Oscillation at Point (Infimum) equals Oscillation at Point (Limit)

From ProofWiki
Jump to navigation Jump to search


Theorem

Let $f: D \to \R$ be a real function where $D \subseteq \R$.

Let $x$ be a point in $D$.

Let $N_x$ be the set of open subset neighborhoods of $x$.

Let $\map {\omega_f} x$ be the oscillation of $f$ at $x$ as defined by:

$\map {\omega_f} x = \inf \set {\map {\omega_f} I: I \in N_x}$

where $\map {\omega_f} I$ is the oscillation of $f$ on a real set $I$:

$\map {\omega_f} I = \sup \set {\size {\map f y - \map f z}: y, z \in I \cap D}$

Let $\map {\omega^L_f} x$ be the oscillation of $f$ at $x$ as defined by:

$\map {\omega^L_f} x = \displaystyle \lim_{h \mathop \to 0^+} \map {\omega_f} {\openint {x - h} {x + h} }$


Then:

$\map {\omega_f} x \in \R$ if and only if $\map {\omega^L_f} x \in \R$

and, if $\map {\omega_f} x$ and $\map {\omega^L_f} x$ exist as real numbers:

$\map {\omega_f} x = \map {\omega^L_f} x$


Proof

Lemma

Let $f: D \to \R$ be a real function where $D \subseteq \R$.

Let $x$ be a point in $D$.

Let $N_x$ be the set of open subset neighborhoods of $x$.

Let $\map {\omega_f} x$ be the oscillation of $f$ at $x$ as defined by:

$\map {\omega_f} x = \displaystyle \inf \set {\map {\omega_f} I: I \in N_x}$

where $\map {\omega_f} I$ is the oscillation of $f$ on a real set $I$:

$\map {\omega_f} I = \displaystyle \sup \set {\size {\map f y - \map f z}: y, z \in I \cap D}$

Let $\map {\omega^L_f} x$ be the oscillation of $f$ at $x$ as defined by:

$\map {\omega^L_f} x = \displaystyle \lim_{h \mathop \to 0^+} \map {\omega_f} {\openint {x - h} {x + h} }$


Let $\map {\omega^L_f} x \in \R$.

Let $\map {\omega_f} x \in \R$.


Then $\map {\omega^L_f} x = \map {\omega_f} x$.

$\Box$


Necessary Condition

Let $\map {\omega_f} x \in \R$.

We need to prove:

$\map {\omega^L_f} x \in \R$
$\map {\omega^L_f} x = \map {\omega_f} x$

where:

$\map {\omega^L_f} x = \displaystyle \lim_{h \mathop \to 0^+} \map {\omega_f} {\openint {x - h} {x + h} }$
$\map {\omega_f} {\openint {x - h} {x + h} } = \sup \set {\size {\map f y - \map f z}: y, z \in \openint {x - h} {x + h} \cap D}$
$\map {\omega_f} x = \inf \set {\map {\omega_f} I: I \in N_x}$
$\map {\omega_f} I = \sup \set {\size {\map f y - \map f z}: y, z \in I \cap D}$


Let $\epsilon \in \R_{>0}$.

Then an $I \in N_x$ exists such that:

$\map {\omega_f} I - \map {\omega_f} x < \epsilon$ by Infimum of Set of Oscillations on Set is Arbitrarily Close

Let $I$ be such an element of $N_x$.

We observe in particular that $\map {\omega_f} I \in \R$.


A neighborhood in $N_x$ contains an open subset that contains the point $x$.

So, $I$ contains such an open subset as $I \in N_x$.

Therefore, a $\delta \in \R_{>0}$ exists such that $\openint {x - \delta} {x + \delta}$ is a subset of $I$.

Let $h$ be a real number that satisfies: $0 < h < \delta$.

We observe that $\openint {x - h} {x + h} \subset I$.

Also, $\openint {x - h} {x + h} \in N_x$.


