Dimensional reduction near the deconfinement transition

Transcription

1 Dimensional reduction near the deconfinement transition Aleksi Kurkela ETH Zürich Wien

2 Outline Introduction Dimensional reduction Center symmetry

3 The deconfinement transition: QCD has two remarkable properties: At small momenta / Large distances: Confinement At T = 0, µ B = 0: Hadronic matter At large momenta / Small distances: Asymptotic freedom At T and/or µ : Quarks and gluons

4 The deconfinement transition: QCD has two remarkable properties: At small momenta / Large distances: Confinement At T = 0, µb = 0: Hadronic matter At large momenta / Small distances: Asymptotic freedom At T and/or µ : Quarks and gluons

5 Preliminaries: Quantum fields in thermal equilibrium Theoretical challenge: Fireball described by relativistic hydrodynamics T µν = [ p(t ) + e(t )]u µ u ν p(t )g µν, µ T µν = 0 Hydrodynamic equations take the equation of state as input Problem in equilibrium thermodynamics Quantities of interest: Equation of state Screening lengths Quasi-particle spectral functions Transport coefficients

6 Preliminaries: Quantum fields in thermal equilibrium Thermodynamical information is encoded in the partition function: Can be written as a path integral (here at zero baryon number density) Z = Tr exp [ Ĥ/T ] = Dφ exp( S E ) S E = 1/T d 3 x dτl E 0 φ(0,x)=φ(1/t,x) Bosons periodic b.c., fermions anti-periodic b.c. Quantites of interest: p(t ) = (T ln Z) = T/V ln Z V s(t ) = (T ln Z) T E(T ) = P V + T S

7 The lattice approach: Tr 0 SB /T /T 4 3p/T 4 p4 asqtad p4 asqtad 2 T [MeV] Bazavov et al

8 The perturbative approach: High T : Renormalized coupling g(t ) 1/ ln T Periodicity in coordinate space makes p 0 discrete 1 p 2 1 p 2 + ωn 2, Bosons: ω n = (2n)πT Static mode: n = 0 Fermions: ω n = (2n + 1)πT 4d integrals become 3+1d sum-integrals d 4 p (2π) 4 T d 3 p (2π) 3 n To get (ln Z): Sum of 1PI vacuum diagrams

9 The perturbative approach: General structure of the perturbative expansion g 2 (T ) 1/ log(t ): Leading order: Gas of non-interacting quarks and gluons ( p SB /T 4 = π2 Nc N ) cn f 45 4 Recipe to compute higher order corrections: Vacuum diagrams p(t ) p SB (T )(1 + #g 2 + #g 3 + #g 4 ln g + #g 4 + #g 5 + #g 6 ln g +...) Non-analytic terms originate from IR divergences of static modes resummations Note: Number of loops power in g 2

10 The perturbative approach: Kajantie et al. hep-ph/ p/p SB _ T/ MS g 2 g 3 g 4 g 5 g 6 (ln(1/g)+0.7) 4d lattice p(t ) p SB (T ) ( 1 + g 2 + g 3 + g 4 ln g + g 4 + g 5 + g 6 ln g + g ) Not straightforward to improve!! g 6 -term non-perturbative. n-loop diagram g 6 (g 2 T/m) (4 l) with m g 2 T

11 Outline Introduction Dimensional reduction Center symmetry

12 Dimensional reduction At high T : For long distance properties ( x 1/T ), the system looks 3d. Degrees of freedom are static modes φ 0 (x) φ(x, τ) = T n= exp(iω nτ)φ n (x) Effective action: Integrate out non-static modes Z = Dφ 0 Dφ n exp( S 0 (φ 0 ) S n (φ 0, φ n )) = Dφ 0 exp( S 0 (φ 0 ) S eff (φ 0 )) 4d theory 3d theory In practice: Need a scale separation between static and non-static modes to give a truncation to the effective action

13 Where does Dimensional Reduction work? Scales in hot Yang-Mills: Perturbatively: Hard scale: 2πT Typical thermal momentum, non-static modes Soft scale: m E gt Debye screening, static electric modes, A 0 Ultra-soft scale: m M g 2 T color magnetic screening, static magnetic modes, A i Asymptotic dimensional reduction Non-perturbatively: m E (T c ) 3T c? 2πTc 4 m D /T N f =0 N f = T/T c Kaczmarek,Zantow No scale separtion below T c

