Non-scalar operators and logarithmic correlation functions for the - - PowerPoint PPT Presentation

non scalar operators and logarithmic correlation
SMART_READER_LITE
LIVE PREVIEW

Non-scalar operators and logarithmic correlation functions for the - - PowerPoint PPT Presentation

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions Non-scalar operators and logarithmic correlation functions for the Potts model in arbitrary dimension RGP 2016 - IHP Romain Couvreur


slide-1
SLIDE 1

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Non-scalar operators and logarithmic correlation functions for the Potts model in arbitrary dimension

RGP 2016 - IHP Romain Couvreur

Laboratoire de Physique Théorique - École Normale Supérieure Institut de Physique Théorique - CEA Saclay

October 2016

Collaborators : Jesper L. Jacobsen, Romain Vasseur

1 / 13

slide-2
SLIDE 2

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Introduction

  • Scale Invariance : power-law and logarithmic correlation
  • c = 0 CFTs : percolation, disordered systems (IQHE,. . . )
  • Logarithmic minimal models
  • Jordan cell in the dilatation operator L0
  • Two dimensions collide J. Cardy 2013

φ(0)φ(r) =

  • i

Ci r2∆i , C1 ∼ −C2 → ∞, C1(∆1 − ∆2) stays finite contribution r−2∆1 log r

  • Insight from discrete symmetries

2 / 13

slide-3
SLIDE 3

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Potts model

  • Q-state Potts model : discrete SQ symmetry

Z =

  • {σ}
  • (i,j)∈E

eKδσi,σj , σ = 1, . . . , Q

  • Fortuin-Kasteleyn clusters (Q ∈ R) :

Z =

  • A⊆E

(eK − 1)|A|Qk(A)

3 / 13

slide-4
SLIDE 4

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Operators acting on 1 spin σ : R. Vasseur, J. L. Jacobsen 2014

  • General form : O(σ) =

Q

  • a=1

Oaδσ,a

  • Action of the symmetric group SQ :

(pO)(σ) =

Q

  • a=1

Oaδσ,p(a) =

Q

  • a=1

Oaδp−1(σ),a

4 / 13

slide-5
SLIDE 5

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Operators acting on 1 spin σ : R. Vasseur, J. L. Jacobsen 2014

  • General form : O(σ) =

Q

  • a=1

Oaδσ,a

  • Action of the symmetric group SQ :

(pO)(σ) =

Q

  • a=1

Oaδσ,p(a) =

Q

  • a=1

Oaδp−1(σ),a

  • Two irreps :

t(σ) =

Q

  • a=1

δσ,a, ta(σ) = δσ,a − 1 Q

  • Identity and magnetization operator
  • decomposition with Young diagrams L(1)

Q = [Q] ⊕ [Q − 1, 1]

4 / 13

slide-6
SLIDE 6

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

General setup (short version) :

Operator acting on N spins : Operators are Q × Q × . . . = QN tensors : L(N)

Q

= Span

  • Oa1,...,aN (σ1, . . . , σN) =

N

  • i=1

δσi,ai

  • Action of p ∈ SQ : O{ai} = O{p(ai)}

SQ symmetry :

  • Choose Young tableau λQ with at least Q − N boxes in the first row
  • Compute Young symmetrizer eλQ
  • May have to specify SN representation
  • Generate invariant subspace of operators

5 / 13

slide-7
SLIDE 7

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Result for L(2)

Q , operators acting on σ1 and σ2 :

  • Subspace with σ1 = σ2 isomorphic to L(1)

Q

  • Symmetric operators under σ1 ↔ σ2

t[Q](σ1, σ2) = δσ1=σ2

...

t[Q−1,1],[2]

a

(σ1, σ2) = δσ1=σ2

  • δσ1,a + δσ2,a − 2

Q

  • ...

t[Q−2,2]

a,b

(σ1, σ2) = δσ1=σ2

  • δσ1,aδσ2,b + δσ1,bδσ2,a

...

− 1 Q − 2

  • t[Q−1,1],[2]

a

+ t[Q−1,1],[2]

b

2 Q(Q − 1)t[Q]

6 / 13

slide-8
SLIDE 8

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Result for L(2)

Q , operators acting on σ1 and σ2 :

  • Subspace with σ1 = σ2 isomorphic to L(1)

Q

  • Symmetric operators under σ1 ↔ σ2

t[Q](σ1, σ2) = δσ1=σ2

...

t[Q−1,1],[2]

a

(σ1, σ2) = δσ1=σ2

  • δσ1,a + δσ2,a − 2

Q

  • ...

t[Q−2,2]

a,b

(σ1, σ2) = δσ1=σ2

  • δσ1,aδσ2,b + δσ1,bδσ2,a

...

