A formula for the local Dirichlet integral (Q1191399)

For a nonnegative finite Borel measure \(\mu\) on the unit circle \(\mathbb{T}\), let \(\varphi_ \mu\) be the associated harmonic function on the unit disc \(\mathbb{D}\) and let the Dirichlet space \(D(\mu)\) be the subspace of \(H^ 2(\mathbb{D})\) of consisting the functions \(f\) for which the integral (1) \(\iint_ \mathbb{D} | f'(z)|^ 2 \varphi_ \mu(z)dA(z)\) is finite. If \(\mu\) is the normalized Lebesgue measure on \(\mathbb{T}\), then \(D(\mu)\) is the classical Dirichlet space. For an \(H^ 1\)-function \(f\) and a fixed \(\zeta\) on \(\mathbb{T}\), the function given by \(| (f(e^{it})- \alpha)/(e^{it} -\zeta)|^ 2\) will be integrable on \(\mathbb{T}\) or at most one complex \(\alpha\). Then \(D_ \zeta(f)\), the local Dirichlet integral of \(f\), is defined to be this finite integral (and to be \(\infty\) if no such \(\alpha\) exists). The authors show that if \(f\) is in \(H^ 2\), then \(\int_ \pi D_ \zeta(f)d\mu(z)\) is equal to the integral (1). J. Douglas used this formula (for the classical Dirichlet space) in his study of minimal surfaces. The authors also obtain a formula for \(D_ \zeta(f)\) which takes account of the canonical factorization of \(f\). Their formula implies that of \textit{L. Carleson} [``A representation theorem for the Dirichlet integral'', Math. Z. 73, 190-196 (1960; Zbl 0090.286)]. We mention two further applications of their results and methods: 1. If \(\varphi\) is a complex valued function on \(\mathbb{D}\) such that \(f\), \(\varphi f\), and \(\varphi^ 2 f\) are in \(D(\mu)\) for some nonzero function \(f\), then the 2-isometric relationship \(\| \varphi^ 2 f\|_ \mu^ 2- \| \varphi f\|_ \mu^ 2= \| \varphi f\|_ \mu^ 2- \| f\|^ 2\) is necessary and sufficient in order that \(\varphi\) is an inner function. 2. A function \(f\) in \(D(\mu)\) is called a cyclic vector if the polynomial multiples of \(f\) are dense in \(D(\mu)\). If \(g\) is cyclic in \(D(\mu)\), if \(f\) is in \(D(\mu)\), and if \(\| f(z)|\geq| g(z)|\) for all \(z\) in \(\mathbb{D}\), then \(f\) must be cyclic. Finally, if \(f\) and \(g\) are bounded functions in \(D(\mu)\), then \(f\) and \(g\) are cyclic in \(D(\mu)\) if and only if \(fg\) is. These results answer questios posed by \textit{L. Brown} and \textit{A. Shields} [Trans. Am. Math. Soc. 285, 269-304 (1984; Zbl 0537.30040)].