Recipe
Integrate by u-substitution
Use when: the integrand looks like f(g(x))\cdot g'(x) -- an outer function wrapped around an inner one, times something that could be that inner function's derivative
Goal: an antiderivative in terms of x, found by temporarily rewriting the whole integral in terms of u
Taught in Choosing u: recognizing a substitution · a later section extends this recipe with more cases
Quick recipe
-
1
Let u be the inner function -- usually whatever's inside a radical, a trig function, an exponent, or raised to a power.
Example
For \int 2x\cos(x^2)\,dx: let u=x^2.
Why: the inner function is exactly the piece whose derivative the chain rule would have produced if this integral came from differentiating something -- u-substitution is that process in reverse.
-
2
Differentiate u with respect to x, then treat du and dx as if they could be separated.
Example
u=x^2 gives du=2x\,dx.
Why: du is what tells you whether the "leftover" part of the integrand actually matches (up to a constant) what's left in the integral once u is substituted in.
-
3
Confirm the rest of the integrand is a constant multiple of du.
- it matches exactly (or by a constant factor) — continue to Step 4.
- it doesn't match at all — this choice of u doesn't work; try a different inner function.
Why: if the leftover doesn't correspond to du, the integral can't be fully rewritten in terms of u alone, and the substitution stalls halfway through.
-
4
Replace every occurrence of x (including inside dx) so the integral contains only u and du.
Example
\int 2x\cos(x^2)\,dx=\int\cos(u)\,du.
Why: this is the actual payoff of Steps 1–3 — a integral that may have looked unfamiliar in x often becomes one of the basic antiderivative forms once written purely in u.
-
5
Find the antiderivative in terms of u, then replace u with its original expression in x.
Example
\int\cos(u)\,du=\sin(u)+C=\sin(x^2)+C.
Why: the original problem asked for an antiderivative in terms of x. u was only ever a bookkeeping device to get there.
Common mistakes
- Forgetting to substitute back to x at the very end, leaving u in the final answer.
- Choosing a u whose leftover factor doesn't actually match du, and forcing the substitution anyway.
- Dropping a constant factor between the leftover and du instead of carrying it through.
- Substituting u into part of the integrand but leaving dx unconverted.