Chapter 08 · Integrals
Your notes
Bookmarks
Calculator
08.01
Approximating Area
Chop a region into thin rectangles and add up their areas. Left, right and midpoint sums give estimates, sigma notation writes them compactly, a table of rates gives a total, and more rectangles give a better answer.
About 40 minutes
You'll learn to
- →Compute left, right and midpoint Riemann sums
- →Read and evaluate sums in sigma notation
- →Tell whether an estimate is too high or too low
- →Estimate a total (distance, volume, people) from a table of rates
- →Write a Riemann sum with n rectangles and see how more rectangles improve the estimate
Summary video coming soon
A short video recap of the whole lesson is on its way.
1. Rectangles under a curve
Video coming soon
A short video for this part is on its way. Everything you need is in the text below.
How much area lies under between and ? There's no geometry formula for a curved region, so we approximate it with shapes we can handle: rectangles.
Split into strips of equal width, build a rectangle on each strip, and add up their areas.
When is cut into equal pieces, each piece (a subinterval or strip) has width . The cut points are .
Each rectangle has width . What's left is its height, which is the function's value at one sample point in its strip. Which point you choose gives three different estimates:
uses the left end of each strip as the sample point: the heights are .
uses the right end of each strip: the heights are .
uses the middle of each strip: the heights are at .
Each is a Riemann sum, a sum of over the strips.
A sum of rectangle areas over the strips of , where is the sample point chosen in the -th strip.
Take on with . Then , and the cut points are .
Open in grapher (opens in a new tab)
- Left: .
- Right: .
- Midpoint (at ): .
The true area is exactly (you'll see how in section 08.04). The endpoint sums miss by a lot, one low and one high; the midpoint sum is much closer.
The only choice in a Riemann sum is the height of each rectangle. The width is always .
- Using the wrong number of heights. Left and right sums each use values of ; the cut points are numbers, and the left sum leaves out , the right sum leaves out .
- Dividing by . strips need . The is the number of cut points.
- Forgetting to multiply by . Adding up the heights gives a number with the wrong size; each height must be multiplied by the width.
Split into strips, build a rectangle on each, and add the areas. The height comes from the left end (left sum), the right end (right sum) or the middle (midpoint sum).
on , : , and the left sum is .
2. Sigma notation for sums
Video coming soon
A short video for this part is on its way. Everything you need is in the text below.
A Riemann sum with rectangles has terms. To write one down, and to work with it, we need a compact notation. You met it in section 01.12; here is what we need for Riemann sums.
means : the sum of as the index runs from up to .
For example, . The index can start anywhere, and the number of terms is : the left sum has terms, and so has the right sum .
Two rules make sums manageable, and two formulas finish the job:
Sum rules
A constant factor pulls out, a sum of sums splits, and adding the constant a total of times gives .
Evaluate .
- Split: .
- Use the formulas with : .
- Add: . (Check by listing the terms: .)
Evaluate .
- Split and pull out the : .
- The first sum is ; the second is .
- .
Sigma notation is only shorthand. If a sum is small, write out the terms to check your work.
- Dropping the constants. subtracts times, not once.
- Squaring the sum. : , but .
- Counting terms wrong. From to there are terms, not .
3. Too high or too low?
Video coming soon
A short video for this part is on its way. Everything you need is in the text below.
An estimate is more useful when you know which way it errs. The direction comes from how behaves on each strip.
Suppose is monotonic on .
- If is increasing, then .
- If is decreasing, then .
Why it’s true
An increasing function is smallest at the left end of each strip and largest at the right end. So on each strip, the left rectangle fits inside the region under the curve and the right rectangle covers it. Adding up the strips gives . For a decreasing function the heights are largest on the left and smallest on the right, which swaps the two.
This also gives a bracket: for a monotonic function, the true area lies between and .
Averaging the two sums gives an estimate that is usually better than either:
. It's the sum of the areas of trapezoids: each strip's top is the slanted line joining the curve at the strip's two ends.
Take on with , so . This is decreasing.
Open in grapher (opens in a new tab)
- Left: . Too high: the picture shows the rectangles poking above the curve.
- Right: . Too low.
- Trapezoid: .
So the area is between and . (It is .)
If is increasing, is the trapezoid estimate too high, too low, or neither?
