The number of phases used in the Complete solution is characterized by the following theorem.
Proof:If
, the number of phases used in the Complete solution
is at most that used in the Simple solution, i.e.
.
Thus, it suffices to prove that if a distribution
,
then at most
phases are needed (recall
by Lemma 1).
If
, then
.
Thus, by (2.7), the number of phases used is
.
If
,
then the number of phases given by (2.8) is exactly
,
except for input distribution
with
and
(i.e. exponential
distribution possibly mixed with
) for which (2.8) gives 1.
Thus, the number of phases used in the Complete solution
for
is
. width 1ex height 1ex depth 0pt
Theorem 6 implies that any distribution in
is well-represented by an EC distribution of
phases.
In the proof, we prove a stronger property that any distribution in
, which is a superset of
, is well-represented
by an EC distribution of
phases, which implies the
following corollary: