5

Given $g: \mathbb{R} \rightarrow \mathbb{R}$ is a differentiable function with bounded derivative i.e. satisfying $|g'(x)|\leq K>0 , \forall x \in \mathbb{R}$, I am trying to show that for some constant $\epsilon>0$ small enough, the function $f: \mathbb{R} \rightarrow \mathbb{R}$ defined by $f(x):=x+\epsilon g(x)$ is invertible.

My idea is to show that $f$ is bijective. It occurred to me that $f$ is injective since its derivative can be made to be strictly positive so that it is strictly increasing if $\epsilon<1/K \implies f'=1+\epsilon g'>1+\epsilon(-K)>1-1>0$. However, how can I prove $f$ is surjective?

Hanul Jeon
  • 27,376
Jay
  • 268

3 Answers3

2

You could restrict the codomain of $f$ to its range:

Then $f:\mathbb{R}\to f(\mathbb{R})$ is surjective.

Update: It turns out we can show $f(\mathbb{R})=\mathbb{R}$.

For $x>0$,

Consider $f(x)-f(0)=x+\epsilon(g(x)-g(0))=x+\epsilon(xg'(c_x))$ for some $c_x\in (0,x)$ by Mean Value Theorem.

Then $f(x)=f(0)+x(1+\epsilon g'(c_x))\geq f(0)+x(1+\epsilon (-K))$ for all $x>0$.

Since $\epsilon<\frac 1K$, $(1+\epsilon(-K))$ is postive. Then letting $x\to +\infty$ will see that $f(x)\to +\infty$.

Analogously consider $x<0$, see that $f(x)\to -\infty$ as $x\to -\infty$.

yoyostein
  • 19,608
0

By the given construction, $f$ is strictly increasing. It is also continuous: $g$ is differentiable, hence continuous, and $f$ is a linear combination of continuous functions.

If an injective $f$ is bounded for some $\epsilon$, $\lim_{x\to\pm\infty}(\epsilon g(x))'=-1$, and decreasing $\epsilon$ will increase this limit such that $\lim_{x\to\pm\infty}f'(x)>0$, i.e. $f$ diverges. With the monotonicity and continuity of $f$ shown above, $f$ becomes surjective. The injectivity of $f$ is preserved in the lowering of $\epsilon$, so it is always possible to choose $\epsilon$ so $f$ is bijective – and therefore invertible.

Parcly Taxel
  • 103,344
0

The key argument here is to show that $f(x)$ can't be "too far" from $x$, so the fact that $x$ is unbounded above and below implies the same for $f(x)$.

One such method is

$$ \begin{align} |f(x) - x| = \epsilon |g(x)| = \epsilon \left| g(0) + \int_0^x g'(t) \, \mathrm{d}t \right| \leq \epsilon \left( |g(0)| + |x| K \right) \end{align} $$

For $x>0$, we have

$$ x (1 - K\epsilon ) - \epsilon |g(0) | \leq f(x) $$

showing that $f(x)$ goes to $\infty$ as $x \to \infty$, assuming $K\epsilon < 1$.

Similarly, for $x < 0$, we have

$$ f(x) \leq x (1 - K \epsilon) + \epsilon |g(0)| $$