Foundations — Taylor's remainder theorem — error estimation

This page assumes you have seen nothing. Before you can read the parent topic, you must own each symbol below. We build them in order — every later one leans on an earlier one.

1. A function $f(x)$ — the machine

Picture. Look at the blue curve below. Slide your finger to some input $x$ on the flat axis; the height of the curve directly above it is $f(x)$.

Why the topic needs it. The whole game is "copy a curve with a polynomial". No curve, no game.

2. The centre $a$ — where we stand

Picture. In the figure above, $a$ is the orange dot on the horizontal axis, and $f(a)$ is its height (orange). Our copycat polynomial will be perfect right at $a$ and drift as you walk away.

Why the topic needs it. Taylor approximation is local: excellent near $a$, worse far away. The distance you walk from $a$ is called $x-a$, and it controls the whole error.

3. Slope, and the derivative $f^{\prime }(x)$

Before symbols, a picture: zoom into any smooth curve far enough and it looks like a straight line. The steepness of that line is the slope.

Picture. Below, the green line just kisses the blue curve at $a$ (it touches without crossing). Its steepness is $f^{\prime }(a)$.

Why the topic needs it. The first useful copycat is the tangent line $f(a)+f^{\prime }(a)(x-a)$: same height, same slope as $f$ at $a$. Taylor just keeps adding more matching qualities.

4. Higher derivatives $f^{\prime \prime }$, $f^{\prime \prime \prime }$, ..., $f^{(k)}$

If the derivative is itself a machine (slope at every point), we can take its derivative — the rate at which the slope changes. That measures bendiness (curvature).

Picture. Look below: three curves with the same height and slope at $a$ but different bending. $f^{\prime \prime }$ is what tells them apart — it is the first thing the slope alone cannot capture.

Why the topic needs it. A degree-$n$ Taylor polynomial forces $f$ and the polynomial to share $f(a),f^{\prime }(a),\dots ,f^{(n)}(a)$ — all $n+1$ of these matched qualities. The error is then governed by the very next unmatched one, $f^{(n+1)}$.

5. Factorial $k!$ — the fast-growing divisor

Picture. Factorials explode. Below, plotted against $k$, you can see how quickly $k!$ towers over ordinary powers — it crushes whatever sits above it in a fraction.

Why the topic needs it. In the error bound $\frac{M}{(n+1)!}|x-a{|}^{n+1}$, the factorial in the denominator grows so violently that even a big top can be squashed to near zero — this is why adding a few Taylor terms makes the error tiny.

6. Sigma notation $\sum$ — a compact "add all of these"

Why the topic needs it. The Taylor polynomial $P_n(x)={\sum }_{k=0}^n\frac{f^{(k)}(a)}{k!}(x-a{)}^k$ is a whole pile of terms; $\sum$ writes them all in one breath.

7. Absolute value $|\cdot |$ — size, ignore the sign

Why the topic needs it. Error can be positive (undershoot) or negative (overshoot); we only care how big. So the theorem bounds $|R_n(x)|$, the magnitude of the gap.

8. The bound $M$ and "$\le$" — a ceiling we can guarantee

Picture idea. Imagine the curve of $|f^{(n+1)}|$ over the interval; $M$ is a flat horizontal ceiling drawn above its highest peak. We don't need the exact peak — any honest ceiling works.

Why the topic needs it. $M$ is what turns the un-computable exact remainder into a usable, provable error estimate.

9. Putting the symbols together

Now every piece of the headline formula reads in plain words:

Read the [!mnemonic] in the parent — "Next derivative, Next factorial, Next power" — now that you know what each word points to.

Prerequisite map

Equipment checklist

Give the plain-words answer aloud before revealing.