− 1 Q − 2

  • t[Q−1,1],[2]

a

+ t[Q−1,1],[2]

b

2 Q(Q − 1)t[Q]

  • Anti-symmetric operators under σ1 ↔ σ2

t[Q−1,1],[1,1]

a

(σ1, σ2) = δσ1=σ2 (δσ1,a − δσ2,a)

...

t[Q−2,1,1]

a,b

(σ1, σ2) = δσ1=σ2

  • δσ1,aδσ2,b − δσ1,bδσ2,a

...

− 1 Q

  • t[Q−1,1],[1,1]

a

− t[Q−1,1],[1,1]

b

6 / 13

slide-9
SLIDE 9

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Result for L(2)

Q , operators acting on σ1 and σ2 :

  • L(2)

Q

decomposition : L(2)

Q = L(1) Q ⊕ ...

⊕ 2

...

...

...

  • Ok with hook formula

Q2 = Q + 1 + 2(Q − 1) + Q(Q − 3) 2 + (Q − 1)(Q − 2) 2

  • Subtracted operators correspond to Young diagrams with boxes

removed

  • Poles are related to logarithmic features

7 / 13

slide-10
SLIDE 10

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Correlation functions for N = 1, 2, some examples

  • We can compute correlation functions with FK-clusters
  • t[Q−1,1]

a

t[Q−1,1]

b

  • = 1

Q

  • δa,b − 1

Q

  • P
  • For N ≥ 2 each operators act on N spins in the same neighbourhood
  • Symmetric correlation functions N = 2 :
  • t[Q−2,2]

a,b

(0)t[Q−2,2]

c,d

(r)

  • (2δδ) −

1 Q − 2 (4δ) + 2 (Q − 2)(Q − 1)

 P       + P         

  • Anti-symmetric correlation functions N = 2 :
  • t[Q−2,1,1]

a,b

(0)t[Q−2,1,1]

c,d

(r)

  • (2δδ) − 1

Q (4δ)

 P       − P         

  • Correlation functions defined for Q ∈ R for finite size
  • t[Q−2,1,1]

a,b

(0)t[Q−2,1,1]

c,d

(r)

  • (2δδ) − 1

Q(4δ) ˜ Z(Q) Z(Q)

8 / 13

slide-11
SLIDE 11

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Scale invariance

  • Power-law correlation functions
  • t[Q−2,1,1]

a,b

(0)t[Q−2,1,1]

c,d

(r)

  • (2δδ) − 1

Q(4δ)

  • 1

r2∆λ2 (Q)

  • on the cylinder from Jones-Temperley-Lieb representation theory

∆ = h k

N ,N + h− k N ,N

  • 4-leg watermelon operator

...

→ P

  • + P
  • , ∆ = 2h0,2
  • Other observables

...

→ P

  • − P
  • , ∆ = h−1/2,2 + h1/2,2
  • Non-scalar, s = h−1/2,2 − h1/2,2 = 1
  • 2d simplified, some configurations are forbidden : P
  • 9 / 13
slide-12
SLIDE 12

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Numerics

  • Transfer matrix methods
  • Monte-Carlo simulation

10 / 13

slide-13
SLIDE 13

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Logarithmic correlation functions

  • How to deal with poles ? By mixing two operators
  • Percolation : 4-leg watermelon operator diverges at Q → 1

t[Q−2,2]

a,b

= regular part − 2 Q(Q − 1)ǫ, ǫ = t[Q],[2] −

  • t[Q],[2]

φa,b = t[Q−2,2]

a,b

+ 2 Q(Q − 1)ǫ

  • Correlation functions should be finite ∆ǫ = ∆2
  • Indeed in 2d, ∆ǫ = ∆2 = 5/4

Contribution as Q → 1 : 1 Q − 1

  • 1

r2∆ǫ(Q) − 1 r2∆2(Q)

  • ∼ ∆ǫ(Q) − ∆2(Q)

Q − 1 log r r2∆2(1)

  • energy and 4-leg watermelon operator mixed in Jordan cell

11 / 13

slide-14
SLIDE 14

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Logarithmic correlation functions

  • Predict scaling laws on the lattice with universal constant δ

P

  • + P
  • − P (

)2 P

  • + P
  • ∼ δr−2∆2 log r
  • Can be verified with Monte-Carlo
  • Whole new set of Jordan cell with non-scalar operators
  • Conjecture for the position of the poles gives the "full" log structure

12 / 13

slide-15
SLIDE 15

Operator in the Potts model Correlation functions and non-scalar operators Log correlation functions

Conclusion

  • Classification of operators/LogCFTs in any d with discrete symmetries
  • Partial results, other symmetries can be considered
  • Jordan-cell of higher rank ?
  • Bestiary of interesting geometrical observables

13 / 13