1. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 1/17 Elliptic Anisotropy v 2 May Be Dominated by Particle Escape instead of Hydrodynamic Flow Zi-Wei Lin Department of Physics East Carolina University Mostly based on arXiv:1502.05572v3: Liang He, Terrence Edmonds, Zi-Wei Lin, Feng Liu, Denes Molnar, Fuqiang Wang: Anisotropic parton escape is the dominant source of azimuthal anisotropy in transport models

2. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 2/17 Outline Current paradigm of v n development in heavy ion collisions Our method: track all parton collisions Estimate contribution of anisotropic particle escape to v 2 by destroying collective flow (through random- ϕ test) Potential Consequences

3. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 3/17 Alver and Roland, PRC 81 (2010) discovered significant triangular flow using A Multi-Phase Transport (AMPT);  intense developments of event-by-event hydrodynamics. Current Paradigm Early hydro-type collective flow in sQGP converts initial spatial anisotropy into final momentum-space v n Transport models at large-enough cross section will approach hydrodynamics. Since both hydrodynamics and transport models can describe v n data, it is generally believed: for low-P T in high-energy heavy ion collisions, the mechanism of v n development in transport models (via particle interactions) is in principle the same as in hydrodynamics (via pressure gradient). Both hydrodynamics and transport models have been used to study v n : Current Paradigm of v n Development

4. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 4/17 Hydro Pathlength-dependent/ anisotropic energy loss It is generally believed: high-P T observables cannot be described by hydrodynamics, high-P T v 2 is dominated by anisotropic energy loss before hadronization. A different paradigm for high-P T Current Paradigm of v n Development

5. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 5/17 Puzzle for small systems such as p+Pb or d+Au: mean free path may be comparable to the system size; is hydrodynamics still applicable to such small systems? Bzdak and Ma, PRL 113 (2014) using AMPT (String Melting version). Bozek and Broniowski, PLB 718 (2013) using e-by-e viscous hydrodynamics. Small systems: both hydrodynamics and transport can describe flow Current Paradigm of v n Development

6. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 6/17 Our method: Study v 2 development by tracking the complete collision history of each parton, including separating partons into 3 populations: freezeout partons, active partons, all partons v 2 versus Ncoll (number of collisions suffered by a parton) Tracking Parton Collisions in Transport Models Most results shown here are obtained with AMPT (string melting version); some are obtained with MPC (elastic version of the parton cascade). We only investigate the parton stage here. v3 results are qualitative the same.

7. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 7/17 dN/dy of π & K Constraining Parameters of the String Melting AMPT Model p T spectra of π & K (in central collisions) v2 of π & K (AuAu@200AGeV b=7.3fm) Model has been tuned to reproduce low-pt π & K data on dN/dy, p T spectra & v2 in central & mid-central 200AGeV Au+Au collisions: ZWL, PRC 90 (2014). We use the same parameters for this study. Parton cross section σ =3mb

8. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 8/17 Results: v 2 versus collision # of each parton Ncoll: number of collisions suffered by a parton 3 parton populations at any given Ncoll: freezeout partons: freeze out after exactly Ncoll collisions; active partons: will collide further, freeze out after >Ncoll collisions; all partons:sum of the above two populations (i.e. all partons that have survived Ncoll collisions). freezeout partons all partons active partons

9. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 9/17 freezeout partons all partons active partons At Ncoll=1: active partons at Ncoll=0 collide once each, become all partons at Ncoll=1: v 2 ≈ 0 Results: v 2 versus collision # of each parton At Ncoll=0: all partons: v 2 =0 by symmetry; they contain 2 parts: escaped/freezeout: v 2 ≈ 4.5%, active: v 2 < 0. This process repeats itself at higher Ncoll…

10. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 10/17 In simplified picture of elliptic flow At Ncoll=0: escaped partons: v 2 ≈ 4.5%, purely due to anisotropic escape probability (response to geometrical shape only, no contribution from collective flow) At Ncoll>=1: escaped partons: v 2 > 0 due to anisotropic escape probability & (anisotropic) collective flow. Results: anisotropic particle escape How to separate the two contributions?

11. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 11/17 Final v 2 is generated by interactions in presence of spatial anisotropy, which also generate (anisotropic) collective flow. Results: anisotropic particle escape The two are coupled in the actual evolution, so we design a Gedanken random- ϕ test to estimate contribution from 1): we randomize ϕ after each parton scattering to destroy collective flow. Let’s view v 2 as coming from 2 separate but complimentary sources: 1)Pure escape (“the escape mechanism”): v 2 from spatial anisotropy only if there were 0 collective flow, this is due to anisotropic escape probability. 2) Pure flow: v 2 from anisotropic collective flow only if there were no spatial anisotropy. f(x,p,t) f(x,iso p,t) f(iso x,p,t)

12. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 12/17 Strong r-p correlation Results: space-momentum correlation vs collective flow reflects space-momentum correlation, ~ transverse flow velocity ~ Collective flow

13. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 13/17 Strong r-p correlation r-p correlation purely from escape mechanism, not from collective flow Collective flow is destroyed Results: space-momentum correlation vs collective flow reflects space-momentum correlation, ~ transverse flow velocity Random- ϕ test: finite r-p correlation ~ Collective flow

14. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 14/17 Random ϕ Normal AMPT Random ϕ Results: contribution of escape mechanism to final v 2 v 2 from Random Test: purely from the escape mechanism Normal for random- ϕ Ratio ~ contribution from pure escape Au+Au 3.9% 2.7%69% 4.6 (modest) d+Au 2.7% 2.5%93% 1.2 (low)

15. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 15/17 Anisotropic particle escape is dominant contribution of v 2 for small systems & even for semi-central AuAu at RHIC At very large σ or, hydrodynamic collective flow will be the dominant contribution of v 2 MPC: the same qualitative conclusion as AMPT despite many differences (parton initial condition, cross section & dσ/dt, formation time, parton-subdivision) Results: contribution of escape mechanism to final v 2 v 2 Ratio

16. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 16/17 One non-central Au+Au event from AMPT Particles freeze out with ϕ -dependent escape probability, this dominates the final v 2 for systems at low to modest opacity/. Beam axes Potential Consequences New picture of v n development: Particle # vs time

17. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 17/17 Potential Consequences The escape mechanism helps to explain similar anisotropic flows observed in small and large systems: since both are dominated by same mechanism (anisotropic escape probability) The driving force for v 2 at low & high P T is qualitatively the same since both are dominated by anisotropic probability of interactions before escape (scatterings for low P T & energy loss for high P T ) At low to modest opacity/ : transport and hydrodynamics are different. The escape mechanism dominates v n at low to modest opacity/ ; hydro-type collective flow dominates v n at very high opacity/ : which is the case for AA collisions? which is case for pp or pA collisions?

18. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 18/17 Extra Slides

19. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 19/17 Zhang, Gyulassy and Ko, PLB (1999) using elastic parton transport. Heinz, BES Workshop at LBNL 2014 using viscous hydrodynamics. Current Paradigm of v n Development Hydrodynamics has been very successful for global observables, especially flow v n Transport model can describe flow v n : degree of equilibration is controlled by cross section σ Early hydro-type collective flow in sQGP converts initial spatial anisotropy into final momentum-space v n

20. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 20/17 First public release of AMPT source codes: ~ April 2004. Detailed physics descriptions in the long paper: ZWL, Ko, Li, Zhang & Pal, PRC 72, 064901 (2005). "Official" versions v1.11/v2.11 (2004) and v1.21/v2.21 (2008) are available at https://karman.physics.purdue.edu/OSCAR & http://myweb.ecu.edu/linz/ampt/ More versions, including later test (t) versions, are available at http://myweb.ecu.edu/linz/ampt/ A Multi-Phase Transport (AMPT) model

21. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 21/17 A+B Final particle spectra Hadronization (Quark Coalescence) ZPC (parton cascade) Strings melt to q & qbar via intermediate hadrons HIJING1.0: minijet partons (hard), excited strings (soft), spectator nucleons Extended ART (hadron cascade) Each parton freezes out after its last collision Generate parton space-time Structure of AMPT (String Melting version) Hadrons freeze out (at a global cut-off time); then strong-decay all remaining resonances

22. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 22/17 All other particles with PYTHIA flavor codes have no secondary interactions, but they can be produced (from HIJING, string fragmentation, or quark coalescence), e.g. Each hadron has an explicit isospin/charge. K S 0 and K L 0 are also produced at the end of hadron cascade. HIJING1.0Two component model (soft strings + hard minijets) ZPCparton cascade (elastic collisions only) HadronizationLund string fragmentation (Default AMPT) or Quark coalescence (String Melting AMPT) Extended ARTHadron cascade, including secondary interactions for Individual components of AMPT

23. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 23/17 Constraining parameters of the AMPT-SM model HIJING1.0AMPT-SM [1]AMPT-SM in [2]AMPT-SM in this Study Lund string a 0.52.20.50.55 for RHIC, 0.30 for LHC Lund string b (GeV -2 ) 0.90.50.90.15 α s in parton cascade N/A0.470.33 Parton cross sectionN/A~ 6 mb1.5 mb3 mb Model describes pp, …v2 & HBT not dN/dy or p T dN/dy & v2 (LHC) not p T dN/dy, p T & v2 (RHIC & LHC) Goal: fit low-pt (<2GeV/c) π & K data on dN/dy, p T –spectra & v2 in central (0-5%) and mid-central (20-30%) 200AGeV Au+Au collisions (RHIC) and 2760AGeV Pb+Pb collisions (LHC). [1] ZWL, Ko, Li, Zhang and Pal, PRC72; etc. [2] Xu and Ko, PRC83.

24. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 24/17 Results: freezeout vs collision # of a parton =4.6 for Au+Au (modest opacity). =1.2 for d+Au (low opacity). Freezeout in the outer region (~surface emission, but not from a sharp surface); freezeout region moves in with Ncoll. N coll =0, freezeout N coll =0, active N coll =5, freezeout N coll =5, active

25. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 25/17 Results: this is a general feature of transport models MPC gives essentially the same results as AMPT at similar despite differences in parton initial condition (number, density profile, P T spectrum), dσ/dt, formation time, parton-subdivision.

26. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 26/17 Results: this is a general feature of transport models MPC gives essentially the same results as AMPT at similar Time-dependence of 3 parton populations: frozen partons, active partons, & all partons v2 of all partons gradually builds up within ~4 fm/c v2 of active partons is small or negative during most of v2 build-up

27. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 27/17 Results: Anisotropic Particle Escape versus Hydrodynamic Flow When will hydrodynamic flow dominate? σ =3mb σ =20mb σ =40mb At very high or σ : v2 of active partons follows the total v2 closely during the early v2 build-up

28. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 28/17 Results: Anisotropic Particle Escape versus Hydrodynamic Flow AMPT results on integrated v2 (forward-angled/Debye d σ /dt) σ Normal Normal ϕ randomized Ratio ~contribution from escape d+Au3mb1.22.7%2.5%93% Au+Au3mb4.63.9%2.7%69% Au+Au20mb135.9%2.7%46% Au+Au40mb176.0%2.3%38% Au+Au60mb205.7%2.0%34% MPC results on integrated v2 (isotropic d σ /dt) Au+Au5.5mb4.75.5%3.5%64% Au+Au20mb178.6%3.5%41% Au+Au40mb359.8%3.0%31% Au+Au60mb5310.%2.6%26%

29. Elliptic Anisotropy v2 May Be Dominated by Particle Escape Zi-Wei Lin (with L He, T Edmonds, F Liu, D Molnar, FQ Wang) Quark Matter 2015 29/17 Anisotropic Particle Escape versus Hydrodynamic Flow Hydrodynamics describe vn data well. AMPT/transport describes vn data well. Are they essentially the same? How can we differentiate them with experimental observables? Can they be improved to converge toward each other?