Improper Integrals — Evaluating Improper Integrals - Calculus 2

SECTION 1

Historical Context & Motivation

The definite integral, as formalized by Riemann in the nineteenth century, requires two conditions: a finite interval of integration and a bounded integrand. Yet many of the most important problems in mathematics and physics — computing the total probability under a normal distribution, determining gravitational potential energy over infinite distances, or evaluating Laplace and Fourier transforms — demand integration over infinite intervals or through singularities where the integrand becomes unbounded. The concept of the improper integral arose precisely to address these situations, redefining integration as a limiting process that can sometimes yield finite, meaningful values even when the domain or the function itself is unbounded.

1668 • Torricelli's Trumpet

Evangelista Torricelli demonstrated that the solid of revolution formed by rotating y = 1/x about the x-axis for x ≥ 1 has finite volume but infinite surface area — a result that stunned contemporaries and implicitly required integration over an infinite interval.

1730s • Euler's Gamma Function

Leonhard Euler defined the Gamma function Γ(n) as an integral from 0 to ∞, establishing one of the first systematic uses of improper integrals and connecting them to the factorial function for non-integer arguments.

1799 • Laplace and the Gaussian Integral

Pierre-Simon Laplace computed ∫₋∞^∞

e^{−x²}

dx = √π using a celebrated double-integral technique, providing a cornerstone result for probability theory and statistical mechanics.

1854 • Riemann's Integral Formalization

Bernhard Riemann rigorously defined the definite integral for bounded functions on closed intervals. Extending his framework to unbounded cases required the limit-based definition of improper integrals that we use today.

1902 • Lebesgue Integration

Henri Lebesgue's measure-theoretic approach to integration provided a more general framework that subsumes many improper integrals, but the classical limit-based approach remains the standard tool in introductory calculus courses.

The central question this lesson addresses is: when an integral violates the standard hypotheses of the Riemann integral — either through infinite limits or unbounded integrands — can we still assign it a finite value, and if so, how do we compute it? The answer lies in replacing the offending feature with a parameter and taking a limit, a technique that is both elegant and practically indispensable across mathematics, engineering, and the natural sciences.

SECTION 2

Core Principles & Definitions

An integral is called improper when the standard Riemann definition does not directly apply. This occurs in exactly two circumstances: either the interval of integration extends to infinity (a Type I improper integral), or the integrand has an infinite discontinuity within the interval (a Type II improper integral). In either case, the strategy is the same: replace the problematic bound or point with a finite parameter, evaluate the resulting proper integral, and then take a limit. If the limit exists and is finite, we say the improper integral converges; otherwise, it diverges.

Type I — Infinite Limits

One or both limits of integration are ±∞. Evaluate by replacing the infinite limit with a variable t and computing lim(t→∞) or lim(t→−∞) of the resulting proper integral.

Type II — Discontinuous Integrand

The integrand has a vertical asymptote at or within the interval [a, b]. Replace the singular endpoint with a variable t and take the appropriate one-sided limit as t approaches the point of discontinuity.

Convergence vs. Divergence

If the defining limit exists and equals a finite number L, the improper integral converges to L. If the limit is ±∞ or does not exist, the integral diverges and has no finite value.

Splitting at Interior Singularities

When the integrand blows up at an interior point c ∈ (a, b), split the integral into two pieces — from a to c and from c to b — and evaluate each as a separate improper integral. Both must converge for the whole integral to converge.

The p-Integral Tests

The integrals ∫₁^∞ 1/xᵖ dx and ∫₀¹ 1/xᵖ dx serve as fundamental benchmarks. The former converges iff p > 1; the latter converges iff p < 1. These results underpin the comparison test for convergence.

✦

KEY TAKEAWAY

Think of evaluating an improper integral like filling a container whose opening extends to infinity or narrows to a point. You fill the container up to some finite marker t, measure the volume accumulated, and then ask: as t moves toward the problematic boundary, does the total volume settle toward a fixed number, or does it grow without bound? If it settles, the integral converges — the 'infinite container' holds a finite amount. This is precisely the role of the limit in the definition: it quantifies whether the area, despite living on an unbounded domain, is actually finite.

SECTION 3

Visual Explanation — Type I Improper Integral

The curve y = 1/x² is plotted for x ≥ 1. The shaded region extends indefinitely to the right, yet its total area equals exactly 1. The dashed pink line at x = t represents the finite upper bound used during evaluation; taking t → ∞ yields the convergent value.

