Optimization Techniques

S. K. Neogy
Indian Statistical Institute Delhi Centre

1
 Some Generalizations

In our version of the problem we chose n = 3 and p = 2.

However, before doing so, it is useful to first solve (1) with p = 2 and for any n.
Hence we wish to solve

2
We shall prove that (3) is correct by induction. We know that the hypothesis is true
for j = 1, 2, 3. Now we assume that it is true for j = s i.e.,

3
4
5
6
7
8
This completes the induction. Therefore we have established that the solution to (1)

9
Therefore we were able to use the calculus to find that optimum. 10
We turn now to the case where 0 < p < 1. For this case, it can be shown that is a
concave function and since the sum of concave functions is concave, the
objectivefunction is concave.
It is also known that at least one global minimum of a concave function over a
closed convex set will be one of the extreme points of the convex set. The convex set
we will be dealing with is

11
12
13
14
15
Some Generalizations

16
17
18
Hence we see that when maximizing convex functions in the
recurrence relations, the analysis of the problem is simpler and more
general.
19
20
 A Problem in Division

We are given a known and positive quantity b which is to be divided into n

parts in such a way that the product of the n parts is to be a maximum. We
shall solve this problem using a dynamic programming approach.
First we note that n is unspecified. Let us consider the simplest
problem. That is surely the case of n = 1. For this problem, the
maximum value of the product is b . Hence we
define

21
22
23
24
25
26
27
28
29
30
31
32
33