Quantum chaos in the quantum fluid
Raju Venugopalan Brookhaven Na onal Laboratory
Joe Fest, McGill, Montreal, June 12-14, 2012 Smoking Joe and Holy Indian
Lone ranger: I think we are surrounded, Kemo sabe Tonto: What do you mean “we” $@!# Smoking Joe and Holy Indian
Lone ranger: I think we are surrounded, Kemo sabe Tonto: Indeed, Lone Ranger – no worries – they are all here to praise you on your 60th ! Talk
Mo va on: the unreasonable effec veness of hydrodynamics in heavy ion collisions to paraphrase E. P. Wigner
An ab ini o weak coupling approach:
Classical coherence of wee partons in nuclear wavefunc ons
Quantum fluctua ons: Factoriza on, Evolu on, Decoherence
Isotropiza on, Bose-Einstein Condensa on, Thermaliza on ?
4 Ab ini o approach to heavy ion collisions RV, ICHEP review talk, 1012.4699
Color Glass Ini al Glasma sQGP - Hadron Condensates Singularity perfect fluid Gas t τ0 ? Compute proper es of relevant degrees of freedom of wave fns. in a systema c framework (as opposed to a “model”)?
How is ma er formed ? What are its non-equilibrium proper es & life me? Can one “prove” thermaliza on or is the system “par ally” thermal ?
When is hydrodynamics applicable? How much jet quenching occurs in the Glasma? Are there novel topological effects (sphaleron transi ons?) Ab ini o approach to heavy ion collisions RV, ICHEP review talk, 1012.4699
Color Glass Ini al Glasma sQGP - Hadron Condensates Singularity perfect fluid Gas t τ0 ? Compute proper es of relevant degrees of freedom of wave fns. in a systema c framework (as opposed to a “model”)?
How is ma er formed ? What are its non-equilibrium proper es & life me? Can one “prove” thermaliza on or is the system “par ally” thermal ?
When is hydrodynamics applicable? How much jet quenching occurs in the Glasma? Are there novel topological effects (sphaleron transi ons?) Gluon Satura on in large nuclei: classical coherence from quantum fluctua ons
Wee parton fluctua ons me dilated on strong interac on me scales
The gluon density saturates at
a maximal value of ~ 1/αS gluon satura on
Large occupa on # => classical color fields
|P>_pert |P>_classical 2 1/QS Quantum decoherence from classical coherence
Color Glass Initial sQGP - Hadron Glasma Condensates Singularity perfect fluid Gas t
Glasma (\Glahs-maa\): Noun: non-equilibrium ma er between CGC and QGP
Computa onal framework Gelis,RV NPA (2006)
Schwinger-Keldysh: for strong me dependent sources (ρ ~ 1/g) , ini al value problem for inclusive quan es
For eg., Schwinger mechanism for pair produc on, Hawking radia on, … Forming a Glasma in the li le Bang Lumpy classical configura ons
Solu ons of Yang-Mills equa ons produce (nearly) boost invariant gluon field configura ons: “Glasma flux tubes”
Lumpy gluon fields are color screened in
transverse plane over distances ~ 1/QS - Nega ve Binomial mul plicity distribu on.
