8 topics. Each one opens with its key idea, then the detail you need to use it.
1. Sequences and their limits
Key idea: Find a sequence limit with the degree rule or the growth order: $\ln n \ll n^p \ll c^n \ll n! \ll n^n$.
A sequence is an ordered list $a_1, a_2, a_3, \dots$ given by a formula in $n$. It converges if $a_n$ approaches one finite number as $n \to \infty$. Otherwise it diverges, either by growing without bound or by oscillating. $a_n = (-1)^n$ diverges because it jumps between $-1$ and $1$ forever and never settles on either value.
For rational expressions, compare the degrees of the top and bottom. If the degrees match, the limit is the ratio of the leading coefficients, so $\frac{3n^2+1}{5n^2-2} \to \frac{3}{5}$. If the top has the higher degree, the sequence diverges to $+\infty$ or $-\infty$, depending on the signs of the leading coefficients. If the bottom wins, the limit is 0. You can check the equal-degree case by dividing everything by $n$: $\frac{n}{n+1} = \frac{1}{1+1/n} \to 1$.
For everything else, use the growth order. From slowest to fastest: $\ln n$, then powers $n^p$, then exponentials $c^n$, then $n!$, then $n^n$. The faster one wins any ratio, so $\frac{\ln n}{n} \to 0$, $\frac{2^n}{n!} \to 0$ and $\frac{n!}{n^n} \to 0$. Also memorize $\left(1+\frac{1}{n}\right)^n \to e$. One theorem matters too: a sequence that is both bounded and monotonic (always rising or always falling) must converge, because it has nowhere else to go.
2. Geometric series and the nth-term test
Key idea: Geometric: converges iff $|r|<1$, sum $= \frac{\text{first term}}{1-r}$. Terms not going to 0 means divergence; terms going to 0 proves nothing.
A geometric series multiplies by the same ratio $r$ each step: $a + ar + ar^2 + \dots$. It converges exactly when $|r| < 1$, and then its sum is $\frac{a}{1-r}$, where $a$ is the first term. So $\sum_{n=0}^{\infty} \left(\frac{1}{2}\right)^n = \frac{1}{1-1/2} = 2$. The formula comes from the partial sums: $S_n = a\frac{1-r^n}{1-r}$, and $r^n \to 0$ when $|r|<1$.
Watch the starting index. $\sum_{n=1}^{\infty} \left(\frac{1}{2}\right)^n$ starts at $\frac{1}{2}$, so its sum is $\frac{1/2}{1/2} = 1$. Always plug the starting index into the formula to get $a$. Do not just read off the coefficient.
The nth-term test (also called the Divergence Test) is your first check on any series. If $\lim a_n \neq 0$, or the limit does not exist, the series diverges. The reason is simple: if the terms do not shrink to 0, you keep adding chunks of a fixed size and the sum cannot settle. For example, $\sum \frac{n}{n+1}$ diverges because its terms approach 1. The test only works one way. If $\lim a_n = 0$, the test tells you nothing, and you must try another test.
3. The Integral Test, p-series and comparison tests
Key idea: Match your series to a $p$-series or geometric series by its leading terms, then compare. Check the Integral Test's conditions before you use it.
The Integral Test links $\sum f(n)$ to $\int_1^{\infty} f(x)\,dx$. It applies only if $f$ is positive, continuous and decreasing. Then the sum and the integral either both converge or both diverge, because the terms are rectangle areas that sit just above or below the curve. The sum and the integral usually have different values, so the test tells you whether the series converges, not what it adds up to.
The Integral Test proves the $p$-series rule: $\sum \frac{1}{n^p}$ converges if $p > 1$ and diverges if $p \le 1$. The harmonic series $\sum \frac{1}{n}$ ($p=1$) diverges, even though its terms go to 0. Memorize this rule, because it feeds every comparison.
The Direct Comparison Test works for positive terms. If $0 \le a_n \le b_n$ and $\sum b_n$ converges, then $\sum a_n$ converges too. If $a_n \ge b_n$ and $\sum b_n$ diverges, so does $\sum a_n$. The inequality has to point the right way. Being smaller than a divergent series proves nothing.
