“…For a given partition A k ∈ P k , the optimization problem reduces to the maximization of k continuous functions, i.e., Π A i (P i ) with respect to P i for i = 1, ..., k. It suces now to apply lemma 1 for each i = 1, ..., k to obtain a solution. Since the total number of partitions is nite, an optimal prot policy must exist, i.e., a solution to the optimization problem (12) exists. Let (A * k ; P * k ) denotes this optimal prot policy and let us assume for the moment it is unique.…”