the integrable structure of liouville theory
play

The integrable structure of Liouville theory J org Teschner DESY - PowerPoint PPT Presentation

The integrable structure of Liouville theory J org Teschner DESY Hamburg Joint with A. Bytsko What is Liouville theory? I d 2 z z + 4 e 2 ) . S [ ] = 4 ( z Related to uniformization of Riemann


  1. The integrable structure of Liouville theory J¨ org Teschner DESY Hamburg Joint with A. Bytsko

  2. What is Liouville theory? – I d 2 z � z ϕ + 4 πµe 2 ϕ ) . S [ ϕ ] = 4 π ( ∂ z ϕ∂ ¯ Σ Related to uniformization of Riemann surfaces: ds 2 = e 2 ϕ dzd ¯ z has constant negative curvature iff z ϕ = µe 2 ϕ . ϕ satisfies Liouville equation of motion ∂ z ∂ ¯ iff S [ ϕ ] is minimized. Essentially: ⇔ Teichm¨ uller theory . Liouville theory

  3. What is Liouville theory? – II The functional S [ ϕ ] plays an important role in Teichm¨ uller theory. Let ϕ be the unique solution to ∂ ¯ ∂ϕ = 2 πµe ϕ , with the boundary conditions z ) = − 2(1 − η s ) log | z | 2 + O (1) ϕ ( z, ¯ at | z | → z s = ∞ , z ) = − 2 η i log | z − z i | 2 + O (1) ϕ ( z, ¯ at | z | → z i , i = 1 , . . . , s − 1 . Let z s − 1 = 1 , z 1 = 0 ⇒ z 2 , . . . , z s − 2 : complex analytic coordinates for M 0 ,n . Let ϕ ] = 1 � | ∂ z ϕ | 2 + µ cl e ϕ � � S cl X ǫ d 2 z � ǫ 4 π � η i � ǫ s − 1 � � � � � dx ϕ + 2 η 2 − i log ǫ + (1 − η s ) dx ϕ − 2 log ǫ , 2 πǫ 2 π ∂D i ∂D s i =1 where D i = { z ∈ C ; | z − z i | < ǫ } , D s = { z ∈ C ; | z | > 1 /ǫ } , and X ǫ = D s \ � s − 1 i =1 D i . Theorem (Takhtajan-Zograf) Let ∂ ( ¯ ∂ ) be (anti-) holomorphic components of de Rham differential. We have 2 πi∂ ¯ ∂ω = ω WP .

  4. What is quantum Liouville theory? – III z ) = e 2 αφ ( z, ¯ z ) . Basic observables: V α ( z, ¯ Theory fully characterized by correlation functions � � V α n ( z n , ¯ z n ) . . . V α 1 ( z 1 , ¯ z 1 ) The correlation functions can be seen as giving deformations of the K¨ ahler potential in the sense that � − 1 ∼ e − 1 b 2 S [ ϕ ] � 2 (1 + O ( b 2 )) , det(∆ ϕ + 1 � � V α n ( z n , ¯ z n ) . . . V α 1 ( z 1 , ¯ z 1 ) 2 ) where S [ ϕ ] is the action functional defined above. Standard physicists rules for quantization of Lagrangian field theories ⇒ ⇒ ( Takhtajan, Teo ) formal series in b 2 . The correlation functions can be defined non-perturbatively by means of conformal bootstrap ( Belavin, Polyakov, Zamolodchikov; Al.B., A.B. Zamolodchikov; J.T.)

  5. Quantum equivalence Liouville theory ⇔ Teichm¨ uller theory Consider T g,n : Teichm¨ uller space of Riemann surfaces with genus g and n conical singularities, deficit angles η k = bα k . Teichm¨ uller space has a K¨ ahler structure. Let m = ( m 1 , . . . , m 3 g − 3+ n ) be complex analytic coordinates for T g,n . Claim ( J.T. ): There exists a canonical K¨ ahler quantization of T g,n , characterized by the deformed Bergmann kernel B Σ (¯ n , m ) ( Karabegov ⇒ star product etc.). The Bergmann kernel B Σ is uniquely determined by • invariance under mapping class group action defined by Fock; Kashaev; Chekhov, Fock . • boundary conditions at ∂ T g,n . We then have � � B Σ ( ¯ m , m ) = V α n ( z n , ¯ z n ) . . . V α 1 ( z 1 , ¯ z 1 ) Σ m

