The integral test compares an infinite sum of positive terms with the area under a related curve. It is useful when the terms come from a function that is easy to integrate and whose shape can be controlled.
Check the function conditions
Suppose where is positive, continuous, and decreasing for all sufficiently large . Then
have the same convergence behavior over their corresponding infinite tails. The test does not say their numerical values are equal.
The geometric reason is that rectangles built from decreasing positive heights can be trapped between neighboring pieces of the area under the curve. A finite total area forces a finite total sum, and an infinite area forces the sum to diverge.
The conditions need only hold eventually. Removing finitely many starting terms does not change whether the series converges. But positivity and decreasing behavior must be established on the tail being used, not assumed from a few sampled values.
Example: derive the p-series rule
For with use which is positive, continuous, and decreasing for
If the relevant integral uses antiderivative The limit at is finite precisely when At the antiderivative is which grows without bound. Therefore
For the terms do not approach zero, so divergence already follows from the nth-term test. The harmonic series is the borderline case equals 1 and diverges.
Example: a logarithmic denominator
Consider Its related function is positive and continuous for and differentiation shows it decreases there. Use
Thus the series converges. The number is the integral's value, not the series' sum. Replacing the squared logarithm with a single logarithm produces which diverges. A seemingly small exponent change alters the tail behavior.
Integral bounds estimate the remainder
If the positive decreasing series converges and is its first N terms, its remainder satisfies
The lower bound starts at the next integer because of how the tail rectangles sit above and below the curve. Add either bound to to obtain corresponding bounds for the full sum.
For this gives To guarantee an error no greater than 0.01, taking is sufficient. This is a guarantee from the bound, not a claim that the actual error is exactly 0.01.
Avoid inappropriate comparisons
The integral test is designed for positive terms on an eventually decreasing tail. An alternating series may need an alternating-series test or a separate test of absolute convergence. If the function is not decreasing initially, identify a point after which it is and handle earlier terms separately.
A graph can suggest monotonicity but does not prove it for all sufficiently large . A derivative sign or a direct algebraic comparison supplies the needed argument.
An eventual decrease can be shown with a derivative
For differentiation gives It is negative for so the positive terms decrease from a sufficiently large integer onward. Integration by parts shows the corresponding improper integral converges.
The initial failure of decrease near equals one does not invalidate an integral-test argument on the tail. State where the conditions begin, then note that finitely many earlier terms do not change convergence.
Try it yourself
Does converge? Give an upper bound for the remainder after N terms.
Answer: It is a p-series with so it converges. The remainder satisfies
What to remember
Verify positivity, continuity, and eventual decrease. Use the integral for convergence behavior and tail bounds, not as the exact value of the series.