Difficulty: Intermediate | Prerequisites: Definite integrals, limits, basic integration techniques (substitution)
Improper integrals extend definite integration to cases that ordinary Riemann sums cannot handle: intervals that stretch to infinity, or integrands that blow up at a point inside (or at the edge of) the interval. They sit right at the boundary between integration and limits, and you will need both toolkits working together. If you are comfortable evaluating definite integrals by antiderivative and taking limits at infinity, you have everything you need to start.
An improper integral is a definite integral where either the interval is infinite (Type 1) or the integrand has a discontinuity on the interval (Type 2). You evaluate them by turning the troublesome bound into a limit variable, computing the resulting proper integral, and then taking the limit. If the limit exists and is finite, the integral converges; otherwise it diverges.
Improper integral
A definite integral in which at least one limit of integration is infinite, or the integrand is unbounded (has a vertical asymptote) on the interval. In simple terms, it is any integral where the ordinary "plug in the bounds" approach breaks down because something is blowing up or running off to infinity.
Type 1 improper integral (infinite limits)
An integral where one or both bounds of integration are infinite, e.g. the integral from 1 to infinity of 1/x^2 dx. Think of it as: the interval itself never ends.
Type 2 improper integral (discontinuous integrand)
An integral where the integrand has a discontinuity (typically a vertical asymptote) at some point within or at the boundary of the interval, e.g. the integral from 0 to 1 of 1/sqrt(x) dx. Think of it as: the function itself blows up somewhere on the interval, even though the interval is finite.
Convergent integral
An improper integral whose associated limit exists and is a finite real number. In simple terms, even though something infinite is happening, the total area settles to a fixed value.
Divergent integral
An improper integral whose associated limit does not exist or is infinite. The area never settles down to a number.
Type 1 integrals arise when one or both limits of integration are infinite. The strategy is always the same: replace the infinite bound with a finite variable, evaluate the proper integral, then take the limit.
Upper bound is infinite
\int_a^{\infty} f(x)\,dx := \lim_{t \to \infty} \int_a^t f(x)\,dxLower bound is infinite
\int_{-\infty}^b f(x)\,dx := \lim_{t \to -\infty} \int_t^b f(x)\,dxBoth bounds infinite
Split at any convenient point (often 0 or the point where the function is simplest). Both pieces must independently converge for the whole integral to converge.
\int_{-\infty}^{\infty} f(x)\,dx := \int_{-\infty}^a f(x)\,dx + \int_a^{\infty} f(x)\,dxConvergence rule for Type 1
The integral converges if the limit exists and is a finite number.
The integral diverges if the limit does not exist, or if the limit is infinite.
Classifying examples from the source notes
Integral from 0 to infinity of e^(-x) dx: Type 1 (upper bound is infinite)
Integral from 1 to infinity of 1/x dx: Type 1 (upper bound is infinite)
Integral from negative infinity to infinity of 1/(1 + x^2) dx: Type 1 (both bounds infinite)
Type 2 integrals arise when the integrand has a discontinuity (usually a vertical asymptote) at some point c where a <= c <= b. The interval is finite, but the function is not bounded.
Discontinuity at an endpoint or interior point
If f has a discontinuity at c where a <= c <= b, split the integral at c and approach the trouble spot from each side:
\int_a^b f(x)\,dx := \lim_{t \to c^-} \int_a^t f(x)\,dx + \lim_{t \to c^+} \int_t^b f(x)\,dxBoth limits must exist and be finite for the integral to converge.
If the discontinuity is at the left endpoint (c = a), you only need the right-hand piece. If it is at the right endpoint (c = b), you only need the left-hand piece.
Convergence rule for Type 2
The integral converges if every limit involved exists and is finite.
The integral diverges if any one of the limits does not exist or is infinite.
Classifying examples from the source notes
Integral from 0 to 1 of 1/sqrt(x) dx: Type 2 (integrand blows up at x = 0, the left endpoint)
Integral from 1 to 3 of 1/(x - 2) dx: Type 2 (integrand has a discontinuity at x = 2, inside the interval)
Integral from 3 to 5 of 1/(x - 2) dx: Proper (the discontinuity at x = 2 is outside the interval of integration)
Identifying "Type 1 + 2" hybrids
Some integrals are improper for both reasons. The integral from negative infinity to 8 of 1/(x - 2) dx has an infinite lower bound (Type 1) and a discontinuity at x = 2 inside the interval (Type 2). Split into separate pieces, each of which must converge independently.
