I'm taking a first course in Real Analysis and the instructor is using Arthur Mattuck's book. I'm struggling with a particular exercise: prove that the Newton's method of finding roots works for finding the root of $$f(x) = x - \cos(x)$$, i.e., it converges for the right value of $x$. It also gives me the hint to use that $|\sin(x)| < |x|, \forall x$ and $|1-\cos(x)| < \frac{x^{2}}{2}, \forall x$.
The chapter 4, which is the one the exercise belongs to, is all about error term analysis, I mean, if you can show that the limit of a sequence $e_{n} = a_{n} - L$ is $0$, where $a_{n}$ is another sequence and $L$ a real number, then the limit of $a_{n}$ is $L$.
Applying the method, it takes me to: $$x_{n+1} = x_{n} - \frac{x_{n} - \cos(x_{n})}{1+\sin(x_{n})}$$ And it gets like this: $$x_{n+1} = \frac{x_{n}\sin(x_{n}) - \cos(x_{n})}{1+\sin(x_{n})}$$ Writing using error terms: $$e_{n+1} = \frac{(e_{n}+L)\sin(e_{n}+L) - \cos(e_{n}+L)}{1+\sin(e_{n}+L)}$$ I know I should prove that ${e_{n}}$ goes to $0$ but I can't figure out the algebra.
I don't know that the direction you are heading will be very useful. Instead, let me note some things:
I believe a little more effort would show that sequence converges for any $x_0 \in (-\pi/2, 3\pi/2)$. But the sequence becomes undefined for $x_n = 2k\pi - \pi/2$. And for starting values outside $(-\pi/2, 3\pi/2)$, a few will eventually land on one of these undefined numbers. So you cannot prove convergence for all $x_0$.