Complexity in complex analysis. (Q1865277): Difference between revisions

The author had already proved in [J. Anal. Math. 78, 329--344 (1999; Zbl 0934.30005)] that the Bergman kernel for any \(n-\)connected domain \(\Omega\) of the complex plane such that no boundary component is a point can be written as a rational combination of three holomorphic functions in the following sense \[ K(z,w) = A(z) \overline{A(w)} R( G_1 (z), G_2 (z), \overline{G_1 (w)}, \overline{G_2 (w)} ), \] where \(A\), \(G_1\), \(G_2\) are holomorphic functions of one variable and \(R\) is a rational function of four variables. A similar representation holds for the square of the Szegő kernel, \(S(z,w)^2\). We say that those kernels are expressed as a rational combination of only two holomorphic functions in the case where the function \(A\) above reduces to a constant. That property is characterized in a first result: when \(\Omega\) is simply connected, the Bergman (or equivalently, Szegő) kernel can be written as a rational combination of two holomorphic functions if and only if the Riemann mapping from \(\Omega\) to the disc and its derivative are algebraically dependent. In the general case of an \(n-\)connected domain \(\Omega\), this happens if and only if a number of equivalent conditions are met, in particular : (i) there exists a single proper holomorphic map \(f\) from \(\Omega\) onto the unit disc such that \(f\) and \(f'\) are algebraically dependent (or any such map \(f\) verifies this property); (ii) the domain \(\Omega\) can be realized as a subdomain of a compact Riemann surface \(X\) such that the Bergman and Szegő kernels extend to \(X\times X\) as single valued meromorphic functions (which implies that they can be expressed as rational combinations of any two functions on \(\Omega\) which extend to form a primitive pair for \(X\)). The author then widely extends the class of functions which can be combined to produce the standard kernels, by defining a class \(\mathcal A\) which is made up of linear combinations of a number of canonical functions, including derivatives of Green's functions for fixed poles in \(\Omega\) and products of the form \(S(z,a_1)S(z,a_2)\) where \(a_1\) and \(a_2\) are points in \(\Omega\), and canonical functions obtained as the complex derivatives of certain harmonic measure functions. To have a representation as above, it will be enough that \(A \in \mathcal A\) and \(G_1, G_2\) be meromorphic functions extending to \(X\), the double of \(\Omega\), to form a primitive pair on that Riemann surface. Furthermore, any proper holomorphic mapping from \(\Omega\) onto the unit disc can be written as a rational combination of \(G_1, G_2\). Many examples of functions in the class \(\mathcal A\) are given, and in particular, if \(\Omega\) is as above, its Bergman kernel \(K(z,w)\) can be expressed as rational combination of the three functions (i) \(K(z,A_1)\), \(K(z,A_2)\), \(K(z,A_3)\), or (ii) \(S(z,A_1)\), \(S(z,A_2)\), \(S(z,A_3)\), or (iii) \(G_z(z,A_1)\), \(G_z(z,A_2)\), \(G_z(z,A_3)\), where \(A_1\), \(A_2\), \(A_3\) are three fixed points in \(\Omega\) and \(G_z\) denotes the \(z-\)derivative of the Green function \(G(z,w)\). An analogous result holds for the Szegő kernel. Finally, a theorem is given in the case when the boundary of \(\Omega\) consists in finitely many smooth curves, about representation of the Szegő kernel (a function of two complex variables) as a rational combination of a proper holomorphic map from \(\Omega\) to the unit disc and a finite number of \(z-\)derivatives of Szegö kernels \(S(z,a_j)\) and \(S(w,a_j)\) for a finite set of points \(a_j \in \Omega\) (functions of one complex variable). The proofs build upon the methods developed by the author in a monograph [\textit{S. Bell}, ``The Cauchy Transform, Potential Theory, and Conformal Mapping'', CRC Press, Boca Raton, FL, 1992] and a series of previous papers [Duke Math. J. 98, No. 1, 187--207 (1999; Zbl 0948.30015); J. Reine Angew. Math. 525, 1--16 (2000; Zbl 0959.30002); J. Anal. Math. 78, 329--344 (1999; Zbl 0934.30005); Houston J. Math. 26, No. 2, 277--297 (2000; Zbl 0981.30007)] about this type of problems, adding (among other things) some new algebraic ingredients to provide this very general and complete picture of the situation.