Evaluate The Series Or State That It Diverges

You're staring at a summation symbol with an infinity sign on top. The terms look harmless enough — maybe they're fractions, maybe they involve factorials, maybe they alternate signs. The question is always the same: does this thing settle down to a finite number, or does it blow up?

I've watched hundreds of students freeze at this exact moment. Worth adding: not because the math is impossible. Because they're trying to memorize a flowchart instead of understanding what the tools actually do.

Let's fix that.

What It Means to Evaluate a Series

An infinite series is just a sum with infinitely many terms. Written out, it looks like:

$\sum_{n=1}^{\infty} a_n = a_1 + a_2 + a_3 + \dots$

But you can't actually add infinitely many things. Now, not in the literal sense. What we really* mean is: look at the sequence of partial sums.

$S_1 = a_1$ $S_2 = a_1 + a_2$ $S_3 = a_1 + a_2 + a_3$ $\vdots$ $S_n = \sum_{k=1}^{n} a_k$

If the sequence ${S_n}$ converges to some finite limit $L$, we say the series converges to $L$. If ${S_n}$ diverges — goes to infinity, oscillates, or just refuses to settle — the series diverges.

That's it. That's the whole definition. Everything else — every test, every trick, every theorem — is just a way to answer the convergence question without computing partial sums by hand.

The Distinction That Matters

"Evaluate the series" and "determine convergence" sound like the same task. They're not.

• Determine convergence: Yes/no. Does a finite sum exist?
• Evaluate the series: Find the actual value* of that sum.

Most series cannot* be evaluated in closed form. We can prove $\sum_{n=1}^{\infty} \frac{1}{n^2}$ converges. Which means finding that it equals $\frac{\pi^2}{6}$? That took Euler years and a flash of genius. The harmonic series $\sum_{n=1}^{\infty} \frac{1}{n}$ diverges — but the alternating harmonic series $\sum_{n=1}^{\infty} \frac{(-1)^{n+1}}{n}$ converges to $\ln 2$.

Knowing whether* a series converges is a skill you can systematize. Practically speaking, finding the exact sum? That's a different game entirely.

Why This Skill Shows Up Everywhere

You're not learning this to pass a calc exam. Series convergence is the backbone of:

Power series and Taylor expansions — every function you approximate with polynomials (sin, cos, e^x, ln) lives or dies by its radius of convergence. Mess up the endpoint testing? Your approximation fails exactly where you need it most.

Fourier series — signal processing, compression, solving PDEs. All of it relies on orthogonal function expansions that only make sense if the series converge.

Probability and statistics — expected values of discrete distributions are infinite series. The geometric distribution, Poisson, negative binomial — their means and variances only exist if certain series converge.

Numerical analysis — iterative methods, root-finding algorithms, quadrature rules. They all produce sequences. Convergence analysis tells you if the method actually works and how fast.

Physics and engineering — perturbation theory, asymptotic expansions, quantum field theory. Divergent series show up constantly* in physics, and knowing how to handle them (Borel summation, analytic continuation) separates usable results from nonsense.

The pattern: infinite processes appear whenever you model something continuous with discrete steps. Convergence tells you whether the model is lying to you.

The Toolkit: Tests You'll Actually Use

There are a dozen named convergence tests. That's why you need maybe five. The rest are special cases or historical curiosities.

The Nth Term Test (Divergence Test)

If $\lim_{n \to \infty} a_n \neq 0$, the series diverges.

That's the whole test. If the terms don't go to zero, the partial sums can't settle down.

Crucial: The converse is false. Terms going to zero does not guarantee convergence. The harmonic series is the classic counterexample — $\frac{1}{n} \to 0$ but the sum diverges.

Use this first. Always. It takes ten seconds and catches the obvious divergences before you waste time on heavier machinery.

Geometric Series

$\sum_{n=0}^{\infty} ar^n = \frac{a}{1-r} \quad \text{for } |r| < 1$

Diverges for $|r| \geq 1$. Also, this is the only* series in the standard curriculum where you get both a convergence criterion and the exact sum in one package. Memorize it. Recognize it in disguise — $\sum (\frac{2}{3})^n$, $\sum \frac{5^n}{7^{n+1}}$, $\sum 3 \cdot (-\frac{1}{4})^n$.

P-Series

$\sum_{n=1}^{\infty} \frac{1}{n^p}$

Converges if $p > 1$, diverges if $p \leq 1$.

The $p=1$ case is the harmonic series. Worth adding: the $p=2$ case is the Basel problem ($\frac{\pi^2}{6}$). For $p > 1$ you get convergence but usually no closed form.

This is your benchmark for comparison tests. "Behaves like a p-series with $p = \dots${content}quot; is the most useful phrase in your vocabulary.

Integral Test

If $f(x)$ is positive, continuous, and decreasing for $x \geq 1$, and $a_n = f(n)$, then:

$\sum_{n=1}^{\infty} a_n \text{ and } \int_1^{\infty} f(x) , dx$

either both converge or both diverge.

Why it works: The sum is a left Riemann sum for the integral. The integral is the area under the curve. The rectangles and the curve trap each other.

