CALCULUS 2 • SEQUENCES & SERIES

Comparison Tests

Determine whether infinite series converge or diverge by comparing them to series whose behavior is already known.

Historical Context & Motivation

The question of whether an infinite series converges to a finite value or diverges to infinity has occupied mathematicians since antiquity. Archimedes, in his quadrature of the parabola, effectively summed a geometric series to compute an area—one of the earliest instances of convergence reasoning in recorded mathematics. Yet the rigorous framework for deciding convergence in general remained elusive for centuries. The comparison tests arose from a natural intuition: if you can bound one series by another whose behavior you already understand, you can inherit convergence or divergence conclusions without summing the series directly.

During the eighteenth and nineteenth centuries, mathematicians such as Euler, Gauss, Cauchy, and Weierstrass progressively sharpened the notion of convergence. Euler's bold manipulations of series often produced correct results, but sometimes led to paradoxes because rigorous convergence criteria had not yet been established. Cauchy's Cours d'analyse (1821) laid much of the groundwork by formalizing limits and introducing systematic convergence tests. The Direct Comparison Test, in particular, emerged as one of the most natural and frequently used tools, drawing on the simple observation that a series of non-negative terms smaller than a convergent series must itself converge.

c. 250 BCE
Archimedes and Geometric Series
Archimedes computes the area of a parabolic segment by summing an infinite geometric series, demonstrating one of the earliest convergence arguments.
1821
Cauchy's Cours d'analyse
Augustin-Louis Cauchy publishes rigorous definitions of limits and convergence, establishing the foundation upon which comparison arguments are formalized.
1837
Dirichlet and Conditional Convergence
Peter Gustav Lejeune Dirichlet distinguishes absolute from conditional convergence, clarifying the scope and power of comparison-based arguments for non-negative series.
1880s
Limit Comparison Refinements
Weierstrass, Stolz, and others refine comparison techniques into the Limit Comparison Test, providing a more flexible tool when direct inequalities are difficult to establish.

The central question these developments address is both practical and foundational: given a series ∑aₙ whose partial sums resist closed-form evaluation, how can we determine convergence or divergence? The comparison tests answer this by leveraging a library of benchmark series—geometric series, p-series, and others—as reference points against which any new series can be measured.

Core Principles & Definitions

The comparison tests rest on the monotone convergence principle: a non-decreasing sequence that is bounded above must converge. Because the partial sums of a series with non-negative terms form a non-decreasing sequence, bounding those partial sums above guarantees convergence. Conversely, if the partial sums grow without bound, the series diverges. This simple monotonicity argument is the engine that drives both the Direct Comparison Test (DCT) and the Limit Comparison Test (LCT).

1

Non-Negative Terms Assumption

Both comparison tests require aₙ ≥ 0 and bₙ ≥ 0 for all n sufficiently large. This ensures partial sums are non-decreasing, allowing monotone convergence arguments.
2

Direct Comparison Test

If 0 ≤ aₙ ≤ bₙ for all n ≥ N, then: ∑bₙ converges ⇒ ∑aₙ converges, and ∑aₙ diverges ⇒ ∑bₙ diverges.
3

Limit Comparison Test

If lim(aₙ/bₙ) = L where 0 < L < ∞, then ∑aₙ and ∑bₙ either both converge or both diverge. They share the same convergence behavior.
4

Benchmark Series Library

Key references: geometric series ∑rⁿ (converges iff |r| < 1), p-series ∑1/nᵖ (converges iff p > 1), and the harmonic series ∑1/n (diverges).
5

Direction of Inequality Matters

In the DCT, you must bound a series above by a convergent series to prove convergence, or below by a divergent series to prove divergence. Reversing the direction yields no conclusion.
KEY TAKEAWAY
Think of the comparison tests like estimating weight at an airport check-in. If your suitcase is lighter than one that passed the weight limit (convergent benchmark), it will certainly pass too. If your suitcase is heavier than one that was already rejected (divergent benchmark), it will also be rejected. The Direct Comparison Test requires you to physically weigh one bag against the other; the Limit Comparison Test instead checks whether two bags are in the same weight class by looking at their ratio.

