Solution

ID: past-exam-of-the-mathematics-course-of-the-university-of-cambridge/2019/ib/paper-3/6a/solution

Since
for , the second-derivative criterion shows that is strictly convex. Put , so . For dual variable , the stationary equation in the Legendre transform is
It is the unique global maximizer of , and hence
Applying the same calculation with the conjugate exponent gives
Under the Legendre-Fenchel transform on all real dual variables, the same supremum additionally gives for .
Now take , , and apply the Fenchel–Young inequality . This yields Young's inequality for products
for positive and conjugate exponents .

New to topics? Read the docs here!