# Reference request: Extremes of composite functions

Let $g : X \rightarrow Y$ be a homeomorphism, and let $f : Y \rightarrow \mathbb{R}$ be a continuous function. Then the extremes of $f$ in $Y$ are mapped to the extremes of the composition $f \circ g$ in $X$, and if $f$ has a unique minimizer in $Y$, then $f \circ g$ has a unique minimizer in $X$. This seems plausible to me, but does anyone know of a reference where this is proved (and preferably related results)?

Kind regards

Olav

-
In fact, you only need $g$ to be a bijection. – Martin Sleziak Jul 29 '12 at 10:39
Yeah that makes sense. Thanks. – Olav Jul 29 '12 at 10:46

From the definition of "homeomorphism" we need only that $g:\ X\to Y$ is bijective. This means that we have a set $$S:=\bigl\{(x,y)\ \bigm|\ x\in X,\ y=g(x)\in Y\bigr\}$$ of "married pairs". Each point $p=(x,y)\in S$ can be adressed either by its first coordinate $x\in X$ or its second coordinate $y\in Y$, and the only difference between $f$ and $f\circ g$ is whether the function $$F:\ S\to {\mathbb R},\quad p=(x,y)\mapsto F(p):=f(y)=f\bigl(g(x)\bigr)$$ is expressed as $F(p):=f(y)$ or as $F(p):=f\bigl( g(x)\bigr)$.
Looking at the setup in this way both your statements become obvious: The extremes of $f$ resp. $f\circ g$ are the extremes of the encompassing idea $F$.