What Is the Direct Comparison Test?
The direct comparison test decides the fate of a positive-term series by squeezing it against a benchmark series whose behavior is already known. The logic is intuitive:
if 0 ≤ aₙ ≤ bₙ and Σ bₙ converges, then Σ aₙ converges (smaller than something finite is finite), if aₙ ≥ bₙ ≥ 0 and Σ bₙ diverges, then Σ aₙ diverges (bigger than something infinite is infinite).
A series bounded above by a convergent one cannot escape to infinity; a series bounded below by a divergent one is dragged to infinity with it. The art is choosing a good benchmark, almost always a p-series (Σ 1/nᵖ) or a geometric series, whose convergence you know instantly. This calculator verifies the termwise inequality over a large range, classifies the benchmark, and delivers the verdict, or explains why direct comparison fails and the limit comparison test is needed instead.
How to Use the Direct Comparison Test Calculator
Enter your series term aₙ and a benchmark bₙ. The steps check whether aₙ ≤ bₙ or aₙ ≥ bₙ holds consistently, classify the benchmark (convergent p-series, divergent harmonic, geometric, and so on), and combine the inequality direction with the benchmark's behavior to reach a conclusion, or report that no consistent inequality holds. The graph plots both term sequences so the domination is visible. When the curves cross and direct comparison stalls, the limit comparison test is the more forgiving alternative, and the divergence test is the quick first gate.
Worked Example
Test Σ 1/(n² + 3) by comparing with bₙ = 1/n². Since n² + 3 > n², the fraction is smaller:
1/(n² + 3) ≤ 1/n² for all n
The benchmark Σ 1/n² is a convergent p-series (p = 2 > 1, summing to π²/6). Our series is termwise smaller and positive, so it is trapped below a finite total and must converge. The verdict is clean because the inequality holds for every term. The second chip shows the divergent direction: 1/(n − 0.5) ≥ 1/n, and since the harmonic series Σ 1/n diverges, the larger series diverges too. The key discipline is matching the inequality direction to the benchmark, use a convergent benchmark above your series, or a divergent one below it; the other two combinations prove nothing.
Why Only Two Combinations Work
There are four ways to pair an inequality direction with a benchmark's behavior, but only two are logically valid, and knowing which is essential. Being smaller than a convergent series proves convergence (bounded above by something finite). Being larger than a divergent series proves divergence (bounded below by something infinite). But being smaller than a divergent series proves nothing, you can be tiny and still diverge, or tiny and converge. And being larger than a convergent series proves nothing either. This is the direct comparison test's rigidity: it demands not just any comparison but the right-direction comparison, and it demands a genuine termwise inequality holding for all sufficiently large n. That rigidity is also its weakness. Consider 1/(n² − n + 3): it should behave like 1/n² and converge, but the inequality 1/(n² − n + 3) ≤ 1/n² fails for small n because subtracting n makes the denominator smaller. The direct test stumbles on the algebra even though the conclusion is obvious. The limit comparison test fixes exactly this by requiring only that the ratio aₙ/bₙ approach a finite positive limit, waiving the pointwise inequality entirely, which is why it is often preferred in practice.
Common Mistakes to Avoid
- Comparing in the wrong direction. A convergent benchmark must sit above your series, a divergent one below. Smaller-than-divergent and larger-than-convergent prove nothing; getting the direction wrong invalidates the argument entirely.
- Assuming an inequality that fails for small n. The test needs aₙ ≤ bₙ (or ≥) for all large n, but check it holds. Denominators with subtracted terms often violate the naive inequality early, requiring a different benchmark or the limit comparison test.
- Using a benchmark of unknown status. The whole method outsources the verdict to bₙ, so bₙ must be a series you can classify, a p-series or geometric series. Comparing against something you cannot evaluate learns nothing.
- Applying it to series with negative terms. Direct comparison requires non-negative terms. For series with sign changes, compare absolute values (testing absolute convergence) or use the alternating series test.
