Minimum Rule for Continuous Functions

From ProofWiki
Jump to navigation Jump to search

Theorem

Let $\struct {S, \tau}$ be a topological space.

Let $f, g: S \to \R$ be continuous real-valued functions.

Let $\min \set {f, g}: S \to \R$ denote the pointwise minimum of $f$ and $g$.


Then:

$\min \set {f, g}$ is continuous.


Proof

Let $x \in S$.

Let $\epsilon > 0$.

Without loss of generality, assume that $\map f x \le \map g x$.


Case 1 : $\map f x = \map g x$

Let $ y = \map f x = \map g x$


By definition of the min operation, $\min \set {\map f x, \map g x} = y$.


From Continuity Test for Real-Valued Functions applied to $f$:

$\exists U \in \tau : x \in U : \forall z \in U : \map f z \in \openint {y - \epsilon} {y + \epsilon}$

Similarly, Continuity Test for Real-Valued Functions applied to $g$:

$\exists V \in \tau : x \in V : \forall z \in V : \map g z \in \openint {y - \epsilon} {y + \epsilon}$


Thus for all $z \in U \cap V$:

$\map f z \in \openint {y - \epsilon} {y + \epsilon}$
$\map g z \in \openint {y - \epsilon} {y + \epsilon}$

By definition of the min operation, for all $z \in U \cap V$:

$\min \set {\map f z, \map g z} \in \openint {y - \epsilon} {y + \epsilon}$

By the Open Set Axiom $\paren {\text O 2 }$: Pairwise Intersection of Open Sets:

$U \cap V \in \tau$


Since $\epsilon > 0$ was arbitrary then:

$\forall \epsilon > 0 : \exists W \in \tau : x \in W : \forall z \in W : \min \set {\map f z, \map g z} \in \openint {y - \epsilon} {y + \epsilon}$

From Continuity Test for Real-Valued Functions, $\min \set {f, g}$ is continuous at $x$.

$\Box$


Case 2 : $\map f x < \map g x$

Let $\map f x < \map g x$.

Then:

$\min \set {\map f x, \map g x} = \map f x$


Let $\delta = \min \set {\epsilon, \dfrac 1 2 \paren {\map g x - \map f x} }$.

Then $\delta > 0$.


From Continuity Test for Real-Valued Functions applied to $f$:

$\exists U \in \tau : x \in U : \forall z \in U : \map f z \in \openint {\map f x - \delta} {\map f x + \delta}$

Similarly, Continuity Test for Real-Valued Functions applied to $g$:

$\exists V \in \tau : x \in V : \forall z \in V : \map g z \in \openint {\map g x - \delta} {\map g x + \delta}$


Thus for all $z \in U \cap V$:

$\map f z \in \openint {\map f x - \delta} {\map f x + \delta}$
$\map g z \in \openint {\map g x - \delta} {\map g x + \delta}$

Now:

\(\ds \map f z\) \(<\) \(\ds \map f x + \delta\) as $\map f z \in \openint {\map f x - \delta} {\map f x + \delta}$
\(\ds \) \(\le\) \(\ds \map f x + \frac 1 2 \paren {\map g x - \map f x}\) as $\delta \le \dfrac 1 2 \paren {\map g x - \map f x}$
\(\ds \) \(=\) \(\ds \frac 1 2 \paren {2 \map f x + \map g x - \map f x}\)
\(\ds \) \(=\) \(\ds \frac 1 2 \paren {\map g x + \map f x}\)
\(\ds \) \(=\) \(\ds \frac 1 2 \paren {\map g x + \map g x - \map g x + \map f x}\)
\(\ds \) \(=\) \(\ds \frac 1 2 \paren {2 \map g x - \paren{\map g x - \map f x} }\)
\(\ds \) \(=\) \(\ds \map g x - \frac 1 2 \paren {\map g x - \map f x}\)
\(\ds \) \(\le\) \(\ds \map g x - \delta\) as $-\dfrac 1 2 \paren {\map g x - \map f x} \le -\delta$
\(\ds \) \(<\) \(\ds \map g z\) as $\map g z \in \openint {\map g x - \delta} {\map g x + \delta}$

Thus:

$\min \set {\map f z, \map g z} = \map f z$


As

$\map f z \in \openint {\map f x - \delta} {\map f x + \delta}$

and

$\map f x = \min \set {\map f x, \map g x}$

then

$\min \set {\map f z, \map g z} \in \openint {\min \set {\map f x, \map g x} - \delta} {\min \set {\map f x, \map g x} + \delta}$


By the Open Set Axiom $\paren {\text O 2 }$: Pairwise Intersection of Open Sets:

$U \cap V \in \tau$


Since $\epsilon > 0$ was arbitrary then:

$\forall \epsilon > 0 : \exists W \in \tau : x \in W : \forall z \in W : \min \set{\map f z, \map g z} \in \openint {\min \set {\map f x, \map g x} - \delta} {\min \set {\map f x, \map g x} + \delta}$

From Continuity Test for Real-Valued Functions, $\min \set {f, g}$ is continuous at $x$.

$\Box$


In either case, $\min \set {f, g}$ is continuous at $x$.

Since $x \in S$ was arbitrary, then $\min \set {f, g}$ is everywhere continuous.

$\blacksquare$


Also see