14 Electrostatic QCD, Magnetostatic QCD Braaten & Nieto Integrate out the hard scale to get EQCD ( x 1/T ) (=3d Yang-Mills + adjoint Higgs) S EQCD = 1 [ g3 2 d 3 x p E TrF ij 2 }{{}}{{} g 2 spatial gluons T + Tr(D i A 0 ) 2 }{{} adjoint kinetic g 2 T + 1 {}}{ 2 m 2 2 E TrA λ ETrA 4 0 }{{} interactions from integration out ] +... Higher order terms suppressed by powers of the scale difference Eff. theory parameters g 3 (T ), m E (T ), λ E (T ) via perturbative matching, no resummations needed. After integrating out the soft modes A 0, ( x 1/(gT )): S MQCD = 1 [ 1 ] g3 2 d 3 x 2 TrF ij 2 Philosophy: Integrate out heavy modes analytically, simulate low-energy theory numerically.

15 Example: g 6 -coefficient from MQCD The g 6 -resummation can be done numerically in eff. theory framework: Match QCD EQCD to sufficient depth in g g 2 3(T ) = T (#g 2 + #g 4 ) m 2 E (T ) = T 2 (#g 2 + #g 4 ) λ E (T ) = #g 4 p E (T ) = T 4 (1 + #g 2 + #g 4 + #g 6 ) Requires a 4-loop computation in full QCD The matching coefficient at g 6 known only in the N f limit Ipp&Rebhan hep-ph/ ; Gynther, AK, Vuorinen Match EQCD MQCD to sufficient depth in g Kajantie et al. hep-ph/ Match the 3d continuum MS theory with lattice theory to order a 0 done using 4-loop Numerical Stochastic Perturbation Theory Di Renzo et al Measure pressure of lattice theory numerically Hietanen et al. hep-lat/ , Hietanen & AK hep-lat/

16 Example: Spatial string tension from MQCD Spatial string tension (magnetic screening) to T c 0.5 r 0 T T/ s 1/2 (T) N =4 N =6 N = T/T 0 Karsch et al.,

17 Outline Introduction Dimensional reduction Center symmetry

18 T < T c T T c T > T c Center symmetry At finite T the full symmetry group of Yang-Mills: Gauge symmetry A µ λ(τ, x)(a µ + i µ )λ(τ, x) 1, s SU(N c ) Global center symmetry λ(τ + 1/T, x) = zλ(τ, x), z Z Nc Order parameter: Ω(x) = TrW (x) = Tr [ P exp ( ig dτa 0 (τ, x) )]

19 Light fermions break the center symmetry softly Polyakov-loop still good approximate order parameter: L ren Tr p4, N =6 8 asqtad, N = T [MeV] Bazavov et al

20 Center symmetry Effective potential in perturbation theory: EQCD obtained by expanding A 0 around one of the N c deconfining minima. EQCD breaks the center symmetry explicitly. No chance for correct phase structure Quantities sensitive to A 0 not well described near T c

21 Center-symmetric effective theories Goal: Want to construct an effective theory that Preserves the Z N center symmetry Reduces to EQCD at high T Is superrenormalizable Effective theory of Wilson lines not (super)renormalizable Pisarski hep-ph/ non-linear sigma model

22 Center-symmetric effective theories Goal: Want to construct an effective theory that Preserves the Z N center symmetry Reduces to EQCD at high T Is superrenormalizable Effective theory of Wilson lines not (super)renormalizable Pisarski hep-ph/ non-linear sigma model Idea: Construct effective theory for coarse grained Wilson loop Vuorinen&Yaffe hep-ph/ Z(x) = T V Block V d 3 yu(x, y)w (y)u(y, x), / SU(N c )

23 Center-symmetric effective theory for SU(2) de Forcrand, AK, Vuorinen arxiv: For SU(2), sum of matrices proportional to SU(2) Z = λω, Ω SU(2), λ > 0 Z = 1 { } ( 1 Σ1 2 }{{} +i Π a σ }{{ a = 2 Σ + iπ 1 iπ 2 Π 3 } 1 iπ 2 + Π 3 2 Σ iπ 1 Singlet Adjoint scalar Transforms exactly like Wilson line Z λ 1 (x)zλ(x) gauge Z Z center Z 2...but note: (3 d.o.f in W = e igt A 0 ) (4 d.o.f in Z) )