Show the answerHide the answer
Neither can be said from monotonicity alone: , and the average of the two lies in between, so could be on either side of the area. (It is on one side when is concave, which is the subject of section 08.11.)
" is decreasing, so the left sum is an underestimate."
A decreasing function is tallest on the left of each strip. The left rectangle is the tall one, so it sticks out above the curve: an overestimate.
On an increasing function, left sums are too low and right sums are too high. On a decreasing one, it's the other way around.
For on , : (too high), (too low), .
4. Totals from a table of rates
Video coming soon
A short video for this part is on its way. Everything you need is in the text below.
Riemann sums aren't only about shapes. If a car travels at mph for hours, it covers miles: rate times time. When the rate changes, we can't use one number, but we can use one on each short interval. A rate measured at several times gives a rectangle for each interval, whose area is a total amount, not a geometric area.
The same works for any rate: a flow rate (gallons per minute) gives volume, a rate of people entering a stadium gives attendance, and power (watts) gives energy.
The area under a rate graph is the total change in the quantity. Its units are rate-units × time-units: (m/s) × s = m.
A table gives values at certain times, so it gives a Riemann sum: with the left value of each interval you get , with the right value you get , and with a reading in the middle of each interval, .
A runner's speed (m/s) is recorded every seconds:
| (s) | |||||
|---|---|---|---|---|---|
| (m/s) |
Estimate the distance in the seconds.
- There are intervals, each s long.
- Left sum (the speed at the start of each interval): m.
- Right sum (the speed at the end of each): m.
- Average: m.
Her speed isn't monotonic (it rises and then falls), so we can't tell which sum is too big: the best we can say is "about m".
- Adding the rates and stopping. isn't a distance: each rate must be multiplied by how long it lasted.
- Using all the rows. A table with readings has only intervals. A left sum uses the first four readings, a right sum the last four.
- Mixing up units. If the rate is per minute and the time is in hours, convert first.
Rate × time is the amount that piles up. With a table of rates, each row gives a rectangle: rate × length of time.
Speeds m/s at the starts of four -second intervals: left sum m.
5. More rectangles, better estimates
Video coming soon
A short video for this part is on its way. Everything you need is in the text below.
The picture suggests it, and the numbers confirm it: thinner rectangles fit the curve more closely. With a formula for the sum in terms of , we can see exactly what happens.
For on we have and .
- Write the sum: .
- Split it: .
- Use and : .
The sums close in on , and the error is : ten times as many rectangles means ten times less error. The region under is a trapezoid with parallel sides and and width , whose area is . The sums are heading for the right number.
The exact area is the number the sums approach as . Making that precise, as a limit, is the definite integral of section 08.02.
For a curve, the same thing happens, but you can't check it with a formula from geometry. A calculator or computer can compute for huge , and it's what you'll use in section 08.11 to estimate integrals that have no neat answer.
- Thinking an estimate is exact at some . For a curved graph, no finite gives exactly the area (apart from lucky cases); it's the limit that does.
- Expecting the sums to grow forever. More rectangles means each is thinner. The widths shrink as fast as the number grows, so the total settles down.
As grows, , and all squeeze in on the same number: the exact area.
For on , . As , .
Ready for more? The Practice tab has 11 more problems on the whole lesson.
Definitions
- Strip width
When is cut into equal pieces, each piece (subinterval) has width .
- Left-endpoint sum
: the Riemann sum whose rectangles have height at the left end of each strip.
- Right-endpoint sum
: the Riemann sum whose rectangles have height at the right end of each strip.
- Midpoint sum
: the Riemann sum whose rectangles have height at the middle of each strip.
- Riemann sum
A sum of rectangle areas over the strips of , with a sample point chosen in each strip.
- Sigma notation
: the sum of as runs from to .
- Trapezoid estimate
: the average of the left and right sums.
Formulas
Strip width
Left-endpoint sum
heights, from to
Right-endpoint sum
heights, from to
Midpoint sum
Trapezoid estimate
Sum rules
a constant factor pulls out; a constant summed times is
Sum of the first n integers
Sum of the first n squares
Study
Apply it
Real-world problems that take several steps. Work through them one step at a time, with each step checked as you go.
Practice
11 practice problems with instant feedback, hints and worked solutions.
Sign up free to practice