Visual Explanation

The following diagram illustrates the logic of the Direct Comparison Test. Two series with non-negative terms are plotted: a known convergent benchmark series (top) and a smaller series beneath it. Because the partial sums of the smaller series are bounded above by those of the convergent series, the smaller series must also converge. The shaded region between the curves visually represents the 'room' available—the smaller series can never exceed the benchmark.

The violet curve shows the partial sums of the benchmark series ∑bₙ, which converges to L₂. The cyan curve shows the partial sums of ∑aₙ, which satisfies 0 ≤ aₙ ≤ bₙ. Because the cyan partial sums are always below the violet ones, they are bounded above and must converge to some limit L₁ ≤ L₂.

Notice that the gap between the two curves narrows as n increases—the terms of both series are shrinking. The critical observation is not the size of the gap but rather that the cyan curve can never cross above the violet curve. This boundedness is what the monotone convergence theorem converts into a convergence guarantee. If instead ∑bₙ diverged and we knew aₙ ≥ bₙ, the argument would run in reverse: the partial sums of ∑aₙ would be bounded below by a sequence tending to infinity, forcing ∑aₙ to diverge as well.

Mathematical Framework

Direct Comparison Test (DCT)

DIRECT COMPARISON — CONVERGENCE
If 0 ≤ aₙ ≤ bₙ for all n ≥ N and ∑bₙ converges, then ∑aₙ converges.
Here aₙ is the series under investigation, bₙ is the known convergent benchmark, and N is some fixed index beyond which the inequality holds. The requirement that the inequality hold for all n ≥ N (rather than all n ≥ 1) reflects the fact that convergence depends only on the tail behavior of a series.
DIRECT COMPARISON — DIVERGENCE
If 0 ≤ bₙ ≤ aₙ for all n ≥ N and ∑bₙ diverges, then ∑aₙ diverges.
The contrapositive of the convergence statement: if the smaller series diverges, any series with larger terms must also diverge. This is used to prove divergence by comparing against a known divergent series such as the harmonic series ∑1/n.

Limit Comparison Test (LCT)

LIMIT COMPARISON TEST
Let aₙ > 0 and bₙ > 0. If lim (n→∞) aₙ/bₙ = L, where 0 < L < ∞, then ∑aₙ and ∑bₙ either both converge or both diverge.
L is a finite, positive constant. The condition 0 < L < ∞ means aₙ and bₙ are asymptotically proportional: aₙ ~ L·bₙ as n → ∞. This proportionality ensures they share the same convergence behavior. If L = 0 or L = ∞, modified conclusions apply (see below).
Edge Cases of the LCT
If L = 0: convergence of ∑bₙ implies convergence of ∑aₙ (since aₙ grows much slower than bₙ), but divergence of ∑bₙ tells us nothing. If L = ∞: divergence of ∑bₙ implies divergence of ∑aₙ (since aₙ grows much faster than bₙ), but convergence of ∑bₙ tells us nothing. These edge cases reduce the LCT to a one-directional version of the DCT.

The proof of the LCT is instructive. Since lim(aₙ/bₙ) = L with L > 0, for any ε > 0 there exists N such that for all n ≥ N we have (L − ε)bₙ ≤ aₙ ≤ (L + ε)bₙ. Choosing ε = L/2, this gives (L/2)bₙ ≤ aₙ ≤ (3L/2)bₙ. The left inequality shows that if ∑bₙ diverges, so does ∑aₙ (by the DCT applied to (L/2)bₙ ≤ aₙ). The right inequality shows that if ∑bₙ converges, so does ∑aₙ (since aₙ ≤ (3L/2)bₙ and ∑(3L/2)bₙ converges by scalar multiplication). This elegant reduction to the DCT illustrates how the LCT is not a fundamentally new tool but rather a streamlined packaging of the comparison idea.

Benchmark Series & Strategy Selection

The effectiveness of comparison tests hinges on your ability to identify an appropriate benchmark series—one whose convergence or divergence is already established. The two most commonly used families are geometric series and p-series, though other known series (such as convergent telescoping series) can also serve as benchmarks. The following diagram and table summarize the key benchmark series and when each comparison test is most effective.

