Unimodality conditions for Gaussian sums
Alkis Konstantellos · IEEE Transactions on Automatic Control · 1980
In this correspondence multivariable sufficient conditions for the unimodality of Gaussian sums (normal mixtures) with two terms are given. The first corresponds to equal proportions and covariance matrices forx \in R^{n}while the second yields to a general mixture andx \in R^{2}. These conditions may be useful in proving nonhomogeneity of random vectors in the reduction of the number of terms encountered in specific non-Gaussian estimation or in the computation of MAP estimates.