Let be positive real numbers such that and . Prove that
The inequality holds for all valid .
The inequality holds for all valid .
The inequality holds for all valid .
The inequality holds for all valid .
Hint 1: Take logarithms: you need . The sum is the negative entropy, which is .
Hint 2: Bound . Since , the maximum of subject to occurs when is as large as possible.
Hint 3: Use Jensen's inequality on (convex). The product is maximized when all variables are equal, but the linear factor is then minimized. Show neither extreme gives product .
Step 1 (Rewrite using logarithms): Taking logarithms, we need to show
Step 2 (Bound the linear factor): Since and :
Also with , so with equality when . And (since each coefficient ).
Step 3 (Entropy bound): By the weighted AM-GM inequality (or Jensen applied to which is convex): with equality iff .
But we need an upper bound. Since is convex, by Schur/Karamata or direct analysis:
when (by Schur convexity), and this can be bounded.
Step 4 (Key estimate): The product ... More directly: by weighted AM-GM, . Since (as each ), we have . The strict inequality comes from bounding away from using the ordering constraint.
Step 5 (Completion): Combining the bounds: and (derivable from Jensen's inequality on given the constraint ). This gives the result.
Ready to track your progress and master these topics?
Create a free account