This "technique" is used when the integrand is the derivative of a function that you already know. Basically, if you remember the common derivatives, then you already know a lot of integrals.
Power Rule
The derivative of xn is nxn−1, so you immediately get
∫nxn−1dx=xn+C⟹∫xn−1dx=nxn+C.
(I absorbed the n into the C, since n is just a constant.)
Sometimes, you need to calculate a definite integral, but you "can't" integrate something (i.e., you can't express the antiderivative in terms of trig functions, exponentials, polynomials, etc.). In some cases, you can still write down the answer:
Example 1.
Calculate ∫−22sin(x3)dx.
Solution.
You won't be able to integrate f(x)=sin(x3) no matter how hard you try. However, this function has a special property-it's an odd function:
f(−x)=sin((−x)3)=sin(−x3)=−sin(x3)=−f(x).
Moreover, the definite integral we're trying to calculate has symmetric bounds, so the integral of f on [−2,0] will cancel out with the integral on [0,2]. In other words,
∫−22sin(x3)dx=0.
u-substitution
u-substitution is the "reverse chain rule." This means you'll want to try it when you see a function and its derivative (maybe times a constant).
Example 2.
Calculate ∫xsin(x2)dx.
Solution.
Here, you can see x2 and basically its derivative 2x, except we're missing a 2. This means we'll want to try u=x2, which gives du=2xdx, so the integral becomes
and from here, it's more clear that we see lnx and its derivative, so we can set u=lnx to get
∫xlnxdx=∫udu=ln∣u∣+C=ln∣lnx∣+C.
Here's a fun problem that's similar:
Exercise 1.
Calculate ∫x(lnx)ln(lnx)dx.
Integration by Parts
Similar to u-substitution, integration by parts is a "reverse product rule."
Theorem (integration by parts)
∫udv=uv−∫vdu
Equivalently, you can write this as
∫f(x)g′(x)dx=f(x)g(x)−∫f′(x)g(x)dx.
In other words, integration by parts lets you replace f(x) with its derivative f′(x), which can sometimes simplify the integral. For example, if f(x)=lnx, then f′(x)=x1, which is generally easier to work with when integrating.
There are two main scenarios where you want to use integration by parts:
When you want to integrate a product of two "random" functions.
When you want to integrate a "weird" function by itself.
Example 4.
Calculate ∫xcosxdx.
Solution.
Here, we're in scenario 1 above. For this problem, you will want to use the parts u=x and dv=cosxdv, which give
This time, we're in scenario 2. When this happens, we can just let u=arcsinx and dv=dx, so
du=1−x2dxandv=x.
Plugging everything in,
∫arcsinxdx=∫udv=uv−∫vdu=xarcsinx−∫1−x2xdx.
So, we just need to calculate this new integral, which can be done using the substitution w=1−x2. This gives us dw=−2xdx, so
∫1−x2xdx=−21∫wdw=−w+C=−1−x2+C.
Thus,
∫arcsinxdx=xarcsinx+1−x2+C.
Trig Substitution
You usually want to try this technique when you see something similar to 1+x2 or 1−x2 and a u-substitution doesn't work. When this happens, you'll want to choose your substitution as follows:
1−x21+x2⇝x=sinθ⇝x=tanθ
Example 6.
Calculate ∫1+x2dx.
Solution.
Any u-substitution you try won't work here since there's no x in the numerator, so we'll have to try a trig substitution. We're in the 1+x2 case, so we can set x=tanθ, which means dx=sec2θdθ. Then
To finish the problem, we need to undo the substitution. We already know that tanθ=x (this was the substitution we started with), and for secθ, we can use our trig identities:
sec2θ=1+tan2θ=1+x2⟹secθ=1+x2.
Substituting everything back in, we get
∫1+x2dx=ln∣∣1+x2+x∣∣+C.
Example 7.
Calculate ∫1−4x2dx.
Solution.
If you don't remember your trig identities (like the double-angle identities), then it may be helpful to take a look at this review sheet I made for my 31B students last quarter (you can ignore everything starting at Series).
For this problem, we're essentially in the 1−x2 case, though it's not exactly in that form yet. We can write
1−4x2=1−(2x)2,
so replacing x with 2x, we get the substitution 2x=sinθ. Then 2dx=cosθdθ, and we get