Start by identifying the dominant behavior of aₙ for large n. If a direct inequality is easy to establish, use the DCT. If the algebra makes inequalities cumbersome, pivot to the LCT and compute a limit instead.
Common benchmark series used in comparison tests
Benchmark SeriesConverges WhenDiverges WhenTypical Use Case
Geometric: ∑arⁿ|r| < 1|r| ≥ 1aₙ involves exponentials like 2ⁿ, (1/3)ⁿ
p-series: ∑1/nᵖp > 1p ≤ 1aₙ is a rational function of n
Harmonic: ∑1/nNeverAlwaysLower bound for divergence proofs
∑1/(n·ln²n)Always (by integral test)N/ALogarithmic expressions in aₙ

A useful heuristic for choosing the benchmark: strip away all additive constants, lower-order terms, and multiplicative constants from aₙ to isolate its asymptotic skeleton. For instance, if aₙ = (3n² + 5)/(n⁴ − 2n + 7), the dominant behavior for large n is 3n²/n⁴ = 3/n², so bₙ = 1/n² is the natural benchmark. This asymptotic reasoning is exactly what the LCT formalizes: if aₙ/bₙ → 3 (a finite, positive constant), the two series share the same convergence behavior.

Worked Examples

Example 1: Direct Comparison Test

Determine whether ∑(n=1 to ∞) 1/(n² + 3n) converges or diverges.
1
Step 1 — Identify Dominant BehaviorFor large n, the term n² + 3n ≈ n², so aₙ = 1/(n² + 3n) behaves like 1/n². Since ∑1/n² is a convergent p-series (p = 2 > 1), we suspect our series converges.
2
Step 2 — Establish the InequalitySince n² + 3n > n² for all n ≥ 1, we have 1/(n² + 3n) < 1/n². Therefore 0 ≤ aₙ < bₙ where bₙ = 1/n².
0 ≤ 1/(n² + 3n) < 1/n² for all n ≥ 1
3
Step 3 — Apply the Direct Comparison TestThe benchmark series ∑1/n² converges (p-series with p = 2 > 1). Since 0 ≤ aₙ ≤ bₙ and ∑bₙ converges, the DCT guarantees that ∑aₙ converges.
∑ 1/(n² + 3n) converges by the Direct Comparison Test.

Example 2: Limit Comparison Test

Determine whether ∑(n=1 to ∞) (2n + 1)/(3n³ − n + 4) converges or diverges.
1
Step 1 — Identify Asymptotic BehaviorFor large n, the numerator behaves like 2n and the denominator like 3n³, so aₙ ≈ 2n/(3n³) = 2/(3n²). We choose bₙ = 1/n² as our benchmark.
2
Step 2 — Compute the Limit of the RatioCompute lim(n→∞) aₙ/bₙ = lim(n→∞) [(2n + 1)/(3n³ − n + 4)] · n² = lim(n→∞) (2n³ + n²)/(3n³ − n + 4). Dividing numerator and denominator by n³: lim(n→∞) (2 + 1/n)/(3 − 1/n² + 4/n³) = 2/3.
L = 2/3, which satisfies 0 < L < ∞
3
Step 3 — Apply the Limit Comparison TestSince L = 2/3 is finite and positive, the LCT tells us ∑aₙ and ∑bₙ share the same convergence behavior. Since ∑1/n² converges (p-series, p = 2 > 1), our original series also converges.
∑ (2n + 1)/(3n³ − n + 4) converges by the Limit Comparison Test.

Strengths & Limitations: DCT vs. LCT

Both the Direct Comparison Test and the Limit Comparison Test serve the same fundamental purpose, but they differ in ease of application and flexibility. The DCT is more elementary and can yield sharper bounds (e.g., if you know ∑bₙ = S, then ∑aₙ ≤ S), but it requires you to produce an explicit inequality. The LCT trades that requirement for a limit computation, which is often easier when the general term involves sums, differences, or complicated algebraic expressions.

Head-to-head comparison of the DCT and LCT
FeatureDirect Comparison TestLimit Comparison Test
What you must showAn explicit inequality: aₙ ≤ bₙ or aₙ ≥ bₙA limit: lim(aₙ/bₙ) = L with 0 < L < ∞
Easiest whenaₙ is clearly ≤ or ≥ a simple benchmark (e.g., removing positive terms from denominator)aₙ is a rational function of n, or involves messy additive/subtractive terms
Provides bounds?Yes — can bound the sum of the seriesNo — only determines convergence or divergence
Common pitfallWrong direction: bounding above by a divergent series or below by a convergent series proves nothingGetting L = 0 or L = ∞ and forgetting the restricted conclusions
LimitationFinding the right inequality can be algebraically difficultRequires both aₙ > 0 and bₙ > 0 (strictly positive)
🔧 WHEN TO USE WHICH
Think of the DCT as a wrench and the LCT as an adjustable wrench. The DCT is simpler and more direct when it fits—you can see the inequality immediately. The LCT is the tool you reach for when the bolt is an odd size: the algebra of establishing a direct inequality is too cumbersome, but the limit of the ratio simplifies cleanly. In practice, try the DCT first; if the inequality goes the wrong direction or is hard to prove, switch to the LCT.

Connection to Other Convergence Tests

The comparison tests sit within a broader ecosystem of convergence tests in Calculus 2. Understanding how they relate to other tests helps you develop an efficient strategy for determining convergence. The Integral Test also uses comparison—between a series and an improper integral—and is closely related in spirit. The Ratio and Root Tests are essentially internal comparison tests: they compare a series to a geometric series via the limit of consecutive-term ratios or nth roots. This geometric comparison is built into the test itself, so you never need to choose a benchmark explicitly.

How common convergence tests relate to comparison reasoning
TestBest ForRelationship to Comparison
DCT / LCTRational functions of n, algebraic expressionsThe foundational comparison technique
Integral Testf(n) positive, continuous, decreasingCompares ∑f(n) to ∫f(x)dx via area inequalities
Ratio TestFactorials, exponentials, productsImplicitly compares to a geometric series via lim|aₙ₊₁/aₙ|
Root TestTerms raised to the nth powerImplicitly compares to a geometric series via lim|aₙ|^(1/n)
Alternating Series TestSeries with (−1)ⁿ factorNot a comparison test; uses cancellation, not bounding

As you advance into real analysis and complex analysis, comparison ideas generalize further. The Weierstrass M-test extends comparison to series of functions, establishing uniform convergence by bounding |fₙ(x)| ≤ Mₙ where ∑Mₙ converges. The Abel and Dirichlet tests handle more nuanced comparison arguments for products of sequences. In each case, the core philosophy remains the same: leverage what you know about one mathematical object to draw conclusions about another.

Practice Problems

PROBLEM 1CONCEPTUAL
A student claims: "Since 1/(n² − 5) > 1/n² and ∑1/n² converges, the series ∑1/(n² − 5) also converges by the Direct Comparison Test." Identify the error in this reasoning and explain why the conclusion may still be correct.
PROBLEM 2BASIC CALCULATION
Use the Direct Comparison Test to determine whether the series ∑(n=1 to ∞) 1/(n³ + n) converges or diverges.
PROBLEM 3INTERMEDIATE
Determine whether the series ∑(n=2 to ∞) (n + 4)/(n³ − n) converges or diverges. Choose an appropriate comparison test and justify your choice of benchmark.
PROBLEM 4APPLIED
In a numerical analysis application, the error after n iterations of an algorithm satisfies eₙ = (3 + sin n)/(2ⁿ + n). Does the total accumulated error ∑(n=1 to ∞) eₙ converge? Use a comparison test to justify your answer.
PROBLEM 5CRITICAL THINKING
Prove or disprove: if aₙ > 0, bₙ > 0, and lim(n→∞) aₙ/bₙ = 0, then ∑aₙ converges. If the statement is false, provide a counterexample and state what additional hypothesis would make it true.

Comparison Tests — Summary

The comparison tests are fundamental tools for determining the convergence or divergence of infinite series with non-negative terms. The Direct Comparison Test requires an explicit inequality between the series under investigation and a benchmark series of known behavior: bound above by a convergent series to prove convergence, or bound below by a divergent series to prove divergence. The Limit Comparison Test replaces the inequality with a limit: if lim(aₙ/bₙ) = L with 0 < L < ∞, then the two series share the same convergence behavior.

The key to both tests is choosing the right benchmark. For rational functions of n, the natural benchmark is a p-series ∑1/nᵖ; for exponential expressions, the benchmark is a geometric series ∑rⁿ. Strip away lower-order terms to identify the asymptotic skeleton of the general term—this reveals the correct benchmark and guides your choice between the DCT and LCT.

Varsity Tutors • Calculus 2 • Comparison Tests