Rising sun lemma

From HandWiki
An illustration explaining why this lemma is called "Rising sun lemma".

In mathematical analysis, the rising sun lemma is a lemma due to Frigyes Riesz, used in the proof of the Hardy–Littlewood maximal theorem. The lemma was a precursor in one dimension of the Calderón–Zygmund lemma.[1]

The lemma is stated as follows:[2]

Suppose g is a real-valued continuous function on the interval [a,b] and S is the set of x in [a,b] such that there exists a y∈(x,b] with g(y) > g(x). (Note that b cannot be in S, though a may be.) Define E = S ∩ (a,b).
Then E is an open set, and it may be written as a countable union of disjoint intervals
[math]\displaystyle{ E=\bigcup_k (a_k,b_k) }[/math]
such that g(ak) = g(bk), unless ak = aS for some k, in which case g(a) < g(bk) for that one k. Furthermore, if x ∈ (ak,bk), then g(x) < g(bk).

The colorful name of the lemma comes from imagining the graph of the function g as a mountainous landscape, with the sun shining horizontally from the right. The set E consist of points that are in the shadow.

Proof

We need a lemma: Suppose [c,d) ⊂ S, but d ∉ S. Then g(c) < g(d). To prove this, suppose g(c) ≥ g(d). Then g achieves its maximum on [c,d] at some point z < d. Since zS, there is a y in (z,b] with g(z) < g(y). If yd, then g would not reach its maximum on [c,d] at z. Thus, y ∈ (d,b], and g(d) ≤ g(z) < g(y). This means that dS, which is a contradiction, thus establishing the lemma.

The set E is open, so it is composed of a countable union of disjoint intervals (ak,bk).

It follows immediately from the lemma that g(x) < g(bk) for x in (ak,bk). Since g is continuous, we must also have g(ak) ≤ g(bk).

If aka or aS, then akS, so g(ak) ≥ g(bk), for otherwise akS. Thus, g(ak) = g(bk) in these cases.

Finally, if ak = aS, the lemma tells us that g(a) < g(bk).

Notes

  1. Stein 1998
  2. See:

References