I was looking for a rational function $f: \mathbb{R} \to \mathbb{R}$ that looks like $\arctan$, in that it is

  • injective
  • not surjective
  • well-defined on all $x\in \mathbb{R}$ (no vertical asymptotes)

I believe I have proven that one can't exist.

My thinking is this: it can't be a polynomial, because an odd-degree polynomial is surjective and an even degree polynomial is not injective. Therefore it must have a nontrivial denominator. The denominator can't be an odd degree polynomial, or else it has a zero somewhere. If the numerator is a polynomial of degree greater than the denominator $N>D$, then we get surjectivity (if $N-D$ is odd), or we lose injectivity (if $N-D$ is even). If $N\leq D$, then we lose injectivity, since the function will approach the same limit as $x \to \pm \infty$ (I claim that together with continuity, this implies that we lose injectivity).

Do you agree? Is there a general result about such functions?

  • $\begingroup$ $1/(x^2+1)$ is not surjective. So, "The numerator must be a polynomial of the same degree, or else we get surjectivity" is not true. $\endgroup$ – lhf Jun 16 '14 at 15:05
  • $\begingroup$ Consider the function $$f(x)=\int_{0}^{x}\frac{dt}{1+t^{4}}$$ $\endgroup$ – TheOscillator Jun 16 '14 at 15:07
  • $\begingroup$ @lhf You are correct. I attempted to fix this $\endgroup$ – Eric Auld Jun 16 '14 at 15:07
  • $\begingroup$ @TheOscillator That is not a rational function. $\endgroup$ – Eric Auld Jun 16 '14 at 15:07
  • 1
    $\begingroup$ Looks good to me. $\endgroup$ – TonyK Jun 16 '14 at 15:15

This looks good to me as well, though the various cases could be cleaned up a bit: the key point is that the range of an injective continuous function is the interval between its two limits, and that for a rational function $f(x)/g(x)$ the two limits can only differ when $\deg f > \deg g$ in which case the interval is $(- \infty, \infty)$.

It's hard to say how to generalize this result: in $\mathbb{C}$ for instance all continuous rational functions are polynomials, and all injective polynomials are surjective. Over any proper subinterval of $\mathbb{R}$, you can find an injective rational functions with whatever values you like on the two boundaries.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.