Techniques of integration · Choose and check integration by parts
Choose parts when the remaining integral improves
Inspect the inner derivative before treating a product by parts. Choose what to differentiate and integrate, then check whether the new integral is simpler or returns to one you can solve for.
Check for a reverse-chain match first
A product alone does not decide the method. Simplify and compare a composite function with its inner derivative; a constant mismatch can be compensated.
- Reverse chainA matching derivative can avoid parts
For , the factor x matches the exponent’s derivative up to a constant. Substitution gives . Do not multiply by a missing function of x as if it were a constant adjustment.
- Parts candidateA leftover factor may suit differentiation
For , integrating the exponential alone leaves the factor x. Parts with and reduces the polynomial degree. Other expressions may need algebra, identities or partial fractions before either method helps.
These are useful structural clues, not an exhaustive or exclusive decision tree. Some integrals admit several valid methods, and a supplied substitution or preceding result can guide the intended route.
Choose u and dv by the work they leave
Choose u to differentiate and dv to integrate. Make sure v can be found and inspect the remaining integral before committing to repeated steps.
- Useful choiceAim for simpler remaining work
For , choosing , leaves . Reversing the roles leaves , which is harder. A valid use of the formula can still be an inefficient choice.
- LIATE guideUse the mnemonic with judgement
Logarithmic, inverse trigonometric, algebraic, trigonometric and exponential functions give a useful preference for u. There are exceptions. Compare the resulting integral rather than treating the order as a compulsory rule.
A standalone function can be written as a product with 1. For , choose and on ; integration is not the same operation as differentiation.
Repeat purposefully and check the result
Keep the minus sign and coefficients. Repeated parts may lower a polynomial degree or bring back the original integral, which can then be collected algebraically.
- Repeated partsTrack the integral you started with
If it returns, name it and move its multiple to the same side before solving. Combine arbitrary constants consistently. Repeating the same unsuccessful choice without inspecting the new integral does not establish progress.
- VerificationRecover the complete original integrand
Differentiate the final expression using product and chain rules. Carry the original domain restrictions, and include an arbitrary constant for an indefinite integral. A product is not integrated by multiplying its separate antiderivatives.
The worked cases below concern indefinite integrals. Definite integrals also require consistent evaluation of the boundary term and the remaining integral on their stated interval.
Try a worked example
Read a case, make a prediction, then reveal the reasoning. The examples keep their stated values.
Example 17(c): make the factor 1 visible
For on , decide which factor to differentiate after writing the integrand as .
Show reasoning
1. Choose factors that simplify
Differentiating ln x gives 1/x, while integrating 1 gives x. Those factors cancel in the remaining integral.
2. Apply parts and retain the domain
The remaining integrand is 1. The logarithm and the original integral are defined here on x>0.
3. Differentiate instead of guessing
The product rule recovers ln x. The expression 1/x is its derivative, not its antiderivative.
Example 17(e): solve the returning integral
For , choose and . Predict what will happen after applying parts a second time.
Show reasoning
1. Name the remaining integral
The exponential integrates to . Let J denote an antiderivative of ; retain the factor 1/2 and an arbitrary constant.
2. Apply parts again and keep the sign
Choose for J. Its derivative is , so subtracting the remaining integral produces a positive multiple of I.
3. Collect the original integral
Substitute for J, move −I/4 to the left and combine the arbitrary constants. The original integral has returned; it is now an algebraic equation rather than another instruction to repeat parts.
4. Check both product-rule terms
Let F be the final antiderivative without C. Its cosine terms cancel and the sine coefficients add to 1, recovering the original integrand for every real x.