7.3: u-substitution
( \newcommand{\kernel}{\mathrm{null}\,}\)
Recall the chain rule:
ddxf(g)=f′(g)⋅g′
The method of “u-substitution” is a way of doing integral problems that undo the chain rule. It also helps deal with constants that crop up.
- Identify an “inside” function whose derivative is multiplied on the outside, possibly with a different constant. Call this “inside” function u.
- Compute dudx and solve for dx.
- Use substitution to replace x→u and dx→du, and cancel any remaining x terms if possible.
- Integrate with respect to u. If at this point you still have any xs in your problem, either you made a mistake or the method of u-substitution will not work for this problem.
- Substitute back the xs back into the answer before evaluating the definite integral.
Let’s do some examples.
We will follow the steps of u substitution.
- In this case, the “inside function” is u=−5x.
- If we compute dudx, we see the derivative of −5x is −5. Hence dudx=−5, and we have solving for dx
dudx=−5du=−5dx−15du=dx
- Going back to the original problem and using substitution u=−5x and dx=−15du, we thus have:
∫30e−5xdx∫30eudx=∫30eu(−15du)=∫30−15eudu.
- Integrating with respect to u,
∫30−15eudu=−15∫30eudu=−15eu|30
- We now substitute u=−5x. We have
−15eu|30=−15e−5x|30=(−15e−15)−(−15e−5(0))=−e−155+15=1−e−155≈0.2
There we go.
Again, we will go through the steps of u-substitution.
- The inside function in this case is x2+1. We can see that the derivative is 2x, and this is good since there is an x multiplied out in front (the 2 is just a constant we can deal with.) Set u=x2+1.
- We see dudx=2x, and hence solving for dx we have du2x=dx.
- Subbing in u=x2+1 and dx=du2x, we have
∫x(x2+1)7dx=∫xu7dx=∫xu7(du2x)=∫x2xu7du=∫12u7du
Great, the xs are all gone!
- We can now integrate with respect to u.
∫12u7du=12∫u7du=12u88=u816
- Finally, we sub in the xs again using u=x2+1.
u816=(x2+1)816+C
Since this is an indefinite integral, we add the constant of integration.
Again, we will go through the steps of u-substitution.
- The inside function in this case is ln(x). We can see that the derivative is 1x, and this is good since there is an x dividing the rest of the problem. Set u=ln(x).
- We see dudx=1x, and hence solving for dx we have du1x=dx.
- Subbing in u=ln(x) and dx=du1x, we have
∫8(ln(x))3xdx=∫8u3x⋅du1x=∫8u3x⋅1xdu=∫8u31du=∫8u3du
Great, the xs are all gone!
- We can now integrate with respect to u.
∫8u3du=8⋅14u4=2u4
- Finally, we sub in the xs again using u=ln(x).
∫8(ln(x))3xdx=2(ln(x))4+C.
If the inside function is linear, the u-substitution is much simpler, and there is even a formula for it (just like in the ∫e−5xdx example above). By the chain rule with g=mx+b and g′=m, we have
ddx1mf(mx+b)=1mddxf(mx+b)=1m(f′(g)⋅g′)=1mf′(mx+b)⋅m=mmf′(mx+b)=f′(mx+b)
If we integrate both sides of this equation, we have the following useful rule which I call the “chain rule shortcut”:
∫f′(mx+b)dx=1mf(mx+b)
This is especially important if f(x)=emx. In this case, we have
∫emxdx=1memx
In other words, you integrate just like normal without any u-substitution, and then add a 1m factor for the fact that you have mx+b inside the function instead of just an x.
Let’s see a couple of examples of this:
-
Find ∫(6x−3)4dx.
Using the inverse power rule, we see this becomes (6x−3)55. However, we need that 1m factor since the problem has a 6x−3 in it. So the final answer is
16(6x−3)55=(6x−3)530+C.
You’d get the same thing doing a full u-substitution with u=6x−3. This way, though, you save some time by just multiplying by 16.
-
Find ∫−1−2(3x+5)7dx.
Again, we use the u-sub shortcut — we just need to do power rule and remember a 1m factor, which in this case is 13.
∫−1−2(3x+5)7dx=13(3x+5)88|−1−2=(3x+5)824|−1−2=((3(−1)+5)824)−((3(−2)+5)824)=25624−124=25524=838
-
Find ∫e0.05tdt.
The antiderivative here is just 10.05e0.05t. We can plug 20, so the answer is 20e0.05t+C.