Riemann Sums (Everything to Know for AP Calculus)

What You Need to Know

The big idea

A Riemann sum is a way to approximate the definite integral $\int_a^b f(x)\,dx$ by adding up areas of thin rectangles (or trapezoids) over a partition of $[a,b]$. On the AP Calc AB exam, Riemann sums show up in two main ways:

• Approximating $\int_a^b f(x)\,dx$ from a graph, formula, or table.
• Recognizing that a scary-looking sum is really an integral (and sometimes evaluating it quickly).

Core definition (the one to know cold)

Partition $[a,b]$ into $n$ subintervals:

• $a = x_0 < x_1 < \dots < x_n = b$
• widths $\Delta x_i = x_i - x_{i-1}$
• choose a sample point $c_i$ in each subinterval $[x_{i-1},x_i]$

Then a Riemann sum is:

$\sum_{i=1}^n f(c_i)\,\Delta x_i$

If you refine the partition so the subintervals get very small, the Riemann sums approach the definite integral:

$\int_a^b f(x)\,dx = \lim_{\|P\|\to 0} \sum_{i=1}^n f(c_i)\,\Delta x_i$

(For equal-width partitions, $\Delta x = \frac{b-a}{n}$ and the limit is usually written as $n\to\infty$.)

Rectangles vs “area”

• $\int_a^b f(x)\,dx$ is signed area.
• If $f(x) < 0$ on a region, that part contributes negative value.

Don’t automatically treat integrals as “area” unless the problem specifically asks for area between the curve and the x-axis, which would be $\int_a^b |f(x)|\,dx$.

Step-by-Step Breakdown

A) Approximating an integral using a Riemann sum (LRAM/RRAM/MRAM)

Use this when you’re asked to approximate $\int_a^b f(x)\,dx$ with $n$ rectangles.

1. Identify $a$, $b$, and $n$.

2. Compute the width (equal partitions):

   $\Delta x = \frac{b-a}{n}$

3. Build partition points:

   $x_i = a + i\Delta x$

4. Choose your sample points based on the method:

   • LRAM (Left): $c_i = x_{i-1}$
   • RRAM (Right): $c_i = x_i$
   • MRAM (Midpoint): $c_i = \frac{x_{i-1}+x_i}{2}$

5. Compute the sum:

   $\sum_{i=1}^n f(c_i)\,\Delta x$

Mini-worked setup (generic)

Approximate $\int_a^b f(x)\,dx$ with **RRAM** and $n=4$:

• $\Delta x = \frac{b-a}{4}$
• right endpoints: $x_1=a+\Delta x,\ x_2=a+2\Delta x,\ x_3=a+3\Delta x,\ x_4=b$
• RRAM:

$R_4 = \Delta x\,[f(x_1)+f(x_2)+f(x_3)+f(x_4)]$

B) Approximating from a table (possibly unequal widths)

Use this when the problem gives you values like $f(0), f(1.2), f(2.0),\dots$ where spacing might not be constant.

1. List the subintervals $[x_{i-1},x_i]$.
2. Compute each width $\Delta x_i = x_i - x_{i-1}$.
3. Pick the correct height (left/right/midpoint) as instructed.
4. Add: $\sum f(\text{chosen } x)\,\Delta x_i$.

If widths are not equal, you cannot factor out a single $\Delta x$.

C) Turning a sum into an integral (pattern recognition)

You’ll often see something like:

$\lim_{n\to\infty} \sum_{i=1}^n f\left(a + i\Delta x\right)\,\Delta x$

Do this:

1. Identify $\Delta x$ as something of the form:

   $\Delta x = \frac{b-a}{n}$

2. Identify the sample point pattern inside $f(\cdot)$ (left/right/midpoint form).

3. Rewrite as:

   $\int_a^b f(x)\,dx$

Quick pattern example

If

$\lim_{n\to\infty} \sum_{i=1}^n \sqrt{1+\left(\frac{3i}{n}\right)^2}\,\frac{3}{n}$

then $\Delta x = \frac{3}{n}$ so $b-a=3$. Also the input is $\frac{3i}{n} = i\Delta x$ which matches right endpoints on $[0,3]$. So the integral is:

$\int_0^3 \sqrt{1+x^2}\,dx$

Key Formulas, Rules & Facts

Riemann sum types (equal widths)

Trapezoidal Rule (AP loves this)

Over/under-estimate rules (high-yield)

These are about approximation direction compared to the true integral.

If $f$ is increasing on $[a,b]$

• $L_n$ underestimates
• $R_n$ overestimates

If $f$ is decreasing on $[a,b]$

• $L_n$ overestimates
• $R_n$ underestimates

If $f$ is concave up on $[a,b]$ (think “cup”)

• $T_n$ overestimates
• $M_n$ underestimates

If $f$ is concave down on $[a,b]$

• $T_n$ underestimates
• $M_n$ overestimates

These concavity rules are for trapezoids vs midpoints, not left vs right.

Sigma notation reminders you actually need

Examples & Applications

Example 1: LRAM vs RRAM setup (function given)

Approximate $\int_0^2 x^2\,dx$ with $n=4$ using LRAM.

• $\Delta x = \frac{2-0}{4} = \frac12$
• left endpoints: $x_0=0,\ x_1=\frac12,\ x_2=1,\ x_3=\frac32$
• LRAM:

$L_4 = \frac12\left[f(0)+f\left(\frac12\right)+f(1)+f\left(\frac32\right)\right]$

Since $f(x)=x^2$,

$L_4 = \frac12\left[0^2+\left(\frac12\right)^2+1^2+\left(\frac32\right)^2\right]$

Key insight: because $x^2$ is increasing on $[0,2]$, $L_4$ is an underestimate.