We have:

$I \in N_x$
$\openint {x - h} {x + h} \in N_x$
$\openint {x - h} {x + h} \subset I$
$\map {\omega_f} I \in \R$

from which follows by Oscillation on Subset:

$\map {\omega_f} {\openint {x - h} {x + h} } \in \R$
$\map {\omega_f} {\openint {x - h} {x + h} } \le \map {\omega_f} I$

We have that:

$\map {\omega_f} {\openint {x - h} {x + h} } \in \set {\map {\omega_f} {I'}: I' \in N_x}$

as $\openint {x - h} {x + h} \in N_x$.

Also, $\map {\omega_f} x$ is a lower bound for $\set {\map {\omega_f} {I'}: I' \in N_x}$ by the definition of $\map {\omega_f} x$.

Therefore:

$\map {\omega_f} {\openint {x - h} {x + h} } \ge \map {\omega_f} x$


We find:

\(\displaystyle \map {\omega_f} x \le \map {\omega_f} {\openint {x - h} {x + h} }\) \(\le\) \(\displaystyle \map {\omega_f} I\)
\(\displaystyle \leadsto \ \ \) \(\displaystyle 0 \le \map {\omega_f} {\openint {x - h} {x + h} } - \map {\omega_f} x\) \(\le\) \(\displaystyle \map {\omega_f} I - \map {\omega_f} x\)
\(\displaystyle \leadsto \ \ \) \(\displaystyle 0 \le \map {\omega_f} {\openint {x - h} {x + h} } - \map {\omega_f} x\) \(\le\) \(\displaystyle \map {\omega_f} I - \map {\omega_f} x < \epsilon\) as $\map {\omega_f} I - \map {\omega_f} x < \epsilon$ is true
\(\displaystyle \leadsto \ \ \) \(\displaystyle 0 \le \map {\omega_f} {\openint {x - h} {x + h} } - \map {\omega_f} x\) \(<\) \(\displaystyle \epsilon\)
\(\displaystyle \leadsto \ \ \) \(\displaystyle \size {\map {\omega_f} {\openint {x - h} {x + h} } - \map {\omega_f} x}\) \(<\) \(\displaystyle \epsilon\)

which means that $\displaystyle \lim_{h \mathop \to 0^+} \map {\omega_f} {\openint {x - h} {x + h} }$ exists and equals $\map {\omega_f} x$ by the definition of limit.

In other words, $\map {\omega^L_f} x \in \R$ and $\map {\omega^L_f} x = \map {\omega_f} x$.

$\Box$


Sufficient Condition

Let $\map {\omega^L_f} x \in \R$.

We need to prove:

$\map {\omega_f} x \in \R$
$\map {\omega_f} x = \map {\omega^L_f} x$

where:

$\map {\omega_f} x = \inf \set {\map {\omega_f} I: I \in N_x}$
$\map {\omega_f} I = \sup \set {\size {\map f y - \map f z}: y, z \in I \cap D}$
$\map {\omega^L_f} x = \displaystyle \lim_{h \mathop \to 0^+} \map {\omega_f} {\openint {x - h} {x + h} }$


We have:

$\displaystyle \lim_{h \mathop \to 0^+} \map{\omega_f} {\openint {x - h} {x + h} } \in \R$ as $\map {\omega^L_f} x \in \R$

Therefore, $\map {\omega_f} {\openint {x - h} {x + h} } \in \R$ for a small enough $h$ in $\R_{>0}$ by the definition of limit.

Let $h$ be such a real number.

We observe that $\openint {x - h} {x + h}$ is a neighborhood in $N_x$.


We have:

$\openint {x - h} {x + h} \in N_x$
$\map {\omega_f} {\openint {x - h} {x + h} } \in \R$

Accordingly:

$\map {\omega_f} x \in \R$ by Infimum of Set of Oscillations on Set

$\map {\omega_f} x = \map {\omega^L_f} x$ follows by Lemma.

This finishes the proof of the theorem.

$\blacksquare$