Let $x_0 \neq 0$ lies in the convergence ball of the solution $\alpha=2$ for the equation:
$f(x)=0$ where $f(x):=x^3-5x^2+8x-4$
What will be the convergence speed of Newton's method?
I did it this way:
Let $\phi(x)=x-\frac{f(x)}{f'(x)}=\frac{2x^3-5x^2+4}{3x^2-10x+8}$
$x_{n+1}=\phi(x_n)$
$e_n=x_n-2$
we have:
$e_{n+1}=x_{n+1}-2=\phi(x_n)-\phi(2)=\phi(e_n+2)-\phi(2)=\phi(2)+\phi'(2)e_n+o(e_n)-\phi(2)=\phi'(2)e_n+o(e_n)$
After calculations:
$\phi'(x)=\frac {2(3x^2-8x+5)}{(4-3x)^2}$
$\phi'(2)=\frac 12$
Thus: $e_{n+1}=\frac12 e_n +o(e_n)=e_n(\frac12+\frac {o(e_n)}{e_n})$
when $n \to \infty$ then $e_{n+1} \approx \frac12 e_n$
So we have linear convergence.
It is correct?
For convenience, we shift the variable with $t:=x+2$, and $f(t):=t^3+t^2$. Newton's iterations are
$$t'=\frac{2t^3+t^2}{3t^2+2t}=t\frac{2t+1}{3t+2}\sim\frac t2.$$
For small $t$, convergence is indeed linear (one more exact bit per iteration). This is because the root is double.