- Forgetting the test only needs eventual domination. Convergence depends only on the tail, so the inequality need only hold from some index onward. Fussing over the first few terms is unnecessary, but the inequality must be genuine for large n.
Real-World Applications
The comparison principle, bounding an unknown quantity between known ones, extends far beyond series and is one of the most reusable ideas in mathematics. In computer science, the analysis of algorithm running times uses direct comparison constantly: bounding a complicated cost sum above by a convergent or polynomially-bounded series proves an algorithm terminates efficiently, and big-O notation is comparison reasoning formalized. In numerical analysis, the convergence of iterative solvers and series approximations is established by comparing error terms against geometric benchmarks (proving geometric convergence) or p-series (proving polynomial rates). Physics and engineering bound the energy or amplitude of a perturbation series by a dominating convergent series to guarantee the model is well-behaved. The technique embodies a fundamental proof strategy, the squeeze, that reappears as the squeeze theorem for limits, the comparison test for improper integrals, and dominated convergence in advanced analysis. Learning to pick the right benchmark and align the inequality direction trains the pattern-recognition that mathematicians use to tame unfamiliar sums by relating them to familiar ones. When the strict inequality is inconvenient, its more flexible cousin the limit comparison test takes over, and together they resolve the convergence of the vast majority of series encountered in practice, from the tails of probability distributions to the residuals of Fourier and power series.
Frequently Asked Questions
How does the direct comparison test work?
It bounds your positive-term series against a benchmark. If your terms are smaller than those of a convergent series, yours converges; if larger than those of a divergent series, yours diverges. The benchmark's known behavior transfers to your series through the termwise inequality.
Which inequality direction proves convergence versus divergence?
Smaller than a convergent benchmark proves convergence (trapped below something finite); larger than a divergent benchmark proves divergence (pushed above something infinite). The other two pairings, smaller than divergent, larger than convergent, prove nothing at all.
What benchmarks should I use?
Almost always a p-series (Σ 1/nᵖ, convergent iff p > 1) or a geometric series (Σ rⁿ, convergent iff |r| < 1), because their convergence is known instantly. Choose one whose terms resemble your series' dominant behavior for large n.
What if the inequality fails for small n?
Convergence depends only on the tail, so the inequality need only hold for all sufficiently large n. If it fails even in the tail (common when denominators subtract terms), direct comparison cannot be used, and the limit comparison test, which needs only a ratio limit, is the fix.
How is this different from the limit comparison test?
Direct comparison requires a genuine termwise inequality; limit comparison requires only that the ratio aₙ/bₙ approach a finite positive number. The limit version is more forgiving because it ignores messy small-n behavior, but direct comparison, when it applies, gives a cleaner, more elementary proof.
Can I use the direct comparison test on alternating series?
Not directly, since it requires non-negative terms. You can apply it to the absolute values to test for absolute convergence: if Σ|aₙ| converges by comparison, the original series converges absolutely. For conditional convergence, use the alternating series test.
Why does being smaller than a divergent series prove nothing?
Because "smaller than infinite" is no constraint, you could be a convergent series (like 1/n²) or a divergent one (like 1/(2n)), both smaller than the divergent 1/n. Without an upper bound by a convergent series, the comparison gives no information about your series' fate.
How do I choose the right benchmark for a rational term?
Look at the dominant powers. For a term like 1/(n² + 3), the denominator behaves like n², so compare with 1/n². Match the leading behavior, then check the inequality direction. If the exact inequality is awkward, switch to limit comparison with the same benchmark.
Does the test give the sum of the series?
No, only whether it converges or diverges. Comparison establishes existence of a finite sum without computing it. Finding the actual value requires other methods (telescoping, geometric formulas, or known results), and most convergent series have no elementary closed-form sum.
What is the connection to improper integrals?
The same comparison principle decides improper-integral convergence: if 0 ≤ f ≤ g and ∫g converges, so does ∫f. The integral test bridges the two, comparing a series with an integral, and the comparison strategy is identical, bound the unknown between known quantities.