17
$\begingroup$

A useful property of the logarithm is that it can "convert" multiplication into addition, as in

$\ln(a)+\ln(b)=\ln(ab) \text{ for all } a, b \in \mathbb{R}^+$

Does there exist a function $f$, which holds a similar property for exponentiation?

$f(a)+f(b)=f(a^b) \text{ for all } a, b \in \mathbb{R}^+$

If so, are there any closed-form expressions for such a function?

$\endgroup$
6
  • 31
    $\begingroup$ It can't exist as $f(a)+f(b)$ is symmetrical with respect to swap of $a$ and $b$, while $f(a^b)$ is not. $\endgroup$ Commented Jun 19, 2022 at 8:02
  • 3
    $\begingroup$ E.g. $f(x)=f(x^1)=f(x)+f(1)=f(1)+f(x)=f(1^x)=f(1)$ so $f$ is constant, $f(x)=c$ and then $c=f(1)=f(1^1)=f(1)+f(1)=c+c$, i.e. $c=0$. $\endgroup$ Commented Jun 19, 2022 at 8:08
  • 3
    $\begingroup$ (so, technically speaking it does exist - a constant function $f(x)=0$ - but I doubt that is what you are looking for.) $\endgroup$ Commented Jun 19, 2022 at 11:28
  • $\begingroup$ I thought a little about your question, and asked a follow-up here: mathoverflow.net/questions/425061/… $\endgroup$ Commented Jun 19, 2022 at 18:38
  • 2
    $\begingroup$ Viewed in another way, the difference is that multiplication is commutative while exponentiation isn't. $\endgroup$ Commented Jun 19, 2022 at 20:53

2 Answers 2

36
$\begingroup$

If $f(a)+f(b)=f(a^b)$ for all $a,b\in\mathbb R^+$, then it follows (by interchanging $a$ and $b$) that $f(b)+f(a)=f(b^a)$ for all $a,b\in\mathbb R^+$. Hence, $f(a^b)=f(b^a)$ for all $a,b\in\mathbb R^+$. Setting $b=1$, we see that $f(a)=f(1)$ for all $a\in\mathbb R^+$. Now $f(1)+f(1)=f(1^1)$, so $f(1)=0$. Hence, the only function $\mathbb R^+\to\mathbb R$ with the desired property is the zero function.

$\endgroup$
0
10
$\begingroup$

While your question as stated has been answered, I'd like to give a satisfying answer to a slight alteration of your question. To avoid the issue of commutivity, we can instead look at commutative hyperoperations $F_n(a,b)$. A few of them are:

\begin{align*} F_1(a,b)&=a+b=b+a\\ F_2(a,b)&=ab=ba\\ F_3(a,b)&=a^{\ln b}=b^{\ln a}\\ \end{align*}

Define $\ln^{(k)}(x):=\underbrace{\ln(\ln(\dots\ln(x)))}_{k\text{ times}}$. It is true that, in general:

$$\ln^{(k)}(F_n(a,b))=F_{n-k}(\ln^{(k)}(a),\ln^{(k)}(b))\tag{$*$}$$

for all integers $n\geq 1$ and $0\leq k < n$.

In other words, if we want to "go down" $k$ operations, then we need to use the $\ln$ function $k$ times.

Indeed, this aligns with the standard rule that $\ln(ab)=\ln a +\ln b$, or in our commutative hyperopration notation, $\ln(F_2(a,b)) = F_1(\ln a,\ln b)$.


So, for (my slight alteration of) your question, we want a function $f$ that satisfies

$$f(F_3(a,b))=f(a^{\ln b})=f(a)+f(b)=F_1(f(a),f(b)).$$

Here, $k=2$, so we can set $f(x)=\ln^{(2)}(x)=\ln(\ln(x))$.

$$\ln(\ln(a^{\ln b}))=\ln(\ln(a)\cdot\ln(b))=\ln(\ln(a))+\ln(\ln(b))$$


Proof of $(*)$:

The commutative hyperoperations are defined recursively by

$$F_n(a, b) = \exp(F_{n-1}(\ln(a), \ln(b))).$$

Therefore,

$$F_n(a, b) = \exp^{(k)}(F_{n-k}(\ln^{(k)}(a), \ln^{(k)}(b))).$$

Hence,

$$\ln^{(k)}(F_n(a, b)) = F_{n-k}(\ln^{(k)}(a), \ln^{(k)}(b)).$$

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.