[Paper Review] Trace formula for counting nodal domains on the boundaries of chaotic 2D billiards

Amit Aronovitch, Uzy Smilansky|arXiv (Cornell University)|Jun 29, 2010

Mathematical Dynamics and Fractals2 references3 citations

TL;DR

This paper derives a semi-classical trace formula for counting nodal domains on the boundary of chaotic 2D quantum billiards, linking the boundary intersection count to classical periodic orbits. The formula combines a smooth Weyl-like term with an oscillating part dependent on orbit length, stability, and geometry, validated numerically for the Africa billiard with excellent agreement in the length spectrum.

ABSTRACT

Given a Dirichlet eigenfunction of a 2D quantum billiard, the boundary domain count is the number of intersections of the nodal lines with the boundary. We study the integer sequence defined by these numbers, sorted according to the energies of the eigenfunctions. Based on a variant of Berry's random wave model, we derive a semi-classical trace formula for the sequence of boundary domain counts. The formula consists of a Weyl-like smooth part, and an oscillating part which depends on classical periodic orbits and their geometry. The predictions of this trace formula are supported by numerical data computed for the Africa billiard.

Motivation & Objective

• To develop a trace formula that describes the sequence of boundary nodal domain counts in chaotic 2D quantum billiards.
• To extend semi-classical methods beyond integrable systems by incorporating periodic orbits into nodal domain counting.
• To provide a theoretical framework that captures both the mean value and fluctuations in boundary intersection counts.
• To validate the formula using numerical data from the Africa billiard, demonstrating agreement with semi-classical predictions.

Proposed method

• Derive a density function $ d_{ ho}^{ ext{sm}}(n) $ for the boundary domain count $ \eta_n $, using a smoothed version of the eigenvalue spectrum.
• Construct the oscillating part of the trace formula using periodic orbits, incorporating their length $ L_p $, monodromy matrix $ M_p $, Maslov index $ \nu_p $, and bounce angles $ \psi_i^{(p)} $.
• Define the trigonometric factor $ \Phi_p = \sum_{i=1}^{n_p} (4\cos^2\psi_i^{(p)} - 1) \cdot 2\sin\psi_i^{(p)} $, which suppresses orbits with angles near 60°.
• Use the Gutzwiller trace formula for the spectral density and invert it to express $ k(n) $, enabling transition from $ k $-dependence to $ n $-dependence.
• Apply a Gaussian window function $ W(q) $ and scale the residual to isolate the oscillatory component for Fourier analysis.
• Test the formula by restricting the boundary count to a subset $ \Gamma \subset \partial\Omega $, confirming that only orbits with bounce points in $ \Gamma $ contribute to the length spectrum.

Experimental results

Research questions

• RQ1 How can the sequence of boundary nodal domain counts in chaotic 2D billiards be described using classical periodic orbits?
• RQ2 What is the functional form of the oscillating part of the boundary domain count, and how does it depend on orbit geometry and stability?
• RQ3 Can a trace formula for boundary nodal domains be derived that includes both smooth and fluctuating components, analogous to those in spectral theory?
• RQ4 To what extent do individual periodic orbits contribute to the observed fluctuations in the boundary intersection count?
• RQ5 Does restricting the boundary region of interest suppress contributions from orbits not intersecting that region, as predicted by the theory?

Key findings

• The trace formula (1) accurately predicts the mean value and fluctuations in the boundary nodal domain count $ \eta_n $, with a smooth part proportional to $ \mathcal{L}q/(2\pi) $ and a correction term involving $ \mathcal{L}^2 - 6\pi\mathcal{A} $.
• The oscillating part of the formula depends on periodic orbits through $ \Phi_p $, which suppresses orbits with bounce angles near 60°, consistent with numerical observations.
• Numerical computation of 20,000 eigenfunctions in the Africa billiard shows excellent agreement between the semi-classical length spectrum and the theoretical prediction.
• Orbits with both bounce points in an excluded boundary region $ \partial\Omega \setminus \Gamma $ vanish from the length spectrum of the partial count $ \eta_\Gamma $, confirming the theory’s locality.
• Orbits with only one or a few bounce points in $ \Gamma $ are significantly suppressed, with one such orbit dropping below the noise floor in the numerical spectrum.
• The inclusion of complex periodic orbits with small imaginary parts had negligible effect, indicating that real classical orbits dominate the spectral fluctuations.