The chain rule differentiates composite functions: the derivative of sin(x2) is cos(x2)⋅2x — differentiate the outer function, then multiply by the derivative of the inside. Integration asks the reverse: given cos(x2)⋅2x, find the original function. That's what u substitution does — it reverses the chain rule.
The Core Intuition
When an integral looks like "a function times the derivative of its inside," substitute the inside with u and the derivative of the inside with du. Consider:
∫2xcos(x2)dx
Here 2x is the derivative of x2, and x2 is the inside of cos(x2). Let u=x2, so du=2xdx:
∫cos(u)du=sin(u)+C=sin(x2)+C
Check: the derivative of sin(x2) is cos(x2)⋅2x.
The Precise Statement
∫f(g(x))⋅g′(x)dx=∫f(u)duwhere u=g(x),du=g′(x)dx
Valid provided g is differentiable and the resulting integral in u is simpler.
The Step-by-Step Method
Identify a function g(x) whose derivative g′(x) also appears (possibly up to a constant factor).
Setu=g(x), compute du=g′(x)dx.
Rewrite the entire integral in u and du — every x and dx must be replaced.
Integrate with respect to u.
Substitute backu=g(x).
Watch out
You cannot mix variables. If any x remains after substitution, you chose the wrong u (or must solve for x in terms of u — rare).
A Second Example (with a constant factor)
Evaluate ∫xx2+1dx. Let u=x2+1, so xdx=21du:
∫u⋅21du=21⋅32u3/2+C=31(x2+1)3/2+C
When Does It Work?
When the integrand is something times the derivative of something inside. Common patterns:
The key idea is to notice that the numerator 4x+1 is exactly the derivative of the denominator’s radicand 2x2+x−3. This makes the integral a direct application of ∫udu=2u+C. The final answer is 22x2+x−3+C.
When you see an integral like this, the first instinct should be to check if the numerator is the derivative of the expression inside the square root. Why? Because the derivative of u is 2u1⋅u′, so if the numerator matches u′, the integral collapses into something simple.
Here, the denominator is 2x2+x−3. Let’s call u=2x2+x−3. Then du=(4x+1)dx. That’s exactly the numerator! So the integral becomes ∫udu, which is a standard power rule.
Mistake 1: Reaching for completing the square out of habit.
Why it's wrong: here the numerator is exactly the derivative of the radicand, so the problem is a one-line u-substitution — completing the square wastes time and invites errors. Correct approach: always check dxd(radicand) against the numerator first.
Mistake 2: Forgetting the factor of 2.
Why it's wrong: ∫u−1/2du=2u1/2, not u1/2. Correct approach: apply the power rule ∫undu=n+1un+1 with n=−21. …