Calculus · Applications of the definite integral
The definite integral was invented to measure areas, and that is still the best way to understand it. This post starts from the area under a curve, then shows how the same idea, "cut the figure into thin strips and add them up", handles areas between curves, curves given parametrically, and curves in polar coordinates, where the strips become thin slices of pizza.
Why. Cut the figure by vertical lines into thin strips of width \(dx\). Each strip is almost a rectangle of height \(f(x)\), so its area is about \(f(x)\,dx\). Adding all strips gives a Riemann sum, and as the strips get thinner, the sum tends to the integral. The error disappears in the limit because the "curved tops" of the strips become negligible compared to the strips themselves.
This "element of area" way of thinking, \( dP = (\text{height})\cdot(\text{width}) \), is the key to every formula below. Instead of memorizing formulas, ask: what does a thin strip look like, and what is its area?
The condition \( f(x) \ge 0 \) in the formula matters. Where \( f(x) < 0 \), the "height" \(f(x)\) of a strip is negative, so the integral counts that part of the area with a minus sign. The integral is a signed area: parts above the axis count positively, parts below negatively.
If \( f(x) \ge g(x) \) on \([a, b]\), a vertical strip between the curves has height \( f(x) - g(x) \), whatever their position relative to the \(x\)-axis:
\[ P = \int_a^b\big[f(x) - g(x)\big]\,dx. \]The formula of Section 1 is the special case \(g = 0\). The limits are usually the \(x\)-coordinates of the intersection points, found by solving \( f(x) = g(x) \).
If the curves cross inside the interval, the "upper" and "lower" curves swap. Then split at the crossing points, or integrate \( |f(x) - g(x)| \).
Nothing forces the strips to be vertical. If the region is bounded on the left and right by curves \( x = g(y) \) and \( x = h(y) \), with \( h(y) \ge g(y) \) for \( c \le y \le d \), horizontal strips of thickness \(dy\) and length \( h(y) - g(y) \) give
\[ P = \int_c^d\big[h(y) - g(y)\big]\,dy. \]If the curve is given by \( x = x(t) \), \( y = y(t) \), substitute in \( P = \int y\,dx \): \( dx = x'(t)\,dt \), so
\[ P = \int_{t_1}^{t_2}y(t)\,x'(t)\,dt, \]where \(t_1\) and \(t_2\) correspond to \(x = a\) and \(x = b\). (If the parameter runs "backwards", from right to left, the result comes out negative; take the absolute value.)
For a curve given in polar coordinates, \( \rho = f(\varphi) \), vertical strips are awkward. The natural pieces are thin sectors from the pole \(O\).
Why the \(\frac12\) and the square? A circular sector of radius \(\rho\) and angle \(\theta\) has area \( \frac12\rho^2\theta \) (a fraction \( \frac{\theta}{2\pi} \) of the disk \(\pi\rho^2\)). Divide the region into thin sectors of angle \(d\varphi\). Each is almost a circular sector of radius \( \rho = f(\varphi) \), with area
\[ dP = \frac12\rho^2\,d\varphi = \frac12\big[f(\varphi)\big]^2\,d\varphi, \]and adding them up gives the formula. The \(\frac12\rho^2\) plays the role that "height" \(f(x)\) played for vertical strips.
| Region | Element of area | Formula |
|---|---|---|
| Under \(y = f(x) \ge 0\) | vertical strip \(f(x)\,dx\) | \( \int_a^b f(x)\,dx \) |
| Between \(g(x) \le f(x)\) | \( [f(x) - g(x)]\,dx \) | \( \int_a^b[f(x) - g(x)]\,dx \) |
| Between \(g(y) \le h(y)\) | horizontal strip \( [h(y) - g(y)]\,dy \) | \( \int_c^d[h(y) - g(y)]\,dy \) |
| Parametric \(x(t), y(t)\) | \( y\,dx = y(t)x'(t)\,dt \) | \( \Big|\int_{t_1}^{t_2}y(t)x'(t)\,dt\Big| \) |
| Polar \(\rho = f(\varphi)\) | thin sector \( \frac12\rho^2\,d\varphi \) | \( \frac12\int_\alpha^\beta f(\varphi)^2\,d\varphi \) |
Try each one before opening the solution. More solved problems from this topic are in Applications of Definite Integrals.
\( \ln x \ge 0 \) for \(x \ge 1\): \( \int_1^e\ln x\,dx = \big[x\ln x - x\big]_1^e = 0 - (-1) = 1 \).
They meet at \(x = -1, 0, 1\). By symmetry, \( P = 2\int_0^1(x - x^3)\,dx = 2\Big(\frac12 - \frac14\Big) = \frac12 \). (Note that \( \int_{-1}^1(x - x^3)\,dx = 0 \): the signs cancel.)
An improper integral: \( \int_0^\infty e^{-x}\,dx = 1 \). An infinitely long region can have a finite area.
This is the circle of radius \(a\) through the pole, traced once for \( -\frac\pi2 \le \varphi \le \frac\pi2 \): \( P = \frac12\int_{-\pi/2}^{\pi/2}4a^2\cos^2\varphi\,d\varphi = 2a^2\cdot\frac\pi2 = \pi a^2 \).
One petal: \( 0 \le \varphi \le \frac\pi3 \), area \( \frac12\int_0^{\pi/3}a^2\sin^2 3\varphi\,d\varphi = \frac{a^2}{2}\cdot\frac\pi6 = \frac{\pi a^2}{12} \). Three petals: \( \frac{\pi a^2}{4} \).
\( \int_{-2}^2(4 - x^2)\,dx = 16 - \frac{16}{3} = \frac{32}{3} \). Check with Archimedes' rule: \( \frac23 \) of the enclosing rectangle \(4\times4\), i.e. \( \frac{32}{3} \).
All the area formulas are one idea in different coordinates: cut the figure into thin pieces whose area you know, and add them up. In Cartesian coordinates the pieces are rectangles, \(f(x)\,dx\); in polar coordinates they are circular sectors, \(\frac12\rho^2\,d\varphi\). The practical skill is choosing the slicing that makes the boundaries simplest, keeping an eye on signs, and using symmetry. The same idea will later give volumes, arc lengths and surface areas.
Related posts: integrating trigonometric functions (used in the polar examples) and improper integrals.