1

The students are taught the well known change base rule of logarithm:

\begin{align}\log_a b = \frac{\log_c b}{\log_c a}\end{align} Most text books proves it by invoking $(a^x)^y=a^{xy}$ to show: \begin{align}\log_c a\times\log_a b=\log_c b\end{align}

Question:

Why don't we call it Chain Rule of Logarithm (at least as a second name)? Is it because the way we use it is almost alway in the change of base format or something else? \begin{align} \log_a b\times\log_b c\times\log_c d\times\log_d e\times\cdots\times\log_y z=\log_a z\end{align}

It is easy to remember and the students can have fun to continue chaining it (similar to the change rule of derivative).

Star Bright
  • 2,338
  • 1
    Seems your product should have $\log_c d$ as third factor [your product skips that]. – coffeemath Mar 13 '21 at 19:20
  • 1
    @coffeemath A chain with a missing link is a broken chain, thanks :-). – Star Bright Mar 13 '21 at 19:24
  • 4
    A few moments of googling (delete "derivative", delete "calculus" to avoid many false-positive hits) led to this question and the usage in these books, so clearly others have anticipated you. My guess for why this phase (or a similar one) hasn't been adopted is that no one needs to apply the base change over and over again in a single base change situation -- just change from the base you have to the base you want, without going through many intermediary bases. – Dave L. Renfro Mar 13 '21 at 19:29
  • @Joe At some places, people call 12:30AM 0:30AM or 12:30 PM 0:30PM. Not sure what the reason is. – Star Bright Mar 13 '21 at 19:29
  • Why not call the usual “chain rule” a “catenary rule”, instead? Because people don’t. As to proving it, the way I do it is by noting that if $\log_a(b) = r$, then $a^r=b$, so $b=c^{\log_c(a^r)} = c^{r\log_a(c)}$, so $\log_c(b) = r\log_c(a) = \log_a(b)\log_c(a)$. – Arturo Magidin Mar 13 '21 at 19:35
  • 1
    Your displayed equation does not match the formula you say you are trying to prove. The second displayed equation leads to $\log_b(c) = \log_a(c)/\og_a(b)$. – Arturo Magidin Mar 13 '21 at 19:38
  • @arturo thanks! My intention was to have a,b,c flow there to illustrate the chain characteristic. If it is confusing, I will change it. – Star Bright Mar 13 '21 at 19:45
  • I think your first two displays should be compatible. Right now, you claim that you will use method B to prove formula A, but you aren’t proving formula A, you are proving a version of formula A with variables permuted without saying so. It’s like saying “I’m going to prove that $y=x^2$“, and then having a formula that leads to $x=y^2$. – Arturo Magidin Mar 13 '21 at 19:56
  • @aarturo It does not try to prove, but show that when people prove it, it usually come out in product form(as you illustrated), rather than quotient form. The feature of chain is right here during the proof process. – Star Bright Mar 13 '21 at 20:03
  • It wasn’t about that; it was about the confusion generated when you have two expressions that are clearly meant to be connected, but they are not because in one of them you’ve permuted the variables. But you’ve fixed that. – Arturo Magidin Mar 13 '21 at 23:34

3 Answers3

2

A side note:

The argument of the current factor must be equal to the base of following factor. I´ve colored the corresponding parameters.

$$\begin{align}\log_{\color{blue}{a}} \color{red}{b}\times\log_{\color{red}{b}} \color{orange}{c}\times\log_{\color{orange}{c}} \color{green}{d}\times\cdots\times\log_y \color{yellowgreen}{z}=\log_{\color{blue}{a}} \color{yellowgreen}z\end{align}$$

I hope you see the difference to your term. The chain rule is about derivatives and the concept is very different from the rule you´ve posted.

