[Paper Review] Pleijel's nodal domain theorem for free membranes

Iosif Polterovich|ArXiv.org|May 11, 2008

Quantum chaos and dynamical systems8 references9 citations

TL;DR

This paper proves an analogue of Pleijel's nodal domain theorem for planar domains with Neumann boundary conditions, confirming a 1956 conjecture. By combining Pleijel's original method with a sharp bound on boundary zeros of Neumann eigenfunctions (from Toth–Zelditch), it establishes that the asymptotic upper bound for the ratio of nodal domains to eigenfunction index remains ≤ 4/j₁² ≈ 0.691, even under Neumann conditions.

ABSTRACT

We prove an analogue of Pleijel's nodal domain theorem for piecewise analytic planar domains with Neumann boundary conditions. This confirms a conjecture made by Pleijel in 1956. The proof is a combination of Pleijel's original approach and an estimate due to Toth and Zelditch for the number of boundary zeros of Neumann eigenfunctions.

Motivation & Objective

• To resolve a long-standing conjecture by Pleijel (1956) regarding the asymptotic behavior of nodal domains under Neumann boundary conditions.
• To extend Pleijel’s classical nodal domain theorem—originally proven for Dirichlet conditions—to the Neumann case.
• To address the technical challenge that nodal domains adjacent to the boundary do not satisfy pure Dirichlet conditions, making standard Faber–Krahn inequalities inapplicable.
• To establish that the limsup of the nodal domain count relative to eigenvalue index remains bounded by 4/j₁² under Neumann conditions, despite this added complexity.

Proposed method

• Decompose the nodal domains into two classes: those adjacent to the boundary (∂Ω) and those with pure Dirichlet conditions on their boundary.
• Use the Toth–Zelditch estimate to bound the number of boundary zeros of Neumann eigenfunctions by O(√λₖ), which controls the number of boundary-adjacent nodal domains.
• Apply Weyl’s law (λₖ/k → 4π/Area(Ω)) to show that the proportion of boundary-adjacent nodal domains vanishes asymptotically (limsup mₖ/k = 0).
• For non-adjacent nodal domains, apply the Faber–Krahn inequality: λ₁(D)Area(D) ≥ πj₁², where j₁ ≈ 2.4 is the first zero of J₀.
• Sum the Faber–Krahn inequality over all non-adjacent nodal domains to derive an upper bound on lₖ, the count of such domains.
• Combine the vanishing proportion of boundary-adjacent domains with the bounded ratio for interior domains to conclude the full limsup bound.

Experimental results

Research questions

• RQ1 Can Pleijel’s nodal domain theorem, originally proven for Dirichlet boundary conditions, be extended to Neumann boundary conditions?
• RQ2 What is the asymptotic behavior of the number of nodal domains relative to the eigenfunction index for Neumann eigenfunctions on planar domains?
• RQ3 How does the presence of boundary-adjacent nodal domains—where the boundary condition is mixed (Neumann on ∂Ω, Dirichlet on the nodal domain boundary)—affect the nodal domain count?
• RQ4 Can the bound 4/j₁² ≈ 0.691 be preserved under Neumann conditions, despite the failure of the Faber–Krahn inequality for boundary-adjacent domains?
• RQ5 Is the limsup of nₖ/k strictly less than 1 for Neumann eigenfunctions, and can this bound be quantitatively estimated?

Key findings

• The proportion of nodal domains adjacent to the boundary satisfies limsupₖ→∞ mₖ/k = 0, due to the O(√λₖ) bound on boundary zeros from Toth–Zelditch.
• The number of interior nodal domains (with pure Dirichlet conditions) satisfies limsupₖ→∞ lₖ/k ≤ 4/j₁² ≈ 0.691, via the Faber–Krahn inequality and Weyl’s law.
• The total number of nodal domains nₖ = mₖ + lₖ satisfies limsupₖ→∞ nₖ/k ≤ 4/j₁², confirming Pleijel’s bound for Neumann eigenfunctions.
• The result holds for piecewise real analytic planar domains satisfying the internal cone condition, ensuring discrete spectrum.
• The proof establishes that the classical Courant bound (nₖ ≤ k) is asymptotically improvable under Neumann conditions, just as in the Dirichlet case.
• The bound 4/j₁² is not sharp; the paper suggests the optimal bound may be 2/π ≈ 0.636, which would be attained for separable eigenfunctions on rectangles.