large nc confinement, universal shocks, and random matrices

20
Large Nc confinement, universal shocks, and random matrices Jean-Paul Blaizot, IPhT- Saclay EMMI - Workshop February 19, 2009

Upload: others

Post on 02-Jun-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Large Nc confinement, universal shocks, and random matrices

Large Nc confinement, universal shocks, and random

matrices

Jean-Paul Blaizot, IPhT- Saclay

EMMI - Workshop

February 19, 2009

Page 2: Large Nc confinement, universal shocks, and random matrices

Outline

- Phase transition in large Nc Yang-Mills theories, in d=2 (Durhuus-Olesen), possible universality and simulations in d=3,4 (Narayanan-Neuberger)

-(Fluid)-dynamics of eigenvalues of Wilson loops, (complex) Burgers equation - Dyson’s fluid

- Further developments : finite Nc effects as viscosity corrections - relation with random matrix theory.

Work done in collaboration with Maciej Nowak(arXiv: 0801.1859 - PRL 101:102001 (2008)and arXiv: 0902.2223)

Page 3: Large Nc confinement, universal shocks, and random matrices

Small loops - Short distance physics - weak interactions

large loops

long distance physics

strong interactions Cross-over

Becomes a sharp transition as

“Confinement-deconfinement” transition (?)

Universal properties in d=2,3,4 (?)

The Wilson loop in SU(N) gauge theories

Page 4: Large Nc confinement, universal shocks, and random matrices

Durhuus-Olesen phase transition(P. Durhuus and P. Olesen, NPB 1981)

Spectral density as a function of the area A (d=2)

Page 5: Large Nc confinement, universal shocks, and random matrices

The moments of the spectral density

The moments are related to Laguerre polynomials

Remarkable properties at large n

Much studied

Olesen, Durhuus, Rossi, Kazakov, Douglas, Gross, Gopakumar, Matytsin, Migdal, etc.

Page 6: Large Nc confinement, universal shocks, and random matrices

(R. Janick and W. Wieczorek, math-ph/0312043)

A random matrix perspective

A model for the Wilson loop

Resolvent and spectral density

Evolution equation

Page 7: Large Nc confinement, universal shocks, and random matrices

Evolution from small areas to large areas (1)

(from R. Janick and W. Wieczorek, math-ph/0312043)

Page 8: Large Nc confinement, universal shocks, and random matrices

Evolution from small areas to large areas (2)

(from R. Janick and W. Wieczorek, math-ph/0312043)

Page 9: Large Nc confinement, universal shocks, and random matrices

Closing of the gap is universal

(R.Narayanan and H.Neuberger, arXiv:0711.4551)

Numerical studies on the lattice

Double scaling limit

Average characteristic polynomial

Page 10: Large Nc confinement, universal shocks, and random matrices

(Fluid) dynamics of eigenvalues and the Burgers equation

Define

(Hilbert transform)

The function obeys the Burgers equation

(Olesesn, Gross, Gopakumar, etc)

Page 11: Large Nc confinement, universal shocks, and random matrices

Navier-Stokes equation

Euler equation

Burgers equation and equations of fluid dynamics

Important: F is here complex

Burger=s equation is often used as a model for turbulence (shocks)

(Hopf-Burgers eqn, free random variable calculus, Voiculescu …)

Page 12: Large Nc confinement, universal shocks, and random matrices

Solution of the complex Burgers equation

Characteristics (= Lines of constant ‘velocity’)

Singularity when

Can be solved with Characteristics

Page 13: Large Nc confinement, universal shocks, and random matrices

The qualitative behavior of the solution is determined by the location of the singularities in the complex plane

Asymptotic behavior of moments

Expanding near a singularity

One can invert the characteristic equation and get

Page 14: Large Nc confinement, universal shocks, and random matrices

Gapped phase

Characterisics

singularity

Vicinity of the singularity

Page 15: Large Nc confinement, universal shocks, and random matrices

Closing the gap

Page 16: Large Nc confinement, universal shocks, and random matrices

Gapless phase

In vicinity of the singularity

Start with initial conditions

singularity

In the second case, the singularity hits the real axis in a finite time A1

Two cases

Page 17: Large Nc confinement, universal shocks, and random matrices

〈δxi〉 = E(xi)∆t 〈(δxi)2〉 = ∆t

E(xj) =∑

i!=j

(1

xj − xi

)

∂P

∂t=

12

i

∂2P

∂x2i

−∑

i

∂xi(E(xi)P )

P (x1, · · · , xN , t) = C∏

i<j

(xi − xj)2 e−P

i

x2i

2t

Approaching

Dyson’s Brownian motion (hermitian matrices)

Fokker-Planck equation for the joint probability

Whose solution reads

(arXiv 0902.2223 and work in progress)

P (x1, · · · , xN , t)

(1)

Page 18: Large Nc confinement, universal shocks, and random matrices

ρ̃(x, t) =∫ N∏

k=1

dxk P (x1, · · · , xN , t)N∑

l=1

δ(x− xl)

∂ρ̃(x, t)∂t

=12

∂2ρ̃(x, t)∂λ2

− ∂

∂λPV

∫dy

ρ̃(x, y, t)x− y

ρ̃(x, y) = ρ̃(x)ρ̃(y) + ρ̃con(x, y)

ρ̃(x) = Nρ(x)

∂ρ(x)∂τ

+∂

∂xρ(x)PV

∫dy

ρ(y)x− y

=1

2N

∂2ρ(x)∂x2

+PV∫

dyρcon(x, y)

x− y

τ = Nt

Approaching

Average density of eigenvalues (“one-particle density”)

Infinite hierarchy of equations

To study the large N limit, rescale

and get

(2)

Page 19: Large Nc confinement, universal shocks, and random matrices

G(z, τ) =⟨

1N

Tr1

z −H(τ)

⟩=

∫dy

ρ(y, τ)z − y

∂τG(z, τ) + G(z, τ) ∂zG(z, τ) = 0

ρ(x, τ)

G(z, τ) =⟨

1N

Tr1

z −H(τ)

⟩=

∂z

⟨1N

Tr ln (z −H(τ))⟩

=∂

∂z

⟨1N

ln det (z −H(τ))⟩

F (z, τ) =∂

∂z

1N

ln 〈det (z − H(τ))〉

〈det (z − H(τ))〉 =N∏

i=1

(z − x̄i)

F (z, τ) ≈ G(z, τ) as N →∞

∂τF (z, τ) + F (z, τ) ∂zF (z, τ) = − 12N

∂2zF (z, τ)

Approaching (3)

Resolvent

Equation for reduces to (inviscid) Burgers eqn. for G (in large N limit)

F fulfills the Viscid Burgers equation(EXACTLY !)

Average of the characteristic polynomial

(see also Neuberger, arXivO806.0149 , 0809.1238)

Note that

Page 20: Large Nc confinement, universal shocks, and random matrices

Conclusions

- Many features of the large Nc transition are coded in the solution of a Simple

Burgers equation (universal shocks, etc). - Provides a simple understanding for the remarkable universality that is emerging from lattice calculations

- Finite Nc corrections appears as « viscous » effects in the fluid of eigenvalues

- A general picture emerges in the framework of random matrix theory