The Limit Comparison Test avoids the inequality. Compute $\lim \frac{a_n}{b_n}$. If it is a finite positive number, the two series behave the same way. To test $\sum \frac{2n+1}{n^3+4}$, compare it to $\sum \frac{1}{n^2}$, built from the leading terms. The limit is 2, and $p=2>1$, so the series converges.
4. Ratio Test and choosing a test
Key idea: Factorials and exponentials mean Ratio Test. $L=1$ means you need a different test, not that the series diverges.
The Ratio Test computes $L = \lim \left|\frac{a_{n+1}}{a_n}\right|$. If $L < 1$ the series converges absolutely, if $L > 1$ it diverges, and if $L = 1$ the test gives no answer. It works because a series with $L<1$ eventually shrinks at least as fast as a geometric series with some ratio below 1.
Use it whenever you see factorials or exponentials, because they cancel cleanly in the ratio. For $\sum \frac{n!}{n^n}$, the ratio simplifies to $\left(\frac{n}{n+1}\right)^n$, which approaches $\frac{1}{e} \approx 0.37$. That is below 1, so the series converges. Trying the Integral Test here would mean integrating a factorial, which you cannot do. The Ratio Test always returns $L=1$ on $p$-series and rational expressions, so do not waste time on it there.
The Root Test, $\lim \sqrt[n]{|a_n|}$, uses the same cutoffs. It is not required by the course framework, but it is handy when the whole term is raised to the $n$th power.
A fast strategy works on most problems:
1. Do the terms go to 0? If not, it diverges.
2. Is it geometric or a $p$-series? Use the known rule.
3. Factorials or $c^n$? Use the Ratio Test.
4. Alternating signs? Use the Alternating Series Test.
5. Looks like a rational function? Use Limit Comparison.
6. Easy to integrate and decreasing? Use the Integral Test.
5. Alternating series and absolute vs. conditional convergence
Key idea: Three possible answers: absolutely convergent, conditionally convergent, or divergent. Test $\sum |a_n|$ first, then the signed series.
An alternating series has terms that switch sign, like $\sum (-1)^{n} b_n$ with $b_n > 0$. The Alternating Series Test says it converges if $b_n$ decreases and $b_n \to 0$. The partial sums step back and forth, and each step is shorter than the last, so they close in on one value. That is why $\sum \frac{(-1)^n}{n}$ converges. Its sum is $-\ln 2$, not 0. That value comes from the series $\ln(1+x) = x - \frac{x^2}{2} + \frac{x^3}{3} - \dots$ at $x=1$, with every sign flipped. Terms going to 0 tells you nothing about the value of the sum.
Next, ask what happens if you remove the signs. If $\sum |a_n|$ converges, the series converges absolutely. If $\sum a_n$ converges but $\sum |a_n|$ diverges, it converges conditionally. For $\sum \frac{(-1)^n}{n}$, the absolute version is the harmonic series, which diverges. So the series converges conditionally. In contrast, $\sum \frac{(-1)^n}{n^2}$ converges absolutely, since $p=2$.
Absolute convergence is the stronger result, because it implies plain convergence. A conditionally convergent series depends on its signs to cancel, and only that cancelling keeps it finite.
The Alternating Series Test also gives an error bound. If you stop after $n$ terms, the error $|S - S_n|$ is at most $b_{n+1}$, the first term you left out. The true sum also lies between any two consecutive partial sums. So for $\sum \frac{(-1)^{n+1}}{n}$, using four terms gives an error under $\frac{1}{5}$.
6. Taylor and Maclaurin polynomials
Key idea: Coefficient $= \frac{f^{(k)}(a)}{k!}$; to recover a derivative, multiply the coefficient by $k!$.
The Taylor polynomial of degree $n$ for $f$ centered at $x=a$ is
$P_n(x) = f(a) + f'(a)(x-a) + \frac{f''(a)}{2!}(x-a)^2 + \dots + \frac{f^{(n)}(a)}{n!}(x-a)^n$.
A Maclaurin polynomial is the special case $a=0$. Each coefficient is $\frac{f^{(k)}(a)}{k!}$. The $k!$ is there because differentiating $(x-a)^k$ a total of $k$ times produces $k!$. Dividing by it makes the $k$th derivative of $P_n$ equal $f^{(k)}(a)$ exactly.
