3.6: Limit Superior and Limit Inferior of Functions
- Page ID
- 49111
\( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)
\( \newcommand{\id}{\mathrm{id}}\) \( \newcommand{\Span}{\mathrm{span}}\)
( \newcommand{\kernel}{\mathrm{null}\,}\) \( \newcommand{\range}{\mathrm{range}\,}\)
\( \newcommand{\RealPart}{\mathrm{Re}}\) \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)
\( \newcommand{\Argument}{\mathrm{Arg}}\) \( \newcommand{\norm}[1]{\| #1 \|}\)
\( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)
\( \newcommand{\Span}{\mathrm{span}}\)
\( \newcommand{\id}{\mathrm{id}}\)
\( \newcommand{\Span}{\mathrm{span}}\)
\( \newcommand{\kernel}{\mathrm{null}\,}\)
\( \newcommand{\range}{\mathrm{range}\,}\)
\( \newcommand{\RealPart}{\mathrm{Re}}\)
\( \newcommand{\ImaginaryPart}{\mathrm{Im}}\)
\( \newcommand{\Argument}{\mathrm{Arg}}\)
\( \newcommand{\norm}[1]{\| #1 \|}\)
\( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\)
\( \newcommand{\Span}{\mathrm{span}}\) \( \newcommand{\AA}{\unicode[.8,0]{x212B}}\)
\( \newcommand{\vectorA}[1]{\vec{#1}} % arrow\)
\( \newcommand{\vectorAt}[1]{\vec{\text{#1}}} % arrow\)
\( \newcommand{\vectorB}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vectorC}[1]{\textbf{#1}} \)
\( \newcommand{\vectorD}[1]{\overrightarrow{#1}} \)
\( \newcommand{\vectorDt}[1]{\overrightarrow{\text{#1}}} \)
\( \newcommand{\vectE}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash{\mathbf {#1}}}} \)
\( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \)
\( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)
We extend to functions and concepts of limit superior and limit inferior.
Let \(f: E \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Recall that
\[B_{0}(\bar{x} ; \delta)=B_{-}(\bar{x} ; \delta) \cup B_{+}(\bar{x} ; \delta)=(\bar{x}-\delta, \bar{x}) \cup(\bar{x}, \bar{x}+\delta) .\]
The limit superior of the function \(f\) at \(\bar{x}\) is defnied by
\[\limsup _{x \rightarrow \bar{x}} f(x)=\inf _{\delta>0} \sup _{x \in B_{0}(\bar{x} ; \delta) \cap D} f(x) .\]
Similarly, the limit inferior of the function \(f\) at \(\bar{x}\) is defineid by
\[\liminf _{x \rightarrow \bar{x}} f(x)=\sup _{\delta>0} \inf _{x \in B_{0}(\bar{x} ; \delta) \cap D} f(x) .\]
Consider the extended real-valued function \(g:(0, \infty) \rightarrow(-\infty, \infty]\) defined by
\[g(\delta)=\sup _{x \in B_{0}(\vec{x} ; \delta) \cap D} f(x)\]
It is clear that \(g\) is increasing and
\[\limsup _{x \rightarrow \bar{x}} f(x)=\inf _{\delta>0} g(\delta) .\]
We say that the function \(f\) is locally bounded above around \(\bar{x}\) if there exists \(\delta > 0\) and \(M > 0\) such that
\[f(x) \leq M \text { for all } x \in B(\bar{x} ; \delta) \cap D .\]
Clearly, if \(f\) is locally bounded above around \(\bar{x}\), then \(\limsup _{x \rightarrow \bar{x}} f(x)\) is a real number, while \(\limsup _{x \rightarrow z} f(x)=\infty\) in the other case. Similar discussion applies for the limit inferior.
