STATISTICAL CHARACTERIZATIONS OF GAUSSIAN MEASURES ON A
HILBERT SPACE
Abstract: Let
be i.i.d. random vectors with values in a real separable Hilbert
space. We consider the problem of estimating the mean of
under quadratic loss and
discuss analogues of characteristic properties of normally distributed real random
variables. It is shown that there exists an equivariant sufficient linear statistic iff
is
Gaussian. Further the optimality of the sample mean
in the class of all equivariant
or unbiased estimators is a characteristic property of Gaussian random vectors.
2000 AMS Mathematics Subject Classification: Primary: -; Secondary: -;
Key words and phrases: -