“Glasma flux tubes” have non-trivial longitudinal color E & B fields at early mes --generate Chern-Simons topological charge
See Bjoern Schenke’s talk from yesterday Their quantum descendents
Color Glass Initial Glasma sQGP - Hadron Condensates Singularity perfect fluid Gas t
Two kinds of important quantum fluctua ons: a) Before the collision: pη=0 modes factorized into the wavefunc ons - responsible for energy/rapidity evolu on of wavefunc ons a) A er the collision pη ≠ 0; hold the key to early me dynamics - responsible for decoherence, isotropiza on, thermaliza on Quantum fluctua ons in classical backgrounds: I Gelis,Lappi,RV: 0804.2630, 0807.1306,0810.4829
Suppressed Factorized into energy evolu on of wavefunc ons
JIMWLK factoriza on: pη=0 (small x !) modes that are coherent with the nuclei can be factorized for inclusive observables
W’s are universal “func onal density matrices” describing distribu on of large x color sources ρ1 and ρ2 of incoming nuclei; can be extracted from DIS or hadronic collisions Ini al condi ons for quantum evolu on
For large nuclei, general considera ons about the color structure of higher dimensional representa ons of color charge density ρa probed give as an ini al condi on for evolu on (MV model) a a 2 ρ (x )ρ (x ) Wx [ρ]=exp d x ⊥ ⊥ 0 − ⊥ 2 µ2 µ2 = Color charge squared per unit area ~ A1/3
Other (sub-leading in A) contribu ons to these ini al condi ons
Jeon, RV Dumitru,Jalilian-Marian,Petreska These ini al condi ons give eikonalized gluon distribu ons at leading log in x and Q2 Higher order correc ons can be modeled by introducing x and b dependence (Basis of IP-sat model discussed in Bjoern’s talk) Solu on of B-JIMWLK hierarchy
∂WY [ρ] = [ρ] WY [ρ] ∂Y H a a ρ α = 2 ∇⊥ B-JIMWLK hierarchy of correlators contains all leading info on charge, rapidity and transverse posi ons of n-gluon correla ons in incoming wavefunc ons
This is now solved ! Correlator of Light-like Wilson lines Tr(V(0,0)V^dagger (x,y))
Rummukainen,Weigert (2003) Dumitru,Jalilian-Marian,Lappi,Schenke,RV, PLB706 (2011)219 Quantum fluctua ons in classical backgrounds: II
η Romatschke,Venugopalan p ≠ 0 (generated a er collision) modes grow exponen ally Fukushima,Gelis,McLerran n g exp QSτ Quant. fluct. grow exponen ally a er collision
As large as classical increasing seed size field at 1/Qs !
2500 T µν (x) T µν (x)
Parametrically suppressed Exponen ate and resum these The first fermi: a master formula
Also correlators of Tμν From solu ons of B-JIMWLK
µν T LLx+Linst. = [Dρ1][Dρ2] WY Y[ρ1] WY +Y[ρ2] beam− beam [da(u)] F [a] T µν [A (ρ , ρ )+a] × init LO cl 1 2
Gauge invariant Gaussian spectrum 3+1-D solu ons of of quantum fluctua ons Yang-Mills equa ons
Expression computed recently-numerical evalua on in progress
Dusling,Gelis,RV This is what needs to be matched to viscous hydrodynamics, event-by-event
All modeling of ini al condi ons for heavy ion collisions includes various degrees of over simplifica on rela ve to this “master” formula Big Bang Little Bang
Hot Era WMAP data (3x105 years) QGP Inflation
CGC/ Glasma
Plot by T. Hatsuda 17 Glasma spectrum of ini al quantum fluctua ons
Path integral over small fluctua ons equivalent to
1 dν iνη A(x , τ, η)=Acl.(x , τ)+ dµk cνk e χk(x ) Hiν (λkτ)+c.c ⊥ ⊥ 2 2π ⊥ Gaussian random variables
Berry conjecture: High lying quantum eigenstates of classically chao c systems, linear superposi ons of Gaussian random variables
Yang-Mills is a classically chao c theory B. Muller et al.