  6. The conformal structure of Liouville theory The general solution to to ∂ ¯ ∂ϕ = 2 πµe ϕ , can be written as | ∂ z A ( z ) | 2 �� � ϕ ( z, ¯ z ) = log 2 πµ , (1 + | A ( z ) | 2 ) 2 It is parameterized by a holomorphic function A ( z ) , which describes the uniformizing mapping. Consider instead theory on cylinder with coordinates t, σ , Minkowskian signature ∂ + ∂ − ϕ = 2 πµe ϕ , where ∂ ± = 1 2 ( ∂ t ± ∂ σ ) . Solution 2 πµ ∂ + A ( x + ) ∂ − ¯ �� � A ( x − ) ϕ ( z, ¯ z ) = log , (1 + A ( x + ) ¯ A ( x − )) 2 In other words: Liouville theory is most easily solved by means of its conformal structure : Factorization into left-movers ( A ( x + ) ) and right-movers ( ¯ A ( x − ) ).

  7. The integrable structure of Liouville theory — 0 Why are we interested in the integrable structure of Liouville theory ? • Explain integrability of 2d gravity !!! • Integrable structure of CFT • Model for conformal field theories where conformal symmetry is not enough to solve them. A lot of work was devoted to the integrable structure of Liouville theory ( Faddeev, Volkov, Kashaev ). Missing: Link to the solution via conformal bootstrap. This is what we can now provide.

  8. The integrable structure of Liouville theory — I Starting point: Sinh-Gordon model, ( ∂ 2 t − ∂ 2 x ) ϕ ( σ, t ) = − 8 πµb sinh(2 bϕ ( x, t )) . zero curvature condition, [ ∂ t − V ( x, t ; λ ) , ∂ σ − U ( x, t ; λ ) ] = 0 , with U , V being − im ( λe − bϕ − λ − 1 e bϕ ) b � � 2 Π U ( x, t ; λ ) = − im ( λe bϕ − λ − 1 e − bϕ ) − b 2 Π − im ( λe − bϕ + λ − 1 e bϕ ) b 2 ϕ ′ � � V ( x, t ; λ ) = − im ( λe − bϕ + λ − 1 e bϕ ) − b 2 ϕ ′ and m chosen such that m 2 = πb 2 µ . Hamiltonian formalism: { Π( σ ) , ϕ ( σ ′ ) } = 2 π δ ( σ − σ ′ ) , The time-evolution of an arbitrary observable is then given as ∂ t O ( t ) = { H , O ( t ) } , with Hamiltonian density � R dx Π 2 + ( ∂ σ ϕ ) 2 + 8 πµ cosh(2 bϕ ) � � H ShG = . 4 π 0

  9. The integrable structure of Liouville theory — II Interesting limit: Let m → 0 , ϕ → ϕ − ξ , ξ → ∞ such that me 2 bξ → µ ⇒ � R dx H Liou = Π 2 + ( ∂ σ ϕ ) 2 + 4 πµe − 2 bϕ � � . 4 π 0 Can be performed on the level of the Lax-pair if combined with shift of spectral parameter ( Faddeev-Tirkkonen ) b − imλe − bϕ � � 2 Π U ( x, t ; λ ) = − im ( λe bϕ − λ − 1 e − bϕ ) − b 2 Π b 2 ϕ ′ − imλe − bϕ � � V ( x, t ; λ ) = − im ( λe − bϕ + λ − 1 e bϕ ) − b 2 ϕ ′ Question: Does the monodromy matrix generate enough conserved quantities?

