8 topics. Each one opens with its key idea, then the detail you need to use it.
1. Riemann sums and the trapezoidal rule
Key idea: Name the sample points before you compute: $n$ pieces means $n$ heights for a rectangle sum, and $n+1$ values for a trapezoidal sum.
To estimate $\int_a^b f(x)\,dx$, cut $[a,b]$ into $n$ pieces of width $\Delta x=\frac{b-a}{n}$ and build a rectangle on each one. The only decision is where to measure the height. A left sum uses each piece's left endpoint, a right sum uses the right endpoint, and a midpoint sum uses the center.
Take $\int_0^4 x^2\,dx$ with 4 pieces: $[0,1],[1,2],[2,3],[3,4]$. The left sum samples at $0,1,2,3$ and never touches 4, giving $1(0+1+4+9)=14$. The right sum samples at $1,2,3,4$, giving $30$. The exact value is $\frac{64}{3}\approx 21.3$. The left sum is low because $x^2$ is increasing, so each left height is the smallest on its piece. In general, left sums underestimate an increasing function and right sums overestimate it, and the reverse holds when the function is decreasing.
A trapezoidal sum averages the left and right heights on each piece. Here it gives $\frac{14+30}{2}=22$. The general form is $\frac{\Delta x}{2}\left[f(x_0)+2f(x_1)+\cdots+2f(x_{n-1})+f(x_n)\right]$, where interior values count twice because two trapezoids share them. Since $x^2$ is concave up, the straight tops of the trapezoids sit above the curve, so 22 is an overestimate. When $f$ is concave down, the trapezoids sit below the curve, so the trapezoidal sum underestimates. A midpoint sum does the reverse of the trapezoidal rule: it underestimates when $f$ is concave up and overestimates when $f$ is concave down. Exam tables often use unequal widths. In that case, find each piece's area separately and add them, rather than trusting a formula built for equal widths.
2. The definite integral as a limit and as signed area
Key idea: In a limit of a sum, $\Delta x$ gives you the length of the interval and $x_k$ gives you the starting point and the integrand.
As $n$ grows, the rectangles get thinner and every Riemann sum approaches the same number. That limit is the definite integral: $\int_a^b f(x)\,dx=\lim_{n\to\infty}\sum_{k=1}^{n} f(x_k)\,\Delta x$.
The exam often hands you the limit and asks for the integral. Read it in parts. In $\lim_{n\to\infty}\sum_{k=1}^{n}\left(1+\frac{2k}{n}\right)^3\frac{2}{n}$, the factor $\frac{2}{n}$ is $\Delta x$, so the interval has length 2. The sample point $x_k=1+\frac{2k}{n}$ starts just past 1 and ends at 3. So the limit equals $\int_1^3 x^3\,dx$. If you instead let $x=\frac{2k}{n}$, the same limit is $\int_0^2 (1+x)^3\,dx$. Both are correct, which is why answer choices sometimes show either form.
Because the integral is signed area, you can evaluate many integrals with geometry and no antiderivative at all. For example, $\int_{-3}^{3}\sqrt{9-x^2}\,dx$ is the top half of a circle of radius 3, so it equals $\frac{9\pi}{2}$. When the exam gives a graph of $f$ made of lines and semicircles, split the interval wherever the graph crosses the axis. Then find each region's area with a triangle, rectangle or circle formula, and give it a minus sign if it lies below the axis.
3. Properties of definite integrals
Key idea: Integrals split over intervals and over sums, and constants factor out, but products never split.
A handful of rules let you combine given integral values without knowing the function. They all follow from thinking of the integral as signed area.
- $\int_a^a f(x)\,dx=0$, because a region with no width has no area.
- $\int_b^a f(x)\,dx=-\int_a^b f(x)\,dx$, because reversing direction makes $\Delta x$ negative.
- $\int_a^b f(x)\,dx+\int_b^c f(x)\,dx=\int_a^c f(x)\,dx$, because adjacent regions add.
- $\int_a^b [k f(x)+g(x)]\,dx=k\int_a^b f(x)\,dx+\int_a^b g(x)\,dx$.
