Factorization of solutions of convolution equations. II (Q1316619)

The main result in the paper is the following: Let \(\mu_ 1,\mu_ 2,\mu: \widehat\mu= \widehat\mu_ 1\cdot \widehat\mu_ 2\) be analytic functionals supported on convex compact sets \(K_ 1\), \(K_ 2\) and \(K\) respectively (\(\widehat\mu\) stands for the Fourier-Borel transform of \(\mu\)). \(\Omega\supset K\) is an open convex domain. Let \(W_ 1\), \(W_ 2\), \(W\) denote respectively the closed subspaces of the solutions to the convolution equations \[ \mu_ 1 * f=0,\quad \mu_ 2* f=0,\quad \mu* f= 0, \] where \(f\in H(\Omega):=\) holomorphic functions in \(\Omega\). Assume that \(\widehat\mu_ 2\) is an entire function (of exponential type, of course) of completely regular growth. Then every solution \(f\in H(\Omega)\) of the equation \(\mu* f=0\) has the representation \(f= f_ 1+ f_ 2\), where \(f_ i\) denotes a solution of the equation \(\mu_ i* f= 0\), \(i= 1,2\) if and only if the convolution operator \(f\mapsto \mu_ 2* f\) restricted to the subspace \(W_ i\) maps it onto itself. This and other theorems in the paper extend to the \(H(\Omega)\) setting previous results of a similar nature by the same authors obtained for the space \(H(C):=\) entire functions.