The function $g$ is strictly positive. Let the function $f$ be defined as

$$f(x) = \int_0^x g(u) du$$

Is there a way to express $f^{-1}(x)$ in terms of $g$?

  • 1
    $\begingroup$ In general $f$ need not be invertible, i.e., $f^{-1}$ need not exist. Do you mean to suppose, e.g., that $g$ is strictly positive (in which case $f$ is strictly increasing and hence invertible)? $\endgroup$ – Travis Willse Oct 6 '15 at 22:22
  • $\begingroup$ Yes, $g$ can be assumed to be strictly positive. $\endgroup$ – Halbort Oct 6 '15 at 22:23

If $g$ takes on both negative and positive values, or is zero on some interval, then $f$ is not invertible, as mentioned in comments.

Assume $g$ is strictly positive (or strictly negative), hence $f^{-1}$ exists and is differentiable by inverse function theorem.

Then $f(f^{-1}(x))=x$, so by differentiating, we get that $f'(f^{-1}(x))(f^{-1})'(x) =1$, i.e, $g(f^{-1}(x))(f^{-1})'(x) = 1$.

Thus we see that $f^{-1}$ satisfies the differential equation $$y' = \frac{1}{g(y)}$$

For some functions $g$ this can be solved exactly (for example, $g(y) = e^y$ or $g(y) = y^2+1$), while for others it cannot be solved exactly (for example $g(y) = e^{y^2}$). Hence, you can get this differential formula but no explicit solution in general.

  • $\begingroup$ To be clear, if $g$ is strictly negative, $f$ is strictly decreasing and so is again invertible. $\endgroup$ – Travis Willse Oct 6 '15 at 22:47
  • $\begingroup$ @Travis: Added, thanks. $\endgroup$ – shalop Oct 6 '15 at 23:13
  • $\begingroup$ This is a concise and complete solution! Well done. +1 $\endgroup$ – Mark Viola Oct 6 '15 at 23:27
  • $\begingroup$ This is exactly what I was looking for. Thank you: +1 $\endgroup$ – Halbort Oct 7 '15 at 0:13

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.