Net change theorem: Visual walkthrough — Maths deep dive

The cast of characters (before Step 1)

We need two objects and one honest picture, or nothing below makes sense.

The whole theorem connects a picture of $F$ (a curve going up and down) to a picture of $F^{'}$ (an area under a different curve). Look at both side by side:

On the left board, the blue curve is $F$ ; the total vertical rise from $a$ to $b$ — the pink bracket $F (b) - F (a)$ — is the net change we want. On the right board is $F^{'}$ , the slope-o-meter reading of that same curve. The claim we will prove is that the yellow area on the right equals the pink height on the left. Different pictures, same number.

Step 1 — Slice the interval into thin strips

WHAT. Chop $[a, b]$ into $n$ equal little pieces. Mark the cut points $a = x_{0} < x_{1} < x_{2} < \dots < x_{n} = b .$ Here $x_{0}$ is just another name for $a$ , $x_{n}$ another name for $b$ , and each $x_{i}$ is the $i$ -th fencepost. Each strip has the same width $Δ x = total length \div number of pieces \frac{b - a}{n} .$ The symbol $Δ$ (Greek "delta") means "a change in," so $Δ x$ literally reads "a small change in $x$ " — the strip width.

WHY. Change over the whole long interval is hard to reason about — $F^{'}$ wiggles. But over a strip so thin that $F$ barely bends, change is easy: it is almost just (slope) $\times$ (width). Slicing trades one hard problem for $n$ easy ones.

PICTURE.

The board shows the interval cut into $n$ equal chalk strips, each of width $Δ x$ . Notice the fenceposts $x_{0}, \dots, x_{n}$ : there are $n + 1$ posts making $n$ gaps. That off-by-one is the classic trap — count the gaps, not the posts.

Step 2 — Total change is a telescoping sum

WHAT. The full rise $F (b) - F (a)$ is just the sum of every strip's own little rise: $F (b) - F (a) = \sum_{i = 1}^{n} rise across strip i [F (x_{i}) - F (x_{i - 1})] .$ The $\sum$ (Greek "sigma") means "add these up," and $i$ runs from $1$ to $n$ , once per strip. The term $F (x_{i}) - F (x_{i - 1})$ is how much $F$ climbed across strip number $i$ alone.

WHY. Write the sum out longhand and watch the middle values annihilate: $(F (x_{1}) - F (x_{0})) + (F (x_{2}) - F (x_{1})) + (F (x_{3}) - F (x_{2})) + \dots + (F (x_{n}) - F (x_{n - 1})) .$ Every interior height appears once with a $+$ and once with a $-$ , so it cancels. Only the outermost survivors $F (x_{n}) = F (b)$ and $- F (x_{0}) = - F (a)$ remain. This "everything cancels through the middle" pattern is a telescoping sum — like a collapsing telescope. It is exact, not an approximation.

PICTURE.

Each pink vertical bar is one strip's rise $F (x_{i}) - F (x_{i - 1})$ . Stack them head to tail up the staircase and the total height of the whole staircase is exactly the pink bracket $F (b) - F (a)$ — nothing lost, nothing double-counted.

Step 3 — Trade each little rise for slope × width (Mean Value Theorem)

WHAT. For a single thin strip, replace its rise by a slope times its width. The Mean Value Theorem guarantees there is some point $x_{i}^{*}$ inside strip $i$ where the curve's instantaneous slope equals the strip's average slope: $F (x_{i}) - F (x_{i - 1}) = slope at that point F^{'} (x_{i}^{*}) \cdot strip width Δ x .$ The star on $x_{i}^{*}$ just tags "the special point inside strip $i$ ." Rearranged, this says $\frac{rise}{run} = F^{'} (x_{i}^{*})$ — a slope, by definition.

WHY. We want to bring the rate $F^{'}$ into the story, because $F^{'}$ is what the theorem is about. The Mean Value Theorem is the bridge: it converts a rise we cannot compute into a rate (which we know) times a width (which we chose). It works because over a strip the curve must, at least once, be as steep as its own start-to-end average — you cannot get from bottom to top without matching the average slope somewhere.

PICTURE.

Zoom into one strip. The pink chord connects the two ends; the yellow tangent line, drawn at $x_{i}^{*}$ , is parallel to it. Same slope. So the true rise (chord's vertical drop) equals $F^{'} (x_{i}^{*}) Δ x$ — a genuine rectangle of height $F^{'} (x_{i}^{*})$ and width $Δ x$ .

Step 4 — Recognise a Riemann sum of rectangles

WHAT. Substitute Step 3's replacement into Step 2's sum: $F (b) - F (a) = \sum_{i = 1}^{n} rectangle height F^{'} (x_{i}^{*}) rectangle width Δ x .$ Each term $F^{'} (x_{i}^{*}) Δ x$ is the area of a rectangle: height $F^{'} (x_{i}^{*})$ , width $Δ x$ .

WHY. This is the exact shape of a Riemann sum — a pile of skinny rectangles sitting under the graph of $F^{'}$ . We engineered it on purpose: the left side is still the exact net change of $F$ , and now the right side is manifestly an approximation of the area under $F^{'}$ .

