logo

Proof of Young's Inequality 📂Lemmas

Proof of Young's Inequality

Theorem

Given constants 1p+1q=1\displaystyle {{1} \over {p}} + {{1} \over {q}} = 1 that satisfy the condition and two positive numbers p,qp,q and a,ba,b which are greater than 1,

abapp+bqq ab \le { {a^{p}} \over {p} } + {{b^{q}} \over {q}}

Description

Apart from the aesthetically pleasing algebraic aspect, this inequality is not often mentioned except when proving the Hölder inequality.

Proof

Since both aa and bb are positive, there exists a real number A,BA,B that satisfies a=eA,b=eBa = e^A, b = e^B.

Convexity and the Second Derivative

Let’s say that ff is twice differentiable at II. For ff to be convex at II, f(x)>0f '' (x) >0 is a necessary and sufficient condition.

Meanwhile, since ex>0 e^x>0, the second derivative is always positive, therefore, it is convex at R\mathbb{R}.

Jensen’s Inequality

For a convex f:IRf : I \to \mathbb{R} at IRI \subset \mathbb{R} and for k=1nλk=1,λk>0\displaystyle \sum_{k=1}^{n} \lambda_{k} = 1, \lambda_{k}>0, f(λ1x1+λ2x2++λnxn)λ1f(x1)+λ2f(x2)++λnf(xn)f( \lambda_{1} x_{1} + \lambda_{2} x_{2} + \cdots + \lambda_{n} x_{n} ) \le \lambda_{1} f( x_{1}) + \lambda_{2} f( x_{2}) + \cdots + \lambda_{n} f( x_{n} ).

By Jensen’s inequality,

e1ppA+1qqB1pepA+1qeqB e^{{{1} \over {p}} p A + {{1}\over {q}} q B} \le {{1} \over {p}} e^{pA} + {{1} \over {q}} e^{qB}

In summary, eA+B1p(eA)p+1q(eB)q e^{A+B} \le {{1} \over {p}} (e^{A})^p +{{1} \over {q}} (e^{B})^q

Since it was a=eA,b=eBa = e^A, b = e^B,

abapp+bqq ab \le { {a^{p}} \over {p} } + {{b^{q}} \over {q}}