Srednicki: Systems that sa sfy Berry’s conjecture exhibit “eigenstate thermaliza on” Also, Jarzynski, Rigol, … Hydrodynamics from quantum fluctua ons
Dusling,Epelbaum,Gelis,RV (2011) scalar Φ4 theory:
Energy density and pressure without averaging over fluctua ons
Energy density and pressure a er averaging over fluctua ons
Converges to single valued rela on “EOS” Hydrodynamics from quantum fluctua ons Dusling,Epelbaum,Gelis,RV (2011)
Anatomy of phase decoherence:
ΔΘ = Δω t
Tperiod = 2π / Δω
Tperiod ≅ 18.2 / g ΔΦmax
Different field amplitudes from different ini aliza ons of the classical field T µ = dφ dφρ˙ (φ, φ˙) T µ(φ, φ˙) dE dθ ρ˜ (E,θ) T µ(E,θ) µ t µ ≡ t µ ≈ µ t dE ρ˜t(E) dθ Tµ (E,θ) = 0 →∞ μ Because Tμ for scalar theory is 2π t+T a total deriva ve dθ T µ(E,θ)= dτ T µ(φ(τ), φ˙(τ)) = 0 µ T µ and φ is periodic t Quasi-par cle descrip on? Epelbaum,Gelis (2011)
Spectral func on ρ (ω,k)
Plasmon mass
ω k Energy density of free quasi-par cles
At early mes, no quasi-par cle descrip on Energy density on the la ce
May have quasi-par cle descrip on at late mes. Effec ve kine c “Boltzmann” descrip on in terms of interac ng quasi-par cles at late mes ? Quasi-par cle occupa on number 4 10 4 10 Gelis,Epelbaum (2011) t = 0 3 20 3 10 10 40 60 2 10 80 100 1 120 2 10 10 140 0 10
-1 1 10 k
f 10
-2 10 0 0.5 1 1.5 2 2.5 0 10
-1 Ini al 10 mode distribu on
-2 10 0 0.5 1 1.5 2 2.5 3 3.5 k
4 t = 0 150 2000 10
50 300 5000 T/(tk-!)-1/2
System becomes over occupied rela ve to a thermal distribu on… Proof of concept: isotropiza on of longitudinally expanding fields in scalar Φ4
101 2PT + PL Dusling,Epelbaum,Gelis,RV: on arXiv, Friday ε
PT 100 PL
10-1
10-2
10-3 0 50 100 150 200 250 300 τ
QCD – similar framework – more challenging computa onally and conceptually Dusling, Gelis, RV: arXiv 1106.3927
Numerical development underway – results hopefully very soon Bose-Einstein Condensa on in HI Collisions ? Blaizot,Gelis,Liao,McLerran,RV: arXiv:1107.5295v2
Cold rubidium atoms in a magne c trap Gell-Mann’s Totalitarian Principle of Quantum Mechanics: Everything that is not forbidden is Compulsory
Possible phenomenological consequences… Mickey Chiu et al., 1202.3679
Bose-Einstein Condensa on and Thermaliza on
Blaizot,Gelis,Liao,McLerran,RV: arXiv:1107.5295v2
Assump on: Evolu on of “classical” fields in the Glasma can be matched to a quasi-par cle transport descrip on See also, Mueller, Son (200)2 Jeon (2005)
All es mates are “parametric”: αS << 1
3 4 System is over-occupied: n ≈ QS /αs ; ε = QS /αS -3/4 1/4 n ε ≈ 1/αS >> 1
In a thermal system, n ε-3/4 = 1
If a system is over-occupied near equilibrium and elas c sca ering dominates, it can generate a Bose-Einstein condensate Known in context of infla on: Khlebnikov, Tkachev (1996) Berges et al. (2011) Bose-Einstein Condensa on and Thermaliza on
neq = feq(p); εeq = ωp feq(p) p p 1 feq(p)= In a many-body system, gluons develop a mass β(ωp µ) 1 1/2 e − − ωp=0 = m ≈ αS T
If over-occupa on persists for μ = m, system develops a condensate
3 1 feq(p)=ncδ (p)+ β(ω m) 1 e p− −
3 Qs 1/4 nc = 1 αS As αS 0, most par cles go into the condensate αS − It however carries a small frac on 1/4 4 4 ε = mn α T << T of the energy density… c c ≈ S Transport in the Glasma df p ∂ f z ∂ f = C[f] dt ≡ τ − τ pz “Landau” equa on for small angle 2 2 sca ering: df Λ2 Λ df α S ∂ p2 + S f(p)(1 + f(p)) dt |coll ∼ p2 p dp Λ S