Type 1 definitions
\int_a^{\infty} f(x)\,dx = \lim_{t \to \infty} \int_a^t f(x)\,dx\int_{-\infty}^b f(x)\,dx = \lim_{t \to -\infty} \int_t^b f(x)\,dx\int_{-\infty}^{\infty} f(x)\,dx = \int_{-\infty}^a f(x)\,dx + \int_a^{\infty} f(x)\,dxType 2 definition
\int_a^b f(x)\,dx = \lim_{t \to c^-} \int_a^t f(x)\,dx + \lim_{t \to c^+} \int_t^b f(x)\,dx \quad \text{(discontinuity at } c \text{)}p-test for Type 1 (essential benchmark)
\int_1^{\infty} \frac{1}{x^p}\,dx \quad \text{converges if } p > 1, \quad \text{diverges if } p \le 1This is the single most useful reference point for quick convergence checks. The integral of 1/x^2 (p = 2) converges to 1. The integral of 1/x (p = 1) diverges.
Example 1: Integral of 6/(x + 4)^2 from 0 to infinity (Type 1, converges)
Substitute u = x + 4, du = dx. The integral becomes the limit as t goes to infinity of the integral from 4 to t + 4 of 6/u^2 du.
The antiderivative of 6/u^2 is -6/u. Evaluating: lim (t to infinity) of [-6/(t + 4) + 6/4] = 0 + 3/2.
Result: converges to 3/2.
Notice this is a p = 2 situation in disguise (after the substitution), so convergence is expected.
Example 2: Integral of 1/x^2 from 1 to infinity (Type 1, converges)
The antiderivative is -1/x. Evaluating: lim (t to infinity) of [-1/t + 1] = 0 + 1 = 1.
Result: converges to 1. This is the p-test with p = 2.
Example 3: Integral of 1/x from 1 to infinity (Type 1, diverges)
The antiderivative is ln|x|. Evaluating: lim (t to infinity) of [ln(t) - ln(1)] = infinity.
Result: diverges. This is the p-test with p = 1. Even though 1/x approaches 0, it does so too slowly for the area to be finite.
Example 4: Integral of x * e^(-x^2) from negative infinity to infinity (Type 1, converges)
Split at 0. Use substitution u = -x^2, du = -2x dx on each piece.
The antiderivative is -1/2 * e^(-x^2). The left piece (negative infinity to 0): lim (t to negative infinity) of [-1/2 * e^0 + 1/2 * e^(-t^2)] = -1/2 + 0. The right piece (0 to infinity): lim (t to infinity) of [-1/2 * e^(-t^2) + 1/2 * e^0] = 0 + 1/2.
Total: -1/2 + 1/2 = 0. Converges to 0.
This one converges because the exponential decay crushes the linear factor.
Example 5: Integral of sin(x) from 0 to infinity (Type 1, diverges)
The antiderivative is -cos(x). Evaluating: lim (t to infinity) of [-cos(t) + cos(0)] = lim [-cos(t) + 1].
The limit does not exist because -cos(t) oscillates forever between -1 and 1.
Result: diverges (the limit DNE). This is a divergence that is neither infinite nor negative infinite. The function never settles.
Students often assume that if the integrand approaches 0, the integral must converge. It does not. The integral of 1/x from 1 to infinity diverges even though 1/x tends to 0. The function must approach 0 fast enough.
Students sometimes evaluate the doubly-infinite integral (negative infinity to infinity) as a single limit, e.g. lim (t to infinity) of the integral from -t to t. This is the Cauchy principal value, which is a different (and weaker) concept. The proper definition requires two independent limits.
Students forget to check for discontinuities inside the interval. The integral from 0 to 3 of 1/(x - 1) dx looks like a straightforward definite integral but is actually improper (Type 2) because the integrand blows up at x = 1.
Some students think that divergence always means the integral "goes to infinity." It can also diverge by oscillation, as with the integral of sin(x) from 0 to infinity, where the limit simply does not exist.
⚠️ Classification is typically the first exam question: given an integral, identify whether it is proper, Type 1, Type 2, or both. Know how to spot discontinuities inside an interval, not just infinite bounds.
⚠️ The p-test for 1/x^p from 1 to infinity (converges if p > 1, diverges if p <= 1) is the most commonly tested benchmark. You will often need it to justify convergence arguments even when the problem uses a different function.
⚠️ Be ready to evaluate and state convergence or divergence with full working. Partial credit typically requires showing the limit setup, the antiderivative, and the limit evaluation.
⚠️ The sin(x) example (oscillating divergence) is a favourite exam question because it tests whether you understand that "diverges" does not only mean "goes to infinity."
Real-world applications
Improper integrals appear throughout probability and statistics: the entire theory of continuous probability distributions rests on integrals over infinite intervals (the normal distribution's bell curve is integrated from negative infinity to infinity). In physics, they arise in computing total energy radiated by a source, gravitational potential, and electric fields of infinite charge distributions.
True or false: If f(x) approaches 0 as x approaches infinity, then the integral of f(x) from 1 to infinity converges. (Answer: False. Counterexample: 1/x.)
Fill in the blank: The integral of 1/x^p from 1 to infinity converges when p is ____. (Answer: greater than 1.)
True or false: The integral from 0 to 1 of 1/sqrt(x) dx is a Type 1 improper integral. (Answer: False. It is Type 2, because the integrand blows up at x = 0.)
True or false: An integral can diverge even if its limit is not plus or minus infinity. (Answer: True. Example: the integral of sin(x) from 0 to infinity diverges by oscillation.)
Fill in the blank: To evaluate the integral from negative infinity to infinity, you must split it into ____ independent limits. (Answer: two.)
Q: Classify the following integral and determine whether it converges or diverges: the integral from 1 to infinity of 1/x^3 dx.
A: Type 1 (infinite upper bound). The antiderivative is -1/(2x^2). The limit as t goes to infinity of [-1/(2t^2) + 1/2] = 1/2. Converges to 1/2.
Q: Evaluate the integral from 0 to infinity of e^(-x) dx, or show that it diverges.
A: Type 1. The antiderivative is -e^(-x). The limit as t goes to infinity of [-e^(-t) + e^0] = 0 + 1 = 1. Converges to 1.
Q: Does the integral from 0 to 1 of 1/x dx converge or diverge? Justify your answer.
A: Type 2 (the integrand has a vertical asymptote at x = 0). The antiderivative is ln|x|. The limit as t goes to 0+ of [ln(1) - ln(t)] = 0 - (-infinity) = infinity. Diverges.
Q: Explain why the integral from negative infinity to infinity of sin(x) dx diverges.
A: The antiderivative is -cos(x). Neither lim (t to infinity) of -cos(t) nor lim (t to negative infinity) of -cos(t) exists, because cosine oscillates between -1 and 1 indefinitely. Since neither piece converges, the integral diverges.
Q: The integral from 1 to 3 of 1/(x - 2) dx appears to be a standard definite integral. Why is it improper, and what type is it?
A: The integrand 1/(x - 2) has a vertical asymptote at x = 2, which lies inside the interval [1, 3]. This makes it a Type 2 improper integral. You must split it at x = 2 and take one-sided limits from each side.
Improper integrals connect directly to sequences and series: the Integral Test for series convergence works by comparing a series to an improper integral of the same function. If you can show the integral of f(x) from 1 to infinity converges, then the sum of f(n) from n = 1 to infinity converges too.
They also connect to probability: every continuous probability density function must integrate to 1 over its entire support, which often means evaluating an improper integral over an infinite interval.
In later coursework (differential equations, Laplace transforms), improper integrals become a core computational tool. The Laplace transform is itself defined as an improper integral from 0 to infinity.
Improper integrals, Type 1 improper integral, Type 2 improper integral, infinite limits of integration, discontinuous integrand, convergent integral, divergent integral, p-test, p-series integral test, vertical asymptote in integral, Cauchy principal value, unbounded integrand, integral to infinity, comparison test for integrals, limit comparison test, M408D, Calc 2, integral calculus, University of Texas calculus