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then \(\ell=\limsup _{x \rightarrow \bar{x}} f(x)\) if and only if the following two conditions hold:
- For every \(\varepsilon > 0\), there exists \(\delta > 0\) such that
\[f(x)<\ell+\varepsilon \text { for all } x \in B_{0}(\bar{x} ; \delta) \cap D ;\]
- For every \(\varepsilon > 0\) and for every \(\delta > 0\), there exists \(x_{\delta} \in B_{0}(\bar{x} ; \delta) \cap D\) such that
\[\ell-\varepsilon<f\left(x_{\delta}\right)\]
- Proof
-
Suppose \(\ell=\limsup _{x \rightarrow \bar{x}} f(x)\). Then
\[\ell=\inf _{\delta>0} g(\delta) ,\]
where \(g\) is defined in (3.10). For any \(\varepsilon > 0\), there exists \(\delta > 0\) such that
\[\ell \leq g(\delta)=\sup _{x \in B_{0}(\bar{x} ; \delta) \cap D} f(x)<\ell+\varepsilon .\]
Thus,
\[f(x)<\ell+\varepsilon \text { for all } x \in B_{0}(\bar{x} ; \delta) \cap D ,\]
which proves conditions (1). Next note that for any \(\varepsilon > 0\) and \(\delta > 0\), we have
\[\ell-\varepsilon<\ell \leq g(\delta)=\sup _{x \in B_{0}(\bar{x} ; \delta) \cap D} f(x) .\]
Thus, there exists \(x_{\delta} \in B_{0}(\bar{x} ; \delta) \cap D\) with
\[\ell-\varepsilon<f\left(x_{\delta}\right) .\]
This proves (2).
Let us now prove the converse. Suppose (1) and (2) are satisfied. Fix any \(\varepsilon > 0\) and let \(\delta > 0\) satisfy (1). Then
\[g(\delta)=\sup _{x \in B_{0}(\bar{x} ; \delta) \cap D} f(x) \leq \ell+\varepsilon .\]
This implies
\[\limsup _{x \rightarrow \bar{x}} f(x)=\inf _{\delta>0} g(\delta) \leq \ell+\varepsilon .\]
Since \(\varepsilon\) is arbitrary, we get
\[\limsup _{x \rightarrow \bar{x}} f(x) \leq \ell .\]
Again, let \(\varepsilon > 0\). Given \(\delta > 0\), let \(x_{\delta}\) be as in (2). Therefore,
\[\ell-\varepsilon<f\left(x_{\delta}\right) \leq \sup _{x \in B_{0}(\bar{x} ; \delta) \cap D} f(x)=g(\delta) .\]
This implies
\[\ell-\varepsilon \leq \inf _{\delta>0} g(\delta)=\limsup _{x \rightarrow \bar{x}} f(x) .\]
It follows that \(\ell \leq \limsup _{x \rightarrow \bar{x}} f(x)\). Therefore, \(\ell=\limsup _{x \rightarrow \bar{x}} f(x)\). \(\square\)
Suppose \(\ell=\limsup _{x \rightarrow \bar{x}} f(x)\). Then there exists a sequence \(\left\{x_{k}\right\}\) in \(D\) such that \(\left\{x_{k}\right\}\) converges to \(\bar{x}\), \(x_{k} \neq \bar{x}\) for every \(k\), and
\[\lim _{k \rightarrow \infty} f\left(x_{k}\right)=\ell .\]
Moreover, if \(\left\{y_{k}\right\}\) is a sequence in \(D\) that converges to \(\bar{x}\), \(y_{k} \neq \bar{x}\) for every \(k\), and \(\lim _{k \rightarrow \infty} f\left(y_{k}\right)=\ell^{\prime}\), then \(\ell^{\prime} \leq \ell\).
- Proof
-
Let \(\delta_{k}^{\prime}=\min \left\{\delta_{k}, \frac{1}{k}\right\}\). Then \(\delta_{k}^{\prime} \leq \delta_{k}\) and \(\lim _{k \rightarrow \infty} \delta_{k}^{\prime}=0\). From (2) of Theorem 3.6.1, there exists \(x_{k} \in B_{0}\left(\bar{x} ; \delta_{k}^{\prime}\right) \cap D\) such that
\[\ell-\varepsilon_{k}<f\left(x_{k}\right) .\]
Moreover, \(f\left(x_{k}\right)<\ell+\varepsilon_{k}\) by (3.11). Therefore, \(\left\{x_{k}\right\}\) is a sequence that satisfies the conclusion of the corollary.