The diagram above illustrates the essential idea behind a Type I improper integral. Although the region under y = 1/x² from x = 1 to x = ∞ stretches infinitely to the right, the curve decays so rapidly that the accumulated area converges to a finite value. The evaluation strategy is visible in the diagram: first compute the proper integral from 1 to t, obtaining the antiderivative −1/x evaluated from 1 to t, which gives 1 − 1/t. Then take the limit as t → ∞. Since 1/t → 0, the total area is exactly 1. This convergence depends critically on the rate at which the integrand approaches zero — a theme we will formalize through the p-integral test.

SECTION 4

Mathematical Framework

Type I Definitions

TYPE I — UPPER LIMIT INFINITE

∫ₐ^∞ f(x) dx = lim(t→∞) ∫ₐᵗ f(x) dx

If f is continuous on [a, ∞), replace the infinite upper limit with a parameter t and evaluate the limit. If the limit exists and is finite, the integral converges; otherwise, it diverges.

TYPE I — LOWER LIMIT INFINITE

∫₋∞^b f(x) dx = lim(t→−∞) ∫ₜᵇ f(x) dx

Analogously, when the lower limit is −∞, replace it with t and let t → −∞.

TYPE I — BOTH LIMITS INFINITE

∫₋∞^∞ f(x) dx = ∫₋∞^c f(x) dx + ∫_c^∞ f(x) dx

Split at any convenient point c (often c = 0). Both pieces must converge independently for the whole integral to converge. This splitting is not optional — the two limits must be taken separately, not as a single symmetric limit, to avoid erroneous cancellations.

Type II Definitions

TYPE II — SINGULARITY AT ENDPOINT b

∫ₐ^b f(x) dx = lim(t→b⁻) ∫ₐᵗ f(x) dx

When f is continuous on [a, b) but has a vertical asymptote at x = b, approach b from the left through the parameter t.

TYPE II — SINGULARITY AT ENDPOINT a

∫ₐ^b f(x) dx = lim(t→a⁺) ∫ₜᵇ f(x) dx

When f is continuous on (a, b] but has a vertical asymptote at x = a, approach a from the right through the parameter t.

⚠

Common Pitfall

Never attempt to evaluate ∫₋∞^∞ f(x) dx as lim(t→∞) ∫₋ₜᵗ f(x) dx. This symmetric limit is called the Cauchy principal value and may exist even when the improper integral diverges. For example, ∫₋∞^∞ x dx diverges (both halves diverge), but the Cauchy principal value is 0 due to symmetry cancellation. Always split at a finite point and require independent convergence of each piece.

The p-Integral Reference Results

P-INTEGRAL — INFINITE INTERVAL

∫₁^∞ (1/xᵖ) dx converges ⟺ p > 1 (value = 1/(p−1))

For p ≤ 1 the integral diverges. At the critical boundary p = 1, this becomes ∫₁^∞ 1/x dx = lim(t→∞) ln t = ∞.

P-INTEGRAL — SINGULARITY AT ZERO

∫₀¹ (1/xᵖ) dx converges ⟺ p < 1 (value = 1/(1−p))

For p ≥ 1 the integral diverges. Notice that the convergence condition is reversed compared to the infinite-interval case — a function that decays too slowly at infinity may be perfectly integrable near a singularity, and vice versa.

SECTION 5

Detailed Classification & Convergence Tests

Not every improper integral can be computed in closed form with an antiderivative. In many cases, we need to determine convergence or divergence without evaluating the integral explicitly. The comparison test and the limit comparison test allow us to deduce the behavior of a difficult integral by comparing it to one whose convergence is already known. The p-integrals from Section 4 serve as the primary benchmarks for these comparisons.

This decision flowchart guides you through classifying an improper integral as Type I or Type II, applying the appropriate limit definition, and determining convergence. When no closed-form antiderivative exists, the comparison tests (orange box) offer an alternative path.

Comparison Test

Suppose 0 ≤ f(x) ≤ g(x) for all x ≥ a. If ∫ₐ^∞ g(x) dx converges, then ∫ₐ^∞ f(x) dx also converges (the smaller function's integral is bounded above). Conversely, if ∫ₐ^∞ f(x) dx diverges, then ∫ₐ^∞ g(x) dx also diverges (the larger function's integral is bounded below by a divergent quantity). Think of it as a ceiling argument: if the ceiling has finite area, everything below it must too.