24 Center-symmetric effective theory for SU(2) de Forcrand, AK, Vuorinen arxiv: For SU(2), sum of matrices proportional to SU(2) Z = λω, Ω SU(2), λ > 0 Z = 1 { } ( 1 Σ1 2 }{{} +i Π a σ }{{ a = 2 Σ + iπ 1 iπ 2 Π 3 } 1 iπ 2 + Π 3 2 Σ iπ 1 Singlet Adjoint scalar Most general superrenormalizable Lagrangian with A i and Z: L Z(2) = 1 ( ) 2 Tr F ij 2 + Tr D i Z D i Z +V (Z) }{{}}{{} spatial gluons Adjoint Kinetic V (Z) = b 1 Σ 2 + b 2 Π 2 a + c 1 Σ 4 + c 2 (Π 2 a) 2 + c 3 Σ 2 Π 2 a }{{} interactions from integration out Higher order terms suppressed by scale difference m E /T. )

25 Perturbative matching of the parameters Parameters of the effective theory {g 3, b 1, b 2, c 1, c 2, c 3 } can be matched to the full theory parameters {g, T } at high T : Split the potential in to hard and soft V = V h + g 2 3 V s Hard potential describes T scales, coarse graining Forces Z SU(2) V h = h 1 Tr (Z Z) + h 2 (Tr Z Z) 2 Soft potential encodes the physics of small fluctuations, EQCD V s = s 1 Tr Π 2 + s 2 ( Tr Π 2 ) 2 + s3 Σ 4

26 Matching at T Parameters can be mached in perturbation theory (series in g2 (7T ) 16π 2!): b 1 = 1 4 r2 T 2, b 2 = 1 4 r2 T g 2 T 2, c 1 /g 2 3 = r g 2, c 2 /g 2 3 = r g 2, c 3 /g 2 3 = r 2, g 2 3 = g 2 T Parameters functions of full theory parameters (g, T ) and r rt : mass of fluctuation away from SU(2) manifold rt = Cutoff of the effective theory continuum limit = r

27 Results from simulations: Z 2 -restoring phase transition β = 6, n = 64, r 2 = 5

28 Results from simulations: Phase diagram resembles the full theory (unlike in EQCD). Insensitive to r > 1 Phase transition at correct g(t )!

29 Effective theory for SU(3) For SU(3) no special relations degree of freedom Z GL(3,C) with L = 1 2 TrF ijf ij + Tr(D i Z D i Z) + V 0 (Z) + g 2 3V 1 (Z) V 0 = c 1 TrZ Z + c 2 ( det Z + det Z ) + c 3 Tr(Z Z) 2 g 2 zv 1 = d 1 TrM M + d 2 Tr(M 3 + M 3 ) + d 3 Tr(M M) 2 where M = Z 1 3 1TrZ

30 Effective theory for SU(3) For SU(3) no special relations degree of freedom Z GL(3,C) L = 1 2 TrF ijf ij + Tr(D i Z D i Z) + V 0 (Z) + g 2 3V 1 (Z) with V 0 = c 1 TrZ Z + c 2 ( det Z + det Z ) + c 3 Tr(Z Z) 2 g 2 zv 1 = d 1 TrM M + d 2 Tr(M 3 + M 3 ) + d 3 Tr(M M) 2 where M = Z 1 3 1TrZ The operator list is not exhaustive Similar splitting of action Hard potential keeps Z near unitary, has superfluous symmetry Soft potential encodes gt physics

31 Summary Dimensional reduction provides a bridge between lattice computations and perturbation theory in the deconfined phase. In DR setup, one can analytically deal with the heavy modes (non-static bosons and fermions) to get a theory which is more amenable to numerical simulations. Incorporating the (approximate) center symmetry to EQCD leads to correct phase diagram Lots of things to do: Check accuracy near T c : Domain wall tension Spatial string tension Screening masses Make predictions: Quarks: Z N breaking terms Finite chemical potential (Correct phase transitions?)