Now let \(\left\{y_{k}\right\}\) be a sequence in \(D\) that converges to \(\bar{x}\), \(y_{k} \neq \bar{x}\) for every \(k\), and \(\lim _{k \rightarrow \infty} f\left(y_{k}\right)=\ell^{\prime}\). For any \(\varepsilon > 0\), let \(\delta > 0\) be as in (1) of Theorem 3.6.1. Since \(y_{k} \in B_{0}(\bar{x} ; \delta) \cap D\) when \(k\) is sufficiently large, we have
\[f\left(y_{k}\right)<\ell+\varepsilon\]
for such \(k\). This implies \(\ell^{\prime} \leq \ell+\varepsilon\). It follows that \(\ell^{\prime} \leq \ell\). \(\square\)
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Suppose \(\limsup _{x \rightarrow \bar{x}} f(x)\) is a real number. Define
\[A=\left\{\ell \in \mathbb{R}: \exists\left\{x_{k}\right\} \subset D, x_{k} \neq \bar{x} \text { for every } k, x_{k} \rightarrow \bar{x}, f\left(x_{k}\right) \rightarrow \ell\right\} .\]
Then the previous corollary shows that \(A \neq \emptyset\) and \(\limsup _{x \rightarrow \bar{x}} f(x)=\max A\).
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then
\[\limsup _{x \rightarrow \bar{x}} f(x)=\infty\]
if and only if there exists a sequence \(\left\{x_{k}\right\}\) in \(D\) such that \(\left\{x_{k}\right\}\) converges to \(\bar{x}\), \(x_{k} \neq \bar{x}\) for every \(k\), and \(\lim _{k \rightarrow \infty} f\left(x_{k}\right)=\infty\).
- Proof
-
Suppose \(\limsup _{x \rightarrow \bar{x}} f(x)=\infty\). Then
\[\inf _{\delta>0} g(\delta)=\infty ,\]
wehre \(g\) is the extended real-valued function defined in (3.10). Thus, \(g(\boldsymbol{\delta})=\infty\) for every \(\delta > 0\). Given \(k \in \mathbb{N}\), for \(\delta_{k}=\frac{1}{k}\), since
\[g\left(\delta_{k}\right)=\sup _{x \in B_{0}\left(\bar{x} ; \delta_{k}\right) \cap D} f(x)=\infty ,\]
there exists \(x_{k} \in B_{0}\left(\bar{x} ; \delta_{k}\right) \cap D\) such that \(f\left(x_{k}\right)>k\). Therefore, \(\lim _{k \rightarrow \infty} f\left(x_{k}\right)=\infty\).
Let us prove the converse. Since \(\lim _{k \rightarrow \infty} f\left(x_{k}\right)=\infty\), for every \(M \in \mathbb{R}\), there exists \(K \in \mathbb{N}\) such that
\[f\left(x_{k}\right) \geq M \text { for every } k \geq K .\]
For any \(\delta > 0\), we have
\[x_{k} \in B_{0}(\bar{x} ; \delta) \cap D .\]
This implies \(g(\delta)=\infty\), and hence \(\limsup _{x \rightarrow \bar{x}} f(x)=\infty\). \(\square\)
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then
\[\limsup _{x \rightarrow \bar{x}} f(x)=-\infty\]
if and only if for any sequence \(\left\{x_{k}\right\}\) in \(D\) such that \(\left\{x_{k}\right\}\) converges to \(\bar{x}\), \(x_{k} \neq \bar{x}\) for every \(k\), it follows that \(\lim _{k \rightarrow \infty} f\left(x_{k}\right)=-\infty\). The latter is equivalent to \(\lim _{x \rightarrow \bar{x}} f(x)=-\infty\).
Following the same arguments, we can prove similar results for inferior limits of functions.
- Proof
-
Add proof here and it will automatically be hidden
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then \(\ell=\liminf _{x \rightarrow \bar{x}} f(x)\) if and only if the following two conditions hold:
- For every \(\varepsilon > 0\), there exists \(\delta > 0\) such that
\[\ell-\varepsilon<f(x) \text { for all } x \in B_{0}(\bar{x} ; \delta) \cap D ;\]
- For every \(\varepsilon > 0\) and for every \(\delta > 0\), there exists \(x \in B_{0}(\bar{x} ; \delta) \cap D\) such that
\[f(x)<\ell+\varepsilon .\]
- Proof
-
Add proof here and it will automatically be hidden
Suppose \(\ell=\liminf _{x \rightarrow \bar{x}} f(x)\). Then there exists a sequence \(\left\{x_{k}\right\}\) in \(D\) such that \(x_{k}\) converges to \(\bar{x}\), \(x_{k} \neq \bar{x}\) for every \(k\), and
\[\lim _{k \rightarrow \infty} f\left(x_{k}\right)=\ell .\]
Moreover, if \(\left\{y_{k}\right\}\) is a sequence in \(D\) that converges to \(\bar{x}\), \(y_{k} \neq \bar{x}\) for every \(k\), and \(\lim _{k \rightarrow \infty} f\left(y_{k}\right)=\ell^{\prime}\), then \(\ell^{\prime} \geq \ell\).
