Calculus › Section E · Integrals › Topic E.3
Riemann sums and the definite integral
The strip-summing idea from last topic now gets its official name, its official symbol, and the precise limit that turns estimates into the exact signed area under a curve.
The idea gets a name
In E.2 we sliced a region into rectangles of equal width, used left, right or midpoint heights, and let the strips grow thin to find an area. Now loosen both choices: allow strips of different widths and any sample point in each strip. The sums are then called Riemann sums, after Bernhard Riemann, and their limit, when it exists, is the definite integral. This topic fixes the notation and the algebraic rules of the new object, ready for E.4 to connect it to antiderivatives.
Riemann sums, precisely
A partition of \([a, b]\) chops it into subintervals, not necessarily equal, of widths \(\Delta x_1, \ldots, \Delta x_n\). In each subinterval pick any sample point \(x_i^*\), and form the Riemann sum \(\sum_{i=1}^n f(x_i^*)\,\Delta x_i\). The norm of the partition, written \(\|\Delta\|\), is the width of the widest strip. The definite integral is the limit of these sums as the norm tends to zero:
Requiring the widest strip to vanish is stronger than merely sending \(n \to \infty\): it forbids any region from staying coarse while others refine. When the limit gives the same number regardless of how strips and samples are chosen, \(f\) is integrable on \([a, b]\).
Reading the symbol
Unpack \(\int_a^b f(x)\,dx\) once, slowly, because the notation will be with you for the rest of mathematics. The elongated S stands for sum. The \(a\) and \(b\) are the limits of integration, endpoints of the interval. The integrand \(f(x)\) plays the role of the strip heights, and \(dx\) is the infinitesimal width, the ghost of \(\Delta x_i\). So the symbol compresses the entire slice-add-shrink story into one glyph cluster. A definite integral evaluates to a number, unlike the indefinite integral of E.1, which names a family of functions.
Who is integrable
Not every function yields a definite integral, but our working functions are safe: every function continuous on \([a, b]\) is integrable there. More is true: a bounded function with only finitely many jumps on the interval is also integrable, so piecewise cost and billing functions, the ones with B.4's jump discontinuities, integrate perfectly well, jump points simply contribute no area. What integrability forbids is pathology like unboundedness within the interval or oscillation so dense that the sums cannot settle. For this course: continuous functions and tame piecewise ones are yours; anything stranger waits for a real analysis course.
Worked example: an area from pure geometry
Evaluate \(\int_{-3}^{3} \sqrt{9 - x^2}\,dx\) without any summing. Recognise the integrand: \(y = \sqrt{9 - x^2}\) is the upper semicircle of radius \(3\), since squaring gives \(x^2 + y^2 = 9\). The region under it over \([-3, 3]\) is exactly that semicircle, whose area is \(\frac{1}{2}\pi r^2 = \frac{1}{2}\pi(9) = \frac{9\pi}{2}\). So the integral equals \(\frac{9\pi}{2}\), about 14.14, read off a picture with no antiderivative needed. The lesson is durable: when the region is a known shape, the integral is a known formula, and no Riemann machinery is needed at all.
Signed area: the axis as an accountant
If the curve dips below the x-axis, its heights are negative, and every strip there contributes a negative area. The definite integral therefore computes signed area: positive above the axis, negative below. Many real quantities net out the same way. The flow of air into a room and back out, a bank balance with deposits and withdrawals, a car reversing down the drive: the integral nets them all. Questions that want total geometric area or total distance travelled are answered differently, by splitting at the zeros and adding pieces in absolute value, as the worked example below shows.
The algebraic properties
A handful of rules, each a one-line consequence of how sums behave: an integral over a single point is zero, \(\int_a^a f = 0\); swapping the limits flips the sign, \(\int_b^a f = -\int_a^b f\); constants pull out and sums split, exactly as with derivatives; and adjacent intervals reassemble, \(\int_a^b f + \int_b^c f = \int_a^c f\). Then the comparison rules: if \(f \ge 0\) on \([a, b]\), its integral is non-negative; and if \(f \le g\) throughout, then \(\int_a^b f \le \int_a^b g\). These let you bound, split, flip and combine integrals before you can evaluate a single one. For example, since \(0 \le x^2 \le x\) on \([0, 1]\), \(\int_0^1 x^2\,dx\) lies between 0 and \(\int_0^1 x\,dx = \frac{1}{2}\), the area of a triangle.
Worked example: displacement versus distance
A walker on a track has velocity \(v(t) = t - 2\) m/s for \(0 \le t \le 4\). Find the displacement and the total distance travelled, geometrically. The velocity graph is the line from \((0, -2)\) to \((4, 2)\), crossing zero at \(t = 2\). Under the axis on \([0, 2]\): a triangle of area \(\frac{1}{2}(2)(2) = 2\), signed \(-2\). Above the axis on \([2, 4]\): another triangle, area \(2\), signed \(2\). The integral, the net signed area, is \(-2 + 2 = 0\): the walker ends where they started. But total distance sums the two absolute pieces: \(2 + 2 = 4\) metres. Same graph, two questions, two answers, and the difference is the whole concept of signed area.
Common mistakes to avoid
The classics: forgetting that below-axis area subtracts, so the integral of a dip is not its geometric area; answering 'distance' with the net displacement; believing \(\int_a^b f\) must be positive whenever \(f\) takes positive values somewhere; swapping the limits without flipping the sign; treating \(dx\) as decoration, it marks the variable, and later substitutions will convert it; and applying geometry shortcuts to unknown curves, the semicircle trick worked because the region was recognisably circular, not because every curve is. Also: a definite integral depends on the interval, not the letter used, \(\int_a^b f(x)\,dx = \int_a^b f(t)\,dt\).
Key takeaways
A Riemann sum is any strip-add \(\sum f(x_i^*)\Delta x_i\); the definite integral is its limit as the widest strip vanishes, producing a number: the signed area from \(a\) to \(b\). Continuous functions, and bounded ones with finitely many jumps, are all integrable. The notation \(\int_a^b f(x)\,dx\) compresses the whole process, and the algebra of it lets you split, flip, scale and compare integrals, with area below the axis counting negative. Geometric regions can be read directly. The net-versus-total distinction is signed versus absolute area. What is still missing is a fast way to evaluate most integrals, and that is the fundamental theorem, E.4.
Flashcards
Flashcard
Σ f(xᵢ*) Δxᵢ over a partition
Test your understanding
Question 1 of 5
For any integrable f, what is ∫ from a to a of f(x) dx?
Question 2 of 5
If ∫ from 0 to 5 of f(x) dx = 7, what is ∫ from 5 to 0 of f(x) dx?
Question 3 of 5
What is the value of ∫ from −3 to 3 of √(9 − x²) dx?
Question 4 of 5
A walker has velocity v(t) = t − 2 m/s for 0 ≤ t ≤ 4. What is their displacement?
Question 5 of 5
The integral of f(x) from 0 to 2 is 3, and from 2 to 5 it is −4. What is the integral from 0 to 5?
Standards: 6
