Young's Inequality: Let p,q>1p,q>1 satisfy 1p+1q=1\frac1p+\frac1q=1. Then for all x,yRx,y\in\mathbb{R},

xyxpp+yqq.|xy|\le \frac{|x|^p}{p}+\frac{|y|^q}{q}.
Examples
  • If p=q=2p=q=2, then xyx22+y22|xy|\le \frac{x^2}{2}+\frac{y^2}{2}.
  • If p=3p=3 and q=32q=\frac32, then xyx33+2y3/23|xy|\le \frac{|x|^3}{3}+\frac{2|y|^{3/2}}{3}.
Remarks

This inequality is a standard tool behind , , and many estimates in . In the lecture notes it is obtained from the applied to a=xpa=|x|^p and b=yqb=|y|^q.

A small coefficient for absorption

Replacing xx by ε1/px\varepsilon^{1/p}x and yy by ε1/py\varepsilon^{-1/p}y gives, for ε>0\varepsilon>0,

xyεpxp+εq/pqyq.|xy|\le\frac{\varepsilon}{p}|x|^p+ \frac{\varepsilon^{-q/p}}q|y|^q.

For quadratic estimates one often writes the equivalent form xyεx2+y2/(4ε)|xy|\le\varepsilon x^2+y^2/(4\varepsilon), which also follows by expanding (εxy/(2ε))20(\sqrt\varepsilon |x|-|y|/(2\sqrt\varepsilon))^2\ge0.