Postgraduate Seminar: Jenny Zhu -- Investigating the smallest class which a Gaussian location-mixture density belongs to

Abstract: Classes of Gaussian location-mixture densities of a fixed component
variance are nested and decreasing. As a consequence, any density
belonging to such a class (which can be shown to be quite rich) may be
reexpressed as a mixture density with a smaller component variance.
Whether or not it is possible to reexpress a density using a larger
component variance is a question which currently does not have a
satisfactory answer.