  10. The integrable structure of Liouville theory — III To answer this question, go to discrete versions of Sinh-Gordon / Liouville. Variables ϕ n ≡ ϕ ( n ∆) , Π n ≡ ∆Π( n ∆) , Lax matrices: µ − 1 V 2 µ − 1 V − 1 � U n + µ ¯ � n U n µ V n + ¯ n L ShG ( µ, ¯ µ ) = , µ V − 1 µ − 1 V n U − 1 µ − 1 V − 2 n U − 1 n + ¯ n + µ ¯ n b 2 Π n , V n = e − bϕ n , µ ≡ − iλm , ¯ µ ≡ − iλm − 1 . Liouville Lax matrix: where U n = e µ − 1 V 2 � U n + µ ¯ � n U n µ V n L Liou ( µ, ¯ µ ) = , µ V − 1 µ − 1 V n U − 1 n + ¯ n The corresponding monodromy matrix, T Liou ( λ ) = tr( L Liou N ( λ ) · · · L Liou ( λ )) 1 gives only L + 1 conserved quantities if N = 2 L + 1 , not enough .

  11. The integrable structure of Liouville theory — IV But consider also: µ − 1 V 2 µ − 1 V − 1 � U n + µ ¯ � n U n µ V n + ¯ ¯ n L Liou ( µ, ¯ µ ) = . µ − 1 V n U − 1 ¯ n The corresponding monodromy matrix, T Liou ( λ ) = tr(¯ ¯ N ( λ ) · · · ¯ L Liou L Liou ( λ )) 1 Poisson-commutes with T Liou ( λ ) and gives another L + 1 conserved quantities. One of these coincides with a quantity from T Liou ( λ ) , ”zero mode”. Together: Number of conserved quantities = Number of degrees of freedom .

  12. The integrable structure of Liouville theory — V Recall 2 πµ ∂ + A ( x + ) ∂ − ¯ �� A ( x − ) � ϕ ( z, ¯ z ) = log (1 + A ( x + ) ¯ A ( x − )) 2 Claim : T Liou ( λ ) : Conserved quantitites constructed from A ( x + ) , T Liou ( λ ) : Conserved quantitites constructed from ¯ ¯ A ( x − ) .

  13. Quantum integrable structure of Liouville theory — I Quantization: U n , V n → operators, algebra U n V n = q V n U n , q = e πib 2 . Realize on L 2 ( R N ) as U n = e πb (2 x n + p n ) , V n = e πb p n , [ p n , x m ] = (2 πi ) − 1 δ nm . Quantum Lax-matrices: µ − 1 V n U n V n µ − 1 V − 1 � U n + µ ¯ � µ V n + ¯ n L ( λ ) = , µ V − 1 µ − 1 V n U − 1 µ − 1 V − 1 n U − 1 n V − 1 n + ¯ n + µ ¯ n Transfer matrices: T ( λ ) = tr(¯ ¯ L N ( λ ) · · · ¯ T ( λ ) = tr( L N ( λ ) · · · L 1 ( λ )) , L 1 ( λ )) . Mutual commutativity: [ T ( λ ) , ¯ [¯ T ( λ ) , ¯ [ T ( λ ) , T ( λ ′ )] = 0 , T ( λ ′ )] = 0 , T ( λ ′ )] = 0 .

  14. Quantum integrable structure of Liouville theory — II Key tool for study of the spectrum of T ( λ ) , ¯ T ( λ ) : The Baxter Q-operators. Solution to conditions T ( λ ) Q ( λ ) = a ( λ ) Q ( qλ ) + d ( λ ) Q ( q − 1 λ ) � � (i) T ( λ )¯ ¯ a ( λ )¯ Q ( qλ ) + ¯ d ( λ )¯ Q ( q − 1 λ ) Q ( λ ) = ¯ T, ¯ T, Q, ¯ (ii) Q commute for arbitrary values of the spectral parameter. This means that T, ¯ T, Q, and ¯ Q can be simultaenously diagonalized. The eigenvalues T ( λ ) , ¯ T ( λ ) and Q ( λ ) , ¯ Q ( λ ) must satisfy the Baxter equations T ( λ )¯ ¯ a ( λ ) ¯ Q ( qλ ) + ¯ d ( λ ) ¯ T ( λ ) Q ( λ ) = a ( λ ) Q ( qλ ) + d ( λ ) Q ( q − 1 λ ) , Q ( q − 1 λ ) Q ( λ ) = ¯ If we know the Q-operators explicitly ⇒ Analytic and asymptotic properties of Q ( λ ) , ¯ Q ( λ ) ⇒ Definition of the space of admissible solutions of Baxter’s equations ⇒ Quantization conditions.

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend