Riemann Sum Table for Data Values

When you have a table of x and f(x) values with unequal interval widths, the width Δx is no longer a single number you multiply every height by. You must compute a separate Δx for each subinterval. This is the single most common mistake on AP Calculus free-response questions involving a Riemann sum table. The AP Calculus AB/BC released free-response questions (2000 AB3 through 2024 AB3, and the BC equivalents) consistently test this. The College Board AP Calculus CED, Topic 6.2, requires you to approximate areas with Riemann sums from tables, and the scoring guidelines award points for a correct setup with the correct widths. You do not need the function's formula. You only need the data points and the ability to identify left, right, midpoint, or trapezoidal rules.

A Riemann sum from a table means you are working from discrete data points, not a continuous formula. You cannot evaluate f(x) at arbitrary points. You can only use the x-values and f(x)-values the table gives you. This affects every method differently.

The Definite Integral And The Finite Partition

The definite integral is the limit of Riemann sums as the number of subintervals goes to infinity. A table is a finite partition. You cannot take the limit. You are stuck with an approximation, and the error depends on how coarsely the function is sampled.

For a non-uniform partition, the mesh (the width of the widest subinterval) determines the worst-case error. You cannot use the standard error bound formulas because they require the second derivative of f, which you do not have from a table. You are trusting that the table's points capture the function's behavior well enough.

Left and right sums with a table use the same logic as with a formula: take the height from the left or right endpoint of each subinterval and multiply by that subinterval's width.

Worked Example 1: A Table of Speeds

A car's velocity is recorded at irregular times. The table gives t (seconds) and v(t) (feet per second). The data: (0, 0), (2, 10), (5, 25), (10, 40). Interval widths: from 0 to 2 is 2 seconds, from 2 to 5 is 3 seconds, from 5 to 10 is 5 seconds.

Left Riemann sum: use v(0)=0 on [0,2], v(2)=10 on [2,5], v(5)=25 on [5,10]. Computation: 0(2) + 10(3) + 25(5) = 0 + 30 + 125 = 155 feet.

Right Riemann sum: use v(2)=10 on [0,2], v(5)=25 on [2,5], v(10)=40 on [5,10]. Computation: 10(2) + 25(3) + 40(5) = 20 + 75 + 200 = 295 feet.

The true distance traveled is unknown from the table alone. The left sum is an underestimate if the function is increasing (velocity is rising). The right sum is an overestimate. This over/underestimate justification is what the AP Calculus scoring guidelines expect: reference monotonicity, not concavity.

A midpoint sum requires the function value at the midpoint of each subinterval. A table that only gives values at the partition points does not give you midpoints. You cannot compute a midpoint sum unless the table explicitly provides f(x) at those midpoints.

On the AP exam, tables are designed so that midpoint sums are possible only when the subintervals are uniform and the table includes the midpoints, or when the problem states "use the midpoint of each subinterval" and the midpoint coincides with a table value. If the table has non-uniform intervals and does not list midpoints, you cannot do a midpoint sum. The AP Calculus CED Topic 6.2 lists midpoint sums as a tested method, but the exam only asks for them when the data supports it.

Avoid the trap of averaging the endpoints and pretending that is the midpoint. The midpoint value f((x_i-1 + x_i)/2) is not the average of f(x_i-1) and f(x_i) unless the function is linear. You do not know if it is linear. Use only what the table gives you.

A trapezoidal sum from a table works exactly like the trapezoidal rule with a formula, but each trapezoid's height is the average of the two endpoint heights. For a non-uniform partition, you still use the same formula: area of each trapezoid is ((f(x_i-1) + f(x_i))/2) * Δx_i.

Worked Example 2: Unequal Interval Trapezoidal Sum

Use the same velocity table: (0, 0), (2, 10), (5, 25), (10, 40).

First trapezoid: [0,2], heights 0 and 10, width 2. Area = ((0+10)/2)*2 = 10.

Second trapezoid: [2,5], heights 10 and 25, width 3. Area = ((10+25)/2)*3 = (35/2)*3 = 52.5.

Third trapezoid: [5,10], heights 25 and 40, width 5. Area = ((25+40)/2)*5 = (65/2)*5 = 162.5.

Total: 10 + 52.5 + 162.5 = 225 feet.

The trapezoidal sum is exactly the average of the left and right sums only when using the same partition. Here, (155 + 295)/2 = 225. That holds because the partition is the same for all three sums. If the partition differed, the average would not be defined.

The trapezoidal sum is exact for linear functions. Since the true velocity function is unknown, you cannot say whether the trapezoidal sum is an overestimate or underestimate without knowing concavity. The AP scoring guidelines accept a statement about concavity if the function is known to be concave up or down, but from a table alone, you cannot justify it.

