The engine under everything is u-substitution. Substitution needs the integrand to look like
∫f(u)dxdudx.
So we want a chunk that is "the inside" (u) and another chunk that is "its derivative" (du).
For trig, the two derivative facts that drive everything:
And the two Pythagorean identities that convert the leftover:
WHY parity? An odd power means you can split off one factor and still leave an even power, and even powers are exactly what the identities sin2/cos2 can rewrite (they only know squares).
WHY these splits? Because du(tan)=sec2xdx needs a spare sec2, and du(sec)=secxtanxdx needs a spare sectan. The identities supply only eventan or evensec replacements, so we save the factor that leaves even leftovers.
u=sinx; save one cosx for du, convert rest with cos2=1−sin2.
∫sinmcosn, both powers even — method?
Power reduction: sin2=21−cos2x,cos2=21+cos2x.
∫tanmsecn, secant power even — which u?
u=tanx; save sec2x, convert via sec2=1+tan2.
∫tanmsecn, tangent power odd — which u?
u=secx; save secxtanx, convert via tan2=sec2−1.
∫tanxdx=?
ln∣secx∣+C (i.e. −ln∣cosx∣+C).
∫secxdx=?
ln∣secx+tanx∣+C (multiply by secx+tanxsecx+tanx).
∫sin2xdx=?
2x−4sin2x+C.
∫tan2xdx=?
tanx−x+C using tan2=sec2−1.
Which identity rebuilds the leftover in tan/sec when saving secxtanx?
tan2x=sec2x−1.
Recall Feynman: explain to a 12-year-old
Imagine the integral is a Lego tower of sin and cos bricks. To take it apart neatly, you pull off one special brick that matches the "machine" (du). If you have an odd number of cosine bricks, pull one cosine off — now an even pile is left, and we have a magic rule (cos2=1−sin2) that turns even cosine piles into sine piles. Then we just rename sinx as u and it becomes baby algebra (u2−u4 etc.). Same trick for tan/sec: pull off the brick that equals the derivative, convert the rest. If everything is even and nothing peels off cleanly, we use the "cut the power in half" rule (power reduction) instead.
Dekho, trig integrals ka pura khel ek hi cheez par tika hai: u-substitution. Substitution ke liye humein integrand mein ek "andar wala part" (u) aur uska "derivative" (du) chahiye. Toh hum sirf yeh decide karte hain ki kaunsa factor peel karke du banega, aur baaki ko kaunsi Pythagorean identity se convert karenge.
sinmcosn ke liye rule simple hai: jiski power odd hai, usme se ek factor nikaal lo (du banane ke liye), aur jo bacha woh even power identity (sin2=1−cos2 ya cos2=1−sin2) se convert ho jaata hai. Agar dono powers even hain, toh koi factor saaf du nahi banta — tab power reduction (sin2=21−cos2x wagairah) use karo. Yaad rakhne ka mantra: "Odd one out goes du."
tanmsecn mein bhi wahi logic, bas anchors badle: dxdtanx=sec2x aur dxdsecx=secxtanx. Toh agar sec ki power even hai, ek sec2x bachao aur u=tanx rakho. Agar tan ki power odd hai, ek secxtanx bachao aur u=secx rakho; baaki tan2=sec2−1 se convert. Aur do classic results ratne ki zaroorat nahi — derive kar lo: ∫tanxdx=ln∣secx∣, aur ∫secxdx=ln∣secx+tanx∣ (numerator-denominator trick se).
Sabse common galti: ∫tan2xdx ko 3tan3x likh dena — galat! Power rule yahan kaam nahi karta kyunki dx=d(tanx). Sahi: tan2=sec2−1, toh answer tanx−x+C. Bas parity dekho, factor peel karo, identity lagao — har sawaal mechanical ho jaayega.