For some [[Function]] $f: \C \to \C$ in which we are attempting to find $f(p)=y_{p}$ (where $y_p$ is a target value, for example $0$), the error, denoted $\epsilon$, of an aproximation $x_i$ is the distance from $x_i$ to the wanted input $p$.
$\huge \epsilon = |x_{i}-p| $
$\huge \begin{align}
y_{i} &= f(x_{i}) \\
\end{align}$
If we can assume that this [[Function]] is [[Continuous]] and infinitely [[Derivative|Differentiable]].
$\huge y_{i}' = f'(x_{i}) \\ $
Assuming $p$ is the [[Roots|root]] of $f$,
$\huge \begin{align}
f(x) &= f(x_{i}) + f'(\xi)(x-x_{i}) \\
0 &=f(x_{i}) + f'(\xi)(p-x_{i}) \\
\end{align}$
$\huge p \leq \xi \leq x_{i} $
Where $\xi,f(\xi)$ relates to the [[Truncated Taylor Series]].
$\huge \begin{align}
f'(\xi)(p-x_{i}) &= -f(x_{i}) \\
p &= x_{i} - \frac{f(x_{i})}{f'(\xi)} \\
\epsilon &= |x_{i} - p |\\
\epsilon&= \left| \frac{f(x_{i})}{f'(\xi)} \right|
\end{align}$
We can turn this concrete error into an easier to reason about inequality if we have the [[Diminutive]] of $f'$.
$\huge \begin{align}
\underbrace{f'_{\dim}}_{\text{Diminutive}} &= \min |f'(x)| \\
\end{align}$
$\huge
\epsilon \leq \left| \frac{f(x_{i})}{f'_{\dim}} \right|
$
>[!example]
>$\huge f(x): \R \to \R $
>$\huge \forall x: f'(x) \geq 3 $
>$\huge f(3) = 0.02 $
>
>If $p$ is the root of $f$, calculate a bound for the error $\epsilon =|3-p|$.
>
>
>Because we know that $f'$ is at minimum $3$, and our current guess $x_i$ yields $0.02$, we can say the error is:
>
>$\huge \boxed{ \epsilon \leq \frac{0.02}{3}} $
>
>Even though $f'(\xi)$ is unknown, all other possible values of $f'(\xi)>3$ therefore would give us a lower $\epsilon$, thus the inequality holds.
>
>[!example]