Integration Techniques
When an integral isn't in the table, reshape it until it is.
Before this Lesson
Differentiating is a machine: feed it any formula, turn the crank on the rules, and the answer comes out. Integrating has no crank. Here are three integrals built from the same two parts:
The first takes two lines. The second takes a different trick altogether. The third has no formula you can write with ordinary functions, though your calculator will still give you a number for it. They look like near-twins, and telling them apart is most of the job.
Reading the table
Every integral formula is a derivative read backward, and the Handbook's Indefinite Integrals list collects the useful ones. Four of its lines cover the exponential, trig, and integrals the exam asks about. Spot the standard form, then copy the line with the right constants.
| Integrand | Integral (add C) |
|---|---|
Adapted from the FE Reference Handbook, Indefinite Integrals
The . Differentiating brings a 3 down by the chain rule, so integrating has to divide one back out. There's a picture behind this, too: is the sine wave squeezed to half its width. Its arch is just as tall and half as wide, so it holds half the area.
The Handbook writes its sine and cosine lines for plain , so the is yours to supply, as in .
The minus sign on . The sine arch sits above the axis, so its area has to come out positive, and it does: . Lose the minus and you'd get .
A radian is arc length over radius. A full turn is radians, so 180° is radians and one radian is about 57.3°. For more, see Areas of Plane Figures.
Radians. The trig lines reverse derivatives that hold only in radians, so a limit of 1 means 1 radian. This is the angle-mode habit: keep the calculator in radians for calculus and in degrees for geometry. In degree mode, reads 0.0175 instead of 0.841, with no warning.
The line. In Integrals and Area, was the one power the rule couldn't touch, since it would divide by 0. But you've since met the function whose slope is , the natural log. With and , the table's last line gives:
The bars are there because exists for negative too.
Substitution: the chain rule run backward
Most integrands don't arrive looking like a table line, so you reshape them. Take a derivative you know, and read it backward:
That is the fingerprint the chain rule leaves behind. When an integrand holds an inner function and that function's derivative as a factor, it's a chain rule result, and you can undo it. The Handbook's Integral Calculus section names this method, Integration by Substitution, and prints no procedure for it, but the bookkeeping is just a change of name. Call the inside . Then , and the integrand collapses into a table line:
Now the Hook's first integral, . It's short a factor of 2, but a constant can slide through the integral sign, so and the answer is . The third integral, , is another story. It's short a factor of , and isn't a constant, so it can't slide anywhere. A missing constant can be fixed. A missing variable can't.
The limits move too
With limits, runs along its own axis, so the limits have to move there with it. In , the upper limit becomes . Every thin slice on the graph has a partner on the graph, with its height divided by and its width multiplied by . The slices hold the same area, so the whole regions do too.
So the integral is , and you never need to go back to . Or go back, putting in for , and use the original limits. Either route works. A mix, with the formula in and the limits in , gives a wrong answer that's probably among the options.
Integration by parts: the product rule run backward
Now the Hook's second integral, . Substitution has nothing to grab here, because there's no inner function, only two unrelated factors. Products come from the product rule, so run that one backward.
In Differentiation Rules, a growing rectangle with width and height gained two strips with every nudge, and . Add up all the strips as it grows, and they fill the final rectangle: . Move one integral across, and you have the Handbook's line for integration by parts:
Here's that trade for , out to . Take and , which makes . The curve runs corner to corner and divides the rectangle in two.
Below the curve is , which you can do. Above it is , which you want, and it's whatever the rectangle has left: . You never integrate the hard thing. You swap it for an easier one, and all the skill is in choosing . Here's the same integral without limits:
It works because turned into 1 and got no worse. Choose the other way, and , and the formula is just as true but hands you , a higher power than you started with. So make the factor that gets simpler when you differentiate it, and a factor you can integrate. If the new integral looks worse than the old one, swap them.
That rule also cracks , which seems to have only one factor. You can't integrate , since that's the question, but its derivative is a simple . So take , and is what's left, plain :
The Handbook prints the finished answers for , , , and , so when an integrand matches one exactly, read the line. Parts is for when the form shifts, as in or .
Which tool, and when to skip them all
The exam won't name the technique, so ask, in this order:
- Is it a table line, maybe with a constant? Read it off.
- Is there an inner function with its derivative beside it, give or take a constant? Substitute.
- Is it a product of unrelated factors, like or ? Use parts.
For a definite integral, there's a shortcut past all three: your calculator's numerical integration. It adds up thin slices and never needs a formula, so it handles even the Hook's third integral: in radian mode. On the exam, that's often the fastest route to a number, and a good check on one you worked by hand.
It can't hand you a formula, though. To choose an indefinite integral, differentiate the options, and take the one that returns the integrand.