Integration by Parts: What People Actually Mean by the UV Rule
When someone says "the UV rule" in a calculus class, they're almost always talking about integration by parts. It's the integral equivalent of the product rule for derivatives, and it comes straight from reversing the product rule. The formula is u dv = uv - v du. That's it. Not magic, not a shortcut, just algebra applied in reverse. The reason this shows up in every calc II course is because a lot of integrals refuse to work with basic substitution. You hit something like x · e^x dx or ln(x) dx and you realize plain u-substitution isn't going to untangle it. That's when integration by parts enters the picture. I've seen students treat it like a last resort, but honestly it's often the fastest path if you set it up correctly the first time.
The Uv Rule In Calculus Unlocking The Secrets Of Limits
Here's the part most textbooks gloss over: integration by parts is fundamentally a limit-based operation. The uv term you subtract at the end comes from evaluating [uv] at your bounds, which is itself a limit process. When you're working with improper integrals — say from 0 to of x·e^(-x) dx — the evaluation step becomes where things actually break. I spent an entire semester debugging student work where they'd apply the formula perfectly but then dropped the limit notation on the boundary terms. The answer was wrong by a factor of two every single time because they evaluated uv at infinity without actually taking the limit. The fix is mechanical but non-negotiable. Write out the full antiderivative with a limit variable first. For the improper integral above, you'd set it as lim(b) [ -x·e^(-x) - e^(-x) ] evaluated from 0 to b. Then plug in b and take the limit separately. The term -b·e^(-b) goes to zero, which isn't obvious until you actually check it with L'Hopital's rule. Skip that step and you're just guessing. For definite integrals, the formula becomes from a to b of u·dv/dx dx = [uv] from a to b minus from a to b of v·du/dx dx. The bracket notation [uv] from a to b means you evaluate uv at both endpoints and subtract. This is where most mistakes happen — people forget the bounds apply to the whole thing, not just the first term.
The LIATE rule is the standard heuristic for picking u and dv: Logarithmic, Inverse trig, Algebraic, Trigonometric, Exponential. Whatever comes first in that list becomes your u. It works about 80% of the time. The other 20% is when you have something like arctan(x) dx, where u = arctan(x) and dv = dx, which sounds backwards but is actually the right move. Most students try to split it the other way and end up making the integral harder. I ran into a genuinely tricky case recently involving sec³(x) dx. It's a classic recursive integral where applying integration by parts once gives you back the original integral on the other side of the equation. The trick is treating it as an algebraic equation: solve for the integral rather than trying to push forward. A lot of people get confused here and think they've made a mistake when the integral reappears. It's not an error, it's the intended path. There's also the tabular method for repeated applications, which is worth knowing even if your professor doesn't cover it. When you have something like x³·sin(x) dx, you set up two columns and differentiate u repeatedly until you hit zero, then integrate dv repeatedly. You draw diagonal arrows and alternate signs. It cuts a problem that would take three separate applications of the formula down to about thirty seconds of writing. I use it whenever the polynomial part is degree three or higher — below that the regular method is faster because setting up the table takes more time than just doing it by hand.
Get the Full Details
One counter-intuitive thing about integration by parts: sometimes you need to apply it twice to the same integral, not to different pieces. Take e^x·sin(x) dx. Apply it once and you get another integral involving e^x·cos(x). Apply it again and you're back to e^x·sin(x). At that point you move that term to the left side, combine like terms, and divide by two. Students regularly stop at the second application and write down an incomplete answer because they don't recognize the circular pattern. The real bottleneck with this method isn't the formula — it's knowing when not to use it. If an integral can be solved with substitution, substitution is almost always faster and less error-prone. Integration by parts introduces an extra term and an extra integral, which means more chances for sign errors and coefficient mistakes. A good rule of thumb: if the integrand is a simple product of two functions where one clearly differentiates to something simpler, try integration by parts. If it's a composition or a rational function, look elsewhere first. Another edge case that trips people up involves integrals with no obvious product structure, like ln(x) dx. There's no visible product here, but you can treat ln(x) as ln(x)·1 and apply the rule with u = ln(x) and dv = dx. It feels forced at first but it's the standard approach and it works cleanly. The same logic applies to arctan(x) dx and similar inverse function integrals.
If you're working with definite integrals that have discontinuities inside the interval, integration by parts still works but you need to split the integral at each discontinuity first. I once had a problem with from -1 to 1 of x/(1-x²) dx where a student applied the formula across the singularity at x = 0 and got a nonsensical answer. The integral needed to be split into two separate improper integrals before any formula could be meaningfully applied. The method itself wasn't wrong, the setup was. For the recursive cases and the tabular shortcuts, practice is the only real path to fluency. The formula itself is easy to memorize. Recognizing which integrals benefit from it, spotting when you've hit a recursive loop, and catching sign errors before they propagate — that comes from doing enough problems that the patterns become automatic. I'd recommend working through at least fifteen to twenty problems of varying difficulty before you feel confident with this. Most of the confusion in exams comes from rushing through the setup rather than misapplying the formula.
