I don't have an extensive formal training in calculus, but I'm doing quite a lot of differential calculus work at the moment and there's something which bothers me.

Say I have the differential

$\dfrac{d\ln (P(x))}{d\ln(x)}$

Where $P$ is some unknown function

By defining $u = \ln(x)$ I can re-write this as

$\dfrac{d\ln (P(e^u))}{du}$

Which I would then solve (using prime notation) as follows


$f(z) = \ln(z)$ , $f'(z) = \dfrac{1}{z}$

$g(z) = P(z)$, $g(z)' = P'(z)$

$h(z) = e^z$, $h'(z) = e^z$

So using the chain rule (note I'm being super explicit in my steps)

$\dfrac{f(g(h(u)))}{du} = f'(g(h(u)) \times g(h(u))' $

$g(h(u))' = g'(h(u)) \times h'(u)$

$\dfrac{f(g(h(u)))}{du} = f'(g(h(u)) \times g'(h(u)) \times h'(u) $

$\dfrac{f(g(h(u)))}{du} = \dfrac{1}{P(e^u)} \times P'(e^u) \times e^u $

$\dfrac{f(g(h(u)))}{du} = \dfrac{P'(e^u)e^u }{P(e^u)} $

$\dfrac{f(g(h(x)))}{dx} = \dfrac{xP'(x)}{P(x)}$

$\dfrac{d\ln (P(x))}{d\ln(x)} = \dfrac{x}{P(x)}P'(x)$

1) is this kind of mixing and matching of Leibniz and prime notation even "legal"?

If not, should I really re-write the differential as

$V'(\ln(x)) = \dfrac{d\ln (P(x))}{d\ln(x)} \text{ where } \Bigg(V(x) = \ln (P(x))\Bigg)$ (i.e. is this correct?)

2) I prefer to use prime notation, but does this have any significant drawbacks (other than being less "clean" in places, i.e. you can treat $\dfrac{df(x)}{dx}$ as a fraction in certain conditions).

3) given that P is a function (say P is defined as $P = kx + cx^2$), what is the difference between

$\dfrac{P(x)}{dx}$ and $\dfrac{P}{dx}$

I'm trying to work out if the stuff I'm reading is just using inconsistent notation, or if there are crucial functional/semantic differences I'm not understanding.

UPDATE: Dealing with $\dfrac{d\ln(P(x))}{d\ln(x)}$

For a concrete example, say P(x) = 2x+5, so, as mentioned above, we define $u = \ln(x)$

We would re-write $\dfrac{d\ln(2x+5)}{d\ln(x)}$ as $\dfrac{d\ln(2e^u+5)}{du}$, then solve this getting


And finally substitute $\ln(x)$ back in for $u$


  • 2
    $\begingroup$ I'm confused by the very first of your notation: $\frac{d\ln(P(x))}{d\ln(x)}$. Can you really write $d\ln(x)$ in the denominator? I have never come across this notation so far ... $\endgroup$ – kafman Jan 21 '13 at 19:25
  • $\begingroup$ I don't see any reason why not? I mean it's maybe not a common way of writing a differential, but pops up a lot in engineering/thermodynamics (although for the record I'd never seen it before either until fairly recently). $\endgroup$ – Alex Jan 21 '13 at 19:28
  • 1
    $\begingroup$ What does $\frac{d\ln{P(x)}}{\ln{x}}$ mean? How is it defined? $\endgroup$ – user39280 Jan 21 '13 at 19:31
  • $\begingroup$ Does $\frac{\mathrm d \ln(P(x))}{\mathrm d \ln(x)}$ mean that $u = \ln(x)$ and we wish to find $\frac{\mathrm d \ln(P(x))}{\mathrm d u}$? $\endgroup$ – George V. Williams Jan 21 '13 at 19:52
  • 1
    $\begingroup$ The notation of the first statement is fine. It means you wish to find the derivative of ln(P(x)) with respect to ln(x). $\endgroup$ – daniel.wright Feb 20 '13 at 3:39

I have a simpler method:


| cite | improve this answer | |
  • $\begingroup$ That's sneaky! I like it :-D $\endgroup$ – Alex Mar 21 '13 at 23:10

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.