[Paper Review] Scattering theory on graphs

Christophe Texier, Gilles Montambaux|HAL (Le Centre pour la Communication Scientifique Directe)|Jul 5, 2001

advanced mathematical theories4 citations

TL;DR

This paper develops a systematic scattering theory framework for the Schrödinger operator on metric graphs with potentials, deriving two equivalent expressions for the scattering matrix: one using arc-based matrices and another using vertex-based matrices. The key contribution is a compact, general formalism that incorporates vertex coupling parameters and potential scattering, enabling efficient analysis of quantum transport and spectral properties in disordered or mesoscopic networks with arbitrary topology and local potentials.

ABSTRACT

We consider the scattering theory for the Schrödinger operator $-\Dc_x^2+V(x)$ on graphs made of one-dimensional wires connected to external leads. We derive two expressions for the scattering matrix on arbitrary graphs. One involves matrices that couple arcs (oriented bonds), the other involves matrices that couple vertices. We discuss a simple way to tune the coupling between the graph and the leads. The efficiency of the formalism is demonstrated on a few known examples.

Motivation & Objective

• To establish a general and systematic formalism for scattering theory on metric graphs with local potentials.
• To derive two equivalent expressions for the scattering matrix—one based on arcs (oriented bonds) and another on vertices—enabling flexible analysis of quantum transport.
• To incorporate tunable coupling between the graph and external leads through adjustable vertex parameters, enhancing physical relevance for mesoscopic systems.
• To generalize existing results for the Laplacian on graphs to the Schrödinger operator with potential, including scattering and spectral determinant relations.
• To provide a compact, computationally efficient framework for studying quantum chaos, weak localization, and Aharonov-Bohm effects in networked systems.

Proposed method

• Define the Schrödinger operator $ H = -\partial_x^2 + V(x) $ on a metric graph composed of one-dimensional wires connected at vertices with specified boundary conditions.
• Use continuity of wavefunctions and current conservation (via a vertex parameter $ \lambda_\alpha $) to formulate the scattering problem at vertices.
• Derive the scattering matrix using arc matrices that encode transmission and reflection coefficients between connected bonds, with phase shifts from potential and magnetic flux.
• Reformulate the scattering matrix using vertex matrices $ M $, which compactly encode the graph's topology and local scattering properties, with $ M $ shown to be anti-Hermitian.
• Introduce a tunable coupling between leads and the graph via adjustable $ \lambda_\alpha $, allowing control over vertex transparency and enabling realistic modeling of external contacts.
• Apply the formalism to known examples, including single-barrier scattering and looped graphs, and derive explicit expressions for $ M $ in terms of transmission, reflection, and phase parameters.

Experimental results

Research questions

• RQ1 How can the scattering matrix of a quantum graph with local potentials be systematically derived using both arc-based and vertex-based formulations?
• RQ2 What is the role of the vertex coupling parameter $ \lambda_\alpha $ in controlling the transparency of vertices and enabling tunable coupling to external leads?
• RQ3 How do transmission, reflection, and phase shifts from local potentials and magnetic fluxes affect the scattering matrix in a general graph?
• RQ4 Can the formalism be extended to describe loops and self-interacting bonds in a minimal vertex representation?
• RQ5 What is the relationship between the scattering matrix and spectral determinants, and how does it generalize known results for the Laplacian on graphs?

Key findings

• The scattering matrix is derived in two equivalent forms: one using arc matrices that couple oriented bonds, and another using vertex matrices $ M $, which offer a more compact and physically intuitive representation.
• The vertex matrix $ M $ is shown to be anti-Hermitian ($ M^\dagger = -M $), ensuring unitarity of the scattering matrix and consistency with current conservation.
• For a single barrier on a bond, the scattering matrix is expressed in terms of transmission $ T_{\alpha\beta} \in [0,1] $, phase $ \Phi_{\alpha\beta} $, and magnetic flux $ \theta_{\alpha\beta} $, with $ T_{\alpha\beta} = T_{\beta\alpha} $ and $ \theta_{\alpha\beta} = -\theta_{\beta\alpha} $.
• In the limit of zero potential ($ V=0 $), the formalism reduces to the standard free-particle scattering with $ T_{\alpha\beta} = 1 $ and $ \Phi_{\alpha\beta} = k l_{\alpha\beta} $, recovering known results.
• For graphs with loops, the diagonal element $ M^\text{loop}_{\alpha\alpha} $ is derived explicitly, accounting for both forward and backward scattering on the loop, and expressed in terms of reflection and transmission coefficients.
• The loop contribution simplifies to $ M^\text{loop}_{\alpha\alpha} = 2i \frac{\cos\Phi_a - \sqrt{T_a}\cos\theta_a}{\sin\Phi_a - \sqrt{1-T_a}\cos\varphi_a} $, showing explicit dependence on potential asymmetry and magnetic flux.