When to use it: Terms look like something you can integrate. $\frac{1}{n \ln n}$, $\frac{1}{n (\ln n)^2}$, $\frac{\ln n}{n^2}$. The integral is often easier than direct comparison.

Trap: The function must* be eventually decreasing. $\frac{\sin n}{n}$ fails this — the integral test doesn't apply. (That series converges, by the way. Dirichlet's test.)

Comparison Tests

Direct Comparison

If $0 \leq a_n \leq b_n$ for all $n$:

• $\sum b_n$ converges $\implies \sum a_n$ converges
• $\sum a_n$ diverges $\implies \sum b_n$ diverges

Limit Comparison

If $a_n, b_n > 0$ and $\lim_{n \to \infty} \frac{a_n}{b_n} = c$ where $0 < c < \infty$, then both series do the same thing (both converge or both diverge).

Limit comparison is almost always better. You don't need to find an inequality that holds for all $n$ — just the asymptotic behavior. The limit captures "behaves like" precisely.

Your go-to $b_n$: A p-series or geometric series. Always. If you're comparing to something else

...something else like a known convergent or divergent series whose terms are simple to manipulate. Here's a good example: when you encounter a term that contains a factorial, compare it to a geometric series; when you see logarithms or powers of (n), a (p)-series often serves as the benchmark.

Example 1 – Limit Comparison with a (p)-series
Consider (\displaystyle \sum_{n=2}^{\infty}\frac{1}{n(\ln n)^2}).
Take (b_n=\frac{1}{n^p}) with (p=1+\varepsilon) for a small (\varepsilon>0). Compute

[ \lim_{n\to\infty}\frac{a_n}{b_n} =\lim_{n\to\infty}\frac{1/(n(\ln n)^2)}{1/n^{1+\varepsilon}} =\lim_{n\to\infty}\frac{n^{\varepsilon}}{(\ln n)^2}=0 . ]

If you found this helpful, you might also enjoy how many centimeters are in a nanometer or 1 gallon of water is how many oz.

Since the limit is (0) and (\sum b_n) converges for any (\varepsilon>0), the limit comparison test tells us that (\sum a_n) also converges. (A cleaner choice is (b_n=\frac{1}{n(\ln n)^2}) itself, but the illustration shows how the test works with a simple (p)-series.)

Example 2 – Limit Comparison with a Geometric Series
Examine (\displaystyle \sum_{n=0}^{\infty}\frac{3^n}{5^n+2^n}).
Here the dominant term in the denominator is (5^n), so set (b_n=\left(\frac{3}{5}\right)^n). Then

[ \lim_{n\to\infty}\frac{a_n}{b_n} =\lim_{n\to\infty}\frac{3^n/(5^n+2^n)}{(3/5)^n} =\lim_{n\to\infty}\frac{5^n}{5^n+2^n}=1 . ]

Because the limit is a positive finite number and (\sum b_n) converges (geometric with ratio (3/5<1)), the original series converges as well.

---

Ratio Test

For a series (\displaystyle \sum a_n) with (a_n\neq0), compute

[ L=\lim_{n\to\infty}\left|\frac{a_{n+1}}{a_n}\right|. ]

• If (L<1), the series converges absolutely.
• If (L>1) (or the limit is infinite), the series diverges.
• If (L=1), the test is inconclusive.

The ratio test excels when factorials, exponentials, or powers of (n) appear, because the ratio often simplifies dramatically.

Example: (\displaystyle \sum_{n=0}^{\infty}\frac{n!}{n^n}).
[ \left|\frac{a_{n+1}}{a_n}\right| =\frac{(n+1)!/(n+1)^{n+1}}{n!/n^n} =\frac{n+1}{(n+1)^{n+1}},n^n =\left(\frac{n}{n+1}\right)^n!\frac{1}{n+1} \longrightarrow 0 . ] Since (L=0<1), the series converges absolutely.

---

Root Test

When the (n)‑th term involves an (n)‑th power, the root test can be more direct. Define

[ L=\lim_{n\to\infty}\sqrt[n]{|a_n|}. ]

The same conclusions as the ratio test hold: convergence if (L<1), divergence if (L>1), inconclusive if (L=1).

Example: (\displaystyle \sum_{n=1}^{\infty}\left(\frac{2n}{3n+1}\right)^n).
[ \sqrt[n]{|a_n|} =\frac{2n}{3n+1}\longrightarrow\frac{2}{3}<1, ] so the series converges.

---

Alternating Series Test (Leibniz)

For an alternating series (\displaystyle \sum_{n=1}^{\infty}(-1)^{n-1}b_n) with (b_n\ge0):

• If (b_n) is decreasing eventually, and
• (\displaystyle \lim_{n\to\infty}b_n=0),

then the series converges. Note that this

Note that this test requires the terms (b_n) to be eventually monotone decreasing; in practice one often checks the derivative of the corresponding function or simply verifies that (b_{n+1}\le b_n) for all sufficiently large (n). A classic illustration is the alternating harmonic series

[ \sum_{n=1}^{\infty}\frac{(-1)^{,n-1}}{n} ;=;1-\frac12+\frac13-\frac14+\cdots . ]

Here (b_n=1/n) clearly satisfies (b_{n+1}<b_n) and (\displaystyle\lim_{n\to\infty}b_n=0), so the alternating series test guarantees convergence. Also worth noting, the series is conditionally convergent: the series of absolute values (\sum 1/n) diverges (the harmonic series), while the original alternating series converges.