- Proof
-
Add proof here and it will automatically be hidden
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Suppose \(\liminf _{x \rightarrow \bar{x}} f(x)\) is a real number. Define
\[B=\left\{\ell \in \mathbb{R}: \exists\left\{x_{k}\right\} \subset D, x_{k} \neq \bar{x} \text { for every } \mathrm{k}, x_{k} \rightarrow \bar{x}, f\left(x_{k}\right) \rightarrow \ell\right\} .\]
Then \(B \neq \emptyset\) and \(\liminf _{x \rightarrow \bar{x}} f(x)=\min B\).
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then
\[\liminf _{x \rightarrow \bar{x}} f(x)=-\infty\]
if and only if there exists a sequence \(\left\{x_{k}\right\}\) in \(D\) such that \(\left\{x_{k}\right\}\) converges to \(\bar{x}\), \(x_{k} \neq \bar{x}\) for every \(k\), and \(\lim _{k \rightarrow \infty} f\left(x_{k}\right)=-\infty\).
- Proof
-
Add proof here and it will automatically be hidden
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then
\[\liminf _{x \rightarrow \bar{x}} f(x)=-\infty\]
if and only if there exists a sequence \(\left\{x_{k}\right\}\) in \(D\) such that \(\left\{x_{k}\right\}\) converges to \(\bar{x}\), \(x_{k} \neq \bar{x}\) for every \(k\), it follows \(\lim _{k \rightarrow \infty} f\left(x_{k}\right)=\infty\). The latter is equivalent to \(\lim _{x \rightarrow \bar{x}} f(x)=\infty\).
- Proof
-
Add proof here and it will automatically be hidden
Let \(f: D \rightarrow \mathbb{R}\) and let \(\bar{x}\) be a limit point of \(D\). Then
\[\lim _{x \rightarrow \bar{x}} f(x)=\ell .\]
if and only if
\[\limsup _{x \rightarrow \bar{x}} f(x)=\liminf _{x \rightarrow \bar{x}} f(x)=\ell .\]
- Proof
-
Suppose
\[\lim _{x \rightarrow \bar{x}} f(x)=\ell .\]
Then for every \(\varepsilon > 0\), there exists \(\delta > 0\) such that
\[\ell-\varepsilon<f(x)<\ell+\varepsilon \text { for all } x \in B_{0}(\bar{x} ; \delta) \cap D .\]
Since this also holds for every \(0<\delta^{\prime}<\delta\), we get
\[\ell-\varepsilon<g\left(\delta^{\prime}\right) \leq \ell+\varepsilon .\]
It follows that
\[\ell-\varepsilon \leq \inf _{\delta^{\prime}>0} g\left(\delta^{\prime}\right) \leq \ell+\varepsilon .\]
Therefore, \(\limsup _{x \rightarrow \bar{x}} f(x)=\ell\) since \(\varepsilon\) is arbitrary. The proof for the limit inferior is similar. The converse follows directly from (1) of Theorem 3.6.1 and Theorem 3.6.6. \(\square\)
Exercise \(\PageIndex{1}\)
Let \(D \subset \mathbb{R}\), \(f: D \rightarrow \mathbb{R}\), and \(\bar{x}\) be a limit point of \(D\). Prove that \(\liminf _{x \rightarrow \bar{x}} f(x) \leq \limsup _{x \rightarrow \bar{x}} f(x)\).
- Answer
-
Add texts here. Do not delete this text first.
Exercise \(\PageIndex{2}\)
Find each of the following limits:
- \(\limsup _{x \rightarrow 0} \sin \left(\frac{1}{x}\right)\).
- \(\liminf _{x \rightarrow 0} \sin \left(\frac{1}{x}\right)\).
- \(\limsup _{x \rightarrow 0} \frac{\cos x}{x}\).
- \(\liminf _{x \rightarrow 0} \frac{\cos x}{x}\).
- Answer
-
Add texts here. Do not delete this text first.