Exam problems often give you a table of derivatives instead of a formula. Suppose $f(2)=3$, $f'(2)=-1$, $f''(2)=4$ and $f'''(2)=12$. Then $P_3(x) = 3 - (x-2) + \frac{4}{2}(x-2)^2 + \frac{12}{6}(x-2)^3$, which simplifies to $3 - (x-2) + 2(x-2)^2 + 2(x-2)^3$. The most common slip is forgetting to divide by the factorial.
You can also read derivatives back out. If a series has coefficient $c_k$ on $(x-a)^k$, then $f^{(k)}(a) = k!\,c_k$. Questions ask for things like $f^{(4)}(0)$ this way.
The polynomial also shows what $f$ is doing at the center. The sign of the $(x-a)^2$ coefficient matches the sign of $f''(a)$, so it tells you the concavity. If $f'(a)=0$ and $f''(a) \neq 0$, the sign of the $(x-a)^2$ coefficient tells you whether $a$ is a local min (positive) or a local max (negative). If that coefficient is 0, this test gives no answer, and $a$ could be neither, as with $f(x)=x^3$ at $a=0$.
7. Taylor series and power series operations
Key idea: Build new series from $e^x$, $\sin x$, $\cos x$ and $\frac{1}{1-x}$ by substituting, multiplying, differentiating or integrating.
Memorize four Maclaurin series:
$e^x = 1 + x + \frac{x^2}{2!} + \frac{x^3}{3!} + \dots$ (all $x$)
$\sin x = x - \frac{x^3}{3!} + \frac{x^5}{5!} - \dots$ (all $x$)
$\cos x = 1 - \frac{x^2}{2!} + \frac{x^4}{4!} - \dots$ (all $x$)
$\frac{1}{1-x} = 1 + x + x^2 + x^3 + \dots$ for $|x|<1$.
You rarely build a new series from derivatives. You build it from these by four operations. Substitution: replace $x$ with an expression. For $e^{-x^2}$, put $u=-x^2$ into $e^u$ to get $1 - x^2 + \frac{x^4}{2!} - \frac{x^6}{3!} + \dots$. Wrap the whole expression in brackets so the negative sign survives, because $(-x^2)^2 = +x^4$. Multiplication: $x\cos x$ is $x$ times each term of $\cos x$. Differentiation and integration: work term by term. Differentiating $\frac{1}{1-x}$ gives $\frac{1}{(1-x)^2} = 1 + 2x + 3x^2 + \dots$. Integrating $\frac{1}{1+x} = 1 - x + x^2 - \dots$ gives $\ln(1+x) = x - \frac{x^2}{2} + \frac{x^3}{3} - \dots$.
Differentiating or integrating never changes the radius of convergence. It can change what happens at the endpoints, though, so check them again.
8. Radius and interval of convergence; error bounds
Key idea: Ratio Test for the open interval, then test both endpoints separately. Lagrange uses the $(n+1)$th derivative; the alternating bound uses the first omitted term.
A power series $\sum c_n (x-a)^n$ converges on an interval centered at $a$. To find it, run the Ratio Test with $x$ left in. Take $\sum \frac{(x-3)^n}{n\,2^n}$. The ratio limit is $\frac{|x-3|}{2}$, and setting it below 1 gives $|x-3| < 2$. The radius is 2, so the series converges on $1 < x < 5$.
The Ratio Test gives $L=1$ at the endpoints, so test each one by hand. At $x=5$ you get $\sum \frac{1}{n}$, which diverges. At $x=1$ you get $\sum \frac{(-1)^n}{n}$, which converges by the Alternating Series Test. The interval is $[1, 5)$. Skipping the endpoint checks is the most common way to lose this point.
The Lagrange error bound covers any Taylor polynomial: $|f(x) - P_n(x)| \le \frac{M}{(n+1)!}|x-a|^{n+1}$. Here $M$ is the largest value of $|f^{(n+1)}|$ between $a$ and $x$. Notice it uses the next derivative after the last one in your polynomial.
For example, $\sin(0.2) \approx 0.2 - \frac{0.2^3}{6} \approx 0.198667$. Every derivative of $\sin$ is at most 1 in size, so $M=1$. Lagrange then gives an error at most $\frac{0.2^4}{4!} \approx 0.0000667$. The series alternates too, so the alternating bound also applies. It gives the first omitted term, $\frac{0.2^5}{5!} \approx 0.0000027$. Use the alternating bound only when the series truly alternates with decreasing terms.