Showing that the Taylor expansion of $\ln(1-x)$ about $x=0$ converges over $(-1,1)$

1.1k Views Asked by At

I'm a little bit lost here.

I'm in need of showing that $f(x)= \ln(1 - x)$ is equal to $$ -\sum_{t=1}^{\infty}\frac{x^{t}}{t} $$ for all $x\in (-1,1)$.

I know how to show that the $n$-th degree Taylor polynomial of $f$ about $x=0$ is equal to $$ T_n(x)=-\sum_{t=1}^{n}\frac{x^{t}}{t},$$ but I don't know how to show that the error in this approximation goes to zero as $n\rightarrow \infty $. I tried to use the Lagrange remainder like this:

*We know that $f$ and all its derivatives are continous on any interval $[a,x] \subset (-1,1)$. Then there is a number $c\in (a,x)$ such that the error in the $n$-th degree taylor approximation is equal to $$ E(x)=\frac{f^{(n+1)}(c)}{(n+1)!}(x-a)^{n+1} $$ Now, $$ f^{(n+1)}(x)=\frac{(-1)^{n+2}n!}{(x-1)^{n+1}}\forall n \in \Bbb N $$ which gives the expression: $$ E(x)=\frac{(-1)^{n+2}n!}{(n+1)!(c-1)^{n+1}}(x-a)^{n+1}=\frac{(-1)^{n}(x-a)^{n+1}}{(n+1)(c-1)^{n+1}} $$ Now, the $(-1)^n$ part of this expression is of no interest, as we are only interested in the size of this error. This means that we need to prove that $$ \lim_{n\to \infty}\frac{1}{n+1}[\frac{x-a}{c-1}]^{n+1}=0 $$ Unfortunately, I can't find any argument that it does. My thoughts are that we can chose $a$ and $x$ such that $a-x$ is as close to $2$ as we want, and the number $c$ might be really close to $1$, such that the term $\frac{x-a}{c-1}$ is greater than one and hence explodes to infinity as $n$ gets large.

Can anybody tell me where my reasoning breaks down, and if there is any better way to do this?

1

There are 1 best solutions below

6
On BEST ANSWER

The point here is that you don't have to take $a$ and $x$ close to $-1$ and $1$ simultaneously. Quite the opposite, you can take your time and pick $a, x$ as close to each other as you like, as soon as your argument works for a number of intervals that covers all of $(-1,1)$ in the end.

If $x\geq 0$, take $a<x$ such that $|x-a|<|1-x|$. Therefore in the end, your quantity above does tend to $0$, which proves your equality on that interval.

The same trick works on the other side (close to $-1$).


An easier way since you asked, would be to use the normal (hence uniform) convergence of the derivative on any interval $[a,b]\subset (0,1)$ and the fact that this allows you to integrate term by term.

The derivative of the series on the right-hand side is the well-known Taylor expansion of $\dfrac{1}{1-x}$ (or rather its negative), whose primitive that vanishes at $x=1$ is exactly the function in your left-hand side.