Improper Integrals
You can't integrate all the way to infinity, or right up to a spot where the function blows up, in one go. So integrate to a finite stand-in and let it slide toward the trouble spot. If the area settles on a number, the integral converges; if not, it diverges.
What the symbols mean
- A moving endpoint
A finite stand-in for ∞: integrate up to t, then let t grow without bound.
What the symbols mean
- The power
How fast the integrand dies off. Faster than 1/x (p > 1) leaves a finite area.
Explanation
An improper integral is simply an integral over a region that goes on forever, either because an endpoint is or because the function shoots up to infinity somewhere. "Improper" doesn't mean wrong; it just means the integral breaks the usual rules (a finite interval and a function that stays finite), so it needs an extra step. It's good for questions like the total work to escape Earth's gravity or a drug's total exposure over all time. All we're really asking with is: take (an ordinary area, stopping at some finite point ), push farther and farther out (), and see whether the area levels off at a number (that's the answer) or keeps growing (no answer). It's an idea you already know from limits, just applied to area; bear with it.
The Fundamental Theorem of Calculus needs a finite interval and a function that doesn't blow up, so neither nor an integrand with a vertical asymptote fits. But you already know how calculus handles "you can't plug that in": take a limit. Integrate up to a point where everything is fine, then let that point slide toward the trouble.
Type I: an infinite limit. Integrate to a finite , then take the limit:
Think about how strange that is. The region under is infinitely long, yet its area is exactly 1. Now compare :
so diverges even though . Getting smaller isn't enough; the function has to shrink fast enough that the extra slivers of area stop adding up. There's an animation further down that shows the two areas side by side as grows.
If both limits are infinite, split at any convenient point: . The whole thing converges only if both pieces do.
Type II: a vertical asymptote. If blows up at the endpoint , approach it from the inside:
If blows up somewhere inside , split there.
A common mistake. For , it's tempting to just plug in the limits and get , because it looks like an ordinary definite integral. It isn't. There's an asymptote at , right in the middle, and the Fundamental Theorem doesn't apply across it. You can tell something's off anyway: the integrand is positive, so a negative answer is impossible. Split at and each half diverges, so the integral diverges.
See it
Worked examples
Evaluate .
- 1
The integrand blows up at , so you can't plug that in. Start at a small instead and let it slide toward from the right: .
Evaluate .
- 1
First find the antiderivative, by parts: .
Practice
Type your answer and check it. Solve them all to complete the lesson.
Evaluate .
Evaluate .
Evaluate .
Evaluate .
Evaluate .
Does converge?
How much work does it take to lift a kg probe from Earth's surface ( m) completely free of Earth's gravity? The force at distance from Earth's center is , with m³/s².
The hardest one here. Get it, and you've got this lesson.
Evaluate , which is improper at both ends.
Where it's used
Escape velocity comes from an improper integral: the work to lift a rocket from Earth's surface to infinity, , is finite, so a fast enough rocket never falls back. In medicine, a drug's total exposure is the area under its concentration curve from 0 to ∞.