From points to probability measures A finite number of points x1, x2, . . . , xk
in Rn can be regarded as a discrete probability measure with each point having probability 1/k. It is natural to define a hub for a probability measure ρ as a point having
minimal expected distance to points distributed according to ρ. Then Theorem 1 has
the following generalization and can be proved in essentially in the same way. (An even more general result can be found in [10].)