Renormalization Group Equations

Transcription

1 Renormalization Group Equations Idea: study the dependence of correlation functions on the scale k Various options: k = Λ: Wilson, Wegner-Houghton, coarse graining constant physics, cutoff/uv insensitivity k = µ: Gell-Mann-Low constant physics, fixing scale insensitivity k = m R : Callan, Symanzik (CS) compare theories with different masses

2 Renormalization Group Equations Idea: study the dependence of correlation functions on the scale k Various options: k = Λ: Wilson, Wegner-Houghton, coarse graining constant physics, cutoff/uv insensitivity k = µ: Gell-Mann-Low constant physics, fixing scale insensitivity k = m R : Callan, Symanzik (CS) compare theories with different masses Advantage of CS idea: start with massive theories = suppressed fluctuations connect with differential equation to small mass theories = regime with strong correlations

3 Renormalization Group Equations Callan-Symanzik: m 2 m 2 + k 2 Flow of master formula: ( k Z k [J] = Dφ 1 2 ) d 4 x( k k 2 )φ(x)φ(x) e S[φ] 1 k φ2 + Jφ

4 Renormalization Group Equations Callan-Symanzik: m 2 m 2 + k 2 Flow of master formula: ( k Z k [J] = Dφ 1 2 = 1 2 = 1 2 Tr[ ( k k 2 )G (2) ] k ) d 4 x( k k 2 )φ(x)φ(x) e S[φ] 1 k φ2 + Jφ d 4 x( k k 2 )G (2) k (x, x)

5 Renormalization Group Equations Callan-Symanzik: m 2 m 2 + k 2 Flow of master formula: ( k Z k [J] = Dφ 1 2 = 1 2 = 1 2 Tr[ ( k k 2 )G (2) ] k ) d 4 x( k k 2 )φ(x)φ(x) e S[φ] 1 k φ2 + Jφ d 4 x( k k 2 )G (2) k (x, x) Legendre transformation to Γ[φ]: k Γ k [φ] = 1 2 Tr [ ( k k 2 ) Γ (2) k + k 2 ] Problem: still UV divergent in D=4

6 Wetterich Equation Idea: replace mass deformation by momentum dependent function 1 (m 2 + k 2 )φ 2 1 d 4 p φ( p) (m 2 + R k (p)) φ(p) 2 2 RG flow equation: t Γ k k k Γ k = 1 2 Tr tr k (Γ (2) k (WETTERICH 93) + R k ) 1 = Exact Renormalization Group

7 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: Γ k =Λ = S micro UV: k Λ k IR: k 0

8 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: UV: k Λ k IR: k 0

9 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: UV: k Λ k IR: k 0

10 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: Γ k 0 = Γ UV: k Λ k IR: k 0

11 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k regulator function R k RG trajectory: Γ k=λ = S bare Γ k=0 = Γ 1PI eff

12 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k role of the regulator R k (O) e.g., chiral symmetry OK! R ψ k = Rψ k (i / )

13 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k role of the regulator R k (O) e.g., chiral symmetry OK! R ψ k = Rψ k (i / )

14 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k role of the regulator R k (O) e.g., chiral symmetry OK! R ψ k = Rψ k (i / )

15 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k role of the regulator R k (O) e.g., chiral symmetry OK! R ψ k = Rψ k (i / )

16 Lesson RG flow equation... exact equation RG flow equation (+ b.c.)... can serve as definition of QFT Wilsonian momentum-shell integration... treats physics scale by scale key element: scale-dependent exact propagator: 1 G k (x, y) = (x, y) Γ (2) k + R k

17 A Simple Example: dimensional QFT classical action of the Euclidean anharmonic oscillator ( 1 S[x] = dτ ẋ 2 ω2 x ) 24 λx 4 truncated effective action Γ k [x] = dτ ( ) V k (x), V k (x) = V 0,k + 1 2ẋ 2 ω k 2 x λ k x 4... second functional derivative Γ (2) k [x] = 2 τ + V k (x)