This is sa sfied by a distribu on where
1 1 ΛS f ; p<ΛS ; ΛS
ΛS and Λ are dynamical scales determined by the transport equa on Transport in the Glasma
1/αS At τ ~ 1/QS
p ΛS=Λ=QS
1/α At τ > 1/Q S S f(p)
p ΛS Λ
When ΛS = αS Λ, the system thermalizes; 1/2 one gets the ordering of scales: Λ =T, m = Λ ΛS = α T, Λ S = αST Thermaliza on: from Glasma to Plasma
3 Fixed box: Energy conserva on gives Λ ΛS = constant 2 From moments of transport eqn., , τcoll = Λ / ΛS ~ t t 3/7 t 1/7 From these two condi ons, Λ Q 0 Λ Q S ∼ s t ∼ s t 0 1 1 7/4 Thermaliza on me: t therm. ∼ Q α S S
Also, Kurkela, Moore (2011)
3 3 Entropy density s = Λ increases and saturates at ttherm as T
3 2 Nquark ~ Λ = Ngluon (Λ ΛS/αS) at ttherm when ΛS = αS Λ Thermaliza on: from Glasma to Plasma
Expanding box : ma er is strongly self interac ng for fixed anisotropy 1+δ t0 ε (t) ε(t ) 0 < δ ≤ 1/3 g ∼ 0 t (4+δ)/7 (1+2δ)/7 t0 t0 ΛS QS Λ Q ∼ t ∼ S t 7/(3 δ) 1 τ0 − Thermaliza on me ttherm = QS τ
For δ= -1, recover fixed box results…
A condensate can s ll form in the expanding case for δ > 1/5 What about plasma instabili es ? Kurkela, Moore(2011) Schlichtling,Mclerran, RV, in prepara on
Relevant case for us: in general need careful numerical study to gauge impact on scaling solu ons Summary
Presented ab ini o picture of mul -par cle produc on and thermaliza on in heavy ion collisions
Thermaliza on is a subtle business even in weak coupling
Hydrodynamics may be unreasonably effec ve because it requires rapid decoherence of classical fields and strong self-interac ons, not thermaliza on
Exci ng possibility of a transient Bose-Einstein Condensate Happy 60th, Joe !!
• Look forward to the 70th, 80th, … THE END P An Analogy with the Early Universe Mishra et al; Mocsy- Sorensen The Universe HIC distributions and kinetic correlations of freeze-out produced particles hadronization lumpy initial energy density
QGP phase quark and gluon degrees of freedom
Credit: NASA
Δρ/√ρref
Δφ WMAP HIC-ALICE Role of inelas c processes ? Wong (2004) Mueller,Shoshi,Wong (2006)
Power coun ng for n m processes contribu ons to the collision integral n+m-2 Ver ces contribute αS
n+m-2 Factor of (ΛS/αS) from distribu on func ons
n+m-4 Screened infrared singularity: (1/Λ ΛS) Remaining phase space integrals Λn+m-5
2 Net result is τinelas ~ Λ / ΛS = τelas At most parametrically of the same order as elas c sca ering. So a transient Bose-Einstein condensate can form.
Dusling,Epelbaum,Gelis,RV, in progress Numerical simula ons will be decisive Blaizot, Liao, McLerran CGC based models and bulk distribu ons Kowalski, Motyka, Wa e+p constrained fits give good descrip on of hadron data Tribedy, RV: 1112.2445
p+p d+Au
Also:
Collimated long range Rapidity correla ons “the ridge” A+A Di-hadron d+A correla ons Status of CGC art Rummukainen,Weigert (2003) Dumitru,Jalilian-Marian,Lappi,Schenke,RV: arXiv:1108.1764 Numerical solu on of B-JIMWLK hierarchy :
Mul -parton correla ons in nuclear wavefunc ons: Rapidity, number and impact parameter fluctua ons of glue
Recent progress in NLO computa ons: F2, FL, single inclusive hadron produc on,… Balitsky,Chirilli; Kovchegov; Chirilli,Yuan,Xiao; Beuf Phenomenological “Global” analysis of e+p, e+A, p+p, d+A
A+A data in CGC models works well Levin, Rezaiean; Tribedy, RV 1112.2445