Integration by parts
Some integrals refuse every trick you already know. is not a simple substitution, and does not even look like a product. Integration by parts is the technique built exactly for this gap: it turns the integral of a product of two functions into a smaller, easier integral, by running the product rule for differentiation backwards. It is the third core weapon in the CBSE Class 12 Integrals chapter, after substitution and partial fractions, and it is the one examiners lean on hardest — because it rewards students who can choose wisely (which factor to differentiate, which to integrate) rather than just apply a formula mechanically. This article rebuilds the formula from the product rule, gives the ILATE rule for choosing factors, works a single-pass and a repeated example, and flags the traps that cost the most marks.
Integration by parts converts into by reversing the product rule, and the ILATE priority order tells you which factor to call .
The core idea
From the product rule, backwards. For differentiable and , the product rule states . Integrate both sides with respect to and rearrange, and the differentiation formula flips into an integration formula:
where: is the factor you choose to differentiate (it should get simpler under ), is the factor you choose to integrate (it should have an easy antiderivative ), and is the derivative of . In the notation NCERT prefers, calling the two factors (first function) and (second function, easy to integrate), the same formula reads:
Both versions say the same thing: pick a , pick a , and the product's integral becomes (something already known) minus (a hopefully-simpler integral).
Choosing and : the ILATE rule. The single biggest source of marks lost on this topic is picking and the wrong way round, which turns a shrinking problem into a growing one. The standard heuristic ranks function types by priority for being chosen as :
Whichever type sits earlier in this list becomes (the one you differentiate); the other becomes (the one you integrate). The logic behind the order is that differentiating an inverse-trig or log function tends to simplify it dramatically (often down to an algebraic fraction), while differentiating a polynomial repeatedly reduces its degree toward zero — both moves shrink the remaining integral. Differentiating an exponential or a plain trig function, by contrast, just reproduces a similar function, so those are better left to be integrated.
A useful shortcut. When the integrand has the special shape , the answer collapses without any real work:
This is a one-line consequence of applying integration by parts to and watching the leftover term cancel against the extra term already present.
Products that never simplify away. For integrands like , neither factor is reduced to nothing by repeated differentiation, so the formula must be applied twice. On the second pass, the original integral reappears on the right side, and it can then be solved for algebraically, as in the worked example below.
Functions in disguise. and do not look like products at all, but they are — with the second factor being the constant function . Writing them as or makes (or ) and explicit, and this trick is the only way logarithmic and inverse-trig functions get integrated by this method, since they have no natural "easy to integrate" partner.
The goal of this derivation is to show, in three short algebraic steps and without any new assumption beyond the product rule for differentiation, that — the formula that lets a hard product integral be traded for a hopefully-easier one, and that underlies every worked example in this article.
Step 1. Start from the product rule for two differentiable functions and : . This is a statement about derivatives, true for every in the common domain of and .
Step 2. Integrate both sides with respect to . The left side, being the derivative of , integrates back to itself (plus a constant, absorbed later): . This step is the entire trick — integration simply undoes the differentiation the product rule performed.
Step 3. Rearrange for the integral we actually want, : subtract from both sides to get . Here is the factor we chose to differentiate and the factor we chose to integrate; the formula trades the original product integral for the boundary term minus a new integral , which is smaller whenever is simpler than (for instance, a polynomial's derivative has lower degree). Verified: with , , differentiating the claimed antiderivative gives back , matching the original integrand — confirming the formula's direction and sign.
Worked example
Evaluate , then extend to the repeated case .
Step 1. By ILATE, "Algebraic" () outranks "Exponential" (), so take and , giving .
Step 2. Substitute into : . Checking by differentiation, , matching the integrand. So .
Step 3. Now the harder case, , where neither factor disappears. Take , : . Apply the formula again to , keeping the same rule (differentiate the trig factor each time): with , , .
Step 4. Substitute back: , so . Final answer: .
Common traps
- Picking and against ILATE — this can make harder than the original integral instead of simpler.
- Adding the constant of integration after each partial step instead of only once, at the very end.
- Switching which factor is midway through a repeated application (example: differentiating on the second pass instead of the trig factor) — the original integral then fails to reappear correctly.
- Ignoring the shortcut and grinding through the full formula when it is not needed.
- Not noticing that a "single" function like or is secretly a product with , and trying to look it up as a standard integral instead.
Ask the tutor: unsure how to pick and for a tricky product, or want a walkthrough of a "twice around" integration by parts with a different function pair? Ask the AI tutor for guided practice.
Why does ILATE put logarithmic functions ahead of algebraic ones?
Because differentiating produces the simple fraction , which is a large simplification, while integrating directly is not something you can do by inspection — you would need integration by parts to do it anyway. Algebraic functions like simplify under differentiation too (their degree drops) but are also trivial to integrate. ILATE therefore favours turning a hard-to-integrate function like into a manageable derivative , so it is always chosen as over an algebraic partner.
What if I choose u and v the wrong way round by mistake?
Nothing breaks the mathematics, but the resulting integral typically gets more complicated instead of simpler — for example, choosing and in leaves you with , a worse integral than you started with. If your new integral looks harder than the original, that is the signal to go back and swap the choice of and .
How do I know when to stop applying integration by parts and switch to solving algebraically?
Apply it once, and check whether the new integral on the right is a standard one you can finish directly (as with ). If instead the new integral is a scaled copy of the original — same function types, same structure, just possibly a sign flipped — as happens with or , stop applying the formula and instead treat the original integral as an unknown, collect all copies of on one side of the equation, and solve for it algebraically.