables computed at NLO in pqcd DGLAP truncated evolution at order s in Mellin space DIS cross sections computed at leading twist Nuclear smearing for deuterium DIS Heavy quark treatment : ZM-VFN Fitting methodology: IMC based on Bayesian statistics Future: Nested sampling!8

9 Why IMC? Typical PDF parametrization: Multiple local minima! Perform single fit: Parameters difficult to constrain Hessian method for uncertainties Introduces tolerance criteria Unsuitable for simultaneous analysis of collinear distributions Monte Carlo methods: Allows efficient exploration of the parameter space Uncertainties directly obtained from MC replicas JAM18 currently uses an IMC based on a Bayesian approach!9

10 Data vs. theory: DIS 1. Data/Theory F p F p DIS : l +(p, d)! l + X SLAC SLAC BDMPS BDMPS F d F p 1 1 Q 1 1 Q Q Q NMC NMC NC HERA NC HERA F d /F p red (e+ p) F d red (e+ p) NC red (e+ p) 1 1 Q HERA NC red (e+ p) 1 1 Q HERA Q NC red (e p) HERA Q CC red HERA Q Q Q Q!1

11 Data vs. theory: DY E866 DY : p +(p, d)! l l + X M 3 d /dmdxf 1 5 pp! l l Q [4, 45] x F Q [45, 5] x F Q [5, 6] x F 6 4 Q [6, 68] x F 4 3 Q [68, 75] Q [1, 14] Q [14, 16] Q [16, ] x F x F x F x F pd! l l Q [4, 45] x F 1. Q [45, 5] x F Q [5, 6] x F 6 4 Q [6, 68] x F 4 Q [68, 75]. 1.5 Q [1, 14] 1..8 Q [14, 16] 1..8 Q [16, ] x F x F x F x F!11

12 Data vs. theory: SIA SIA : e + + e! h + X log( ) 1 3 = 1 d T dz Argus z = d dz K ± ± Belle z 4 = 1 d T dz BaBar z 4 = 1 d T dz TPC z = 1 d T dz = 1 d T dz = 1 d T dz 4 = 1 d T dz 1 1 TPC(c) z TPC(b) z 4 TASSO z ALEPH z 4 = 1 d T dz = 1 d T dz 4 = 1 d T dz = 1 d T dz 4 DELPHI z 5 1 DELPHI(b) z 4 SLD z SLD(c) z 4 6 = 1 d T dz SLD(b) z = 1 d T dz OPAL z = OPAL(c) z = OPAL(b) z!1

13 Data vs. theory: SIDIS SIDIS : l + p/d! l + h + X First time SIDIS data are included in unpolarized PDFs global fit! COMPASS Di cult to fit low Q data! only Q > 5 GeV data included!13

14 Unpolarized PDFs (preliminary) xf(x) Q = 1. GeV g/1 u v d v 1 - CL ū d s SIDIS supports the strange suppression ū d asymmetry at large x x u Q = 1 GeV v d v ūū f/f The gluon distribution changes significantly with the inclusion of SIDIS f/f.99 JAM18 (DIS, DY) JAM18 (DIS, DY, SIDIS, SIA) d d s CL g !14 x x x

15 FFs (preliminary) u d s + zd h q (z).6.4 K + Q =1.7 GeV.6.4 Q =1.7 GeV.6.4 Q =1.7 GeV c dotted : JAM18 (SIA) solid : JAM18 (DIS, DY, SIDIS, SIA) 1..8 g zd h q (z).6.4 Q = m c 1 - CL.6.4 b + Q = m b.6.4 Q =1.7 GeV z z SIDIS data has large effect on flavor decomposition of FFs! z

16 Summary MC statistical methods are important for a robust extraction of non-perturbative collinear distributions Crucial for future Global TMDs, GPDs analysis First (preliminary) MC fit of PDFs and FFs using DIS, SIDIS and SIA data Strange PDF constrained by SIDIS data Significant effect of SIDIS data on flavor decomposition of FFs Difficulties in incorporating low Q < 5 GeV SIDIS data!16

17 Outlook Impact pf SIDIS data on s vs. s Introduce the HQ treatment: ACOT (GM-VFNS) Use and HERA data GM-VFNS required! F c F b Likelihood sampling methods: Nested sampling Inclusion of polarized DIS and SIDIS and extract PDFs, FFs, and PDFs Simultaneous extraction of all non-perturbative input Strict test of universality!17

18 Backup

19 Iterative Monte Carlo (IMC) JAM15

20 JAM15 Impact of JLab data Q =1GeV u + d + :reduction of the uncertainties Q =1GeV d + less negative due to Jlab data Q =1GeV Q =1GeV Q =1GeV Q =1GeV Harder s + s + positive Q =1GeV Q =1GeV g big uncertainty g positive Q =1GeV Flavor decomposition of twist-3 distributions Twist-4 functions compatible with

21 JAM15 d -moment agrees with experimental data Q =1GeV

22 Iterative Monte Carlo (IMC) JAM16

23 JAM16: iterative convergence

24 JAM16: FFs evolution The gluon peak disappears

25 JAM16: comparison Q =1GeV Q =1GeV Q =1GeV Q = 1 GeV Q = GeV Q =1GeV

26 JAM16: comparison II

27 JAM17: motivation Spin sum rule Released in JAM17

28 Euler Beta function Isospin Penalty term syst + stat (quad.) Normalization (Correlated uncertainties) Nuisance parameters

29 Polarized SIDIS

30 JAM17: Data vs. Theory

31 JAM17: Polarized PDFs

32 JAM17: Lowest moments

33 JAM18: Parametrization

34 Unpolarized PDFs (JAM18).1.8 JAM18 (DIS, DY) JAM18 (DIS, DY, SIDIS, SIA) CL f/f.6 Q = 1 GeV u v.5.1 ūū d v f/f d d.1..3 x.1. s.1..3 x Decrease of strange uncertainties due to.1. K g.1..3 x SIDIS data

35 Unpolarized PDFs (JAM18) Sea distributions d/u Q = 1 GeV CJ15 MMHT14 CT x d/ū JAM18 (DIS, DY) JAM18 (DIS, DY, SIDIS, SIA) x (s + s)/(ū + d) CL x

36 Next steps Using and data? GM-VFNS required! F c F b HERA data: arxiv: [hep-ex] 5 points 7 points c c red : <x<.5 and.5 GeV <Q < GeV b b red : <x<.5 and.5 GeV <Q < GeV!36

37 JAM18