The Szegö kernel and oblique projections: conformal mapping of non-smooth regions (Q6181165)

It is a well-known fact that one can explicitly express the Riemann mapping function through the Szego kernel of a bounded domain with smooth a boundary [\textit{N. Kerzman} and \textit{E. M. Stein}, Math. Ann. 236, 85--93 (1978; Zbl 0419.30012)]. Kerzman-Trummer method [\textit{N. Kerzman} and \textit{M. R. Trummer}, J. Comput. Appl. Math. 14, 111--123 (1986; Zbl 0582.30007); \textit{M. R. Trummer}, SIAM J. Numer. Anal. 23, 853--872 (1986; Zbl 0596.30013)] computes the Riemann mapping functions \(R:\; \Omega \longrightarrow D\) using the fact that the Szego kernel \(S\) for domains \(\Omega \) with smooth (at least \(\mathcal{C}^2\) boundaries) satisfies a Fredholm integral equation of the second kind with a compact kernel function. Later on, \textit{A. D. Thomas} [J. Math. Anal. Appl. 200, No. 1, 162--181 (1996; Zbl 0846.30005)], starting with the work of S.Bell, derives formulas for the Szego kernel for non-smooth domains \(\Omega \) and determines explicitly the corresponding conformal mapping. The present article addresses this same problem for convex domains with corners, introducing a new, more efficient numerical method. Using the idea of a smoothing map, the author introduces a new integral Equation (11) for domains with corners. The solution of (11) is the corresponding Szego kernel. Via the classical formula (1) the author directly derives from the Szego kernel the Riemann mapping function \(R: \Omega\longrightarrow D\). The essential ingredient responsible for the novelty of (11) is a careful choice of the smoothing function \(\mu \). This choice requires that \(\mu(\partial \Omega)\) has no corners and remains 1-1 near the corner. Lemma 1 in the paper illustrates the construction of \(\mu\) when \(\Omega \) is a convex polygon. The method can be extended to domains with piece-wise boundaries. Also, the observation that \(R(z(t))\) is orthogonal to \(R^{\prime}(z)\), where \(z(t)\) is a regular parametrization of the boundary \(\partial \Omega \) is crucial. The new integral equation results in faster convergence rates for the numerical solution.