Limit Comparison Test

If f(x) > 0 and g(x) > 0 for x ≥ a, and lim(x→∞) f(x)/g(x) = L where 0 < L < ∞, then ∫ₐ^∞ f(x) dx and ∫ₐ^∞ g(x) dx either both converge or both diverge. The limit comparison test is often easier to apply than the direct comparison test because you do not need a pointwise inequality — only asymptotic equivalence. In practice, the standard comparison function g is typically a p-integral benchmark 1/xᵖ, where you select p to match the dominant power behavior of f as x → ∞.

SECTION 6

Worked Examples

Example 1 — Type I Improper Integral

1
Step 1 — Identify the Improper Feature

Evaluate ∫₁^∞ (1/x³) dx. The upper limit of integration is ∞, so this is a Type I improper integral. The integrand f(x) = 1/x³ is continuous on [1, ∞), so the only source of impropriety is the infinite upper bound.
2
Step 2 — Replace ∞ with Parameter t

Rewrite the integral as lim(t→∞) ∫₁ᵗ x⁻³ dx. This is now a proper integral for each fixed t > 1.
3
Step 3 — Find the Antiderivative

Using the power rule, ∫ x⁻³ dx = x⁻²/(−2) = −1/(2x²) + C.

Antiderivative: F(x) = −1/(2x²)
4
Step 4 — Evaluate from 1 to t

∫₁ᵗ x⁻³ dx = [−1/(2x²)]₁ᵗ = −1/(2t²) − (−1/2) = 1/2 − 1/(2t²).

= 1/2 − 1/(2t²)
5
Step 5 — Take the Limit

lim(t→∞) [1/2 − 1/(2t²)] = 1/2 − 0 = 1/2. The limit exists and is finite, so the integral converges.

∫₁^∞ (1/x³) dx = 1/2

Example 2 — Type II Improper Integral

1
Step 1 — Identify the Improper Feature

Evaluate ∫₀⁴ 1/√x dx. The integrand f(x) = x⁻¹ᐟ² has a vertical asymptote at x = 0 (the left endpoint), making this a Type II improper integral.
2
Step 2 — Replace the Singular Endpoint

Rewrite as lim(t→0⁺) ∫ₜ⁴ x⁻¹ᐟ² dx. For each t > 0, the integrand is continuous on [t, 4], so the integral is proper.
3
Step 3 — Find the Antiderivative

∫ x⁻¹ᐟ² dx = x¹ᐟ²/(1/2) = 2√x + C.

Antiderivative: F(x) = 2√x
4
Step 4 — Evaluate from t to 4

∫ₜ⁴ x⁻¹ᐟ² dx = [2√x]ₜ⁴ = 2√4 − 2√t = 4 − 2√t.

= 4 − 2√t
5
Step 5 — Take the Limit

lim(t→0⁺) [4 − 2√t] = 4 − 0 = 4. The limit is finite, so the improper integral converges.

∫₀⁴ 1/√x dx = 4

SECTION 7

Common Pitfalls & Practical Tips

Pitfall	Why It Fails	Correct Approach
Applying FTC directly without limits	Substituting ∞ directly into an antiderivative is undefined; the Fundamental Theorem of Calculus requires a closed, finite interval.	Always introduce a parameter t, evaluate on [a, t], and take lim(t→∞).
Missing hidden discontinuities	An integrand like 1/(x−2) on [0, 4] has a singularity at x = 2 that may not be immediately obvious.	Always check for vertical asymptotes inside the interval before integrating.
Using symmetric limits for ∫₋∞^∞	Computing lim(t→∞) ∫₋ₜᵗ f(x) dx may give a Cauchy principal value that exists even when the improper integral diverges.	Split into two independent pieces at a finite c and require both to converge separately.
Confusing the two p-tests	∫₁^∞ 1/xᵖ dx converges for p > 1, but ∫₀¹ 1/xᵖ dx converges for p < 1. Mixing up these conditions leads to incorrect convergence conclusions.	Associate 'infinite interval → need faster decay → p > 1' and 'singularity at 0 → need milder blow-up → p < 1'.
Ignoring that both halves must converge	When splitting ∫₋∞^∞ or splitting at an interior singularity, one convergent piece cannot 'save' a divergent piece.	Evaluate each piece separately; if either diverges, the whole integral diverges.