Here is a typical use. Suppose $\int_0^5 f(x)\,dx=7$ and $\int_0^2 f(x)\,dx=3$. Splitting the interval gives $\int_2^5 f(x)\,dx=4$. Then $\int_2^5 [2f(x)+1]\,dx = 2(4)+3=11$. The $+3$ is where students slip, because $\int_2^5 1\,dx$ is the width of the interval, not 1.
Two more facts save time. If $f$ is odd, then $\int_{-a}^{a} f(x)\,dx=0$, because the area on the left cancels the area on the right. If $f$ is even, the integral is twice $\int_0^a f(x)\,dx$. One thing does not work: the integral of a product is not the product of the integrals.
4. The Fundamental Theorem, Part 1: accumulation functions
Key idea: $\frac{d}{dx}\int_a^{h(x)} f(t)\,dt = f(h(x))\,h'(x)$, and the graph of $f$ is the graph of $g'$.
Define $g(x)=\int_a^x f(t)\,dt$. As $x$ moves right, $g$ collects more signed area under $f$. FTC Part 1 says $g'(x)=f(x)$. Here is why. The difference $g(x+h)-g(x)$ is a thin strip about $f(x)$ tall and $h$ wide, so its area is about $f(x)\cdot h$. Divide by $h$ and let $h\to 0$, and you get $f(x)$.
So $\frac{d}{dx}\int_1^x \sqrt{t^3+1}\,dt=\sqrt{x^3+1}$. You replace $t$ with $x$ and stop. You do not differentiate the integrand, and the constant lower bound adds nothing to the derivative.
When the upper bound is a function of $x$, the chain rule applies: $\frac{d}{dx}\int_0^{x^2}\cos t\,dt=\cos(x^2)\cdot 2x$. Plug in the bound, then multiply by the bound's derivative. If $x$ appears in the lower bound, flip the bounds first and add a minus sign.
The theorem also lets you analyze $g$ from a graph of $f$ alone. Since $g'=f$, $g$ increases where $f>0$ and has a relative maximum where $f$ changes from positive to negative. It is concave up where $f$ is increasing, because $g''=f'$. To find an actual value such as $g(4)$, add the signed areas from $a$ to 4. Remember that $g(a)=0$ every time.
5. The Fundamental Theorem, Part 2, and net change
Key idea: Integral of a rate means change in amount; add the starting value for an amount, and use $|v|$ for distance.
FTC Part 2 says $\int_a^b f(x)\,dx=F(b)-F(a)$ for any antiderivative $F$ of $f$. Any antiderivative works because two antiderivatives differ only by a constant, and that constant cancels in the subtraction. That is why definite integrals never carry $+C$. For example, $\int_1^3 (2x+1)\,dx=\left[x^2+x\right]_1^3=12-2=10$. Always subtract the value at the lower bound, even when it looks like it might be zero.
Read the theorem as a statement about change: $\int_a^b R'(t)\,dt=R(b)-R(a)$. Rearranged, it becomes $R(b)=R(a)+\int_a^b R'(t)\,dt$, which is the most used equation in rate problems. If a tank holds 50 gallons at $t=0$ and water enters at $r(t)$ gallons per hour, it holds $50+\int_0^{6} r(t)\,dt$ gallons at $t=6$.
For a moving particle, $\int_a^b v(t)\,dt$ is displacement and $\int_a^b |v(t)|\,dt$ is total distance. Take $v(t)=2t-4$ on $[0,5]$. Displacement is $\left[t^2-4t\right]_0^5=5$. But $v$ changes sign at $t=2$, so the particle reverses there. Its position goes from 0 to $-4$ and then to 5, so the legs are 4 and 9, and the total distance is 13. To find distance, locate every time $v=0$ with a sign change, find each leg, and add the absolute values.
6. Antiderivatives and initial value problems
Key idea: Rewrite into powers, integrate term by term, check by differentiating, and use the given point to find $C$.
An antiderivative of $f$ is any function whose derivative is $f$. The rules are the derivative rules run backward.
- $\int x^n\,dx=\frac{x^{n+1}}{n+1}+C$ for $n\neq -1$, and $\int \frac{1}{x}\,dx=\ln|x|+C$.
- $\int e^x\,dx=e^x+C$.
- $\int \sin x\,dx=-\cos x+C$ and $\int \cos x\,dx=\sin x+C$.
