Abstract: The
expression for the mutual information for a mixture of two multivariate normal
distributions is derived without the assumption that the two component normals
be widely separated. This is achieved by splitting the sample space of the
distribution into two disjoint subspaces in which the series expansion for is valid.
Keywords and phrases: mutual information,
mixture of two normal distributions, normal linear forms, normal quadratics
forms, moment generating functions, truncated normals, chi-square distribution,
series expansion for