18 A Simple Example: dimensional QFT e.g., linear regulator R k (p 2 ) = (k 2 p 2 ) θ(k 2 p 2 ) (LITIM 00) flow equation flow of the potential t Γ k [φ] = 1 2 Tr [ t R k (Γ (2) k [φ] + R k) 1] = L τ dpτ 2π t V k (x) = 1 π k 2 θ(k 2 p 2 τ ) k 2 + V k (x) k 3 k 2 + V k (x)

19 A Simple Example: dimensional QFT polynomial expansion V k (x) = 1 2 ω k 2 x λ k x 4 + E 0,k + const. k coupling flows ( β functions) (const. k fixed such that d dk E 0,k = 0 for ω k = 0) x 0 : x 2 : x 4 : d dk E 0,k = 1 π d dk ω k 2 = 2 π d dk λ k = 24 π ( k 2 ) k 2 + ω 2 1 k k 2 λ k (k 2 + ω 2 k ) 2 2 k 2 (k 2 + ω k 3 ) 2 ( λk 2 (1) (2) ) (3)

20 A Simple Example: dimensional QFT truncate to (1) (ω k ω) E 0 = E 0,k 0 = 1 2 ω (HO) truncate to (1)+(2) and solve perturbatively E 0 = 1 2 ω + 3 ( ) λ 4 ω 24ω 3 82 ( ) 2 λ 40 ω 24ω compare with perturbation theory E 0 = 1 2 ω + 3 ( ) λ 4 ω 24ω ( ) 2 λ 40 ω 24ω (BENDER&WU) = 1-loop exact, 2-loop 20% error

21 A Simple Example: dimensional QFT truncate to (1) (ω k ω) E 0 = E 0,k 0 = 1 2 ω (HO) truncate to (1)+(2) and solve perturbatively E 0 = 1 2 ω + 3 ( ) λ 4 ω 24ω 3 82 ( ) 2 λ 40 ω 24ω compare with perturbation theory E 0 = 1 2 ω + 3 ( ) λ 4 ω 24ω ( ) 2 λ 40 ω 24ω (BENDER&WU) = 1-loop exact, 2-loop 20% error

22 A Simple Example: dimensional QFT truncate to (1) (ω k ω) E 0 = E 0,k 0 = 1 2 ω (HO) truncate to (1)+(2) and solve perturbatively E 0 = 1 2 ω + 3 ( ) λ 4 ω 24ω 3 82 ( ) 2 λ 40 ω 24ω compare with perturbation theory E 0 = 1 2 ω + 3 ( ) λ 4 ω 24ω ( ) 2 λ 40 ω 24ω (BENDER&WU) = 1-loop exact, 2-loop 20% error

23 A Simple Example: dimensional QFT ground state energy E 0 /E HO vs. λ 24 (ω = 1, m = 1 2 ) exact loop perturb. truncated to (2) truncated to (3)

24 A Simple Example: dimensional QFT strong-coupling limit E 0 = ( ) 1/3 λ [ α 0 + O(λ )] 2/3 24 α 0 = (JANKE& KLEINERT) α 0 (2) = error: 4% α 0 (3) = error: <1%

25 Lesson RG flow equation... encodes perturbative & nonperturbative physics already in simple approximations key element: scale-dependent exact propagator: many applications... 1 G k (x, y) = (x, y) Γ (2) k + R k

26 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: Γ k =Λ = S micro UV: k Λ k IR: k 0

27 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: UV: k Λ k IR: k 0

28 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: UV: k Λ k IR: k 0

29 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k RG trajectory: Γ k 0 = Γ UV: k Λ k IR: k 0

30 RG flow in Theory Space t Γ k = 1 2 Tr 1 t R k + R k Γ (2) k Abstract viewpoint: FRG provides vector field on theory space [PICTURE: F. SAUERESSIG] search for RG trajectories that can be completed to arbitrarily high scales = fundamental theory candidates: fixed points = bare action ( ˆ= microscopic action) is a result rather than an initial condition

