= Hardy-Littlewood maximal inequality
{c}
For $f\in L^1(\mathbb R^d)$ and $\lambda>0$, the <Hardy-Littlewood maximal function> satisfies
$$
|\{x:Mf(x)>\lambda\}|\leq\frac{C_d}{\lambda}\|f\|_1.
$$
This is a <weak type (1,1)> bound: it controls the size of a superlevel set rather than asserting $Mf\in L^1$. The <uncentered maximal weak-type inequality> gives, for example, $C_d=3^d$, since the centred maximal function is no larger than its uncentred counterpart.
For the covering argument, take a compact subset of the superlevel set and choose a finite family of balls on which the average of $|f|$ exceeds $\lambda$. The <Wiener covering lemma> selects disjoint balls whose triple dilations cover that compact set. The volumes of the selected balls sum to at most $\|f\|_1/\lambda$. Their triples supply the factor $3^d$, and <inner regularity of Lebesgue measure> removes the compact-subset restriction. This estimate turns a small $L^1$ approximation error into a small exceptional set of bad local averages.
Back to article page