It is more related to the overall growth rate $r$, if you have n consecutive growth rates ($r_i$), with $r_i=\frac{y_{i+1}}{y_i}-1$.

$$1+r=\frac{y_{1}}{y_0}\cdot \frac{y_{2}}{y_1}\cdot \ldots \cdot \frac{y_{n-1}}{y_{n-2}} \cdot \frac{y_{n}}{y_{n-1}}=\frac{y_{n}}{y_{0}}$$

The rule you´ve posted is related to the concept of the geometric mean.

callculus42
  • 30,550
0

Well, finally I found another user or person who concurs with my idea. I discovered this relation personally.

\begin{align} \log_a b\times\log_b c\times\log_c d\times\log_d e\times\cdots\times\log_y z=\log_a z\end{align}

but this is just a special case of

\begin{align} \log_{b_1}{a_1} \times \log_{b_2}{a_2} \times \log_{b_3}{a_3} \times \log_{b_4}{a_4} \times \cdots \times \log_{b_n}{a_n} = \log_{b_1}{a_{\pi_1}} \times \log_{b_2}{a_{\pi_2}} \times \log_{b_3}{a_{\pi_3}} \times \log_{b_4}{a_4} \times \cdots \times \log_{b_n}{a_{\pi_n}} \end{align}

where ${\pi}$ is any permutation of the subscripts 1, ..., n.

Kav
  • 121
0

Well, I think this "chain form" of base changing is as it another form of another identity in reversing ts proof

$\large n^{\log_bx} = x^{\log_bn}$

One revering its simple proof is:

$\large c^{\log_ab} = b^{\log_ac}$

$\large \Rightarrow \log_bc^{log_ab}=log_bb^{log_ac}$

$\large \Rightarrow \log_ab\cdot \log_bc=\log_bb\cdot \log_ac$

$\large \Rightarrow \log_ab\cdot \log_bc=1\cdot \log_ac$

\begin{align}\log_a b\times\log_b c=\log_a c\end{align}

One simple proof is:

$\log_ab\cdot \log_bc=1\cdot \log_ac$

$\large \Rightarrow \log_ab\cdot \log_bc=\log_bb\cdot \log_ac$

$\large \Rightarrow \log_bc^{log_ab}=log_bb^{log_ac}$

$\large \Rightarrow \large c^{\log_ab} = b^{\log_ac}$

Then how do we proceed our proof from two terms to many terms? Use \begin{align}\log_a b = \frac{\log_c b}{\log_c a}\end{align}

$\log_a b\times\log_b c\times\log_c d\times\log_d e\times\cdots\times\log_y z=\log_a z$

$\large \Rightarrow \large \frac{1}{\log_b a} \times \log_b c \times \frac{1}{log_dc} \times \log_d e \times \cdots\times\log_y z=\log_a z$

$\large \Rightarrow \large \frac{ \log_b c}{\log_b a} \times \frac{\log_d e}{log_dc} \times \frac{\log_f g}{log_fe}\times \cdots\times\log_y z=\log_a z$

$\large \Rightarrow \large \log_a c \times log_ce \times \log_eg \times \log_gi \times \log_ik \times \log_km \times \log_mo \times \log_oq \times \log_qs \times \log_su \times \log_uw \times \log_wy \times\log_y z=\log_a z$

$\large \Rightarrow \large \cdots$

$\large \Rightarrow \large \log_ae \times \log_ei \times \log_im \times \log_mq \times \log_qu \times \log_uy \times\log_y z=\log_a z$

$\large \Rightarrow \large \cdots$

$\large \Rightarrow \large \log_ai \times \log_iq \times \log_qy \times\log_y z=\log_a z$

$\large \Rightarrow \large \cdots$

$\large \Rightarrow \large \log_aq \times \log_qy \times\log_y z=\log_a z$

$\large \Rightarrow \large \cdots$

$\large \Rightarrow \large \log_ay \times\log_y z=\log_a z$

$\large \Rightarrow \large \cdots$

$\large \Rightarrow \large \log_a z=\log_a z$

Kav
  • 121