Example 2: Midpoint rule with a “midpoint-looking” expression

Approximate $\int_2^6 \ln(x)\,dx$ with $n=4$ using MRAM.

• $\Delta x = \frac{6-2}{4}=1$
• subintervals: $[2,3],[3,4],[4,5],[5,6]$
• midpoints: $2.5, 3.5, 4.5, 5.5$

$M_4 = 1\,[\ln(2.5)+\ln(3.5)+\ln(4.5)+\ln(5.5)]$

Key insight: midpoint uses the “half-step” idea: $2+\left(i-\frac12\right)\cdot 1$.

Example 3: Convert a limit of a sum to an integral

Convert and evaluate:

$\lim_{n\to\infty} \sum_{i=1}^n \left(1+\frac{2i}{n}\right)^3 \cdot \frac{2}{n}$

• Recognize $\Delta x = \frac{2}{n}$ so $b-a=2$.
• The inside is $1+\frac{2i}{n} = 1 + i\Delta x$, which matches right endpoints on $[1,3]$.

So this is:

$\int_1^3 x^3\,dx$

Evaluate:

$\int_1^3 x^3\,dx = \left[\frac{x^4}{4}\right]_1^3 = \frac{81-1}{4} = 20$

Example 4: Table-based trapezoidal estimate (unequal widths)

Given:

Approximate $\int_0^4 f(x)\,dx$ using trapezoids.

Compute by intervals:

• On $[0,1]$: width $1$, trapezoid area $\frac{1}{2}(2+5)\cdot 1$
• On $[1,3]$: width $2$, trapezoid area $\frac{1}{2}(5+1)\cdot 2$
• On $[3,4]$: width $1$, trapezoid area $\frac{1}{2}(1+3)\cdot 1$

Total:

$\frac{1}{2}(7) + \frac{1}{2}(6)\cdot 2 + \frac{1}{2}(4) = 3.5 + 6 + 2 = 11.5$

Key insight: with unequal widths, treat each subinterval separately.

Common Mistakes & Traps

1. Wrong $\Delta x$

   • What goes wrong: you use $\Delta x = \frac{b-a}{n-1}$ or forget to divide by $n$.
   • Why it’s wrong: $n$ is the number of rectangles/subintervals.
   • Fix: always write $\Delta x = \frac{b-a}{n}$ first.

2. Using the wrong endpoints (off-by-one error)

   • What goes wrong: for LRAM you accidentally include $x_n$, or for RRAM you include $x_0$.
   • Why it’s wrong: LRAM uses $x_0$ through $x_{n-1}$; RRAM uses $x_1$ through $x_n$.
   • Fix: quickly list the first and last sample points before summing.

3. Midpoint confusion (using endpoints instead of midpoints)

   • What goes wrong: you average the function values instead of using $f\left(\frac{x_{i-1}+x_i}{2}\right)$.
   • Why it’s wrong: midpoint rule samples the function at the midpoint x-value.
   • Fix: compute midpoint x-values first, then plug into $f$.

4. Factoring out $\Delta x$ when widths are unequal

   • What goes wrong: you treat a table with irregular x-spacing like equal partitions.
   • Why it’s wrong: each rectangle has its own width $\Delta x_i$.
   • Fix: sum $f(c_i)\Delta x_i$ interval-by-interval.

5. Forgetting integrals are signed area

   • What goes wrong: you add magnitudes when the graph/table implies negative values.
   • Why it’s wrong: $\int_a^b f(x)\,dx$ counts below the x-axis as negative.
   • Fix: keep signs of $f(x)$; only use absolute value if asked for total area.

6. Over/under-estimate rules applied backward

   • What goes wrong: you say left sum overestimates for increasing functions.
   • Why it’s wrong: for increasing functions, left endpoints are smaller heights.
   • Fix: visualize: increasing graph → rectangles from the left sit below the curve.

7. Misidentifying $a$ and $b$ when converting a sum to an integral

   • What goes wrong: you see $\Delta x = \frac{2}{n}$ and assume $[0,2]$ automatically.
   • Why it’s wrong: the interval depends on the “starting shift” inside $f(\cdot)$.
   • Fix: match the input to $a+i\Delta x$ or $a+\left(i-\frac12\right)\Delta x$.

8. Trapezoidal rule endpoint weights wrong

   • What goes wrong: you multiply every term by $2$, or forget the halving.
   • Why it’s wrong: endpoints appear once, interior points twice.
   • Fix: memorize the pattern $1,2,2,\dots,2,1$ in brackets.

Memory Aids & Quick Tricks

Quick Review Checklist

• You can write the general form $\sum_{i=1}^n f(c_i)\Delta x_i$ and explain what $c_i$ and $\Delta x_i$ mean.
• For equal partitions, you instantly compute $\Delta x = \frac{b-a}{n}$.
• You know exactly which x-values LRAM uses ($x_0$ to $x_{n-1}$) and RRAM uses ($x_1$ to $x_n$).
• You can find midpoint x-values and set up $M_n$ correctly.
• You remember trapezoidal rule and the endpoint weighting pattern.
• You can determine under/over-estimates using increasing/decreasing and concavity rules.
• You can convert $\lim_{n\to\infty} \sum \cdots \Delta x$ into $\int_a^b \cdots dx$ by matching $\Delta x$ and the inside sample-point pattern.
• You don’t forget: definite integrals are signed unless the question asks for total area.

You’ve got this—Riemann sums are all about clean setup and pattern recognition.