A function algebra providing new Mergelyan type theorems in several complex variables (Q2656152)

The authors deal with the following important problem in approximation theory: to get Mergelyan-type theorems for several complex variables. To describe the results, we use the following notations: given a compact subset \(K\) of \(\mathbb{C}^n\), \begin{itemize} \item let \(P(K)\) and \(R(K)\) be the algebras of complex functions that are uniform limits on \(K\) of polynomials and rational functions, respectively; \item let \(\overline{\mathcal{O}}(K)\) be the algebra of complex functions that are uniform limits on \(K\) of holomorphic functions in a neighbourhood of \(K\). \end{itemize} Mergelyan's original result states that if \(n=1\) and \(K\) has connected complement, then the algebra \(P(K)\) coincides with the algebra of functions which are continuous on \(K\) and analytic in the interior of \(K\), denoted by \(A(K)\). With this in mind, the extensions of Mergelyan theorem considered in the paper are theorems giving sufficient conditions to assure the coincidence of the algebra \(A(K)\) with \(P(K)\), \(R(K)\) or \(\overline{\mathcal{O}}(K)\). The authors introduce a novel approach to the subject: instead of looking for characterizations of the compact sets \(K\) where the above classes coincide, they focus on understanding the set of approximable functions. In order to do that, they introduce a new algebra, \(A_D(K)\), that consits of the functions which are continuous on \(K\) and analytic on each analytic disk contained in \(K\). This new framework allows them to prove new approximation theorems. The authors also give a counterexample (very simple in its nature) to an assertion stated by \textit{T. W. Gamelin} and \textit{J. Garnett} in [Trans. Am. Math. Soc. 143, 187--200 (1969; Zbl 0193.03501), Cor. 6.4], which was already used in the later articles [\textit{J. Falcó} and \textit{V. Nestoridis}, Rev. R. Acad. Cienc. Exactas Fís. Nat., Ser. A Mat., RACSAM 111, No. 4, 1147--1155 (2017; Zbl 1380.32004); Stud. Math. 248, No. 1, 93--107 (2019; Zbl 1423.32006); \textit{P. M. Gauthier} and \textit{V. Nestoridis}, Math. Anal. Appl. 433, No. 1, 282--290 (2016; Zbl 1331.46044)], and they show that these later articles can be justified using the results of the present paper. The main results of the paper are Theorems 3.6 and 3.7. The first one states that given a compact set \(K_1\subset \mathbb{C}^n\) with \(A_D(K_1) = \overline{\mathcal{O}}(K_1)\), and a function holomorphic in the interior of \(K_1\) and continuous up to the boundary, then for its graph \(K\) we have equality between \(A(K)\) and \(\overline{\mathcal{O}}(K)\), which is a Mergelyan type theorem. Moreover, they extend the definition of the algebra \(A_D(K)\) to arbitrary products and prove other remarkable result on the density of finite sums or products of certain rational functions in the algebra \(A_D(K)\), where the compact \(K\) is an infinite cartesian product of planar compact sets.