Five common errors when evaluating improper integrals

✦

KEY TAKEAWAY

The evaluation of improper integrals is fundamentally a two-stage process: first, solve a standard integration problem on a finite interval parameterized by t; second, determine whether the result has a finite limit as t approaches the problematic boundary. This separation of 'compute' and 'test' mirrors how engineers handle systems pushed to extreme conditions — you analyze a controlled version first and then study what happens at the boundary. Mastering this workflow prepares you for Laplace transforms, Fourier analysis, and probability distributions, all of which rely on the convergence of improper integrals.

SECTION 8

Connections to Advanced Topics

The techniques for evaluating improper integrals that you have developed in this lesson form the gateway to several powerful areas of mathematics and applied science. The limit-based definition of improper integrals is not merely a pedagogical tool — it is the operational mechanism behind integral transforms, probability theory, and series convergence tests. Understanding convergence of improper integrals directly enables the integral test for series, since the convergence of ∑ f(n) is linked to the convergence of ∫₁^∞ f(x) dx for positive, decreasing functions.

This Lesson	Advanced Extension
∫₁^∞ f(x) dx with limit as t→∞	Integral test for infinite series: if f is positive, continuous, and decreasing, then ∑ f(n) converges iff ∫₁^∞ f(x) dx converges.
p-integral ∫₁^∞ 1/xᵖ dx	p-series ∑ 1/nᵖ: converges for p > 1, diverges for p ≤ 1. The integral and series share the same convergence boundary.
Comparison and limit comparison tests for integrals	Direct comparison and limit comparison tests for series — the logic transfers almost verbatim from integrals to sums.
∫₀^∞ f(x)e⁻ˢˣ dx (convergent improper integral)	Laplace transform ℒ{f(t)} = ∫₀^∞ f(t)e⁻ˢᵗ dt — requires evaluating a Type I improper integral for each s in the domain of convergence.
∫₋∞^∞ f(x) dx converges to 1	Probability density functions: the total probability must be 1, requiring convergence of the integral over (−∞, ∞).

How improper integral techniques extend to advanced topics

In more advanced courses, the classical (Riemann) improper integral is generalized by Lebesgue integration, which handles a broader class of functions and provides cleaner convergence theorems such as the Dominated Convergence Theorem and Monotone Convergence Theorem. However, for the vast majority of computations encountered in Calculus 2, differential equations, and introductory probability, the limit-based approach developed here is both sufficient and computationally efficient.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

Explain why ∫₁^∞ (1/x) dx diverges even though the integrand 1/x approaches 0 as x → ∞. Does every integral whose integrand tends to zero necessarily converge? What additional condition is needed?

Show answer

Although 1/x → 0 as x → ∞, the decay is not fast enough to produce a finite area. Formally, ∫₁ᵗ (1/x) dx = ln t, and lim(t→∞) ln t = ∞, so the integral diverges. The integrand approaching zero is a necessary but not sufficient condition for convergence; the function must approach zero rapidly enough. The p-integral test quantifies this: ∫₁^∞ 1/xᵖ dx converges only when p > 1, meaning the decay must be faster than 1/x. This is analogous to the divergence of the harmonic series ∑ 1/n despite its terms tending to zero.

Question 2: BASIC CALCULATION

Evaluate ∫₂^∞ (1/x²) dx. State clearly whether the integral converges or diverges, and give the exact value if it converges.

Show answer

This is a Type I improper integral. Rewrite as lim(t→∞) ∫₂ᵗ x⁻² dx. The antiderivative is −1/x. Evaluating: [−1/x]₂ᵗ = −1/t − (−1/2) = 1/2 − 1/t. Taking the limit: lim(t→∞) (1/2 − 1/t) = 1/2 − 0 = 1/2. Since the limit is finite, the integral converges, and ∫₂^∞ (1/x²) dx = 1/2. This is consistent with the p-integral formula 1/(p−1) · 1/aᵖ⁻¹ with p = 2 and a = 2, giving 1/(2−1) · 1/2¹ = 1/2.

Question 3: INTERMEDIATE

Evaluate ∫₀³ 1/(3 − x)²ᐟ³ dx. Identify the type of improper integral and compute the exact value.