On the AP Calculus free-response exam, the setup for a Riemann sum table problem is worth more points than the arithmetic. The scoring guidelines for the released FRQs (such as 2022 AB3 and 2023 AB3) award points for:

1. Identifying the correct subinterval widths. Write each Δx_i explicitly. Do not just write a single Δx. Use sigma notation with the correct index to show the sum over the subintervals.

2. Using the correct function values. Write f(x_i) or f(x_i-1) or (f(x_i-1)+f(x_i))/2, depending on the method. A generic "f(x)" without a specific x loses the point.

3. Producing a numerical answer. The final boxed number must match the arithmetic from your setup. Premature rounding loses accuracy. Keep at least three decimal places until the final step.

Templates For Each Method

A setup template for a left Riemann sum from a table with n subintervals: Σ_{i=1}^{n} f(x_{i-1}) * (x_i - x_{i-1}). For a right sum: Σ_{i=1}^{n} f(x_i) * (x_i - x_{i-1}). For a trapezoidal sum: Σ_{i=1}^{n} ((f(x_{i-1}) + f(x_i))/2) * (x_i - x_{i-1}).

The index i runs from 1 to n, where n is the number of subintervals. This is critical because the table gives you n+1 data points. Many students use too many points or miscount the number of rectangles.

The two most frequent errors on a Riemann sum table problem are assuming equal subinterval widths and using the wrong number of points.

Equal Δx mistake: When a table gives points like (0,5), (2,7), (5,3), the widths are 2, then 3. Students who write Δx = (5-0)/2 = 2.5 are wrong because the partition is not uniform. The correct widths are 2 and 3. Check every adjacent pair of x-values. Do not assume the problem is neat.

Counting Points And Indexing Errors

Too many points mistake: A table with 5 points gives 4 subintervals. For a left sum, you use the first 4 function values. For a right sum, you use the last 4. For a trapezoidal sum, you use all 5. Counting errors shift the entire sum by one rectangle, which changes the approximation substantially.

Another failure mode: using the wrong index in sigma notation. Writing Σ f(x_i) Δx when the sum should start at i=1 and end at i=n is an indexing error. The AP scoring guidelines penalize this harshly because it shows a misunderstanding of the Riemann sum definition.

On the AP Calculus AB exam, the typical FRQ with a table (question 3) also asks for an overestimate or underestimate justification. The only correct justification is monotonicity: if the function is increasing, the left sum is an underestimate and the right sum is an overestimate. Do not mention concavity unless the problem asks for trapezoidal or midpoint sum behavior. The College Board scoring guidelines explicitly reject concavity-based justifications for left and right sums.

Of the four methods tested on the AP exam (left, right, midpoint, trapezoidal), only the midpoint sum requires data that is not at the partition points. A midpoint sum from a table is only possible when the table lists f(x) at the midpoints of each subinterval, or when the subintervals are uniform and the midpoints coincide with table entries (for example, a table with values at x=0, 1, 2, 3, 4 allows a midpoint sum using x=0.5, 1.5, 2.5, 3.5 if those are given).

If the table does not provide midpoints, you cannot compute a midpoint sum. Do not fabricate values by averaging. The trapezoidal sum, by contrast, uses only the endpoint values that the table already has. It is always computable from any table that provides f(x) at the partition points.

Simpson's rule is not tested on AP Calculus AB. It appears in BC-only contexts and requires an even number of subintervals of equal width, which tables with unequal intervals do not satisfy.

A Riemann sum from a table where some f(x) values are negative requires careful handling. The signed area means terms from subintervals where the function is below the x-axis contribute negative values. This is not a mistake; it is the definition of the definite integral.

For example, a table with (0, -3), (2, 1), (4, 5) has a negative left-sum term on [0,2] when using the left endpoint f(0)=-3. The product (-3)*2 = -6 correctly represents negative signed area. Students who change the sign or drop the negative term produce an incorrect approximation.

On AP FRQs, rate tables (like water flow rates or velocities) are usually non-negative, but position tables can have negative values. Always check the sign of each f(x) before multiplying.

Method	What You Need From the Table	Works With Unequal Intervals?	Data Points Used for n Subintervals
Left Riemann Sum	Left endpoint of each subinterval	Yes	First n function values
Right Riemann Sum	Right endpoint of each subinterval	Yes	Last n function values
Midpoint Riemann Sum	Midpoint of each subinterval	Only if midpoints are given	n midpoint values
Trapezoidal Sum	Both endpoints of each subinterval	Yes	All n+1 function values

The single thing that most often goes wrong is a student treating a non-uniform table as if it had equal intervals. On the AP exam, the problem will give you intervals of different widths. The scoring guidelines penalize a single Δx because it shows you did not read the table. Check every pair of adjacent x-values. Write each width next to the interval. Then build the sum. This one habit eliminates the majority of point deductions on tabular Riemann sum FRQs.