Integration by parts
Some integrals are products that no single standard formula can touch — , , even a lone . Integration by parts is the tool built exactly for this: it turns the product rule for differentiation inside out, trading one integral for a hopefully simpler one. It sits inside the Integrals unit of ISC Class 12 Mathematics (Calculus, Section A, 32 marks), right after substitution, and it is used constantly afterwards — inside partial fractions, definite integrals, and reduction formulae. The skill has three parts: picking which factor is and which is using the ILATE priority order, applying the formula (sometimes twice), and recognising the special "multiply by 1" trick for lone transcendental functions like or . Master the choice of and the rest becomes routine algebra.
Integration by parts reverses the product rule to convert into , with chosen by the ILATE priority order so that differentiating it simplifies the remaining integral.
The core idea
The formula. If and are differentiable functions of , running the product rule for differentiation backwards and integrating gives
where: is the factor chosen to be differentiated, is the factor chosen to be integrated first, is the derivative of , and is an antiderivative of . In the compact notation you will use most often, writing , this reads .
Choosing : the ILATE rule. With two factors in front of you, decide which is using the priority order Inverse trig, Logarithmic, Algebraic, Trigonometric, Exponential. Whichever type sits earlier in ILATE becomes , because it usually gets simpler on differentiation, while the later type integrates easily and becomes . In , Algebraic beats Exponential, so ; differentiating once kills it, while integrates to itself.
The "multiply by 1" trick. When only one function is visible — or — write it as (function). Take as the function and , so . For : , , giving , , and
The same move handles and : put the inverse-trig function as against , then simplify what remains, often with a quick substitution.
A shortcut worth memorising. When the integrand has the exact shape , skip integration by parts entirely:
where: is any differentiable function and its derivative — recognising this pattern in the integrand saves a full application of the by-parts formula.
The goal is to show, starting only from the product rule for differentiation that a student already knows, exactly why the integration-by-parts formula has a minus sign in front of its second term and why that second term involves the integral of rather than itself — seeing this removes any need to memorise the formula by rote.
Step 1. Start from the product rule. For differentiable functions and , . This is a statement about derivatives; the plan is to integrate both sides with respect to .
Step 2. Integrate both sides. . The left side is simply (integrating a derivative undoes the differentiation), so .
Step 3. Rearrange for the term you actually want. Solving for gives , which is exactly the integration-by-parts formula written with known explicitly (rather than as ).
Step 4. Read off the roles. The minus sign appears because it came from rearranging an equality, not from any arbitrary convention. And the surviving integral is , not again — this is precisely why choosing so that simplifies is the whole strategy: a simpler makes the remaining integral easier than the one you started with.
Worked example
Evaluate .
Step 1. Choose and . By ILATE, Algebraic () outranks Exponential (), so take and .
Step 2. Differentiate and integrate : , and .
Step 3. Substitute into the formula: .
Step 4. Check by differentiating the answer: , which matches the original integrand. Answer: .
Competitive stretch
Extend the same idea to , which needs parts twice. Take , , so , , giving . Applying parts again to (with , ) gives . Combining: . When repeated parts instead brings back a multiple of the original integral — as in — treat the result as an algebraic equation in the unknown integral and solve for it, rather than applying parts forever.
Common traps
- Picking and the wrong way round. Taking in still "works" formally but the resulting integral never simplifies — always check ILATE priority first.
- Dropping the second integral. Students often compute and stop, forgetting the term entirely.
- Sign slips when or involves sine or cosine — track each term's sign carefully rather than trusting memory.
- Giving up after one application. If the new integral is still a product (as in ), it usually needs a second pass of the same rule, not a different method.
- Forgetting to simplify before integrating again. After the first application, the leftover integral is often itself a product; tidy the coefficients first so the second pass of the formula is applied to the cleanest possible expression.
Ask the tutor: Stuck on which factor to pick as for a tricky product, or want another worked example step by step? Ask the AI tutor to walk through it with you.
How do I remember which function is u without second-guessing myself
Use ILATE as a checklist, not a guess: Inverse trig, Logarithmic, Algebraic, Trigonometric, Exponential, in that priority order. Scan the two factors in the integral, find which type appears earlier in that list, and make that one . The reasoning behind it is that the earlier types get algebraically simpler when you differentiate them, while the later types (like or ) integrate cleanly no matter how many times you integrate them, so putting them in the slot never causes trouble.
Why does integrating log x need the multiply-by-1 trick instead of a direct formula
There is no elementary function whose derivative is , so you cannot integrate it directly the way you integrate or . Writing turns it into a product, which integration by parts can handle: take (it differentiates to the simple ) and (which integrates to ). The same trick rescues and , both of which also have no direct antiderivative formula but do have simple derivatives.
I applied integration by parts and the new integral looks just as hard as the original — did I choose wrong
Not necessarily. Some products, like , genuinely need two applications before the integral becomes elementary — check whether the new integral is at least simpler than the last one (lower power of , for instance), and if so keep going. If instead the new integral is identical in difficulty or brings back a multiple of your very first integral, that is the signal for the "solve for " algebraic trick rather than a sign you chose wrong. It is also worth remembering that any valid antiderivative works for the intermediate — the arbitrary constant you could add to it cancels out, so add the single overall constant of integration only at the very end.