On the polynomial convexity of the union of two maximal totally real subspaces of \(\mathbb C^n\) (Q1093053)

An example due to Burns has shown that the union of two totally real planes which meet only at \(0\) in \({\mathbb C}^ 2\) need not have a polynomially convex intersection with the unit ball. This paper contains a complete description of those pairs of totally real \(n\)-dimensional subspaces of \({\mathbb C}^ n\) which meet only at \(0\) whose union has the property that every compact subset is polynomially convex. After a change of coordinates one can reduce to the case that the pair is \(({\mathbb R}^ n,(A+i){\mathbb R}^ n)\) where \(A\) is a real \(n\times n\) matrix such that \((A+i)\) is non-singular. The paper \(({\mathbb R}^ n,(A+i){\mathbb R}^ n)\) has the stated property if and only if \(A\) has no eigenvalue of the form \(t_i\), where \(t>1\). In this case, every continuous function on \({\mathbb R}^ n\cup (A+i){\mathbb R}^ n=X\) can be approximated uniformly on compact sets by polynomials. If \(A\) has \(k\) eigenvalues of the form \(t_i\), \(t>1,\) and if each such eigenvalue is of simple type then there is a \(k\)-parameter family of polyannular regions in \({\mathbb C}^ n\) whose distinguished boundaries lie in \(X\) such that every continuous function on \(X\) which extends analytically to each polyannulus is the uniform limit on compact subsets of \(X\) of polynomials.