Is there a closed-form solution for $\frac {d^n}{dx^n}\frac 1{f(x)}$? I've looked at the first five derivatives in search of some pattern, but I can't identify anything strong enough to give a closed formula.

  • $\begingroup$ Is $f(x)$ a particular function or its just a generalization? $\endgroup$ – AvZ Mar 6 '15 at 12:45
  • $\begingroup$ @AvZ A generalization. $\endgroup$ – Laertes Mar 6 '15 at 13:01

It's not exactly a closed form, but Faà di Bruno's formula

$$\frac{d^n}{dx^n} f(g(x))= \sum_{k_1 + 2k_2 + \ldots + n k_n = n} \frac{n!}{k_1!1!^{k_1}...k_n!n!^{k_n}}\left(\frac{d^{k_1 + \ldots+ k_n}}{dx^{k_1 + \ldots+ k_n}} f\right)(g(x)) \prod_j \left(\frac{d^j}{dx^j} g(x) \right)^{k_j}$$


$$\frac{d^n}{dx^n} \frac{1}{g(x)} = \sum_{k_1 + 2k_2 + \ldots + n k_n = n} \frac{(-1)^k n!({k_1 + \ldots+ k_n})!}{k_1!1!^{k_1}...k_n!n!^{k_n}g(x)^{1+k_1 + \ldots+ k_n}} \prod_j \left({\frac{d^j}{dx^j} g(x)} \right)^{k_j}$$

  • $\begingroup$ Appreciated, but I was hoping for something slightly more closed. $\endgroup$ – Laertes Mar 6 '15 at 13:31
  • 3
    $\begingroup$ @Laertes Aside from perhaps a recurrence relationship, I think this is the best you can hope for I'm afraid! $\endgroup$ – Thomas Russell Mar 6 '15 at 13:53
  • $\begingroup$ Small correction: The $(-1)^k$ in the second formula makes no sense because $k$ is not defined. It should read $(-1)^{k_1+\dots+k_n}$. $\endgroup$ – WillG Oct 18 '19 at 0:37

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.