[Paper Review] Integrable models with boundaries and defects

E. Corrigan|ArXiv.org|Nov 11, 2004

Nonlinear Waves and Solitons23 references3 citations

TL;DR

This paper investigates integrable field theories with boundaries and defects, focusing on how boundary conditions preserve integrability in massive scalar and affine Toda models. It derives exact boundary potentials for the sinh-Gordon and affine Toda models using algebraic constraints from conserved charges, revealing that integrability restricts boundary parameters to discrete sets—especially in simply-laced Lie algebras—where $ b_i^2 = 4n_i $, eliminating continuous freedom and yielding a finite, exact set of allowed boundary conditions.

ABSTRACT

Two lectures given at the UK-Japan Winter School on 'Geometry and Analysis Towards Quantum Theory', Durham, January 2004.

Motivation & Objective

• To understand how integrability is preserved in one-dimensional field theories when boundaries or defects are introduced.
• To derive the precise form of boundary potentials that maintain integrability in affine Toda field theories.
• To investigate the constraints on boundary parameters, especially in simply-laced Lie algebras, and explain their origin through conserved charges.
• To clarify the role of boundary conditions in preserving the algebraic structure of integrable models, particularly through the reflection factor and level-matching conditions.

Proposed method

• Uses the Euler-Lagrange equations applied to a Lagrangian with a boundary delta-function term to derive boundary conditions for scalar fields.
• Applies the condition of integrability by requiring the existence of conserved charges that commute with the Hamiltonian, leading to algebraic constraints on boundary data.
• Derives the boundary potential $ ilde{eta}( heta) = \frac{2}{\beta^2}(b_1 e^{\beta\phi/2} + b_0 e^{-\beta\phi/2}) $ for sinh-Gordon models, ensuring compatibility with bulk integrability.
• Imposes level-matching conditions on the Lax connection components, leading to the equation $ [k_3, m_i E_{-\alpha_i}]_- = [k_1, m_i E_{\alpha_i} + \frac{b_i}{24}[k_1, \alpha_i \cdot \mathbf{H}]_-]_- $, which constrains the boundary parameters.
• Analyzes the resulting algebraic equations using Lie algebra commutation relations, particularly for simply-laced root systems, to derive the condition $ b_i^2 = 4n_i $.
• Performs a case-by-case analysis of the Dynkin diagrams for $ a_r, d_r, e_r $, showing that only discrete choices of $ b_i $ preserve integrability.

Experimental results

Research questions

• RQ1 What form must the boundary potential take to preserve integrability in affine Toda field theories?
• RQ2 Why do the boundary parameters become quantized in simply-laced Lie algebras, and what is the origin of the condition $ b_i^2 = 4n_i $?
• RQ3 How do the conserved charges of the bulk theory constrain the allowed boundary conditions?
• RQ4 What is the role of the reflection factor in boundary integrable models, and how does it relate to the boundary potential?
• RQ5 Why is the number of free parameters in boundary conditions drastically reduced in simply-laced cases compared to non-simply-laced ones?

Key findings

• For the sinh-Gordon model, the boundary potential must be $ \tilde{\cal B}(\phi) = \frac{2}{\beta^2}(b_1 e^{\beta\phi/2} + b_0 e^{-\beta\phi/2}) $ to preserve integrability, with $ b_0, b_1 $ arbitrary real constants.
• In simply-laced affine Toda models (e.g., $ a_r, d_r, e_r $), the boundary parameters are quantized such that $ b_i^2 = 4n_i $ for each $ i=0,1,\dots,r $, eliminating continuous freedom.
• The condition $ b_i^2 = 4n_i $ arises from requiring the vanishing of coefficients of non-root generators in the level-matching equation, enforced by Lie algebra commutation relations.
• For the $ a_1 $ case, no level-three roots exist, so the constraint does not apply, and the parameter space remains continuous.
• In non-simply-laced cases, some free parameters may remain, but the set of allowed boundary conditions is still severely restricted.
• The reflection factor for the free massive scalar field is $ R(k) = \frac{ik + \lambda}{ik - \lambda} $, derived from the boundary condition $ \partial_x\phi = -\lambda\phi $ at $ x=0 $.