31 Quantum field theory gravity Problem of Physics? expected typical scale of QG effects: M Planck GeV early/late universe cosmology? astrophysical singularities? hierarchy problems (gauge hierarchy, cosmological constant & coincidence problem) Minimum requirements: compatibility with observed physics existence of semiclassical GR regime D = 4 = D RG, cr compatibility with observed matter content of the universe

32 QFT Gravity (GOROFF,SAGNOTTI 85 86; VAN DE VEN 92) perturbative quantization fails Γ 2-loop div = ɛ 2880 (16π 2 ) 2 d 4 x gc µνρσ C ρσλτ C λτ µν = Any quantum theory of gravity has to explain the fate of C 3

33 Spacetime Dimensionality (perturbative) QFT: X X δ(γ) = d nei [φi ] + nvα δ(vα ) α i = RG critical dimension: 4 (gauge + matter, Yukawa/Higgs) DRG, cr = 2 (gravity, pure fermionic matter) (macroscopic) universe: D=4 It is not known whether the fact that space time has just four dimensions is a mere coincidence or is logically connected with this property. (J. Z INN -J USTIN, IN QFT AND CRITICAL PHENOMENA )

34 Quantizing Gravity I know of only one promising approach to this problem... (S. WEINBERG, IN CRITICAL PHENOMENA FOR FIELD THEORISTS (1976))

35 Asymptotic Safety

36 Necessity of Renormalizability IR physics well separated from UV physics (... no/mild cutoff Λ dependence) # of physical parameters <... or countably (... predictive power) = realized by perturbative RG... =... and by Asymptotic Safety (WEINBERG 76)

37 Asymptotic Safety g i g 2 Theory Space g 1

38 Asymptotic Safety g i effective action Γ k g 2 Theory Space g 1

39 Asymptotic Safety g i RG step functional RG: (WETTERICH 93) t Γ k = 1 2 Tr tr k (Γ (2) k + R k ) 1 (WILSON 71; WEGNER,HOUGHTON 73; POLCHINSKI 84) g 2 Theory Space g 1

40 Asymptotic Safety g i Fixed Point g 2 Theory Space g 1

41 Asymptotic Safety g i UV repulsive, irrelevant Θ < 0 UV attractive, relevant Θ > 0 g 2 Theory Space g 1

42 Asymptotic Safety g i critical surface S g 2 Theory Space g 1

43 Asymptotic Safety g i = dim S = # Θ s>0 g 2 Theory Space g 1

44 Asymptotic Safety g i (Λ independence ) (# phys. parameters < ) (universality & predictivity ) g 2 Theory Space g 1

45 Asymptotic Safety FP regime: g i t g i = B i j (g j g j ) +... stability matrix g 2 g 1 B i j = β i(g ) g j critical exponents: { Θ } = spect( Bi j ) Theory Space

46 Mechanisms of Asymptotic Safety Dimensional Balancing

47 Non-Gaußian Fixed Points coupling ḡ with canonical dimension δ ḡ (D) in spacetime dimension D: [ḡ] = δ ḡ (D) critical RG dimension: δ ḡ (D RG, cr ) = 0 β g perturbative β function in D RG, crit, e.g.: β ḡ = b 0 ḡ = if b 0 < 0: theory is asymptotically free (and safe)

48 Non-Gaußian Fixed Points away from D RG, cr (+ analyticity in D): β ḡ = b 0(D) k δḡ (D) ḡ dimensionless coupling in units of a given scale k g = ḡ k δ(ḡ;d) RG flow of dimensionless coupling: k d dk g β g = δ ḡ (D)g + b 0 (D) g 2

49 Non-Gaußian Fixed Points RG flow of dimensionless coupling: k d dk g β g = δ ḡ (D)g }{{} dimensional running +b 0 (D) g 2 }{{} fluctuation-induced running = NGFP g for: = g > 0 for: sign(δ ḡ (D)) = sign(b 0 (D)) δ ḡ (D), b 0 (D) < 0 β g g *

50 Example: Fermionic Systems for instance, Nambu Jona-Lasinio / Gross-Neveu in 3 dimensions: Γ k = d 3 x ψi / ψ λ( ψψ) , [ λ] = 1 dim less coupling λ = k λ t λ = λ c λ 2 UV fixed point λ = 1/c critical exponent Θ = 1 = asymptotically safe proven to all orders in 1/N f expansion: (GAWEDZKI, KUPIAINEN 85; ROSENSTEIN, WARR, PARK 89; DE CALAN ET AL. 91)

51 Example: 3D Gross-Neveu model exact mapping to Yukawa model: ( ) Γ[ψ, σ] = d 3 Nf Z σ x 2 ( σ)2 + ψ(z ψ i / + i hσ)ψ + N f U(σ) (STRATONOVICH 58,HUBBARD 59) non-gaußian fixed point in Yukawa model: (BRAUN,HG,SCHERER 10)

52 Example: 3D Gross-Neveu model exact large-n f fixed point effective potential (BRAUN,HG,SCHERER 10) u (ρ) = 2d 8 ( 3d 4 ρ 2F 1 1 d 2, 1; 2 d (d 4)(d 2) ; 2 6d 8 ) d ρ d γ v d, ρ = σ2 2 exact critical exponents: Θ = 1, 1, 1, 3, 5, 7,... = dim S = 1 physical parameter

53 Example: 3D Gross-Neveu model critical exponents beyond large-n f (ν = 1/Θ 1 ) (BRAUN,HG,SCHERER 10) matches even with N f 0 limit (Ising model), excellent agreement with lattice simulations (available for N f = 2) (KARKKAINEN ET AL. 93)

54 Quantum Einstein Gravity (REUTER 96) effective action in Einstein-Hilbert truncation 1 Γ k = d D x g( R + 2Λ k ) 16πG k running dim less Newton s constant in D = 4: g = k 2 G k, Λ k = 0 t g = 2 g c g 2 + O(g 3 ), c = c[r k ] > 0 β g * g (DOU,PERCACCI 97) (SOUMA 99) (LAUSCHER,REUTER 01 02) (REUTER,SAUERESSIG 01) (NIEDERMAIER 02) (LITIM 03) (CODELLO,PERCACCI 06) (CODELLO,PERCACCI,RAHMEDE 07 08) (MACHADO,SAUERESSIG 07) (BENEDETTI,MACHADO,SAUERESSIG 09) (EICHHORN,HG,SCHERER 09) (DONKIN,PAWLOWSKI 12) (CHRISTIANSEN,LITIM,PAWLOWSKI,RODIGAST 12) (CHRISTIANSEN,MEIBOHM,PAWLOWSKI,REICHERT 15) (FALLS 15;CHRISTIANSEN 16) (DENZ,PAWLOWSKI,REICHERT 16)

55 Quantum Einstein Gravity (REUTER 96) effective action in Einstein-Hilbert truncation 1 Γ k = d D x g( R + 2Λ k ) 16πG k (DOU,PERCACCI 97) (SOUMA 99) (LAUSCHER,REUTER 01 02) (REUTER,SAUERESSIG 01) running G k and Λ k in D = 4: g = k 2 G k, λ = Λ k /k 2 e.g., sharp cutoff t g = (2 + η)g [5 ln[1 2λ] 2ζ(3) + 52 η ] t λ = 2(2 η)λ g π anomalous graviton dimension: η = 2g [ ] ln(1 2λ) ζ(2) + 6 6π + 5g 1 2λ (NIEDERMAIER 02) (LITIM 03) (CODELLO,PERCACCI 06) (CODELLO,PERCACCI,RAHMEDE 07 08) (MACHADO,SAUERESSIG 07) (BENEDETTI,MACHADO,SAUERESSIG 09) (EICHHORN,HG,SCHERER 09) (DONKIN,PAWLOWSKI 12) (CHRISTIANSEN,LITIM,PAWLOWSKI,RODIGAST 12) (CHRISTIANSEN,MEIBOHM,PAWLOWSKI,REICHERT 15) (FALLS 15;CHRISTIANSEN 16) (DENZ,PAWLOWSKI,REICHERT 16)

56 RG flow of Quantum Einstein Gravity (REUTER,SAUERESSIG 01) critical exponents: ReΘ 1,2 2

57 From Quantum to Classical Gravity (HG,KNORR,LIPPOLDT 15) RG trajectories interconnecting the transplanckian and classical regimes exist : physical trajectory: gλ k "today"

58 Fate of two-loop counterterm operator expansion with background field method [PICTURE: F. SAUERESSIG]

59 Flow of Einstein-Hilbert + Goroff-Sagnotti effective action Γ k = Γ EH k + Γ GS k (HG,KNORR,LIPPOLDT,SAUERESSIG 16) Einstein-Hilbert Γ EH k = 1 16πG k d D x g( R + 2Λ k ) Goroff-Sagnotti: Γ GS k = σ k d D x g C µνρσ C ρσλτ C λτ µν dimensionless coupling constants g = k 2 G k, λ = Λ k /k 2, σ = σ k k 2

60 Flow of Einstein-Hilbert + Goroff-Sagnotti Fluctuations with GS vertex: Γ GS(2) σc µν αβ + O(R2 ) (HG,KNORR,LIPPOLDT,SAUERESSIG 16) BUT: trc µν αβ = 0 = Two-loop counterterm does not directly feed back into EH = Fixed point in Einstein-Hilbert sector is maintained asymptotic safety

61 Flow of Einstein-Hilbert + Goroff-Sagnotti Einstein-Hilbert: t g = (2 + η N ) g t λ = (η N 2) λ + g 2π ( (HG,KNORR,LIPPOLDT,SAUERESSIG 16) ) 5 1 2λ η N 1 1 2λ Goroff-Sagnotti t σ = c 0 + (2 + c 1 ) σ + c 2 σ 2 + c 3 σ 3 (10 3 ) 3 terms g > 0 = c 3 > 0 = asymtotic safety extends to (irrelevant) GS term

62 Phase portrait (HG,KNORR,LIPPOLDT,SAUERESSIG 16) crossover to classical regime persists

63 Towards apparent convergence in Quantum Gravity (DENZ,PAWLOWSKI,REICHERT 16) systematic vertex expansion n 4 fully dynamical propagators = asymptotically safe fixed point = 3 relevant directions Λ R R 2 R µν R µν irrel. = promising path towards systematic scheme establishing asymptotic safety

64 Lesson FRG facilitates a search for quantizable theories candidates: RG fixed points Wilsonian renormalization extends beyond perturbative realm UV complete and predictive Asymptotic safety in fermion systems and in gravity... perturbatively nonrenormalizable, but nonperturbatively renormalizable Functional RG interconnects all physical scales... from Planck to Hubble

65 Lesson FRG facilitates a search for quantizable theories candidates: RG fixed points Wilsonian renormalization extends beyond perturbative realm UV complete and predictive Asymptotic safety in fermion systems and in gravity... perturbatively nonrenormalizable, but nonperturbatively renormalizable Functional RG interconnects all physical scales... from Planck to Hubble Why D=4?... no news yet

66 The flow of the Renormalization Group holds, I think, the supreme position among the laws of Nature. Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)

67 The flow of the Renormalization Group... If someone points out to you that your pet theory of the universe is in disagreement with Maxwell s equations... Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)

68 The flow of the Renormalization Group then so much the worse for Maxwell s equations. Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)

69 The flow of the Renormalization Group... If it is found to be contradicted by observation... Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)

70 The flow of the Renormalization Group well, these experimentalists do bungle things sometimes. Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)

71 The flow of the Renormalization Group... But if your theory is found to be against the flow of the renormalization group... Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)

72 The flow of the Renormalization Group... I can give you no hope; there is nothing for it but to collapse in deepest humiliation. Sir Athur Eddington (1927) paraphrased by V. Rivasseau (2011)