A useful by‑product of the alternating series test is an explicit bound on the remainder after truncating the series. If (S) denotes the sum and (S_N) the (N)-th partial sum, then

[ |R_N| = |S-S_N| \le b_{N+1}, ]

because the tail consists of an alternating decreasing sequence whose magnitude is bounded by the first omitted term. This estimate is often employed when approximating alternating series to a prescribed accuracy.

---

Integral Test

When a term (a_n) can be expressed as a function (f(n)) that is continuous, positive, and decreasing for (x\ge 1), the series (\sum a_n) and the improper integral (\int_{1}^{\infty}f(x),dx) share the same fate: both converge or both diverge. The intuition is that the area under the curve approximates the sum of rectangles.

Example. Consider (\displaystyle\sum_{n=2}^{\infty}\frac{1}{n(\ln n)^2}). Let (f(x)=1/(x(\ln x)^2)). Since

[ \int_{2}^{\infty}\frac{dx}{x(\ln x)^2} = \Bigl[-\frac{1}{\ln x}\Bigr]_{2}^{\infty}= \frac{1}{\ln 2}<\infty, ]

the integral converges, and therefore the series converges as well. This test is especially convenient for series involving logarithms or rational powers of (n).

---

Dirichlet and Abel Tests

Both the Dirichlet and Abel tests are extensions of the comparison idea that handle series whose terms are products of two sequences. Also, the Dirichlet test states: if the partial sums of (\sum u_n) are bounded and ({v_n}) is a monotone sequence converging to (0), then (\sum u_n v_n) converges. The Abel test relaxes the boundedness requirement: it requires (\sum u_n) to converge and ({v_n}) to be monotone and bounded.

These tools are invaluable for series such as (\displaystyle\sum_{n=1}^{\infty}\frac{\sin n}{n}) (Dirichlet) or (\displaystyle\sum_{n=1}^{\infty}\frac{(-1)^n}{\sqrt{n}}) (Abel), where elementary comparison or ratio tests give no immediate answer.

---

Summary

A mathematician’s toolkit for deciding whether an infinite series converges or diverges contains several complementary methods. And the limit comparison test lets one compare a complicated term with a simple benchmark (a (p)-series or a geometric series). The ratio and root tests excel when factorials, exponentials, or powers of (n) dominate the term, because successive ratios or (n)‑th roots often simplify dramatically.

Another useful device for series of positive terms is Cauchy’s condensation test. If ({a_n}) is a non‑increasing sequence of non‑negative numbers, then

[ \sum_{n=1}^{\infty}a_n \quad\text{converges};\Longleftrightarrow; \sum_{k=0}^{\infty}2^{k}a_{2^{k}} \quad\text{converges}. ]

The proof hinges on grouping together blocks of terms whose lengths double each time; this transformation often turns a seemingly intractable series into one whose terms decay exponentially, making the comparison with a geometric series straightforward. The test is especially handy when the original terms behave like (1/(n(\ln n)^p)) or similar slowly varying functions.

Beyond the purely convergence‑oriented criteria, one also distinguishes between absolute and conditional convergence. A series (\sum a_n) is said to converge absolutely if (\sum|a_n|) converges; in that case the original series converges regardless of any rearrangements of its terms. Still, when only the signed series converges but the series of absolute values diverges, the convergence is termed conditional. Recognizing this distinction is essential, because many of the tests above (e.g., the alternating series test, Dirichlet’s test) guarantee convergence only in the conditional sense.

Understanding how these various tools fit together provides a coherent strategy for tackling any series that appears in analysis or applied mathematics. One typically begins by inspecting the size of the terms — if they resemble a (p)-series or a geometric progression, the limit comparison, ratio, or root tests are the natural first steps. If the terms are alternating or involve bounded partial sums multiplied by a monotone factor, the alternating series test or Dirichlet/Abel criteria become the appropriate choices. For series with slowly varying denominators, condensation or integral comparisons often break the deadlock. Finally, checking for absolute convergence adds a layer of robustness, ensuring that the series behaves well under more delicate operations such as term‑by‑term differentiation or integration.

Simply put, the convergence tests constitute a versatile toolbox: each is built for a particular pattern of term behavior, yet all share the common goal of deciding whether an infinite sum settles to a finite value. Worth adding: mastery of this toolbox not only equips the analyst to handle the myriad series that arise in theory and applications, but also deepens the appreciation of how different notions of “size’’ and “order’’ interact in the infinite realm. With these methods at hand, the question of convergence becomes a matter of systematic inspection rather than guesswork, allowing mathematicians to move confidently from a formal expression to a definitive answer.