- $\int \sec^2 x\,dx=\tan x+C$.
- $\int \frac{1}{1+x^2}\,dx=\arctan x+C$ and $\int \frac{1}{\sqrt{1-x^2}}\,dx=\arcsin x+C$.
The sign on $\sin x$ is the most common slip. Check it by differentiating: the derivative of $-\cos x$ is $\sin x$. The $+C$ matters because every vertical shift of an antiderivative has the same derivative.
There is no product or quotient rule for integrals, so rewrite first. For instance, $\frac{x^2+1}{\sqrt{x}}=x^{3/2}+x^{-1/2}$, which integrates to $\frac{2}{5}x^{5/2}+2x^{1/2}+C$.
An initial value problem gives you $f'$ and one point, and the point pins down $C$. If $f'(x)=3x^2-4$ and $f(2)=5$, then $f(x)=x^3-4x+C$. Substituting gives $8-8+C=5$, so $C=5$ and $f(x)=x^3-4x+5$. You can also write this as $f(x)=5+\int_2^x (3t^2-4)\,dt$. That form is the only option when $f'$ has no elementary antiderivative.
7. u-substitution, indefinite and definite
Key idea: Pick $u$ so that $du$ is already in the integrand, and on a definite integral either convert the bounds or convert back, never half of each.
u-substitution reverses the chain rule. Look for an inner function whose derivative also appears as a factor, give or take a constant. Call the inner function $u$, rewrite everything in terms of $u$, integrate, and convert back.
In $\int x\cos(x^2)\,dx$, let $u=x^2$, so $du=2x\,dx$ and $x\,dx=\frac{1}{2}du$. The integral becomes $\frac{1}{2}\int \cos u\,du=\frac{1}{2}\sin(x^2)+C$. The $\frac{1}{2}$ fixes the constant mismatch. A linear inner function works the same way: $\int e^{3x}\,dx=\frac{1}{3}e^{3x}+C$. If, after substituting, some $x$ is left over that you cannot write in terms of $u$, you chose the wrong $u$.
For a definite integral you have two clean options. You can change the bounds to $u$-values and never return to $x$. Or you can find the antiderivative in $x$ and use the original bounds. For $\int_1^2 \frac{2x}{x^2+1}\,dx$, let $u=x^2+1$. Then $x=1$ gives $u=2$ and $x=2$ gives $u=5$, so the integral is $\int_2^5 \frac{1}{u}\,du=\ln 5-\ln 2=\ln\frac{5}{2}$. The error to avoid is mixing the two: using $x$-bounds on a $u$-antiderivative. Also notice that here $du=2x\,dx$ absorbs the whole factor of $2x$. Multiplying by 2 again double-counts it.
8. Long division, completing the square, and choosing a technique
Key idea: Diagnose the form before integrating: what is on top decides between a logarithm, an arctangent and a division.
Some rational functions match no basic rule until you reshape them. When the numerator's degree is at least the denominator's, divide first. For $\int \frac{x^2+3}{x+1}\,dx$, note that $x^2+3=(x+1)(x-1)+4$, so the integrand is $x-1+\frac{4}{x+1}$. Each piece is now basic, and the answer is $\frac{x^2}{2}-x+4\ln|x+1|+C$.
When the denominator is a quadratic with no real roots and there is no matching $x$ on top, complete the square to reach the arctangent form. $x^2+4x+5=(x+2)^2+1$, so $\int \frac{dx}{x^2+4x+5}=\arctan(x+2)+C$. This is really a u-substitution with $u=x+2$.
To choose a technique, run through a short list in order:
1. Can you expand or rewrite it into powers and basic forms? Then integrate term by term.
2. Is there an inner function whose derivative is also a factor? Use u-substitution.
3. Is it a fraction with the top degree at least the bottom degree? Divide.
4. Is it $\frac{1}{\text{quadratic}}$ with no $x$ on top? Complete the square.
Small differences change the tool completely. $\int \frac{x}{x^2+1}\,dx$ is a u-substitution that gives $\frac{1}{2}\ln(x^2+1)+C$. But $\int \frac{1}{x^2+1}\,dx$ has no $x$ on top to serve as $du$, so it is $\arctan x+C$.