We can now use the imputed profit vector at any time in order to make a choice from the set of efficient points. For example, given competing quantizers , let be a set of allocation processes holding Axiom 1 through Axiom 4. From Axiom 1 and Axiom 3 we have that is a convex set of allocation processes. Hence, by Theorem 1, the concept of ``efficient allocation process'' is now characterized by profit maximization. That is, maximization of with respect to over the set of allocation processes at any given time . Given that is a compact set (by Axiom 1 and Axiom 3), the existence of solution for this maximization problem is guaranteed by the Weierstrass Theorem since the inner product is a continuous function.
From Axiom 1 we have that
Since
we then have that assigns bits to quantizer ,
, with a total bit consumption of
, assuming
:
Then, it follows that the bit allocation set
Then the problem of finding
which maximizes
We shall consider an example of these problems. It solves the bit allocation analysis among three quantizers at a certain time . We have that the respective cost of quantization for each quantizer is: bpp; bpp; and bpp.
The three basic allocation processes are (Table I):
(4) |
For this example we assume that the imputed profit vector is
For this example the bit allocation analysis can be represented by:
By Theorem 1, the efficient combination of allocations up to the bit resource limitation at this time is the solution of this linear programming problem (where represents the efficient allocation for quantizer ): (i) bits; (ii) bits; and (iii) bits. No desired bit allocation for a particular quantizer can be increased without decreasing other desired quantizer allocation or increasing bit consumption, at the given profit vector.