[Paper Review] Singular solutions to the Loewner equation

Dmitri Prokhorov, Alexander Vasil’ev|ArXiv.org|Jul 3, 2008

Advanced Differential Equations and Dynamical Systems6 references3 citations

TL;DR

This paper investigates the conditions under which the Löwner equation generates one-slit maps, focusing on singular solutions with continuous driving terms. It establishes that non-slit solutions arise when the associated subordination chain becomes singular—adding non-zero area at a moment—generalizing Kufarev's example and showing that slit solutions require specific asymptotic behavior of the driving term and slit length, with arc length growing as $ s(t) \sim A\sqrt{t} $ near $ t=0 $. The key result identifies necessary and sufficient conditions on the driving term and slit geometry for one-slit behavior.

ABSTRACT

We consider the Löwner differential equation generating univalent self-maps of the unit disk (or of the upper half-plane). If the solution to this equation represents a one-slit map, then the driving term is a continuous function. The reverse statement is not true in general as a famous Kufarev's example shows. We address the following main problem: to find a criterium for the Löwner equation to generate one-slit solutions. New examples of non-slit solutions to the Löwner equation are presented. Properties of singular slit solutions are revealed.

Motivation & Objective

• To determine the conditions under which the Löwner equation with a continuous driving term generates a one-slit map, addressing the reverse of the classical result where continuous driving terms imply slit maps.
• To analyze Kufarev’s example and construct new examples of non-slit solutions with continuous driving terms, revealing the role of singular subordination chains.
• To compare the dynamics of the Löwner ODE with the Löwner PDE, showing that the PDE framework exhibits different behavior, particularly in the presence of singularities.
• To characterize the asymptotic behavior of the slit length and driving term near $ t=0 $, establishing sharp growth rates for the arc length and the driving term.
• To prove that for analytic slits, the only possible asymptotic behavior yielding a one-slit solution is $ \lambda(t) \sim 0 $, $ h^+(0,t) \sim 2\sqrt{t} $, with $ s(t) \sim A\sqrt{t} $, and that $ s(t)t^{-1/2+\epsilon} \to 0 $ as $ t \to 0^+ $.

Proposed method

• Uses the Löwner ODE as a characteristic equation for the Löwner PDE, analyzing solutions via subordination chains in the upper half-plane.
• Applies the hydrodynamic normalization $ h(z,t) = z + \frac{2t}{z} + O(1/z^2) $ near infinity to study the evolution of slit domains.
• Employs a conformal map $ \zeta = \sqrt{z^2 - 1/4} $ to transform the slit domain into a standard form, enabling analysis of arc-length parameters.
• Analyzes the behavior of the inverse function $ h^{-1}(w,t) $ near the slit tip, deriving bounds on the derivative to estimate the slit length.
• Uses asymptotic expansions and estimates involving coefficients $ b_n(t) $ of the power series of $ h^{-1}(w,t) $ to bound the slit length $ s(t) $.
• Compares the behavior of the arc-length parameter $ \sigma $ in the transformed domain with $ s(t) $, showing $ s^2 \sim \sigma - 1/2 $ near $ \sigma = 1/2 $.

Experimental results

Research questions

• RQ1 Under what conditions on the driving term does the Löwner equation generate a one-slit map, despite the existence of continuous-driving-term solutions that are not slit maps?
• RQ2 How does the singularity of the subordination chain (i.e., non-zero area added at a moment) relate to the non-slit nature of the solution?
• RQ3 What is the precise asymptotic behavior of the slit length $ s(t) $ as $ t \to 0^+ $ for analytic slits, and how does it constrain the driving term?
• RQ4 Why does the sharp bound of 4 for the 1/2-norm of the driving term (from Lind) not fully characterize one-slit solutions, given that Kufarev’s example exceeds this bound?
• RQ5 How does the behavior of the Löwner ODE differ from that of the Löwner PDE in the presence of singularities in the subordination chain?

Key findings

• The only possible asymptotic behavior for a one-slit solution with analytic slit is $ \lim_{t\to 0^+} \frac{\lambda(t)}{\sqrt{t}} = 0 $ and $ \lim_{t\to 0^+} \frac{h^+(0,t)}{\sqrt{t}} = 2 $, with $ s(t) \sim A\sqrt{t} $, $ A \neq 0 $.
• For any $ \epsilon > 0 $, the slit length satisfies $ \lim_{t\to 0^+} s(t) t^{-1/2 + \epsilon} = 0 $, indicating that $ s(t) $ grows slower than any power exceeding $ \sqrt{t} $.
• The bound $ \|u\|_{1/2} \leq 4 $ from Lind is sharp, but Kufarev’s example with $ \|u\|_{1/2} = 3\sqrt{2} \approx 4.24 $ shows that this norm alone does not guarantee a one-slit solution.
• Non-slit solutions arise when the subordination chain becomes singular—i.e., when a non-zero area is added at a finite time—indicating a breakdown in the slit structure.
• The arc-length parameter $ \sigma $ in the transformed domain satisfies $ \sigma - 1/2 \sim A_1 t $, which implies $ s^2 \sim A_1 t $, so $ s(t) \sim \sqrt{A_1 t} $, confirming the $ \sqrt{t} $-growth.
• The result generalizes previous findings on rectilinear and circular slits, showing that the $ \sqrt{t} $-growth of the slit length is universal for analytic slits under the given normalization.