[Figure: FiveStudy logo]
FiveStudy
Calculus BC · Unit 6 · 6.2
AP Calculus · Integration and Accumulation of Change
Approximating Areas with Riemann Sums
Same slices every time — the only choice is where you sample the height, and which way that pushes the error.
key idea$\int_a^b f\,dx\approx\sum_{i=1}^{n} f(x_i^*)\,\Delta x,\quad \Delta x=\frac{b-a}{n}$
The method
Slice $[a,b]$ into $n$ pieces of width $\Delta x=\dfrac{b-a}{n}$ and add the rectangle areas. The methods differ only in where on each slice you sample the height. The CED expects all four representations — graph, table, formula, words.
[Figure: left, right and midpoint Riemann rectangles compared]
Four subintervals on $[0,4]$ for an increasing, concave-up function. Left endpoints under-shoot every rectangle; right endpoints over-shoot; midpoints split the difference.| Method | Height of the $i$-th rectangle | Formula |
|---|
| Left Riemann sum | value at the left edge | $\Delta x\left[f(x_0)+f(x_1)+\dots+f(x_{n-1})\right]$ |
| Right Riemann sum | value at the right edge | $\Delta x\left[f(x_1)+f(x_2)+\dots+f(x_{n})\right]$ |
| Midpoint sum | value at the centre | $\Delta x\left[f(\bar x_1)+\dots+f(\bar x_n)\right]$ |
| Trapezoidal sum | average of the two edges | $\dfrac{\Delta x}{2}\left[f(x_0)+2f(x_1)+\dots+2f(x_{n-1})+f(x_n)\right]$ |
The left and right sums share every interior term; they differ only in the first and last. That is why their difference is $\Delta x\left[f(b)-f(a)\right]$.
| $x$ | $0$ | $1$ | $2$ | $3$ | $4$ |
|---|
| $g(x)$ | $1.0$ | $1.5$ | $3.0$ | $5.5$ | $9.0$ |
The same function as the figure, $g(x)=\tfrac12x^{2}+1$, sampled at the subinterval edges. Every example below uses these values.
Worked examples
EX 1
Left and right sums from the table
$n=4$, $\Delta x=1$Using the table, approximate $\displaystyle\int_{0}^{4} g(x)\,dx$ with a left and then a right Riemann sum, $n=4$.
1
Find $\Delta x$
$\Delta x=\dfrac{4-0}{4}=1$.
2
Left sum — drop the last value
Use $g$ at $x=0,1,2,3$: $L=1\left[1.0+1.5+3.0+5.5\right]=11.0$.
3
Right sum — drop the first value
Use $g$ at $x=1,2,3,4$: $R=1\left[1.5+3.0+5.5+9.0\right]=19.0$.
4
Check the relationship
$R-L=\Delta x\left[g(4)-g(0)\right]=1(9.0-1.0)=8.0$, which matches $19.0-11.0$. A quick way to catch an arithmetic slip.
EX 2
Which side of the true value?
monotonicityThe exact value is $\displaystyle\int_{0}^{4}\left(\tfrac12x^{2}+1\right)dx=\tfrac{x^{3}}{6}+x\Big|_{0}^{4}=\tfrac{64}{6}+4\approx 14.67$. Explain, without computing it, why $L$ had to be an under-estimate and $R$ an over-estimate.
1
Establish the direction
$g'(x)=x\gt 0$ on $(0,4)$, so $g$ is increasing throughout.
2
Reason about the left sum
On each slice the left edge is the lowest point of the function there, so every rectangle sits below the curve. $L$ under-estimates.
3
Reason about the right sum
The right edge is the highest point on each slice, so every rectangle pokes above the curve. $R$ over-estimates.
4
Confirm
$11.0\lt 14.67\lt 19.0$. The reasoning is what earns the marks — the arithmetic only confirms it.
5
State the general rule
For an increasing $f$: left under, right over. For a decreasing $f$: the reverse. Direction, not concavity, decides this one.
EX 3
Trapezoidal sum and the role of concavity
$T=\tfrac{\Delta x}{2}\left[\dots\right]$Approximate the same integral with a trapezoidal sum, $n=4$, and say whether it over- or under-estimates.
1
Apply the formula
$T=\dfrac{1}{2}\left[1.0+2(1.5)+2(3.0)+2(5.5)+9.0\right]$.
2
Compute
$=\dfrac{1}{2}\left[1.0+3.0+6.0+11.0+9.0\right]=\dfrac{30.0}{2}=15.0$.
3
Note the shortcut
$T=\dfrac{L+R}{2}=\dfrac{11.0+19.0}{2}=15.0$. The trapezoidal sum is always the average of the left and right sums.
4
Decide the direction of the error
$g''(x)=1\gt 0$, so $g$ is concave up. Each chord lies above the curve, so every trapezoid over-shoots: $T$ over-estimates.
5
Confirm
$15.0\gt 14.67$. Note the discriminator changed: over/under for trapezoids depends on concavity, not on whether $f$ is increasing.
EX 4
From a verbal description
words → numberWater flows into a tank. The rate is $2$ litres/min at noon, rising to $5$, then $9$, then $14$ litres/min at 3-minute intervals. Estimate the total volume added over the 9 minutes, and say whether your estimate is too big or too small.
1
Recognise the integral
Total volume is $\displaystyle\int_{0}^{9} r(t)\,dt$ where $r$ is the rate in litres/min and $t$ is minutes.
2
Set up the sum
Readings at $t=0,3,6,9$ give three subintervals with $\Delta t=3$.
3
Choose and compute
A left sum uses $t=0,3,6$: $3\left[2+5+9\right]=48$ litres.
4
Judge the direction
The rate is increasing throughout, so a left sum under-estimates. The true volume exceeds $48$ litres.
5
State units
$48$ litres — litres/min $\times$ min. Rate problems live or die on the units of the answer.
How wrong can it be?
Direction decides left/right; concavity decides trapezoid
Increasing ⇒ left under, right over. Concave up ⇒ trapezoid over, midpoint under. Mixing the two criteria is the classic error.
$n$ subintervals means $n+1$ data points
A table with five values supports at most four subintervals. Counting rectangles instead of points gives a sum one term too long.
Unequal widths are allowed
When a table's $x$-values are not evenly spaced you cannot factor out a single $\Delta x$ — compute each rectangle's width separately and add.
More slices, less error
Halving $\Delta x$ roughly halves the error of a left or right sum, and quarters it for the trapezoidal sum. Refinement does not change the over/under direction, only the size of the gap.