PICTURE.

The yellow rectangles march under the $F^{'}$ curve. Their total area equals the pink net-change height from the left board of Step 1 — for any number of strips, because Step 3 was exact strip by strip.

Step 5 — Take the limit: rectangles become the integral

WHAT. Let the strips get infinitely thin, $n \to \infty$ (so $Δ x \to 0$ ). The rectangle-sum sharpens into the exact area, which is what the integral symbol means: $F (b) - F (a) = lim_{n \to \infty} \sum_{i = 1}^{n} F^{'} (x_{i}^{*}) Δ x = \int_{a}^{b} F^{'} (x) d x .$ The tall $\int$ is a stretched "S" for "sum"; the $d x$ is the width $Δ x$ shrunk to a whisker. Reading right to left: sum of (rate $\times$ tiny width) over $[a, b]$ .

WHY. As the strips thin, two things vanish together: the curve's bending inside each strip, and any error in approximating rise by slope $\times$ width. In the limit the wobbly rectangle-tops smooth exactly onto $F^{'}$ , and the sum is the area. The left side never moved — it was $F (b) - F (a)$ the whole time — so the two are equal.

PICTURE.

Three boards: coarse rectangles, finer rectangles, and the smooth filled area. The jagged top settles onto the $F^{'}$ curve; the yellow area is now $\int_{a}^{b} F^{'} (x) d x$ , dead equal to the pink height. $■$

Step 6 — The edge case: when the rate goes negative

WHAT. Nothing above assumed $F^{'} \geq 0$ . If $F^{'} (x) < 0$ on part of $[a, b]$ , that strip's rectangle has negative height, so it contributes negative area — and $F$ goes down there.

WHY. A negative rate means the stuff is decreasing (water draining, car reversing). The proof still holds line for line: Step 3's slope $F^{'} (x_{i}^{*})$ is simply a negative number, so the rise is negative. The sum then nets the ups against the downs. This is exactly why the theorem gives net change, not total distance.

PICTURE.

The $F^{'}$ curve dips below the axis on the middle stretch. Rectangles above the axis are yellow (positive, $F$ rising); rectangles below are pink (negative, $F$ falling). The integral adds them signed: yellow area minus pink area. That signed total is precisely $F (b) - F (a)$ — see the blue $F$ curve rise, then fall, then rise, ending at the pink net height.

Step 7 — The degenerate cases: check the machine doesn't break

WHAT & WHY. Three limiting inputs, each a sanity test.

Empty interval, $a = b$ . Zero strips of zero total width: $\int_{a}^{a} F^{'} d x = F (a) - F (a) = 0$ . No time elapsed, no change. ✓
Constant quantity, $F^{'} \equiv 0$ . Every rectangle has height $0$ , so the area is $0$ and $F (b) - F (a) = 0$ . Flat stuff never changes. ✓
Constant rate, $F^{'} \equiv c$ . Every rectangle has the same height $c$ ; total area $= c \cdot (b - a)$ , a plain rectangle. And indeed net change $= c (b - a)$ — steady rate for a fixed time. ✓

PICTURE.

Three mini-boards showing the collapsed strip ( $a = b$ ), the flat-zero rate, and the single fat rectangle of a constant rate. In every case the yellow area and the pink height agree — the theorem degrades gracefully.

The one-picture summary

Read the arrow chain: the pink rise of $F$ (left) equals the staircase of little rises (telescoping), equals the sum of yellow rectangles (MVT), equals — in the limit — the yellow area under $F^{'}$ (right). One number, seen four ways.

Recall Feynman: tell the whole walkthrough to a 12-year-old

Picture a hill. The height of the hill above sea level is $F$ ; how steeply it slopes right where you stand is $F^{'}$ . I want to know how much higher the far end $b$ is than the near end $a$ — that is the net change.

I slice the walk into tiny steps (Step 1). Across each tiny step the ground barely bends, so the little bit of climb is just "how steep it is here" times "how long the step is" (Step 3 — the Mean Value Theorem promises there's a spot in each step whose steepness is exactly right). If I stack up all those little climbs, the middles cancel and I'm left with only "top minus bottom" (Step 2 — telescoping). Each little climb is a skinny rectangle, height = steepness, width = step length, so the whole climb is a pile of rectangles under the steepness graph (Step 4). Make the steps infinitely small and that pile becomes the smooth area under the steepness curve — the integral (Step 5).

If the hill goes downhill somewhere, that step's climb is negative and gets subtracted, so I get the net height gained, up minus down (Step 6). And if I don't walk at all, or the ground is flat, or it slopes the same everywhere, the machine still gives the obvious answer (Step 7). That's it: add up steepness-times-tiny-width and you recover the total change in height.

Connections

Net change theorem — the parent result these pictures prove.
Fundamental Theorem of Calculus — the same boxed equation, stated abstractly.
Definite Integral as a Riemann Sum — Steps 4–5 are this definition.
Mean Value Theorem — the engine of Step 3.
Telescoping Sums — the cancellation of Step 2.
Displacement vs Distance — the net-vs-total split of Step 6.
Marginal Cost and Revenue — where signed area meets economics.