The problem of maximizing subject to the constraints ,

, and can be done if instead of the AM-GM inequality we use a stronger inequality, called **Jensen’s inequality. It is stated as follows:**

**Theorem. **

Suppose is a twice differentiable, real-valued function on an interval and that for all . Then, for every positive integer m and for all points in , we have

Moreover, equality holds if and only if . A similar result holds if

for all except that the inequality sign is reversed.

What this means is that the value of assumed by the function h at the arithmetic mean of a given set of points in the interval cannot exceed the arithmetic mean of the values assumed by h at these points, More compactly, the value at a mean is at most the mean of values if is positive in the open interval and the value at a mean is at least the mean of values if is negative on it. (Note that is allowed to vanish at one or both the end-points of the interval .)

A special case of Jensen’s inequality is the AM-GM inequality.

Jensen’s inequality can also be used to give easier proofs of certain other trigonometric inequalities whose direct proofs are either difficult or clumsy. For example, applying Jensen’s inequality to the function on the interval one gets the following result. (IITJEE 1997)

If n is a positive integer and for , then

.

More later,

Nalin Pithwa