Show answer

The integrand f(x) = (3 − x)⁻²ᐟ³ has a vertical asymptote at x = 3 (the upper endpoint), making this a Type II improper integral. Rewrite as lim(t→3⁻) ∫₀ᵗ (3 − x)⁻²ᐟ³ dx. Using the substitution u = 3 − x (du = −dx), the antiderivative becomes −(3 − x)¹ᐟ³/(1/3) = −3(3 − x)¹ᐟ³. Evaluating: [−3(3 − x)¹ᐟ³]₀ᵗ = −3(3 − t)¹ᐟ³ − (−3 · 3¹ᐟ³) = 3 · 3¹ᐟ³ − 3(3 − t)¹ᐟ³. As t → 3⁻, (3 − t)¹ᐟ³ → 0, so the limit is 3 · 3¹ᐟ³ = 3⁴ᐟ³. The integral converges to 3⁴ᐟ³ ≈ 4.327. Note that p = 2/3 < 1 here (in the local variable u = 3 − x), consistent with the p-integral test for singularities.

Question 4: APPLIED

In probability, the exponential distribution has density f(x) = λe⁻λˣ for x ≥ 0 (with λ > 0). Show that ∫₀^∞ f(x) dx = 1, confirming this is a valid probability density function. Then compute the expected value E[X] = ∫₀^∞ x · λe⁻λˣ dx.

Show answer

For the normalization: ∫₀^∞ λe⁻λˣ dx = lim(t→∞) ∫₀ᵗ λe⁻λˣ dx = lim(t→∞) [−e⁻λˣ]₀ᵗ = lim(t→∞) (−e⁻λᵗ + 1) = 0 + 1 = 1. This confirms f is a valid PDF. For the expected value: E[X] = ∫₀^∞ xλe⁻λˣ dx. Use integration by parts with u = x, dv = λe⁻λˣ dx, so du = dx and v = −e⁻λˣ. Then E[X] = lim(t→∞) {[−xe⁻λˣ]₀ᵗ + ∫₀ᵗ e⁻λˣ dx} = lim(t→∞) {−te⁻λᵗ + 0 + [−(1/λ)e⁻λˣ]₀ᵗ} = lim(t→∞) {−te⁻λᵗ − (1/λ)e⁻λᵗ + 1/λ}. Since te⁻λᵗ → 0 and e⁻λᵗ → 0, we get E[X] = 1/λ. The expected value of an exponentially distributed random variable is the reciprocal of its rate parameter.

Question 5: CRITICAL THINKING

Consider the integral ∫₁^∞ sin(x)/x² dx. This integral cannot be evaluated in elementary closed form. Use the comparison test to prove that it converges absolutely. Then discuss: does ∫₁^∞ sin(x)/x dx converge? (You may use without proof the fact that it converges conditionally but not absolutely.)

Show answer

For ∫₁^∞ sin(x)/x² dx, note that |sin(x)/x²| ≤ 1/x² for all x ≥ 1, since |sin(x)| ≤ 1. We know ∫₁^∞ 1/x² dx converges (p = 2 > 1). By the comparison test, ∫₁^∞ |sin(x)/x²| dx converges, which means ∫₁^∞ sin(x)/x² dx converges absolutely. For ∫₁^∞ sin(x)/x dx, the situation is fundamentally different. Since |sin(x)/x| behaves like 1/x on average and ∫₁^∞ 1/x dx diverges, the integral does not converge absolutely. However, it does converge conditionally — the alternating positive and negative half-cycles of sin(x) produce cancellations that drive the partial integrals toward a limit (this can be proved using the Dirichlet test for integrals). This example highlights that absolute convergence is a stronger condition than conditional convergence, and that the comparison test can only establish absolute convergence. Functions with oscillatory behavior can produce conditional convergence even when the absolute integral diverges.

SUMMARY

Lesson Summary

Summary

An integral is improper when it involves an infinite interval of integration (Type I) or an unbounded integrand (Type II). In both cases, the evaluation strategy is the same: replace the offending feature with a finite parameter t, compute the resulting proper integral, and take a limit. If the limit is finite, the integral converges; if the limit is infinite or does not exist, the integral diverges. When both limits are infinite or a singularity lies in the interior of the interval, the integral must be split into independent pieces, each of which must converge for the whole integral to converge.

The p-integral tests serve as fundamental benchmarks: ∫₁^∞ 1/xᵖ dx converges iff p > 1, while ∫₀¹ 1/xᵖ dx converges iff p < 1. These results power the comparison test and limit comparison test, which determine convergence without requiring an explicit antiderivative. Mastery of improper integrals is essential for the integral test for series, Laplace transforms, and probability density functions — topics that permeate the rest of your mathematical education.